What oscilloscope specs do I need for Arduino projects?

[DanielC]

Hello,

My partner and I have recently started learning how to make simple projects with the Arduino. We have a nice little workbench with most of the items recommended in , but we have been reluctant to buy an oscilloscope. Even a budget oscilloscope is very costly. Over the past two weeks we made an effort to get informed about oscilloscopes, and I feel that we understand the key concepts. However, I still need help:


1) What problems do you solve with an oscilloscope?

This is not about specs, but I thought I'd ask. It is very hard to get a concrete answer. An answer like "it shows you what's going on in your circuit" may be accurate, but is not helpful. Can you share an example of a problem I might have that I could solve more easily if I have an oscilloscope on my bench?

2) What bandwidth do I need?

I know what the bandwidth is, but I don't know how much I need. Let's see... the Arduino uses an Atmel AVR microcontroller with a 16 MHZ crystal oscillator. I imagine that this means that the fastest signal it can produce must be 8 MHZ (one operation to turn the current one, and one to turn it off). To measure an 8 MHZ square wave I need 5 x 8 = 40 MHZ. If my reasoning is correct, any oscilloscope with at least 40 MHZ bandwidth will correctly capture any signal that an Ardunio can generate...

Can someone confirm that my reasoning is correct? Or have I missed the point?

3) Sample rate?

Following the same reasoning as earlier, I think I need 5 x 8 = 40 MSa/s. But this value actually seems really low compared to what a digital scope does. A budget digital scope is in the range of 500 MSa/s to 1 GSa/s. Have I missed something important?

4) Memory depth?

I understand the relation:

(memory depth) = (sample rate) x (time base)

I really have no idea what is a reasonable time base for a typical Arduino project. That makes it harder to pick a memory depth. I do understand that a longer memory also means that I can pick out more frequencies at the same time and that in general I can get nicer graphs with a higher resolution. But I still wish I had a sense of what time base I am likely to care about (remember that I don't really know what problems I am going to be solving with this thing).

As of this moment, our baseline model is the Siglent SDS1102CML which has a 100 MHZ bandwidth, 1 GSa/s sampling rate, and 2 Mpts memory depth:

http://www.amazon.co.uk/Siglent-SDS1102CML-Digital-Oscilloscope-Channel/dp/B00DR59DQC

But there are cheaper models with lower specs. So I'd like to get a sense of where I can compromise. For example, is it better to go for a cheaper model that has only 70 MHZ bandwidth, or a scope with only 40 Kpts memory?

I would really appreciate some guidance.

[suicidaleggroll]

Honestly, for simple uC projects, you would be MUCH better off getting a decent logic analyzer.  If you already have one, then a scope can help you debug physical problems with your circuit.  For example, a while back I was debugging an I2C comms problem I was having on a project.  For the life of me I could not get the device to ack the address.  I finally stuck a scope on the bus and instantly saw the problem, it was trying to ack, but it was only pulling the bus to half of Vcc, which wasn't enough for the host to register a logic 0.  Turns out another device on the I2C bus was misbehaving and actively driving the SDA line high when it wasn't using it, instead of switching to high Z.  This made the effective pull-up resistance on the line ~200 ohms, which the host had no trouble with but my device couldn't handle it.

[ez24]

You did not indicate what country you are in via your profile other than a UK link.

I suggest you look at the new Rigol DS1054Z  (50mhz) 4 channel scope with optional (maybe) serial protocols.  It seems to me a major thing with the Arduino is serial communications with sensors.

 ()   It might be a good compromise.  Pay attention to the protocol parts (I like the I2C part).  There are 1 million posts on it on this forum  (it is like a religion ha ha).  Additional info (but out of our budget) -->

I was about to spring for a Siglent SDS1102CML until I learned about the serial stuff of the Rigol.  What I did was to download both manuals and browse through them.  There might be something in one of their manuals that catches your eye.  (it was the I2C in the Rigol manual that got my attention)

This is well respected (and I wish I had one):

[url=https://www.youtube.com/watch?v=Aymumu3mYl8]https://www.youtube.com/watch?v=Aymumu3mYl8[/url]

A good deal if you can get the student discount otherwise maybe a DS1054Z.

I look forward to what the pros say.

[SL4P]

Depending on what you're developing & programming - no oscilloscope or analyser would be a better start!
Learn about the target platform first, then depending on what your application is - save $300 until you actually need it.

I worked on minis and micros since the PDP-8 / 11 / VAX and 2650 / Z80 etc in the late 70s, and pretty much only 'needed' a personal scope of my own for a project that involved analyzing fast pulses about 4-5 years ago.  I've been developing PIC and Atmel projects for almost ten years...  for web, telnet, serial, I2C, SPI and others...

Yes - I could borrow one occasionally back in the 80s & 90s, but a simple logic probe ($20) was all I needed for a long time - to check levels and tri-state pins.

If you think about it, a blinking LED (status) and a serial port with some debugging messages on a terminal will get you through probably 80% of development problems - without the headache of carting out the probes, etc.

If a source signal is unknown - then the best solution is finding the spec for the signal. and only then pulling it apart - otherwise it's wasted effort that someone else has probably already done.

[Psi]

For simple arduino projects pretty much any scope will be useful. 

I would stay above 20Mhz bandwidth but apart from that get anything you can find that is cheap and works.

[rob77]

a logic probe with integrated pulse counter is way more usefull for small uC projects.

[DanielC]

Thanks for the help. I think you guys just saved me $300.

Reply to ez24: I live in Sweden. I generally buy equipment either locally or have it shipped from England or Germany.

The consensus opinion seems to be that I should not be buying an oscilloscope right now (and maybe not ever!). So I'm glad I asked before I bought anything. As I said earlier, I don't really know what I would do with one if I had it. This thread has also made me realize that I need to learn a lot about serial buses. I only have the faintest notion of what I2C and SPI mean.

So for now I'll just focus on learning about my platform, but I will get myself a logic probe to complement my digital multimeters.

[SL4P]

You will go far my son

[agehall]

Keep jour eyes peeled at Tradera and Blocket. I picked up a really old (but pretty nice) scope for 400 sek not too long ago. Felt that was a too good of a deal to not jump on...

[rob77]

So for now I'll just focus on learning about my platform, but I will get myself a logic probe to complement my digital multimeters.

i would suggest to build a logic probe yourself. don't buy an off-the-shelf one - those are overpriced. build one with a relatively fast input stage , adjustable high & low thresholds, indication of high-zi or undefined state  and definitely a pulse counter (counting low-to-high transitions). i built such a probe - costs were well under 10eur + my own time to design & build it. it was lm393 input stage + 4 stages of 4bit binary counters (74hc193) with LEDs - you have to "read binary" to read the numbers

[picandmix]

Many suggestions above, but for the price something like one of these lo cost  Saleae logic analyzers Clones are probably all you need for Arduino work for now.
http://www.ebay.co.uk/itm/USB-saleae-Logic-Analyzer-Device-Set-USB-Cable-24MHz-8CH-24MHz-for-MCU-ARM-FPGA-/251939297329?pt=LH_DefaultDomain_3&hash=item3aa8c09831

Got one the other day and it works fine as you can see from the pics, it may look complicated but plenty of info on how to use them ot there.


You can even try before you buy with Saleaes own software and built in simulator.
http://downloads.saleae.com/betas/1.1.34/Logic Setup 1.1.34.exe

[tautech]

As of this moment, our baseline model is the Siglent SDS1102CML which has a 100 MHZ bandwidth, 1 GSa/s sampling rate, and 2 Mpts memory depth:

http://www.amazon.co.uk/Siglent-SDS1102CML-Digital-Oscilloscope-Channel/dp/B00DR59DQC

But there are cheaper models with lower specs. So I'd like to get a sense of where I can compromise. For example, is it better to go for a cheaper model that has only 70 MHZ bandwidth, or a scope with only 40 Kpts memory?

I would really appreciate some guidance.

Others have suggested what might be better tools for what you are doing ATM.
But there is nothing that can compete with "seeing" waveforms, especially as you gain experience and knowledge.
Rest assured, the SDS1102CML is a good choice, better than entry level, proven and bug free.
If and when you get a scope, the likes of this model could be all you ever need.

[JackP]

When a scope comes in Useful is during repairs. Easily check the supply lines for ripple, verify control signals,the times where you were not involved in the design, so you don't know where the designer cocked up, or a component malfunctioned, or damage was caused to the instrument. I know you don't really do repairs, just giving you an answer to point No. 1

Edit: to emphasise suicidaleggroll's point, a scope makes diagnostics much easier, and his problem would have been very difficult to debug without it. However, you have to be experienced to do this (obviously he was knowledgeable about his circuit, and knew what should have been happening).

[daybyter]

The hantek 6022be is a bargain in my eyes. Take a look at the thread in this forum.

[DanielC]

The hantek 6022be is a bargain in my eyes. Take a look at the thread in this forum.

An issue with any PC-based scope is that I am running Linux. This might not be a complete show stopper - I could try to run their software under emulation, or maybe try the "OpenHantek" project. But the lack of official Linux support makes me hesitate. There is a PC based scope called BitScope that officially supports Linux. Someone posted about it in this forum four years ago, and Dave was concerned about the 40 MS/s sampling rate. But for an Arduino project maybe that's enough. Comparing these two, the Hantek is cheaper than the BitScope, but the BitScope has Linux support and added features like wave generation and a logic analizer.

[DanielC]

When a scope comes in Useful is during repairs. Easily check the supply lines for ripple, verify control signals,the times where you were not involved in the design, so you don't know where the designer cocked up, or a component malfunctioned, or damage was caused to the instrument. I know you don't really do repairs, just giving you an answer to point No. 1

Edit: to emphasise suicidaleggroll's point, a scope makes diagnostics much easier, and his problem would have been very difficult to debug without it. However, you have to be experienced to do this (obviously he was knowledgeable about his circuit, and knew what should have been happening).

Thanks. That does make things clearer. At this point my circuits are extremely simple, so maybe the value of the oscilloscope will become clearer when my circuits start to become complex.

[tggzzz]

Those are thoughtful questions, so it is worth replying...

... Arduino
...Over the past two weeks we made an effort to get informed about oscilloscopes, and I feel that we understand the key concepts. However, I still need help:
1) What problems do you solve with an oscilloscope?
Can you share an example of a problem I might have that I could solve more easily if I have an oscilloscope on my bench?

Potential tools are scope, logic analyser, LED+switch, printf/single-stepping.

Use a scope to check analogue signals (obviously!) and the analogue behaviour of digital signals. You should realise that all circuits are analogue (except photon counting and similar), but that the circuits interpret the signal as being one of two levels. A principal use of a scope is to verify "signal integrity" in terms of level and transition time.

Once you have ensured signal integrity, move into the digital domain as fast as possible.

If the signal is generated or received under software control, first use a LED+switch+test program to ensure your program is "connected" to the correct i/o line, then drop back into single-stepping and/or printf statements.

If the signal's timing is too fast to be validated in software, then use a logic analyser.

2) What bandwidth do I need?
Can someone confirm that my reasoning is correct? Or have I missed the point?

The maximum frequency contained in a digital signal is completely unrelated to its period. All that matters is the transition time. The standard rule of thumb is that to see a rise/falltime of t_r the oscilloscope's analogue bandwidth, BW, should be >0.35/t_r. That's the important value for signal integrity.

3) Sample rate?
Following the same reasoning as earlier, I think I need 5 x 8 = 40 MSa/s. But this value actually seems really low compared to what a digital scope does. A budget digital scope is in the range of 500 MSa/s to 1 GSa/s. Have I missed something important?

There is absolutely zero correlation between the sampling rate and the bandwidth, despite what manufacturers would have you believe. All that matters is the scope's analogue bandwidth. To make that point clear, years ago I used very expensive 1GHz scope to check sub-nanosecond risetimes; the scope sampled at 25MS/s - but that was fine for looking at a repetitive signal.

However, for non-repetitive signals and modern scopes, the analogue front-end bandwidth is what matters with non-repetitive signals.

4) Memory depth?
I really have no idea what is a reasonable time base for a typical Arduino project.

You understand the calculations, but I suggest you think about what you need to observe and then you will be able to choose the best tool.

[picandmix]

Even though the Arduino IDE is available for Linux, most genre of micros and test equipment are primarliy Windows based.

Any reason you could not dual boot with a Windows partition or similar,    soley for micro development work ?

For any serious work with Arduino then you will really appreciate Atmels Studio 6 with the Visual Micro plug in for Arduino, which also allows for debugging that would solve much of your needs for hardware test equipment.


Even with the basic Arduino ide you can do lots of basic debugging and testing  by using your code via the serial monitor, test leds and your multimeter, the need for a scope during normal projects is not essential, its for developing and fault finding more unsual projects. (many such faults actually being software caused anyway)

I would suggest you save your money for now and enjoy developing your knowledge of Arduino and C++
As you skills increase so will your awareness of what test equipement will best serve you in the future.

 

[LabSpokane]

For simple Arduino work, you really don't need much more than an LED and a resistor.

Unless you start designing your own boards and need to validate power or signal integrity, you don't *need* a scope.  As time goes on, you may run into an issue that only a scope can solve, but there's a couple dozen things that I would get before the scope.

[daybyter]

An issue with any PC-based scope is that I am running Linux. This might not be a complete show stopper - I could try to run their software under emulation, or maybe try the "OpenHantek" project. But the lack of official Linux support makes me hesitate. There is a PC based scope called BitScope that officially supports Linux. Someone posted about it in this forum four years ago, and Dave was concerned about the 40 MS/s sampling rate. But for an Arduino project maybe that's enough. Comparing these two, the Hantek is cheaper than the BitScope, but the BitScope has Linux support and added features like wave generation and a logic analizer.

Same here. I also run Linux and prefer VMs for most of my work. But so far no luck to run the 6022 software in a VM. That's why I still have an XP partition on my old laptop. But this new python project looks promising. It still doesn't work in a VM here, but it should work on plain Linux.
For logic analyzing, I got me one of those cheap 10$ units from dealextreme. But after I used it once for an arduino project, I realized, that 8 inputs are not really enough. Ok, you can workaround this limitation somehow, but changing the inputs all the time is annoying. So I'm looking for an alternative in the long run.

[Howardlong]

You will go far my son

Looking back on it, I developed sh!t loads of stuff without any scope at all. My tools for many years from the mid 70 for my mostly digital world were a crappy 1kOhm/v moving coil multimeter, and an LED and resistor. You can do an awful lot with those.

The first scope I owned was about 20 years or so ago, a 20MHz dual channel CRO. While I've moved up over the years, it's the law of diminishing returns. By using simple test equipment, even as simple as an analogue multimeter and an LED, you learn an awful lot and become more analytical in your approaches.

However there are times though when a scope will solve things that a meter and LED won't. It will also usually give you a much faster diagnostic and resolution, once you've learned how to drive it.

[ez24]

How about this for $10?

[url=http://www.aliexpress.com/item/EISTAR-Digital-Logic-Probe-Pen-DTL-TTL-Digital-Logic-Probe-DC-18V-Max-DTL-TTL-Cmos/2014261457.html]http://www.aliexpress.com/item/EISTAR-Digital-Logic-Probe-Pen-DTL-TTL-Digital-Logic-Probe-DC-18V-Max-DTL-TTL-Cmos/2014261457.html[/url]

seems to get good reviews

[tggzzz]

The hantek 6022be is a bargain in my eyes. Take a look at the thread in this forum.

An issue with any PC-based scope is that I am running Linux. This might not be a complete show stopper - I could try to run their software under emulation, or maybe try the "OpenHantek" project. But the lack of official Linux support makes me hesitate. There is a PC based scope called BitScope that officially supports Linux. Someone posted about it in this forum four years ago, and Dave was concerned about the 40 MS/s sampling rate. But for an Arduino project maybe that's enough. Comparing these two, the Hantek is cheaper than the BitScope, but the BitScope has Linux support and added features like wave generation and a logic analizer.

Well, not quite anything. The Digilent Analog Discovery has a scriptable Linux beta variant. It includes two AWGs and pattern generator/logic analyser to 100Mb/s/channel.
The scope's bandwidth is only 10MHz or so, though, i.e. sufficient for basic debugging, but not for digital signal integrity.

[eas]

Honestly if you develop a good rigorous practice of design and analysis you're not likely to NEED a scope for years.  The scope could quickly help you "quickly" figure out lots of avoidable problems that you really SHOULDN'T have gotten yourself into to begin with, but that's not so much of a justification for buying an expensive scope as more of a justification for "know what you're doing before you do it" methodology of design.

This sounds like wise advice. I'd argue that it isn't. In order to know what you are doing before you do it, you have to know what you are doing. In order to know what you are doing, you have to learn it. How does one learn? Often it involves making mistakes. Sometimes the fact that you've made a mistake is obvious, because something doesn't produce the expected result. However, the nature of the mistake isn't always obvious. Reasoning out the mistake is an important way to learn, but often it is easier to reason out a mistake if you can make observations of the non-working system. Tools that help with those observations can be quite valuable for learning. I think this is particularly true in the early stages of learning, when you can't really be confident in your understanding. An oscilloscope can be quite valuable in shortening the time between trying something, and understanding the result.

I found having a scope quite helpful when I was making a basic arduino circuit and code to read data out of laptop batteries over SMBus. I was pretty confident in both my circuit, and my code, but nothing was working. I tried various experiments, but they didn't improve my understanding. Then I decided to hook up my then-new scope to the clock and data lines. I could immediately tell that something was wrong (there was a changing signal on the "data line" and a flat signal on the clock line), and guess at the solution. I thought I had the correct pin naming for my Arduino variant. I didn't. The documentation was poorly organized and confusing. Knowing that I had been lead astray made it clear to me that there was other documentation, somewhere, and that finding it was worthwhile. I'm sure I would have figured it all out eventually without the scope, but with the scope, I figured it out and fixed it in ~1 hour. I'm sure it would have taken much longer otherwise.

That doesn't mean that a scope is a smart buy for you at this point, but its not a bad one. In my case, an inexpensive logic analyzer would have helped me track down my Arduino problem just as well as the scope. It wouldn't have been very helpful for the RS-232 problem, both because the voltages in play would have been out of spec, and because seeing the analog waveform was much more informative than a trace of on/off states

On a related note, my new-to-me 20 year old bench DMM was having trouble with RS-232 communications. I checked everything with a multimeter, which was helpful, but being able to get a trace made it much clearer what was going wrong, and what a work around solution was. I was able to avoid the problem by modifying a cable, where as other people were just suffering through unreliable communications, replacing a surface-mount level shifters of sending in for service. Even more interesting, to me, I couldn't find anyone else online who managed to find the root cause (a design issue), and also evidence that the "bug" remained in the design, and without being addressed in firmware for a decade or more.

[DanielC]

Any reason you could not dual boot with a Windows partition or similar,    soley for micro development work ?

For any serious work with Arduino then you will really appreciate Atmels Studio 6 with the Visual Micro plug in for Arduino, which also allows for debugging that would solve much of your needs for hardware test equipment.

The only reason is that I don't like Windows and I hate rebooting. I currently reboot roughly once or twice a year. Using a Windows partition for micro work would mean rebooting daily.But I would consider running Windows on a virtual machine if needed, and I could run Atmel Stuido inside that. I do agree that the baseline Arudiono "IDE" is incredibly minimalist and must be replaced. There are some alternatives that work on Linux. There are plugins for Eclipse, Code::Blocks and Sublime Text 2. Right now I'm using a program called UECIDE and I'm quite happy with it.

I would suggest you save your money for now and enjoy developing your knowledge of Arduino and C++
As you skills increase so will your awareness of what test equipement will best serve you in the future.

Thanks.

[tggzzz]

Honestly if you develop a good rigorous practice of design and analysis you're not likely to NEED a scope for years.  The scope could quickly help you "quickly" figure out lots of avoidable problems that you really SHOULDN'T have gotten yourself into to begin with, but that's not so much of a justification for buying an expensive scope as more of a justification for "know what you're doing before you do it" methodology of design.

Has it occurred to you that, for a beginner, that is a chicken and egg situation?

A beginner is unlikely to be able understand all of the hardware/firmware/software documentation at once, particularly when the data sheets are wrong or internally inconsistent.

e.g. Learn NOT to hook the polarity of the power supply backwards.  Learn NOT to have your power supply at the wrong voltage (a half decent DVM/DMM could tell you these things).   

Even with half a century of experience, I've just done that again. Doh 

Learn NOT to design circuits without well placed decoupling capacitors. 

That is more subtle that you might believe - you can get resonances in the ground plane if you add too many capacitors of the inappropriate combinations of values.

Learn NOT to use long wires and solderless breadboards for precision or high speed of high sensitivity or high impedance circuits. 

Sound advice. I hate those damn things and will always use dead-bug techniques in preference.

Simplify the problem and prove the SW and HW piece by piece, step by step.

Sound advice.

Do follow good ESD control practices, wrist straps, grounded mat, etc.  Debugging a semi-fried PCB is not useful. Do learn things like Ohm's law, what capacitors, inductors, transistors, logic gates do.  Do experiment with counters, timers, PWM, LEDs, resistors, voltage regulators, speakers, buzzers, etc. 

If you don't do anything until after you fully understand those, a beginner will never get anywhere.

But keep it simple, one or two things at a time, there will be less variables to debug in SW / HW at once.

Sound advice.

Beyond that, really, mostly you just won't NEED a scope.  Things will either work or they won't, and if they don't it is probably a stupid mistake, typo, something you didn't read / connect right, a SW bug, etc.  Not too hard to track down piece by piece.

...providing you are omniscient. If you are using C (or worse, C++), then even people that have been programming since the 60s and using C since the 70s cannot predict what will happen with different compilers (or even vesions ) and architectures. If you don't believe that, read comp.arch or see http://yosefk.com/c++fqa/ (the section on const is one of my favourites), or where p1 and p2 are pointers, define what will happen with

  if (p1 < p2)
     return 1;
  else
     return 1;

Hint: it is nasal demon territory, depending on whether or not the compiler can prove that the pointers are to the same object. (And defining "object" is more subtle than you believe.)

As a simple example, try writing memove() portably, i.e. so that it will work on all architectures with all compilers.

Get a DMM you can trust, get two even so you can double check if one doesn't make sense.  Low batteries or such can get bad results.  Get a decent breadboard, jumpers, sockets, clip leads, cables.  Poor ones can be worse than useless if they make the whole thing flaky or not work and you spend hours debugging a flaky connection.  Look at inexpensive tools like the Saleae devices, the Bus Pirate, things like that.

As for scopes, look for old "boat anchor" transistor or vacuum tube units or low end IC based units being given away free or sold inexpensively locally in classified ads for your city.  I see free or $50-$100 US scopes all the time various places.  Nothing fancy, but maybe 20MHz, 50 MHz, some very old 100 MHz ones, etc.  More than enough to get a sense of the waveforms and what the essential state of an Arduino circuit is.  Mostly not that useful for debugging SPI / I2C / JTAG / UART related digital bus stuff, but that's what debugging SW is for, blinking LEDs, writing to a little LCD screen, writing to your debug UART, etc. is for, as well as Bus Pirate / Saleae, etc. Scope will be more useful for looking at analog stuff that is not DC (your DMM will do for much DC work), and digital pulses if you're not even sure if they're even on or running at the right frequency or not, or if the waveform is glitching/distorted, but mostly you should be able to figure that stuff out other ways.

I wouldn't argue with that, but affording, understanding and buying all those tools is non-trivial.

[picandmix]

Any reason you could not dual boot with a Windows partition or similar,    soley for micro development work ?

For any serious work with Arduino then you will really appreciate Atmels Studio 6 with the Visual Micro plug in for Arduino, which also allows for debugging that would solve much of your needs for hardware test equipment.

The only reason is that I don't like Windows and I hate rebooting. I currently reboot roughly once or twice a year. Using a Windows partition for micro work would mean rebooting daily.But I would consider running Windows on a virtual machine if needed, and I could run Atmel Stuido inside that. I do agree that the baseline Arudiono "IDE" is incredibly minimalist and must be replaced. There are some alternatives that work on Linux. There are plugins for Eclipse, Code::Blocks and Sublime Text 2. Right now I'm using a program called UECIDE and I'm quite happy with it.

I would suggest you save your money for now and enjoy developing your knowledge of Arduino and C++
As you skills increase so will your awareness of what test equipement will best serve you in the future.

Thanks.

On principle I would also use Linux, but have found on a practical basis when working on micros etc is so difficult not to use Windows.

Did try some of those other Arduino IDEs but they just are no comparison to Studio 6 , plenty of Ytubes on Studio 6  / Visual Micro if you want to view it.

If you hate rebooting , assume its the slowness of it ?   then you will find Windows 8 really does boot up very quickly.
If you add a SSD  then you will be amazed how fast booting and everything else is.

[tggzzz]

The only reason is that I don't like Windows and I hate rebooting. I currently reboot roughly once or twice a year. Using a Windows partition for micro work would mean rebooting daily.But I would consider running Windows on a virtual machine if needed, and I could run Atmel Stuido inside that. I do agree that the baseline Arudiono "IDE" is incredibly minimalist and must be replaced. There are some alternatives that work on Linux. There are plugins for Eclipse, Code::Blocks and Sublime Text 2. Right now I'm using a program called UECIDE and I'm quite happy with it.

Thanks; I'll have a look at that next time I need to do something non-trivial on an arduino.

[EEVblog]

As of this moment, our baseline model is the Siglent SDS1102CML which has a 100 MHZ bandwidth, 1 GSa/s sampling rate, and 2 Mpts memory depth:
But there are cheaper models with lower specs. So I'd like to get a sense of where I can compromise. For example, is it better to go for a cheaper model that has only 70 MHZ bandwidth, or a scope with only 40 Kpts memory?

50MHz is plenty.
More memory is better, but any scope is infinitely better than none.
What is your budget? i.e. how much is too much?

[EEVblog]

Honestly if you develop a good rigorous practice of design and analysis you're not likely to NEED a scope for years.  The scope could quickly help you "quickly" figure out lots of avoidable problems that you really SHOULDN'T have gotten yourself into to begin with, but that's not so much of a justification for buying an expensive scope as more of a justification for "know what you're doing before you do it" methodology of design.

This sounds like wise advice. I'd argue that it isn't.

So would I.
If you are into electronics a scope is the one tool you must have. It lets you see what a circuit is doing. Arguing to do without is just silly.
Even a $50 20MHz dual channel analog scope will open your eyes into another world.

[AG6QR]

Honestly if you develop a good rigorous practice of design and analysis you're not likely to NEED a scope for years.  The scope could quickly help you "quickly" figure out lots of avoidable problems that you really SHOULDN'T have gotten yourself into to begin with, but that's not so much of a justification for buying an expensive scope as more of a justification for "know what you're doing before you do it" methodology of design.

This sounds like wise advice. I'd argue that it isn't.

I'd agree that it's not.

You could make the same argument about a DMM.  If you know what you're doing, you should be able to figure out what the voltages should be without using a DMM.  If you're careful in your implementation, the voltages will be what you predicted them to be, and your circuits will all just work.  There have been times when I've soldered together circuits, applied power, and had a working device on the first try without using a meter.

No reasonable person would suggest you do electronics without a DMM, though.  No matter how good you are, there will be times when things don't match your predictions and don't work.  Or there will be times when you want some insight into what's going on with a device you didn't design or build, or maybe you just plain know that you don't understand the concepts well enough to predict what the voltages are going to be.  Then you've got to start measuring things to see what's going on.  A circuit may have failed due to faulty design, faulty parts, faulty soldering, or a complete misunderstanding of the fundamental concepts, but regardless, a meter will give you "magic eyes" that let you see the invisible electron flow and figure out what's going on, at least in DC.

An oscilloscope is basically a very fast voltmeter that lets you see how things are changing over time, even over very short timescales.  If you're dealing with voltages that change faster than your DMM can respond, a scope will be as useful a tool as a DMM is to the person who is dealing with DC circuits.  And sooner or later, anybody who claims to be working with electronics is likely to deal with voltages that changes faster than the DMM can measure.

When data doesn't seem to be flowing as you expect, you can put a scope on it and figure out why.  True, as some have pointed out, a logic analyzer can sometimes be a more appropriate tool for debugging some issues.  But a logic analyzer doesn't let you see signal integrity issues that may be caused by analog problems in the signal path.

[idpromnut]

..snip..
If you are into electronics a scope is the one tool you must have. It lets you see what a circuit is doing. Arguing to do without is just silly.
Even a $50 20MHz dual channel analog scope will open your eyes into another world.
..snip..

And a clarification: when we're saying a 'scope lets you "see what a circuit is doing", we mean voltage in the time domain. A DMM will let you see the instantaneous voltage, but you have no idea what the shape or values of that voltage over time is. A DMM can sure you the value of AC signals, but only up to a certain frequency, and even then don't tell you anything about the signal in terms of it's minimum/maximum peak value, shape (square versus sine wave) and rise time (which may not be useful).

I picked up a Tektronix 2225 as my first scope (based on the recommendation of Dave no less) and there was one situation that would have stopped me had I not been able to see what the voltage of my ground in my circuit looked like referenced to my power supply (it was on a bread board); there was a ton of noise that was false-triggering a 555 timer and that was caused by a motor. But I wouldn't have thought to start looking down that path if I couldn't observe more closely what my circuit was doing.

[tggzzz]

And a clarification: when we're saying a 'scope lets you "see what a circuit is doing", we mean voltage in the time domain. A DMM will let you see the instantaneous voltage,

Um, some variant of the average voltage over an unspecified period - typically <1s.

A scope is much better at showing the instantaneous voltage.

[tautech]

No reasonable person would suggest you do electronics without a DMM, though. 

Correct, although it is entirely possible with a modern DSO.

But you must have a range of probes, etc to increase functionality.

[DanielC]

As of this moment, our baseline model is the Siglent SDS1102CML which has a 100 MHZ bandwidth, 1 GSa/s sampling rate, and 2 Mpts memory depth:
But there are cheaper models with lower specs. So I'd like to get a sense of where I can compromise. For example, is it better to go for a cheaper model that has only 70 MHZ bandwidth, or a scope with only 40 Kpts memory?

50MHz is plenty.
More memory is better, but any scope is infinitely better than none.
What is your budget? i.e. how much is too much?

Budget: Let's see... I'd say that my hard limit is 500 EUR or 560 USD. My soft limit is 350 EUR or 390 USD because I can get the Siglent SDS1102CML for less than that, so I would need a good reason to go higher.

[LabSpokane]

Daniel

What tools do you have on hand now?  Is your budget for the scope, or for everything?

[DanielC]

Daniel

What tools do you have on hand now?  Is your budget for the scope, or for everything?

I currently have (1) a soldering iron and solder material, (2) two nice digital multimeters, (3) an Arduino starter kit, (4) books, and (5) miscellaneous items like cables, resistors, motors, sensors, capasitors, LEDs, LCDs, breadboard, and basically everything I need to do beginner Arduino projects.

When I began this thread, I was thinking that the oscilloscope was the only item missing. So in that sense, the budget is for "everything else", which I guess probably means an oscilloscope and/or a logic analyzer.

[LabSpokane]

Daniel

What tools do you have on hand now?  Is your budget for the scope, or for everything?

I currently have (1) a soldering iron and solder material, (2) two nice digital multimeters, (3) an Arduino starter kit, (4) books, and (5) miscellaneous items like cables, resistors, motors, sensors, capasitors, LEDs, LCDs, breadboard, and basically everything I need to do beginner Arduino projects.

When I began this thread, I was thinking that the oscilloscope was the only item missing. So in that sense, the budget is for "everything else", which I guess probably means an oscilloscope and/or a logic analyzer.

Daniel,

You can do a lot with what you have now. If you really enjoy doing projects, *then* consider buying a scope or LA. I know that Dave says its a must and will open a whole new world, but make sure its a fun world for you first. Once you know that, then get a scope. And it will open your eyes to new dimensions of electronics. Don't feel pressured to buy immediately because all the other kids are. There's all sorts to do with the Arduino with little more than a DMM, a few basic parts and the IDE's serial monitor.

In the meantime, you can investigate scope brands and features. Then you will be an educated buyer and will understand your own needs. 

[tggzzz]

Daniel

What tools do you have on hand now?  Is your budget for the scope, or for everything?

I currently have (1) a soldering iron and solder material, (2) two nice digital multimeters, (3) an Arduino starter kit, (4) books, and (5) miscellaneous items like cables, resistors, motors, sensors, capasitors, LEDs, LCDs, breadboard, and basically everything I need to do beginner Arduino projects.

When I began this thread, I was thinking that the oscilloscope was the only item missing. So in that sense, the budget is for "everything else", which I guess probably means an oscilloscope and/or a logic analyzer.

Daniel,

You can do a lot with what you have now. If you really enjoy doing projects, *then* consider buying a scope or LA. I know that Dave says its a must and will open a whole new world, but make sure its a fun world for you first. Once you know that, then get a scope. And it will open your eyes to new dimensions of electronics. Don't feel pressured to buy immediately because all the other kids are. There's all sorts to do with the Arduino with little more than a DMM, a few basic parts and the IDE's serial monitor.

In the meantime, you can investigate scope brands and features. Then you will be an educated buyer and will understand your own needs.

That's pretty sound advice.

An engineer will think:

• what do I need to achieve, and when?
• how can I achieve it with whatever is already available?
• if not, what do I need to achieve it in terms of delay, cost, learning curve?
• if I've misjudged what is necessary, what would the delay be?

and in many cases 1 can be achieved with 2. Only if there is a demonstrable 3 should that option be taken, particularly if 4 is small.

[picandmix]

Daniel,

You can do a lot with what you have now. If you really enjoy doing projects, *then* consider buying a scope or LA. I know that Dave says its a must and will open a whole new world, but make sure its a fun world for you first. Once you know that, then get a scope. And it will open your eyes to new dimensions of electronics. Don't feel pressured to buy immediately because all the other kids are. There's all sorts to do with the Arduino with little more than a DMM, a few basic parts and the IDE's serial monitor.

In the meantime, you can investigate scope brands and features. Then you will be an educated buyer and will understand your own needs.

That's pretty sound advice.

+2,   thats the best advice you can get !

There are millions of Arduinos out there, probably only a very  small percentage of owners have  a scope, many not even a dvm,  yet they can get great enjoyment and  produce some excellent work.

Would also suggest you spend more time in Arduinos own forum and concentrate on learning more about how to program and use the basic Arduino hardware, there are loads of  free  Arduino tutorials out there, sure you will find one that suits you ( no scope needed  )

Examples


http://www.ladyada.net/learn/arduino/
http://tronixstuff.com/tutorials/

[DanielC]

Thanks for the advice (to all of you). I will do as you suggest. I will continue exploring the Arduino world with what I have right now, and only buy a scope or LA when there is a demonstrable need for one. The books I have contain dozens of projects that I can do with the tools I have already, and so far the serial monitor has been more than enough to debug my programs. So I'll try to not feel rushed to buy an expensive scope that I don't even need yet.

Cheers.

[ez24]

Maybe this will make you want to get a scope:

http://www.tek.com/sites/tek.com/files/courseware/ST_Arduino_Labs_Combined.pdf

I think you could learn a lot with this and an arduino

Edit: corrected link

[JacquesBBB]

Even if you can delay for the purchase of an oscilloscope, I would recommend to buy a cheap <10$
logic analyzer  compatible with the Saleae software as this one.

http://www.ebay.com/itm/USB-Logic-Analyzer-Device-Set-USB-Cable-24MHz-8CH-24MHz-for-ARM-FPGA-/351339823002?pt=LH_DefaultDomain_0&hash=item51cd7c3f9a

This will allow you to lear a lot more from your arduino projects.

[ldhingra]

Hi Daniel,

It's the month of May and year is 2020.
I am exactly in the same spot and fix today what you were in 5 years ago as to which scope should I buy and moreover do I actually need it for Arduino projects.

How has been your journey since then?

Would you have pointers for a newbie like me?

Regards
Lalit

[tggzzz]

You should read the answers that have been given to this question many times before.

You should understand that often a scope is not the only tool that can be useful to you. Then choose the best combination of tools available to you.

[stafil]

5 years ago

...and 1054Z is still a fine choice

Rigol hit the Jackpot with that scope

[Dubbie]

The thing with a scope is that I think it will help you learn more, faster than not having one.
There are many projects I have made where I would have absolutely gotten stuck and never been able to complete them without a scope.

The other thing is, if you don't have a scope, you don't really know what you would need one for. After you get one, you'll find yourself using it for all sorts of things you never dreamed of before. Thats why there are very few people out there without scopes who are desperate to get one. They don't know what they are missing.

If you are already an electronics expert, then sure, maybe you don't need one, but if you are not, then it will open your eyes.

[wizard69]

Hi Daniel,

It's the month of May and year is 2020.

Yeah reviving this old thread is a bit confusing.

I am exactly in the same spot and fix today what you were in 5 years ago as to which scope should I buy and moreover do I actually need it for Arduino projects.

Well like was expressed in the first few posts I read (before realizing how old the thread was) there are various points of view on need and when to buy.   There is  no right answer for everybody, but for Arduino's I would tend to say get the rest of the bench in order first.

How has been your journey since then?

Would you have pointers for a newbie like me?

Regards
Lalit

Pace yourself!    A scope is a very basic piece of test equipment so they do provide a lot of value to a technician.   However new they are expensive.   I wold tend to suggest following Dave's advice in the first linked video and look for an old analog scope that you can get for free or at a significant discount.   The reasons are 1) Cheap.   2) The potential to learn a lot.    3) Analog scopes fill some niches nicely.     Purchase a decent digital scope once you have some experience under your belt.   The reason is you want to self inform yourself to make a better buying decision.   In the end you should know after a few months to a few years where your interests are going and thus make a good decision on scope features when you buy.

In the end the feature set you need will not be the same as somebody else do audio or RF work.   You may very well find yourself needing a multi channel (more than two) scope capable of doing serial channel decode.   Maybe you will get into motor control other other industrial solutions and find that a scope with isolated channels makes sense.   What I'm getting at is that you can easily buy the wrong scope until you understand what your personal interests are.

If you do go out an buy a new scope a few suggestions:

• At the very minimal the scope should have 100MHz bandwidth.   With modern technology it doesn't make sense to spend good money on anything less.   More is better but is not needed if you stay in the Mega world.
• Ideally the scope should have more than two input channels.   Again it depends upon ones interest but I've run into many occasions where two channels was less than optimal.
• Try to find a scope that isn't completely locked down with excessive fees on enhancement software.    This is perhaps the most frustrating thing when working with the mainline suppliers.   After a bit of time advanced features should be considered base level features.   I don't even have an up to date suggestion on supplier here, just don't pay a lot of money for software upgrades that are pretty mainstream needs these days.
• Make sure the scope offer plenty of connectivity.    One day you might want to graph some data off a scope or hook it to a computer for another reason and having options is golden.   At the very minimal you want a USB port that you can dump data too.    However the more the ports can support the better off you will be
• Dave in is video was a bit down on USB based instruments.   5 years on I'm not sure I agree with him.    yes there are pretty bad solutions out there but there are also good (expensive ones).   More importantly there are modern moderately priced stuff that is good for a beginner.   Some of the low end things like USB logic analyzers are likely as good or better than a scope for someone starting out.    That isn't to say that they will replace a $5000 scope, which is silly to think, just that they can be worth it for a hobbyist.

So in a nut shell, yeah you will need a scope but don't jump into an expensive purchase blindly.   Develop the knowledge to make an informed buy and get by until, with second hand or very low end solutions.

[tggzzz]

The thing with a scope is that I think it will help you learn more, faster than not having one.
There are many projects I have made where I would have absolutely gotten stuck and never been able to complete them without a scope.

The other thing is, if you don't have a scope, you don't really know what you would need one for. After you get one, you'll find yourself using it for all sorts of things you never dreamed of before. Thats why there are very few people out there without scopes who are desperate to get one. They don't know what they are missing.

I don't disagree, but I will note that all tools have a learning curve. For a beginner, the simpler the tool the better. A scope can introduce many traps, especially related to safety with mains voltages (avoidable) and probing technique (unavoidable).

If you are already an electronics expert, then sure, maybe you don't need one, but if you are not, then it will open your eyes.

I disagree with that. In my teens I designed and built my first computer (based on a 6800, 128bytes RAM, TTL, switches, LEDs and home-etched PCBs) with only a multimeter.

That required thinking, understanding, planning, and was an invaluable experience in many ways.

The OP will be able to do many things with printf statements, and a cheap logic analyser - plus thinking. See my .sig

[tggzzz]

Pace yourself!    A scope is a very basic piece of test equipment so they do provide a lot of value to a technician.   However new they are expensive.   I wold tend to suggest following Dave's advice in the first linked video and look for an old analog scope that you can get for free or at a significant discount.   The reasons are 1) Cheap.   2) The potential to learn a lot.    3) Analog scopes fill some niches nicely.     Purchase a decent digital scope once you have some experience under your belt.   The reason is you want to self inform yourself to make a better buying decision.   In the end you should know after a few months to a few years where your interests are going and thus make a good decision on scope features when you buy.

I agree with all that, with one proviso. Make very sure an old analogue scope is in good working condition. A beginner won't know whether something they see is in their circuit, their use of the scope setup, or their probing technique. If you add the possibility of a subtly broken scope, they would be hopelessly confused and discouraged.

The other major advantage of an analogue scope for a beginner is that it is simple and all the controls are visible - not hidden in a menu system.

I've snipped your other sensible points.

Now, over to the traditional set of comments that analogue scopes should be avoided like the plague...

[rstofer]

There are two features of the modern DSO that you just can't get with a CRO and they are terribly important plus the issue of 2 vs 4 channels.

• Decodng:  This can be done with a logic analyzer but it's much more useful to do it with a scope.  You get to see the actual timing of chip select versus clock and things like that.
• The big deal is single shot trigger mode. This mode captures a trace for one  sweep (one buffer full, 1/2 buffer before the trigger and 1/2 after the trigger) and holds it forever.  You can set the trigger point to be anywhere on the screen, not just in the center.  There are analog storage scopes but they don't work nearly as well as the modern DSO in terms of single shot mode.  In my view, this is the most important feature of the modern DSO
• There is also the 4 channel business.  Better to have them and not need them than the other way around.  There are 4 channel analog scopes and if you go that way, try to concentrate on those with 4 channels
  

I see nothing to gain by buying an obsolete CRO and then turning around in a year or so and buying a DSO.  Mind you, I still have my Tek 485 but it's been around for at least 12 years and the DS1054Z for just a couple.

For the very beginner with nothing on their bench, I would highly recommend the Digilent Analog Discovery 2.  Two channel scope, two channel arbitrary waveform generator, dual power supply and 16 digital pins that can be used as a logic analyzer.

https://store.digilentinc.com/analog-discovery-2-100msps-usb-oscilloscope-logic-analyzer-and-variable-power-supply/

I know it doesn't have a 4 channel scope and that's regrettable but what it does have is simply spectacular for breadboarding.

[Dubbie]

In my teens I designed and built my first computer (based on a 6800, 128bytes RAM, TTL, switches, LEDs and home-etched PCBs) with only a multimeter.

I don’t doubt that in the slightest. However with that sort of digital project you can single step through the logic, checking that the state is as you would expect literally step by step.

Imagine a microcontroller project with tight timing requirements and many tasks interacting. How on earth do you figure out where there is a hardware problem or software issue? Plugging in a scope might reveal in 10 seconds that you made an error configuring your DMA, or that your wiring is not up to scratch for the bus speed you are running. something that might take hours or poring through code or messing with hardware to discover otherwise.

[tggzzz]

In my teens I designed and built my first computer (based on a 6800, 128bytes RAM, TTL, switches, LEDs and home-etched PCBs) with only a multimeter.

I don’t doubt that in the slightest. However with that sort of digital project you can single step through the logic, checking that the state is as you would expect literally step by step.

Only partially true. The technique is to test each part of the system (hardware, software, compiler, application) in isolation before interconnecting them. Only the final test requires full-speed operation.

That strategy is necessary for all complex systems, both hardware and software.

Imagine a microcontroller project with tight timing requirements and many tasks interacting. How on earth do you figure out where there is a hardware problem or software issue? Plugging in a scope might reveal in 10 seconds that you made an error configuring your DMA, or that your wiring is not up to scratch for the bus speed you are running. something that might take hours or poring through code or messing with hardware to discover otherwise.

I don't have to imagine that! I am very well acquainted with all those concepts, problems, and the range of techniques and solutions.

Only a scope can measure and detect the "signal integrity" issues you mention. But a beginner won't know what to look for, nor know how poor probing technique will produce misleadingly invalid results.

A scope won't help solve tight timing requirements and task interaction. Printfs and simple cheap logic analysers are sufficient (drive an output high/low at appropriate points in the code). Even a multimeter or LED can be a useful diagnostic technique, e.g. an average voltage can be related to average timings.

A scope won't be any extra help w.r.t. setting up DMA correctly.

Nonetheless, a scope can be used and be helpful - and can distract and mislead!

[rstofer]

A scope won't be any extra help w.r.t. setting up DMA correctly.

Trigger on DMA Request and watch for DMA Grant.  Clearly, this can be done with a logic analyzer but it's not clear to me that the cheapest LAs can resolve much better than 10 ns and that's a lifetime in logic terms.  Unless, of course, the LA is using an external state clock.  An LA without a state clock input is mostly useless.

I think a scope if the better tool for checking setup and hold times even though a 100 MHz scope probably has a rise time of 3.5 ns.  The data and clock signals have the same rise time so their relationship is still easy to understand.

If you want to review a single I2C or SPI transaction, the 4 channel scope is handy because it will also be looking at signal integrity.  If you want to decode "War and Peace", a logic analyzer is a better choice.

In the end, people wind up with both tools.

[tggzzz]

A scope won't be any extra help w.r.t. setting up DMA correctly.

Trigger on DMA Request and watch for DMA Grant.  Clearly, this can be done with a logic analyzer but it's not clear to me that the cheapest LAs can resolve much better than 10 ns and that's a lifetime in logic terms.  Unless, of course, the LA is using an external state clock.  An LA without a state clock input is mostly useless.

That's one of the issues I have against cheap LAs. Expensive LAs finesse the issue by the expense of halving the sample rate and constraining the allowable "clock" width. Vintage LAs did it properly, but slowly!

I think a scope if the better tool for checking setup and hold times even though a 100 MHz scope probably has a rise time of 3.5 ns.  The data and clock signals have the same rise time so their relationship is still easy to understand.

Completely agreed. Scopes are the weapon of choice for signal integrity issues.

If you want to review a single I2C or SPI transaction, the 4 channel scope is handy because it will also be looking at signal integrity.  If you want to decode "War and Peace", a logic analyzer is a better choice.

A decent LA will allow you to skip the boring bits of War and Peace, and only read the key scenes A scope can't do that.

In the end, people wind up with both tools.

Yes, but there is a lot of overlap where it is possible to use either. It is fruitful to understand where one tool cannot be used and the other must be used - i.e. a tools USP

[rstofer]

Yes, there's overlap but, in my view, the scope is useful in a broader area.  Mostly, we're talking digital in this thread but that a limited view of electronics.

I like analog computing, I need some kind of scope to see the analog output.  It doesn't need whiz-bang specs because the time constant for integration is still one second but it does need dual channels and x-y mode.

To Arduino projects, almost everything will be digital but if I could only have a scope OR a logic analyzer, I'm pretty sure I would start with the scope.

Rigol DS1054Z unlocked to 100 MHz  -- what I am using today and it works well
Siglent SDS1104X-E unlocked to 200 MHz -- what I plan to get someday

[rstofer]

As long as we're talking ATmega328P Arduinos, bandwidth isn't such a big deal.  50 MHz and up should cover it.

However, it is worth noting that the Teensy 4.1 uses the Arduino infrastructure and matching libraries.  Application code is mostly portable and it runs at 600 MHz!  I'm not sure of the details but apparently it can toggle pins at 100 MHz. That's blistering fast!

[wizard69]

...snipped

Now, over to the traditional set of comments that analogue scopes should be avoided like the plague...

Ive read a few of those comments and don't believe them for a minute.    A functional scope as you point out is important but there is valuable learning with analog scopes.   I'm not knocking digital scopes just that for a few bucks you can get a good grounding in viewing wave forms.   As you note you need to learn to probe and learn and understand what the host of knobs do.

Knocking analog scopes is kinda like knocking analog VTVM that where the staple of the industry for decades.   The accuracy might not be there but that doesn't imply that they are not useful.   

[tggzzz]

As long as we're talking ATmega328P Arduinos, bandwidth isn't such a big deal.  50 MHz and up should cover it.

What's the risetime? For a scope's USP the clock rate is irrelevant, as I'm sure you are well aware.

What's the risetime of whatever is connected to the arduino, especially transients related to switching currents fast?

However, it is worth noting that the Teensy 4.1 uses the Arduino infrastructure and matching libraries.  Application code is mostly portable and it runs at 600 MHz!  I'm not sure of the details but apparently it can toggle pins at 100 MHz. That's blistering fast!

I've an MCU that will do arbitrary input/output at up to 250Mb/s on many i/o ports simultaneously. It will also count the transistions on one or more 62.5Mb/s input streams, plus do front- panel i/o plus communicate over a USB link to a PC - all at the same time. Now that's blisteringly fast

[james_s]

What's the risetime? For a scope's USP the clock rate is irrelevant, as I'm sure you are well aware.

What's the risetime of whatever is connected to the arduino, especially transients related to switching currents fast?

For a ATMEGA328P based Arduino I measured the rise and fall times for IO pins a while back at around 5ns.

In practice I can't say I've ever really needed to know that though, it was just out of curiosity. I don't normally care what the rise time is, it's negligible compared to the dwell times and when I'm looking at the signal in or out of a microcontroller I usually just want to know if the pin is toggling, if the levels look reasonable, and then does it look like the pattern that I expect to be there. It's not important that the scope might not show all the fine details, if I really need to see that I've got a 1GHz scope on the bench but most of the time for this stuff I'm just going to use the little one with 100MHz probes, good enough.

There are niche cases where it matters, but for the typical hobbyist using something like an arduino the rise time is largely irrelevant, it is what it is, if you have the budget for an exotic scope you probably won't be using an arduino in the first place. More bandwidth is a nice luxury but it's ridiculous and elitist to suggest that one cannot get by without it when so many do.

[tggzzz]

What's the risetime? For a scope's USP the clock rate is irrelevant, as I'm sure you are well aware.

What's the risetime of whatever is connected to the arduino, especially transients related to switching currents fast?

For a ATMEGA328P based Arduino I measured the rise and fall times for IO pins a while back at around 5ns.

In practice I can't say I've ever really needed to know that though, it was just out of curiosity. I don't normally care what the rise time is, it's negligible compared to the dwell times and when I'm looking at the signal in or out of a microcontroller I usually just want to know if the pin is toggling, if the levels look reasonable, and then does it look like the pattern that I expect to be there. It's not important that the scope might not show all the fine details, if I really need to see that I've got a 1GHz scope on the bench but most of the time for this stuff I'm just going to use the little one with 100MHz probes, good enough.

There are niche cases where it matters, but for the typical hobbyist using something like an arduino the rise time is largely irrelevant, it is what it is, if you have the budget for an exotic scope you probably won't be using an arduino in the first place. More bandwidth is a nice luxury but it's ridiculous and elitist to suggest that one cannot get by without it when so many do.

The key point is knowing the theory, so people can predict where it does matter.

I repeatedly point out that, with skill imagination and thought, you can debug a hell of a lot with a multimeter, LEDs, and printf statements. One classic example is using a voltmeter to measure an MCU's average percentage idle time.

[Ian.M]

If you are on a very limited budget, you *DON'T* need that 100MHz 4 channel DSO.   Even a 50MHz dual channel CRO (in good condition) + a USB logic analyser + a cheap DMM is sufficient for hardware debugging of nearly all AVR based Arduino (and other 8 bit MCU) projects, including checking for most signal integrity issues.   Of course I'd prefer a 100MHz delay timebase CRO, but not if it was going to cost me big bucks that I should be saving towards getting a good modern DSO when I actually need one. 

OTOH good CROs are getting harder to find in good condition, and a late CRO era analog scope that needs significant repairs is a mill-stone around the neck of a novice who doesn't have the budget, experience and test equipment to restore it to good working order.  I wouldn't even consider a faulty CRT era DSO or MSO.

Most cheap off-brand handheld scopes and cheap low-end USB scopes are essentially worthless for anything much above audio and ultrasonic frequencies, and the software/firmware is often a reeking pile of sh!te that would fail any independent real world usability testing!  Don't get one unless you need extreme portability, literally a lab in a briefcase

[james_s]

I'll always have a soft spot for analog CROs, but IMO the only reason a hobbyist should consider one is if it's very cheap and budget doesn't allow one of the hobbyist oriented "real" DSOs that you can get for ~$350-$400. There is still value in playing with an analog scope but beyond XY mode they don't offer anything over even a low end DSO. I love my Tek 465B and don't plan to get rid of it but I very rarely use it once I got a decent DSO. A nice older analog scope is better than the toy DSOs but outside of a few niche applications digital is superior.

[tggzzz]

I'll always have a soft spot for analog CROs, but IMO the only reason a hobbyist should consider one is if it's very cheap and budget doesn't allow one of the hobbyist oriented "real" DSOs that you can get for ~$350-$400. There is still value in playing with an analog scope but beyond XY mode they don't offer anything over even a low end DSO. I love my Tek 465B and don't plan to get rid of it but I very rarely use it once I got a decent DSO. A nice older analog scope is better than the toy DSOs but outside of a few niche applications digital is superior.

Given the choice between a working analogue scope from 70s-90s and a digitising scope older than a decade, I would probably choose the analogue scope.

Earlier digitising scope were just too limited for many purposes. Examples... They couldn't show the peak and mean values in a display sample. The number of samples was too short to be able to zoom in on relevant features occurring long after the trigger.

Apart from that, I have a couple of digitising scopes that I use on the few occasions when their USP is valuable (single shot, postprocessing, raw bandwidth). But for the responsiveness, simplicity and ease of use I keep an old analogue scope on my bench. Those emphasised points are important for a beginner.

[Dubbie]

To be honest, I don’t know why people make out oscilloscopes to be this hard to use thing. Once you get the general concept, you can generally feel your way around an unfamiliar one in very short order. It’s the rest of being a good EE that is the hard part! Not using a scope! I don’t think it matters about learning a CRO first then working your way up to a DSO. It’s not rocket science. Start with whatever is most convenient to get your hands on and you will be completely fine I guarantee it. Now if only designing robust and well performing circuits was as easy! 

[tggzzz]

To be honest, I don’t know why people make out oscilloscopes to be this hard to use thing. Once you get the general concept, you can generally feel your way around an unfamiliar one in very short order. It’s the rest of being a good EE that is the hard part! Not using a scope! I don’t think it matters about learning a CRO first then working your way up to a DSO. It’s not rocket science. Start with whatever is most convenient to get your hands on and you will be completely fine I guarantee it. Now if only designing robust and well performing circuits was as easy! 

CRO vs DSO indicates you are confused, which may be significant. CROscilloscope is an output technology, DSOscilloscope is an input and processing technology. I have a LeCroy digitising (and digital) scope that has a cathode ray output.

Note that the context is that the OP is a beginner.

Of course, everything can be learned especially once you have the fundamentals under your belt. Nonetheless, having everything visible on the front panel does make it simpler.

Even experienced professionals are misled and frustrated by subtle important settings hidden three deep in a menuing system. Typical expressions I've repeatedly heard include "I don't believe that" and "what's it doing now".

[RoGeorge]

Since there is no budget limit in the OP, buy the best available one, a Keysight UXR 110GHz BW, 256GS/s, 10-bit Real-Time Oscilloscope



 

[Mr. Scram]

CRO vs DSO indicates you are confused, which may be significant. CROscilloscope is an output technology, DSOscilloscope is an input and processing technology. I have a LeCroy digitising (and digital) scope that has a cathode ray output.

Note that the context is that the OP is a beginner.

Of course, everything can be learned especially once you have the fundamentals under your belt. Nonetheless, having everything visible on the front panel does make it simpler.

Even experienced professionals are misled and frustrated by subtle important settings hidden three deep in a menuing system. Typical expressions I've repeatedly heard include "I don't believe that" and "what's it doing now".

I don't think anyone is confused about what he's saying. CRO is often used as analogue for analog on here. Digital or analogue doesn't matter much. Don't get fooled into a quest to find an analog oscilloscope if a decent digital one presents itself. The reverse is also true. Getting started isn't rocket science either way. I suspect much of the frustration of experienced professionals stems from them growing up with different technology, while younger people take to menu based systems without issue.

[Ian.M]

CRO vs DSO indicates you are confused, which may be significant. CROscilloscope is an output technology, DSOscilloscope is an input and processing technology. I have a LeCroy digitising (and digital) scope that has a cathode ray output.

Note that the context is that the OP is a beginner.

Errr  . . . *NO*.   A CRO is analog all the way from the input or ext trigger BNCs to the CRT deflection plates.  There may^* or may not be some digital readouts for attenuator and timebase settings and maybe even an integrated DMM.  A DSO digitises its inputs (after some analog preprocessing) and has a digital counter timebase.  its display technology is (mostly^#) irrelevant.   A MSO is a combo of a CRT DSO and a CRO on the same CRT.

* As  CROs are ageing beasts, 'pure' CROs without digital 'encrustations' are vastly preferable as they are simpler and thus more maintainable.

# The display technology may influence the display update rate.

Of course, everything can be learned especially once you have the fundamentals under your belt. Nonetheless, having everything visible on the front panel does make it simpler.

Even experienced professionals are misled and frustrated by subtle important settings hidden three deep in a menuing system. Typical expressions I've repeatedly heard include "I don't believe that" and "what's it doing now".

Too true.  If you have enough experience with scopes, and not just a few semesters setting them strictly according to the lab guide for your course, even if you are a hobbyist who gained that experience using a decent CRO as your 'daily driver', you'll have a good idea when a DSO is telling you porkies, know when to dig into the menus and wont be so dependent on its 'autosetup' button.  However that still takes time and a brain firing on all cylinders, so many of us here miss the easy days when all you had to do to get sanity out of a CRO was to push in all pullout knobs, set shift controls to mid-range and everything else to 'cal', and select appropriate input coupling, attenuator, trigger and timebase settings.

@RoGeorge: I doubt I could even afford a day's rental on that beast!

[Wallace Gasiewicz]

OK, this is from someone who has been using old scopes for a while and has not used Arduino technology. (perhaps I should get Arduino for Idiots) I am expecting my second DSO any day now(Agilent 54810A), I sold an old digital scope years ago, I did not think the older digital scopes were competitive with a good analog. I use scopes to fix old stuff. I use a HP 1980 scope system at present and have a Tek 475 also. The 1980 has some digital features but essentially is an analog scope. It also has digital display memory but that is another story.
What frequencies are we talking about with the Arduino? What voltage levels are common digital signals?
I know that I do not know.
It is very difficult to see a non repeating signal on an analog scope. You have to get the circuit to "repeat" to have a decent chance of seeing the signal. Trouble shooting such things as trigger circuits takes a lot of time.
I would think that one shot would help a lot.
Since with an Arduino we are dealing with several digital signals would not a logic analyzer be more helpful? Some old ones are cheap but I do not know if they meet the freq needs etc (that's why I asked about freq requirements)...I sold a HP logic analyzer that was a 200 MHz scope for $100, working, with leads. I hated the display however. Again I do not know what freq we are talking about with the Arduino. If frequency is low enough, maybe one of these old logic analyzers would be the most efficient use of financial resources. A logic analyzer and a 200 MHx scope for $100, And a great huge box. HA! What do you folks think?
On fixing old equipment, it is difficult to connect more than a few probes, it would seem that putting together a fresh Arduino project, it would be far easier to leave access to many signals so you could easily connect a logic analyzer to see many signals in real time.

[tggzzz]

CRO vs DSO indicates you are confused, which may be significant. CROscilloscope is an output technology, DSOscilloscope is an input and processing technology. I have a LeCroy digitising (and digital) scope that has a cathode ray output.

Note that the context is that the OP is a beginner.

Of course, everything can be learned especially once you have the fundamentals under your belt. Nonetheless, having everything visible on the front panel does make it simpler.

Even experienced professionals are misled and frustrated by subtle important settings hidden three deep in a menuing system. Typical expressions I've repeatedly heard include "I don't believe that" and "what's it doing now".

I don't think anyone is confused about what he's saying.

A beginner, such as the OP, looking up the acronyms ought to be confused.

CRO is often used as analogue for analog on here.

You mean "analogue for analogue" And yes, the terms are poorly used here - as are others which measure time in mohs (e.g. 10mS instead of 10ms).

It is all wrong, wrong I tell you, wrong!

Digital or analogue doesn't matter much. Don't get fooled into a quest to find an analog oscilloscope if a decent digital one presents itself. The reverse is also true.

Agreed.

(Unless it is a crap digital scope made by Schlumberger where the user interface was created by softies. The vertical sensitivity was marked "calibration" )

Getting started isn't rocket science either way. I suspect much of the frustration of experienced professionals stems from them growing up with different technology, while younger people take to menu based systems without issue.

I've been using menu systems for 38 years starting on a Sun 1 and continuing on Lisas and Mac. They have never confused me.

The problem is things hidden inside a menu hierarchy, without being very visible at the point of making a measurement.

[tggzzz]

CRO vs DSO indicates you are confused, which may be significant. CROscilloscope is an output technology, DSOscilloscope is an input and processing technology. I have a LeCroy digitising (and digital) scope that has a cathode ray output.

Note that the context is that the OP is a beginner.

Errr  . . . *NO*.   A CRO is analog all the way from the input or ext trigger BNCs to the CRT deflection plates. 

Er, **YES**

A cathode ray oscilloscope (CRO) uses a CRT to display the results. Some of my CROs (e.g. Tek 24x5) are analogue all the way through. Others (e.g. my LeCroy 9310AM) have an analogue front end followed by a digitiser and DSP, and the results are displayed on a pleasant monochrome orange CRT.

Here's a picture of it on auto trigger, to make the time and amplitude samples very visible. Note the 100MS/s above the top right of the CRT:

There may^* or may not be some digital readouts for attenuator and timebase settings and maybe even an integrated DMM.  A DSO digitises its inputs (after some analog preprocessing) and has a digital counter timebase. 

If you believe that, you are wrong. Please explain the picture below of my Tek 2465 scope (ICs have 1984 datecodes) with integral counter timer.

You should take into account that the 24x5s are widely regarded as analogue scopes, from beginning to end.
For the sine wave note the precision of the frequency readout, made by the scope without my setting cursors. (Neither the oscillator nor scope had warmed up; the OCXO is <1ppm)
For the pulses, note that the scope has measured the positive width without my setting cursors.



Amusingly and for the avoidance of doubt, the 2465 counter/timer/word recogniser interface is extremely confusing, even worse than a menu system

its display technology is (mostly^#) irrelevant.   A MSO is a combo of a CRT DSO and a CRO on the same CRT.

Maybe some people think of an MSO that way - but if they wish to have a common language with other people they will have to state their definition at the start of a conversation.

To companies like Keysight a Mixed Signal Oscilloscope (MSO) is something with a few analogue channels and more digital channels. Effectively they are a combination of scope and a simple logic analyser with the same timebase. They can also display multiple digital signals as a single bus, and the like. The first example returned by google: https://www.keysight.com/en/pdx-x201841-pn-MSOX3012A/mixed-signal-oscilloscope-100-mhz-2-analog-plus-16-digital-channels?cc=HU&lc=eng

[tggzzz]

OK, this is from someone who has been using old scopes for a while and has not used Arduino technology. (perhaps I should get Arduino for Idiots) I am expecting my second DSO any day now(Agilent 54810A), I sold an old digital scope years ago, I did not think the older digital scopes were competitive with a good analog. I use scopes to fix old stuff. I use a HP 1980 scope system at present and have a Tek 475 also. The 1980 has some digital features but essentially is an analog scope. It also has digital display memory but that is another story.
What frequencies are we talking about with the Arduino? What voltage levels are common digital signals?

The maximum frequency in the signal is defined by the risetime, not the frequency. The rule of thumb is BW=0.35/t_r. A little theory and a couple of practical measurements are shown here: https://entertaininghacks.wordpress.com/2018/05/08/digital-signal-integrity-and-bandwidth-signals-risetime-is-important-period-is-irrelevant/

I know that I do not know.
It is very difficult to see a non repeating signal on an analog scope. You have to get the circuit to "repeat" to have a decent chance of seeing the signal. Trouble shooting such things as trigger circuits takes a lot of time.
I would think that one shot would help a lot.

Truly one-shot events are the USP for storage scopes, whether digitising storage scopes or analogue storage scopes (may they rest in hell).

Except that a general good principal is to make any situation, either correct operation or buggy operation, repeatable. When that is possible then the storage becomes less valuable.

Since with an Arduino we are dealing with several digital signals would not a logic analyzer be more helpful? Some old ones are cheap but I do not know if they meet the freq needs etc (that's why I asked about freq requirements)...

You can get simple slow (24MS/s/channel??) modern LAs very cheaply. They will be useful for many "Arduino tasks" but certainly not all. If nothing else, they would be a good way to gain experience.

As with all tools, learn to exploit what it can do, and be aware of how it will lie to you

I sold a HP logic analyzer that was a 200 MHz scope for $100, working, with leads. I hated the display however. Again I do not know what freq we are talking about with the Arduino. If frequency is low enough, maybe one of these old logic analyzers would be the most efficient use of financial resources. A logic analyzer and a 200 MHx scope for $100, And a great huge box. HA! What do you folks think?
On fixing old equipment, it is difficult to connect more than a few probes, it would seem that putting together a fresh Arduino project, it would be far easier to leave access to many signals so you could easily connect a logic analyzer to see many signals in real time.

200MS/s per channel would be more than adequate, but make sure the sample depth is sufficiently long.

[Wallace Gasiewicz]

I think I get it, the usefulness of a "scope" in this situation is it's ability to show brief sharp pulses, which are common in computer circuits. I have noticed that one old scope is different from another in it's ability to display a pulse somewhat independent of it's MHz rating
So Perhaps one of the old HP logic analyzers that doubles as a 100-200 MHz scope might be a good choice for the OP.
Might be pretty inexpensive and the OP will also get a usable scope. This might enable further experimentation and descent into TE acquiring habits.

As an aside, My HP 1980 has a digital storage for waveforms that appear on the screen. It takes about 3-4 seconds for it to scan the waveform and then display it as a series of dots. This feature has very limited usefulness. Interesting, but dumb. as compared to today's digital scopes.
Wally

[tggzzz]

I think I get it, the usefulness of a "scope" in this situation is it's ability to show brief sharp pulses, which are common in computer circuits. I have noticed that one old scope is different from another in it's ability to display a pulse somewhat independent of it's MHz rating

All the signals in an Arduino are analogue. However, the receiving logic gates can accurately interpret them as digital signals, if (and only if) they meet the requirements in terms of voltage and timing.

The scopes USP is that it can show you the voltages (>X for a logic 1, <Y for a logic 0) and the relative timings (t_setup and t_hold). For the latter you need two channels at least, and sometimes more is convenient. This is "ensuring signal integrity".

After you have verified signal integrity, you can choose to "flip to the digital domain" and debug digital signals with a logic analyser or printf statements. A logic analyser will have many more inputs than a scope, and can trigger on patterns rather than just voltage levels. It can also ignore "uninteresting" signals, e.g. only record what happens at a clock signal. That can be helpful.

On the other hand, if you look at digital signals with a scope, you can (and sometimes the scope) can interpret the waveforms visually as digital signals.

So, there is overlap between scopes and logic analysers, as well as important differences. Choose a screwdriver for screws, and a hammer for nails

So Perhaps one of the old HP logic analyzers that doubles as a 100-200 MHz scope might be a good choice for the OP.
Might be pretty inexpensive and the OP will also get a usable scope. This might enable further experimentation and descent into TE acquiring habits.

As an aside, My HP 1980 has a digital storage for waveforms that appear on the screen. It takes about 3-4 seconds for it to scan the waveform and then display it as a series of dots. This feature has very limited usefulness. Interesting, but dumb. as compared to today's digital scopes.
Wally

I designed and built my first computer with LEDs and a multimeter. It required care and thought, but that is not a bad thing!

Use the tools you have. When you can't think of a way to use them to see what you need, you will know what you need to obtain.

Simple example. I've just been preparing a Tek 2465DMS scope for fleabay. I had to replace the infamous RIFA capacitors, and then found everything worked except that it wouldn't display current. Now that scope has many advanced (for 1984!) features including a microprocessor, and the problem could have been in many places, digital or analogue. In the end I used a multimeter to find where the signal had not reached, and eventually spotted a wire had separated from an input relay. Resolder it and job done.

[wizard69]

This thread is most interesting but I think people are missing one important point, we are talking about somebody new to electronics.    The reasons I suggest a used scope isn't so much because it would be analog (there are used digital scopes) but rather that it is a low cost investment that allow learning.    In the end when somebody gets into this as hobby the direction that hobby might take is unknown at the start.    For example if one gets into robotics, CNC or other technologies that involve higher power circuits, there might be a need for a scope with isolated inputs.

So in short it is far better to hold off buying anything that is going to hit the budget significantly until you have a better idea as to what you need as a tech/engineer.    It is very easy to throw away a lot of money on an expensive instrument that doesn't really fit ones need.   This is why I think it is better to develop the knowledge to make a good purchase than to rely upon suggestions that reflect somebody else's interests.   Thus the suggestion about a used analog scope (really any used scope that can be had dirt cheap) is more about learning about what you need at the lowest possible cost.   Once somebody knows what they need it is far easier to understand the suggestions being made here. 

[tggzzz]

This thread is most interesting but I think people are missing one important point, we are talking about somebody new to electronics.    The reasons I suggest a used scope isn't so much because it would be analog (there are used digital scopes) but rather that it is a low cost investment that allow learning.    In the end when somebody gets into this as hobby the direction that hobby might take is unknown at the start.    For example if one gets into robotics, CNC or other technologies that involve higher power circuits, there might be a need for a scope with isolated inputs.

So in short it is far better to hold off buying anything that is going to hit the budget significantly until you have a better idea as to what you need as a tech/engineer.    It is very easy to throw away a lot of money on an expensive instrument that doesn't really fit ones need.   This is why I think it is better to develop the knowledge to make a good purchase than to rely upon suggestions that reflect somebody else's interests.   Thus the suggestion about a used analog scope (really any used scope that can be had dirt cheap) is more about learning about what you need at the lowest possible cost.   Once somebody knows what they need it is far easier to understand the suggestions being made here.

I fundamentally agree, with the proviso that it is a working scope. See ,for example, my previous post.

There as many vocal people that disagree, and "push" modern digitising scopes.

[rstofer]

There as many vocal people that disagree, and "push" modern digitising scopes.

Alas, I'm one of them.  Although I have that Tek 485 and built my first analog scope from plans in the ARRL Handbook clear back in the late '50s, I view analog scopes vs digital scopes the same way I view the slide rule versus the HP48GX.  I got through undergrad with a slide rule, I still drag it out from time to time, but I actually use a calculator.

I don't agree that there is anything special to learn from an analog scope.  The primary controls (V/div, t/div, position, trigger type/level) are the same on any scope.  But the digital scope brings so much more to the game and, in particular, Single shot capture.  In either case, the learning curve is going to be about the same but there is so much more capability with the modern DSO.

A used, but  working, analog scope may fill a financial check box but that would be one of only two reasons for buying such a thing.  The other being bandwidth.  It's pretty easy to find a Tek 485 350 MHz scope for a couple of hundred bucks, plus a bunch for shipping.  In the end, it's a great analog scope with a bunch of bandwidth, plenty of features and not a bit of measurement capability.

I'm not sure what the resale would be on a used (again) analog scope but I'll bet the digital will hold its value pretty well.

Today, on eBay, there are a couple of 485s  One is about $200 plus $24 shipping and the other is refurbished, warranted and calibrated for about $500 and $50 shipping.  At the same time, on Amazon, the Rigol DS1054Z is $349 with free shipping.  Kind of in the middle of the cost range between a dubious 485 and a known working/calibrated 485.

If the money isn't there for the Rigol, so be it.  But under no condition is the used analog going to compare with a modern DSO - particularly a 4 channel DSO.  And the resale on the Rigol is going to be around $315 according to eBay.   Not a bad deal if things don't work out in electronics.

And the new DSO works, has a warranty and probably doesn't need a lot of capacitor replacements.  And if you need one of the unobtanium custom ICs, you are just out of luck.

I simply don't see the justification for a used analog scope.  I have one, it works very well, I'm not going to junk it but I actually use the DSO.

[rstofer]

So in short it is far better to hold off buying anything that is going to hit the budget significantly until you have a better idea as to what you need as a tech/engineer.    It is very easy to throw away a lot of money on an expensive instrument that doesn't really fit ones need.   This is why I think it is better to develop the knowledge to make a good purchase than to rely upon suggestions that reflect somebody else's interests.   Thus the suggestion about a used analog scope (really any used scope that can be had dirt cheap) is more about learning about what you need at the lowest possible cost.   Once somebody knows what they need it is far easier to understand the suggestions being made here.

I'm pretty dense, I want to see a picture!  Understanding circuits, to me, requires visualization and simulation doesn't count.  I want to take measurements, with a DMM certainly, but also with a scope.  If I can't see it, I won't really understand it.

I remember pencil whipping Bode' plots in undergrad.  That's not nearly as satisfying as see a plot using the AD2.

Frankly, for the Arduino user, the Analog Discovery 2 is a better choice than either an analog or digital scope.  The 16 channels of digital IO with protocol decoding is a pretty slick feature.  Having a 2 channel scope and 2 channel arbitrary waveform generator brings more capability.  We all understand how PWM works but it's even better to see it in motion.  Then there are the punch-through problems of a discrete H-bridge...  Timing is everything!

If I'm going to use I2C from a uC, it is absolutely guaranteed that it won't work right out of the box.  That protocol is a PITA to get working.  SPI isn't nearly as difficult but you want to be certain the last bit has shifted out before raising CS' in software.  How do you PROVE that if you can't see it?  You can pretty much guess how I KNOW that this  is a problem.

I can't imagine trying to do uC projects without a scope.  A logic analyzer is very helpful, and a lot cheaper, but a scope is my primary tool.  The fact that the modern DSO can do decoding is just icing on the cake.  Let's not forget Single trigger mode!

If I were a newcomer having no equipment and no apparent path, I would buy the AD2 to handle measurements and something in terms of a power supply. There are plenty of used HP power supplies on eBay.

[james_s]

I simply don't see the justification for a used analog scope.  I have one, it works very well, I'm not going to junk it but I actually use the DSO.

The justification is when you can get them for <$100, sometimes much less, I've received several of them for free and given away at least two. They're not as common as they used to be but they are still out there and occasionally burn up. If the price is right, an analog scope can get the job done. No way I'd pay $500 though, not even for a high end low hour unit in excellent condition. If the price is right and funds are limited, you do what the budget allows.

Or if you have a specific need for XY mode, that's one of the few reasons I pull out my 465 anymore, XY mode on an analog scope blows away XY on any DSO I've tried but that's a pretty niche use case.

[rstofer]

I simply don't see the justification for a used analog scope.  I have one, it works very well, I'm not going to junk it but I actually use the DSO.

The justification is when you can get them for <$100, sometimes much less, I've received several of them for free and given away at least two. They're not as common as they used to be but they are still out there and occasionally burn up. If the price is right, an analog scope can get the job done. No way I'd pay $500 though, not even for a high end low hour unit in excellent condition. If the price is right and funds are limited, you do what the budget allows.

If the money isn't there for a DSO, so be it.  Any scope is better than none. 

What isn't justifiable is the idea that there is some kind of 'right of passage' or superior 'learning experience' derived from an analog scope.  They are dinosaurs (other than bandwidth) and should be left in the tar pits.  Including mine.

Or if you have a specific need for XY mode, that's one of the few reasons I pull out my 465 anymore, XY mode on an analog scope blows away XY on any DSO I've tried but that's a pretty niche use case.

I have a pretty specific XY use case when I want to watch the output of an analog computer.  The X axis will be driven by an internal sweep generator and the y axis by the output of the differential equation.  I use it all the time on my DS1054Z and it works very well.  But that's my use case, others may have a different experience.

[tggzzz]

If the money isn't there for a DSO, so be it.  Any scope is better than none. 

Precisely. Even a digitising scope is better than none

What isn't justifiable is the idea that there is some kind of 'right of passage' or superior 'learning experience' derived from an analog scope.  They are dinosaurs (other than bandwidth) and should be left in the tar pits.  Including mine.
 use case.

The point in favour of a working analogue scope is ease of learning and a more gentle learning curve.

Anything else is bunkum.

[Mr. Scram]

The point in favour of a working analogue scope is ease of learning and a more gentle learning curve.

Anything else is bunkum.

As others have noted before, the learning curve on either analogue or digital oscilloscope is already flat as can be. It's not a complicated instrument to get started with. The only exception I see is getting a broken unit or just a flat out terrible one like one of those DIY kits. Any not-horrible two channel oscilloscope in reasonable condition you can get your hands on will do.

[tggzzz]

The point in favour of a working analogue scope is ease of learning and a more gentle learning curve.

Anything else is bunkum.

As others have noted before, the learning curve on either analogue or digital oscilloscope is already flat as can be. It's not a complicated instrument to get started with. The only exception I see is getting a broken unit or just a flat out terrible one like one of those DIY kits. Any not-horrible two channel oscilloscope in reasonable condition you can get your hands on will do.

In my experience, having introduced several groups of amateurs to scopes, the curve is flatter on an analogue scope.

The principal reasons are:

• all tweakable controls and their settings are visible on an analogue scope's front panel
• digitising scopes have more things that can be tweaked, and in some cases need to be tweaked - and therefore explained and understood
• digital scopes have controls hidden, in layers of a menuing system
• all digital scopes vary subtly from each other, in terms of menuing system, cryptic screen icons, and even terminology - which makes it more difficult to swap from one digitising/digital scope to another

All of those points are surmountable, but they cause a steeper learning curve for beginners.

[tautech]

A scope is a scope is a scope !
They all do the same things and their only difference for the beginner is the UI, be it with knobs, buttons, mouse or a touch display.
One overlooks how little settings info is displayed on a CRO yet most DSO's offer substantial settings info on the display so in screenshots we can see what the beginner might be doing wrong and advise them appropriately.

OTOH a CRO must have a pic taken of the whole instrument to examine settings and in such a way to see if knobs are pulled or pushed. 

The beginner is much better served with a modern DSO to come to the forum or their mentor for advice.

[rstofer]

And let's not forget the "Auto" button.  It may well be that the scope powered up with settings you saved months ago and have long forgotten.  Rather than hunt down all the possibilities, push "Auto" and everything will be fine.

I know the pros don't like the feature but I find it quite useful.  Not only do I get a trace somewhere on the screen, it might actually be pretty close to what I want.  A little tweaking of V/div and t/div and I'm there!  Maybe move the position a bit...

Even analog scopes have a "Beam Finder" button to prove that they can draw a line on the screen.  Unfortunately, the line is totally unrelated to the incoming signals, they can still be way off screen.

[Mr. Scram]

In my experience, having introduced several groups of amateurs to scopes, the curve is flatter on an analogue scope.

The principal reasons are:

• all tweakable controls and their settings are visible on an analogue scope's front panel
• digitising scopes have more things that can be tweaked, and in some cases need to be tweaked - and therefore explained and understood
• digital scopes have controls hidden, in layers of a menuing system
• all digital scopes vary subtly from each other, in terms of menuing system, cryptic screen icons, and even terminology - which makes it more difficult to swap from one digitising/digital scope to another

All of those points are surmountable, but they cause a steeper learning curve for beginners.

That's essentially the same point worded in slightly different ways. Even if there were to be some advantage to quite simple versus somehow simpler, it seems to be quickly overtaken by digitizing oscilloscopes being simpler in the sense that they just hand the data to you instead of having to count graticules or inferring it in other ways. No Auto button either. And again, the menu argument seems to stem mostly from people not being used to it. If you watch a teen using an Apple device you'd quickly find that they have no trouble with those at all, despite endless gestures and obscured functions.

Not that it all matters much, the argument seems to be whether it's easier to learn to eat an apple or a banana. People can figure it out either way. Just getting the Arduino tool chain set up tends to be many times more complicated.

[rstofer]

Buying a CRO is simply deferring the cost of a DSO at the extra cost of the CRO.  The new user WILL be buying a DSO just as soon as they see how little the CRO can do when compared to a modern DSO.  Assuming they hang around electronics long enough to find out...

I understand cash flow, I really do!  If the funds aren't there for a DSO, consider buying the least expensive CRO that can possibly do the job because it is certainly not going to be the last scope and there's no point in having a lot of money tied up in it.

[Mr. Scram]

Buying a CRO is simply deferring the cost of a DSO at the extra cost of the CRO.  The new user WILL be buying a DSO just as soon as they see how little the CRO can do when compared to a modern DSO.  Assuming they hang around electronics long enough to find out...

I understand cash flow, I really do!  If the funds aren't there for a DSO, consider buying the least expensive CRO that can possibly do the job because it is certainly not going to be the last scope and there's no point in having a lot of money tied up in it.

To be fair, a CRO could probably be sold at the same price. Then again, a DS1054Z doesn't tend to lose a lot of value either. Obviously, you need to be able to spend it in the first place.

[james_s]

Buying a CRO is simply deferring the cost of a DSO at the extra cost of the CRO.  The new user WILL be buying a DSO just as soon as they see how little the CRO can do when compared to a modern DSO.  Assuming they hang around electronics long enough to find out...

I understand cash flow, I really do!  If the funds aren't there for a DSO, consider buying the least expensive CRO that can possibly do the job because it is certainly not going to be the last scope and there's no point in having a lot of money tied up in it.

I used one for ~15 years of debugging microcontrollers and everything else before I ever had a DSO, they can do the job, some things are not as easy but there are workarounds. Sometimes deferring the cost is a necessity, I mean the mortgage on my house is really only deferring the cost, but financial realities made that a necessity. A CRO is better than no scope at all by a long shot, and if you don't have the budget for a DSO you don't have the budget.

It's strange to me how people seem to have forgotten already that it's only the last 10 years or so that any sort of DSO was within reach of all but the most serious hobbyists and yet somehow we got by and managed to create and debug all sorts of things, a DSO is a nice luxury, it's not a necessity. I wouldn't want to give mine up but I could get by without it if I had to.

[Mr. Scram]

I used one for ~15 years of debugging microcontrollers and everything else before I ever had a DSO, they can do the job, some things are not as easy but there are workarounds. Sometimes deferring the cost is a necessity, I mean the mortgage on my house is really only deferring the cost, but financial realities made that a necessity. A CRO is better than no scope at all by a long shot, and if you don't have the budget for a DSO you don't have the budget.

It's strange to me how people seem to have forgotten already that it's only the last 10 years or so that any sort of DSO was within reach of all but the most serious hobbyists and yet somehow we got by and managed to create and debug all sorts of things, a DSO is a nice luxury, it's not a necessity. I wouldn't want to give mine up but I could get by without it if I had to.

People did fine without cars, running water or electricity until very recently but I wouldn't want to be without them if I can help it. I could, but don't want to.

[james_s]

People did fine without cars, running water or electricity until very recently but I wouldn't want to be without them if I can help it. I could, but don't want to.

Plenty of people get by without cars, some for financial reasons, some because they just don't need one. Buses and bicycles aren't as convenient in many cases but they get the job done. Comparing running water to a DSO is a bit silly, if one is proficient an analog scope can do most of the things a DSO can do.'

Again, if you don't have the budget you don't have the budget. You get by with what you have or can afford.

[tggzzz]

People did fine without cars, running water or electricity until very recently but I wouldn't want to be without them if I can help it. I could, but don't want to.

Plenty of people get by without cars, some for financial reasons, some because they just don't need one. Buses and bicycles aren't as convenient in many cases but they get the job done. Comparing running water to a DSO is a bit silly, if one is proficient an analog scope can do most of the things a DSO can do.'

Again, if you don't have the budget you don't have the budget. You get by with what you have or can afford.

Precisely.

Doing more with less has another benefit: it gets you in the mindset that will enable you to do very advanced work. Simple example: what scope do you use when you are creating the world's fastest scope?

"An engineer is somebody that can do for $1 what any fool can do for $2" - Arthur M. Wellington

[Mr. Scram]

Doing more with less has another benefit: it gets you in the mindset that will enable you to do very advanced work. Simple example: what scope do you use when you are creating the world's fastest scope?

"An engineer is somebody that can do for $1 what any fool can do for $2" - Arthur M. Wellington

Yet you too have the fancy oscilloscope. What that means for being a fool I'll leave for the reader to decide.

[tggzzz]

Doing more with less has another benefit: it gets you in the mindset that will enable you to do very advanced work. Simple example: what scope do you use when you are creating the world's fastest scope?

"An engineer is somebody that can do for $1 what any fool can do for $2" - Arthur M. Wellington

Yet you too have the fancy oscilloscope. What that means for being a fool I'll leave for the reader to decide.

You're going to have to explain what you mean by that cryptic comment.

[wizard69]

Buying a CRO is simply deferring the cost of a DSO at the extra cost of the CRO. 

Exact!   The cost is deferred until the buyer has a better understanding of what he needs.   It is the difference between spending $50 or $500 to $5000.

The new user WILL be buying a DSO just as soon as they see how little the CRO can do when compared to a modern DSO.  Assuming they hang around electronics long enough to find out...

I don't really buy the idea that someone will rush out and buy DSO so quickly.   Especially if they have a functional analog scope.

As for the second point does it really make sense to invest huge bucks into what might be a hobby when you are getting started?    I mean seriously people here are rushing out to buy $300 desktop power supplies ot power up an Arduino that can run off a cheap plug pack.    I just don't see the wisdom in investing in high value stuff right out of the gate.

I understand cash flow, I really do!  If the funds aren't there for a DSO, consider buying the least expensive CRO that can possibly do the job because it is certainly not going to be the last scope and there's no point in having a lot of money tied up in it.

Yes that is the point for somebody starting out in this field as a hobby.   If this was an engineer setting up a "work at home" solution I can see a completely different set of answers being offered up.

One of the reasons I often suggest in these threads for newbies to build their own power supplies is the value in learning.   Even a kit offers plenty of soldering and assembly experience on usually forgiving bits and pieces.   It isn't just soldering of course, the more your power supply is DIY the more there is to learn.

 In the case of an O'scope a newbie can't realistically be expected to make one.   At the same time he doesn't have the background to use an advance DSO with an endless list of bells and whistles.  More importantly a newbie can't even specify his scope needs which is probably the best reason to put off a purchase.

[rstofer]

In the case of an O'scope a newbie can't realistically be expected to make one.   At the same time he doesn't have the background to use an advance DSO with an endless list of bells and whistles.  More importantly a newbie can't even specify his scope needs which is probably the best reason to put off a purchase.

Nobody knows where their hobby will head.  I've been fooling around with electronics since I was a kid - probably starting around '55 or so.  Even then, it hasn't been an active hobby all those years, I had other things to do and electronics would get shoved aside for years on end.  By the time I got my EE, I hated electronics!  Grad school revived it a bit.   You just never know what the future holds.

In the OP, nothing is mentioned about budget, why are we making assumptions?  What was asked is what scope was appropriate for Arduino projects.  The short answer is still:  just about anything rated 10 MHz or more.  To my knowledge, there are no pins on the ATmega328P Arduino that will toggle anywhere near 1 MHz.  The low dollar solution is an old low MHz analog scope.

Moving up to the Teensy 4.1 and its ability to toggle pins at 100 MHz is going to take a lot more scope.  It's debatable whether any of the entry level scopes are going to handle something like that.  The closest would be the Siglent SDS1104X-E unlocked to 200 MHz bandwidth and that's going to come up short.  But it may be good enough.

So, I can see where the hobby moves from ATmega328P to iMXRT1062 with no change in the underlying 'wiring' methodology.  Decades ago, a VP at the company I worked for and I were discussing large scale computer systems and the speed-power product.  He mentioned "Don't bet against technology!"  How right he was.  The same thing will happen with our hobby.  That 100 MHz entry level Rigol will be suboptimal for the next generation of chips.  But how much 'future proofing' can a hobbyist afford?

For the ATmega328P projects, a 10 MHz CRO will be entirely adequate.  Move one step up to the mbed LPC1768 and we're clocking at 60 MHz.

Since cost wasn't specifically mentioned, I don't see why it is being discussed.  Buying the Rigol and then reselling it down the line won't result in much of a loss.  Maybe $100?  Pretty cheap rent for a couple of years of use.

Not a poll:

• How many of the hobbyist respondents to this thread own a scope?  My guess:  all of them!
  
• Of those that own a scope, how many own a DSO?  My guess:  most of them.  There will be some CRO users that have had their CRO for a long time and see no reason to upgrade.  So be it!  It gets to the point of 'marginal utility'.  Are the DSO advantages sufficient to retire the CRO?  For some, yes, for others, no.  Or, in my case, I keep the high bandwidth Tek485 but I use the modern DS1054Z.
  
• Of those who use a scope for troubleshooting uC projects, how many want to step back and debug I2C without a scope?  My guess:  none of them!  Yes, a logic analyzer will do the job but it's pretty useless in the analog domain.  A scope does both...
  
• Of those who own a scope, if they didn't have one, would they be in the market for one in the near future?  My guess:  absolutely!
  

[radiolistener]

1) What problems do you solve with an oscilloscope?

electro debugging and AC or RF measurements

2) What bandwidth do I need?

minimum required bandwidth for hobby projects with 5-10 MHz signals is 100-200 MHz

3) Sample rate?

minimum sample rate for 100-200 MHz bandwidth is 1-2 GS

4) Memory depth?

look for at least 14 megapoints and above.
More memory depth, more detailed picture you can see and more slow time scale you can use with no aliases

[james_s]

Of those who use a scope for troubleshooting uC projects, how many want to step back and debug I2C without a scope?  My guess:  none of them!  Yes, a logic analyzer will do the job but it's pretty useless in the analog domain.  A scope does both...

I use a Bus Pirate to debug I2C and other similar interface, the scope can show me the signal so it's good as a quick check for signal integrity but I've never had a scope that can decode it so beyond verifying that the signal looks ok I don't really use my scope for that. For what it's worth, I did the same thing with an analog scope before I had a DSO, it allowed me to check the levels and verify that the clock rate looked reasonable but obviously it was not gonna decode anything.

[james_s]

1) What problems do you solve with an oscilloscope?

electro debugging and AC or RF measurements

2) What bandwidth do I need?

minimum required bandwidth for hobby projects with 5-10 MHz signals is 100-200 MHz

3) Sample rate?

minimum sample rate for 100-200 MHz bandwidth is 1-2 GS

4) Memory depth?

look for at least 14 megapoints and above.
More memory depth, more detailed picture you can see and more slow time scale you can use with no aliases

Funny, somehow I was able to work with signals >30MHz back when all I had was a 100Ms/sec 100MHz DSO, guess I must be a magician or something? Apparently I can do the impossible? Even now I never use anything even close to 14mpoints, my fingers would fall off scrolling through such a long capture.

[tggzzz]

1) What problems do you solve with an oscilloscope?

electro debugging and AC or RF measurements

2) What bandwidth do I need?

minimum required bandwidth for hobby projects with 5-10 MHz signals is 100-200 MHz

3) Sample rate?

minimum sample rate for 100-200 MHz bandwidth is 1-2 GS

4) Memory depth?

look for at least 14 megapoints and above.
More memory depth, more detailed picture you can see and more slow time scale you can use with no aliases

Funny, somehow I was able to work with signals >30MHz back when all I had was a 100Ms/sec 100MHz DSO, guess I must be a magician or something? Apparently I can do the impossible? Even now I never use anything even close to 14mpoints, my fingers would fall off scrolling through such a long capture.

The first digital scope I used professionally[1], a serious HP54100a boat anchor, was a 1GHz 40MS/s box. Note where the M and G are. People that take much notice of the S/s figure are being bamboozles by salesmen; the only thing that matters is the MHz

I have a couple of Tek1502s where the sampler has ~5GHz of bandwidth, but the sampling rate is ~30kS/s, IIRC. Guess which is the important parameter, and which nobody ever cared about.

[1] it is sobering to think that we are closer to 2050 than to then.

[Mechatrommer]

Funny, somehow I was able to work with signals >30MHz back when all I had was a 100Ms/sec 100MHz DSO, guess I must be a magician or something? Apparently I can do the impossible?

no you are not yet a magician. the real magician is someone here who built a computer with just a DMM, probably the china 50 positions manual range type. iirc its mentioned here or on the other thread,his name cant be pronounced with my tongue at least (sorry cant help it, so pun intended) ps: there you go, he beat me to it

[radiolistener]

I was able to work with signals >30MHz back when all I had was a 100Ms/sec 100MHz DSO

You will be able to see "something" even if you put >250 MHz on the input of 10 MS ADC. But this "something" is not what is present in reality. You will see aliases, you will see waveform which doesn't present in real signal.

Also you will not be able to measure amplitude properly if your signal has highest frequency component higher than about 30% of your scope bandwidth.

I'm talking not about "watch something", I'm talking about real usage, e.g. watch real signal and measurement and not fight with phantoms due to limited sample rate and bandwidth.

guess I must be a magician or something? Apparently I can do the impossible?

no, you just don't understand what you're doing and what you see... 
Because when you connect 30 MHz and more square wave to 100 MS scope, you will see distorted amplitude and a lot of aliases instead of real signal waveform. Aliases may be reduced with antialias filter on the input, but it will give you even worse result, because amplitude will be distorted more dramatically. You will see waveform which doesn't exists in reality with incorrect waveform and incorrect amplitude...

This is why for 30 MHz digital signals it's better to use 300 MHz bandwidth scope with 3 GS sample rate and about 50 megapoints memory. But this is your case if you really want to see what is going on with the signal...

But if you just want to see that "there is present some signal" and don't want to know what is going on in details, then you can use 100 MS scope.

Even now I never use anything even close to 14mpoints, my fingers would fall off scrolling through such a long capture.

it just means that you don't understand how to use oscilloscope properly and how memory depth affects oscilloscope results... 

For example, when your scope has low memory depth, it will reduce sample rate at slow time scale. As result you will see phantom waveform instead of real singal waveform due to aliases which happens when you try to digitize high frequency signal with not enough sample rate.

a 1GHz 40MS/s box. Note where the M and G are. People that take much notice of the S/s figure are being bamboozles by salesmen; the only thing that matters is the MHz

Sampling oscilloscope is not suitable for digital signals like arduino. Sampling oscilloscope can be used for periodic signal, because it uses stroboscopic effect to render waveform in several signal cycles. If signal waveform changes from one cycle to another, like it happens in digital signal, you will see just garbage...

So, S/s is not critical for analog periodic signal with sampling oscilloscope. But when you're want to analyze digital signals it will be useless, because there S/s value plays significant role and you will needs to use oscilloscope working at high sample rate. And there is needs enough memory depth to avoid reduction of sample rate at slow horizontal scale (because small memory cannot fit all screen data at full sample rate for slow scale) 

[james_s]

no, you just don't understand what you're doing and what you see... 
Because when you connect 30 MHz and more square wave to 100 MS scope, you will see distorted amplitude and a lot of aliases instead of real signal waveform. Aliases may be reduced with antialias filter on the input, but it will give you even worse result, because amplitude will be distorted more dramatically. You will see waveform which doesn't exists in reality with incorrect waveform and incorrect amplitude...

I understand it quite well actually. It's called understanding the limitations of the equipment you have. If you need to capture a detailed image of a non-repeating waveform then you need more bandwidth, but this is often not the case. A square wave is repeating and a DSO will perform sampling in order to capture a reasonably accurate picture of the waveform, it's not necesssary to grab the whole thing in one shot, they have video memory and build up a picture through several captures, whether you believe it or not it does work. Sure, in an ideal world we'd all have multi-GHz scopes with multi-GS/s sample rates but reality is that most people have to get by with much less, and it's possible to do so with a bit of creativity and knowledge. It wasn't that long ago that having any oscilloscope was an extravagant luxury and a DSO was something hobbyists only dreamed of and yet people have been able to develop and debug all kinds of stuff. Now you say it's useless, but I can debug something like an arduino with ease using a 100MS/s DSO. The fact that you think it's impossible does not seem to stop me from doing it.

Obviously more bandwidth and higher sampling rates are always nice, but these elitist blanket statements of "you have to have xxx MHz and yyy samples/sec to do xyz" really get on my nerves the same was as people who act like having a new car and the latest smartphone are life's bare necessities, or the time someone flat out insisted to me that it was "impossible" to eat on a daily budget that was less than I typically spend on food without even particularly trying to be cheap. Maybe it's just professionals who do something for a living and are unable to see anything from the perspective of a hobbyist? I don't know. I mean if I were telling the boss the sore of scope I need for a project at work then obviously I'm going to specify an instrument capable of doing the job with ease but if I'm spending my own money on a tool I'm going to be a lot more willing to live with limitations or do tricks to get the instrument to show me what I need to see. For example when all I had was an analog scope and I was trying to debug a microcontroller that was writing to a string of shift registers, I wrote some test code that would cause my shift out routine to produce a repeating pattern. It would have been nice to have a fancy DSO with enough memory to capture the whole string in one go but that wasn't an option back then so I took a different approach and found a way to work around the limitations of the equipment I had and see what I needed to see, which ultimately worked fine and was not all that hard.

[tggzzz]

I was able to work with signals >30MHz back when all I had was a 100Ms/sec 100MHz DSO

You will be able to see "something" even if you put >250 MHz on the input of 10 MS ADC. But this "something" is not what is present in reality. You will see aliases, you will see waveform which doesn't present in real signal.

Here's a picture of a 50ps risetime (140ps falltime) signal on a <100kS/s scope produced in 1972. That pulse represents the length of a 78mm long transmission line.

a 1GHz 40MS/s box. Note where the M and G are. People that take much notice of the S/s figure are being bamboozles by salesmen; the only thing that matters is the MHz

Sampling oscilloscope is not suitable for digital signals like arduino. Sampling oscilloscope can be used for periodic signal, because it uses stroboscopic effect to render waveform in several signal cycles. If signal waveform changes from one cycle to another, like it happens in digital signal, you will see just garbage...

Not true; this is a classic use for scopes, both analogue and digitising - and is an example of where scopes are the only instrument suitable.



BTW the term is "subsampling", and it is important in many circumstances.

So, S/s is not critical for analog periodic signal with sampling oscilloscope. But when you're want to analyze digital signals it will be useless, because there S/s value plays significant role and you will needs to use oscilloscope working at high sample rate. And there is needs enough memory depth to avoid reduction of sample rate at slow horizontal scale (because small memory cannot fit all screen data at full sample rate for slow scale) 

Firstly you appear to be confusing analogue waveforms captured by scopes with digital signals. Converting one into the other requires interpretation by hardware (usually a 1bit ADC) and/or software.

If you want to analyse digital signals, better tools are logic analysers, protocol analysers, and printf statements.

[james_s]

Just as a quick test, I looked at a 50MHz clock signal with three different scopes, a 1GHz 4GS/s DSO, a 200MHz 100MS/s DSO, and a 100MHz CRO. Same signal, same probe, it was just a quick setup and using the ground lead on the probe and no decoupling for the oscillator so I'd expect the signal to look a bit uglier than it actually is but that's probably beneficial for this particular discussion. Unfortunately I had been experimenting with HPGL plots and forgot to set the first scope back to bitmap output but it'll do.

As expected, the GHz scope shows a bit more detail than the 200MHz scope but I would certainly consider both to be acceptable. The CRO is at its maximum sweep rate so it is unable to show the same level of detail as either of the DSOs but again that is an expected limitation of the instrument. I'm going to say that all three of these instruments are capable of providing a useful display of the waveform, all three correctly show the frequency, period and amplitude, they all show the shape of the waveform, they all allow me to see that there's some ringing, they're all suitable to one degree or another. As I've said all along, the high bandwidth high sample rate scope is better, but it also cost FAR more, even today a new GHz scope will set you back thousands of dollars. The lower bandwidth scopes are not as good, but they're certainly good enough for debugging mundane hobbyist circuits like an Arduino. They're all MILES ahead of being blind with no scope at all.

A more interesting comparison would be looking at something like I2C, 1Wire or SPI, I may do that later, for now I just wanted to do a quick example. Suffice to say over the years I have used all three of these instruments to successfully debug those things. The CRO was my only scope for about 14 years and I used it very frequently, it was not nearly as convenient as a DSO but it did the job.

[tggzzz]

That is showing a +5 -2V signal which would damage whatever it is connected to.

Now do the measurements with the the same scopes and probe, but with the ground lead removed and replaced with a 1cm ground spring. I'll bet that ~100MHz ripple disappears.

Even better, use or make a resistive divider probe.

[james_s]

That is showing a +5 -2V signal which would damage whatever it is connected to.

Now do the measurements with the the same scopes and probe, but with the ground lead removed and replaced with a 1cm ground spring. I'll bet that ~100MHz ripple disappears.

Even better, use or make a resistive divider probe.

Except it doesn't damage it, that's the output of a raw TTL oscillator with no load other than the 10X scope probe, I know the oscillator itself is fine and works in circuit. The negative excursion is likely almost entirely ringing due to the long ground clip lead and possibly also due to the complete lack of a bypass capacitor. That's all beside the point though, or rather it helps make my point.

This was not meant to be an example of proper probing technique and the importance of power supply bypassing, rather it was to illustrate that a 100MHz CRO or an old DSO with a paltry 100MS/s sample rate is capable of viewing and diagnosing problems a 50MHz signal. In this case the problems are likely mostly due to inappropriate technique but they could just as easily be real faults in a real circuit and these faults were effectively shown by all three scopes. Yes the higher bandwidth scopes show a more accurate representation of what is going on there than the lower bandwidth scopes. The rise time is visibly faster on the 1GHz scope, the lower bandwidth scope also less accurately shows the peak spike in the amplitude, but being aware of these limitations one could still use any one of these scopes to diagnose the problem.

[radiolistener]

but these elitist blanket statements of "you have to have xxx MHz and yyy samples/sec to do xyz" really get on my nerves

I just showed minimum specs for today's low-end oscilloscope... 
There is no sense to buy oscilloscope with worse specs, because you will pay more and get less.

Here's a picture of a 50ps risetime (140ps falltime) signal on a <100kS/s scope produced in 1972. That pulse represents the length of a 78mm long transmission line.

Can you use it to analyse waveform where each next pulse has different waveform with different rise time?

Does it allow you to capture signal by sync pulse and then analyze what happens with signal?

Firstly you appear to be confusing analogue waveforms captured by scopes with digital signals. Converting one into the other requires interpretation by hardware (usually a 1bit ADC) and/or software.

If you want to analyse digital signals, better tools are logic analysers, protocol analysers, and printf statements.

logic analyser with 1 GS sample rate is really useful, but it cost the same as oscilloscope and cannot replace oscilloscope, because it doesn't show you analog details at high speed.

logic anasyser is more suitable when you don't interesting in details of waveform, but more interesting in protocol debugging.

[radiolistener]

The rise time is visibly faster on the 1GHz scope, the lower bandwidth scope also less accurately shows the peak spike in the amplitude, but being aware of these limitations one could still use any one of these scopes to diagnose the problem.

Yes, oscilloscope with 1 GHz bandwidth will cost you a lot of money and not worth the money if you planning to use for arduino purposes.

But 200 MHz / 1 GS is today's low-end and there is no sense to buy worse oscilloscope, because you will lose more than you save.

And in my experience 100-200 MHz bandwidth sometimes is not enough, for example when you working with high speed controllers and FPGA. So, there is no sense to buy less than 200 MHz bandwidth.

[tggzzz]

That is showing a +5 -2V signal which would damage whatever it is connected to.

Now do the measurements with the the same scopes and probe, but with the ground lead removed and replaced with a 1cm ground spring. I'll bet that ~100MHz ripple disappears.

Even better, use or make a resistive divider probe.

Except it doesn't damage it, that's the output of a raw TTL oscillator with no load other than the 10X scope probe, I know the oscillator itself is fine and works in circuit. The negative excursion is likely almost entirely ringing due to the long ground clip lead and possibly also due to the complete lack of a bypass capacitor.

"would" in the sense of "if it was really there and not an artefact of the ground lead". Not "will" i.e. unconditionally

This was not meant to be an example of proper probing technique and the importance of power supply bypassing, rather it was to illustrate that a 100MHz CRO or an old DSO with a paltry 100MS/s sample rate is capable of viewing and diagnosing problems a 50MHz signal.

Maybe that 50MHz signal, iff you could see the signal without probing artefacts.

But not all 50MHz signals, e.g. the output of a jellybean 74lvc1g* gate toggling at 50MHz. Hint: that has noticeably subnanosecond transition times, even into a 50ohm load.

[james_s]

I just showed minimum specs for today's low-end oscilloscope... 
There is no sense to buy oscilloscope with worse specs, because you will pay more and get less.

You're moving the goalpost. You said that it was the minimum needed to work with those frequencies.

One of the most popular hobbyist oscilloscopes on the market today is limited to 100MHz, it has been popular for years and it still sells in large numbers. There can be lots of sense in buying lower bandwidth scopes, they cost less, sometimes much less. If you can afford more, get more, but the thing that makes no sense is to not buy anything at all after you find you can't afford 200MHz or more.

[james_s]

Maybe that 50MHz signal, iff you could see the signal without probing artefacts.

But not all 50MHz signals, e.g. the output of a jellybean 74lvc1g* gate toggling at 50MHz. Hint: that has noticeably subnanosecond transition times, even into a 50ohm load.

Sure, I have pulse generators I've built out of 74lvc14's, they're very fast, even the 1GHz scope struggles to display the true rise time.

But so what? That logic can be used to justify buying the highest bandwidth scope on the market, having more bandwidth is always better, having more channels is always better, having a higher available sample rate is always better, having deeper memory is always better, up to infinite values. Ultimately one has to decide what is "good enough" for their needs and within their budget because you can't have infinite of anything and unless you are wealthy you can't have more than a few hundred MHz, maybe a couple of GHz if you're like me and don't mind an old boatanchor.

[tautech]

I just showed minimum specs for today's low-end oscilloscope... 
There is no sense to buy oscilloscope with worse specs, because you will pay more and get less.

You're moving the goalpost. You said that it was the minimum needed to work with those frequencies.

One of the most popular hobbyist oscilloscopes on the market today is limited to 100MHz, it has been popular for years and it still sells in large numbers. There can be lots of sense in buying lower bandwidth scopes, they cost less, sometimes much less. If you can afford more, get more, but the thing that makes no sense is to not buy anything at all after you find you can't afford 200MHz or more.

200 MHz DSO's are not unaffordable and in fact SDS1202X-E are my best seller.

[james_s]

200 MHz DSO's are not unaffordable and in fact SDS1202X-E are my best seller.

That depends on your budget, and it's only 2 channels. For many people a 100MHz 4 channel scope is going to have advantages over a 200MHz 2 channel scope. Which is better depends on your needs.

[tautech]

200 MHz DSO's are not unaffordable and in fact SDS1202X-E are my best seller.

That depends on your budget, and it's only 2 channels. For many people a 100MHz 4 channel scope is going to have advantages over a 200MHz 2 channel scope. Which is better depends on your needs.

Yes of course and the reverse also applies. Many don't need a 4ch DSO but want the BW.....which in the old days was always king. 

[tggzzz]

But so what?

The answer to that question is in the context, which you snipped.

(Unhelpful, isn't it!)

[james_s]

But so what?

The answer to that question is in the context, which you snipped.

(Unhelpful, isn't it!)

Indeed. Point taken.

[hamster_nz]

If I wanted that one tool for debugging Arduino / microcontroller designs, it would be a Saleae Logic-8, with the 50% discount for "Enthusiast Pricing"

https://blog.saleae.com/saleae-discounts/

8 channels of digital at 100MS/s, about 1MHz of analog bandwidth on up to 8 channels, 10 bit samples, lots of protocol decoding.

But don't take my word for it...