Linux compatibility - Another newbie looking for an oscilloscope

[tatel]

Hi all.

First post on this forum. As a presentation, I have some 20 years of experience as linux sysadmin. I do have very little knowledge or experience in electronics, but I'm loving to play with arduinos, or, more exactly, with atmel microprocessors. Sensors, motor control, PWM, 1-wire, I2C and SPI communications, perhaps even CAN since that's what my car speaks. I want to be able to decode serial communications as easily as possible, see what my PWMs are doing, and what is going on on my proto boards. That kind of things.

So I think it's time to get an oscilloscope, better than a logic analyzer. Not for professional use, just for fun. I don't expect to get any further than that into electronics. By the way, for logic analysis, I'm looking to BusPirate.

I did some research looking for a cheap oscilloscope, cause it's going to be mainly a  toy. At first, I looked at those dirty cheap USB, PC based things. But it seemed as if those weren't good enough. Then I looked at some things in the 200 € range, cause this is the amount I was comfortable with, and they looked as if they could be good enough. But, it wasn't clear that those things could work with linux. No linux, no fun. Now I realize that, to be sure I'm getting this, quite probably I'll have to get into the 400 € range. I don't want to spend more than that on a just-for-fun thing.

Yesterday I spent all my free time looking some Dave's videos and reading some threads here. After that, I'm considering both the... Rigol DS1054Z and Siglent SDS1202X-E. Don't get me wrong, I hope this will not start any religion war! I have read a thread here that became somewhat hot.

I can get both from Amazon Spain with a difference of just some 25 €, and this looks good, since the price from Aliexpress is about the same after customs. So I'm leaning to Rigol, cause it has 4 channels, can be hacked to 100 MHz and it looks as being supported by linux software like sigrok and dsreport. Good.  But the Siglent looks good too, for it can do 200 MHz and has an ash shell inside.

Now, I'm unable to find any more infos about those two scopes and linux, so it's time to ask for some help from experienced people.  So my questions are:

1- Which one works better with linux?

2- Can I do things as, say, send the data directly to my PC for storage? That could bypass memory limitations on the scope.

3- Can I load and execute shell scripts on these scopes?

4- Can I decode serial communications, not supported by the scopes, on my linux system, using data from the scope?

5- I do realize that four channels are better than two, but the siglent looks as being able to decode most of the mentioned serial protocols with just two channels. Should I buy the siglent, will I in the future have to repent about not getting four channels? Which would be the case for Atmel-related work? And for electronics in a more general way?

6- I do realize that 200 MHz BW is enough to work with Atmel micros working at say, 16-20 MHz, while it does look that doing it with an 100 Mhz hacked scope could be perhaps not good enough, because one has to get the seventh harmonic or more?  Excuse my ignorance, but it has somewhat shocked me. I was expecting 100 MHz should be more than enough. What is this 7th harmonic thing about? Does it matter as much that working at 100 MHz could be annoying?

7- Do you know of any other €400 -or less- scope that could fit my bill?

Any answers/hints/pointers would be greatly appreciated.

Best wishes

[PKTKS]

AFAIK  if you raise your budget above the 200 you will not
need any sort of PC applet...

The only one really working almost perfect (even better than OEM)
are the ones supported by sigrok https://sigrok.org/wiki/Supported_hardware

and among them Hantek USB based with an excellent OpenHantek scope app.

The problem with these new gen gizmos is that crappy USB stack.

On ETH based they can be managed order of magnitudes better..
but at that budget you will not find any..

For breadboard ARDUINO?  go Hantek.  cheapo affordable works ..

For anything above serious go RIGOL (USB is dangerous stuff in scopes)

Paul

Alas.. check MUSTOOL MDS120 hand scope..  cheap works fine... portable..

[nctnico]

Look at the GW Instek GDS-1054B. This one should fit the budget and it can dump data straight onto a Samba share (on a capture per capture basis; not streaming). People are working to figure out how to get the Lua scripting going on this scope but no word on progress yet.

100MHz is enough for working with standard microcontrollers. Seeing the 3rd harmonic is good enough to get an idea of a digital signal. In reality most signals surrounding a microcontroller have a much lower frequency.

[tautech]

tatel
Siglent equipment and their SW doesn't work natively with Linux although the instruments OS are Linux based.

Instead we need to use other tools to interface with Siglent gear over LAN and LXI Tools is a proven solution.
Some resources for you here:
https://siglentna.com/video/quick-remote-computer-control-using-lxi-tools/
https://www.eevblog.com/forum/testgear/open-source-lxi-tools-and-liblxi-v1-0-released-for-gnulinux/
https://www.eevblog.com/forum/beginners/need-help-connecting-siglent-1202x-e-scope-to-lxi-tools/

[tatel]

Thank you all for the help.

I'll probably go rigol since having 200 MHz seems not as useful as four channels for my use case.

[nigelwright7557]

Some scope vendors have optimistic bandwidths.
A 100MHz scope will only be good for 10MHz sine wave.
You need at least 10 points to get anywhere near an accurate sinewave.
Extrapolating from 2 points doesn't distinguish between sine, square and ramp waveforms, it just does sine usually.

I still prefer my analogue scope for most of my analogue work and USB scope for capturing digital data packets.

[nctnico]

Some scope vendors have optimistic bandwidths.
A 100MHz scope will only be good for 10MHz sine wave.
You need at least 10 points to get anywhere near an accurate sinewave.

That has been debunked a long time ago. Read up on signal processing theory. Modern DSOs have enough information from a samplerate which is 2.5 times the bandwidth to show you a sine wave. The problem isn't in the DSO but our brains which can't connect the dots. Throwing a higher samplerate at a sine wave does not give you any extra information.

[tatel]

You need at least 10 points to get anywhere near an accurate sinewave.

Yeah, got it from Dave's videos. Other times he speaks about having at least four points, I think. But, excuse my ignorance: you are speaking about bandwidth, but this is about samplerate instead, it isn't?

The problem isn't in the DSO but our brains which can't connect the dots

Well if I pay for a tool, I want it good enough to help my brain. And, I'm not sure that the matter is to see pure sine waves. So, to draw the actual signal as accurately as possible, probably the more points it can get, the better, I think.

Now, as I understand it, if working with just one channel, so 1 Gsps sample rate, I should be able to get 10 points on a 100 MHz wave, while working with two, that ideal situation remains only until 50 MHz, and working with 3-4 channel it will go just to 25 MHz

If working with arduinos, so 16 MHz max speed, and need to see that curious 3rd harmonic, then I need to be able to do 48 MHz bandwidth. Which I could easily do, while maintaining the said 10 points/wave, if working  with just one channel. I could also do it working with 2 channels as long as that 3rd harmonic remains below 50 MHz, as would be the case if working with Arduino's 16 MHz max speed. And quite good enough still if working with full-blown 20 MHz Atmel speed. But not so much if working with 3-4 channels. Is that right?

Mi reasoning is that, probably, I would need more than 2 channels while working on, maybe, SPI serial communications, but these, as far as I know, aren't so fast, and are digital, so even 250 Ksps could be good enough. And when working on other, analog, faster things, perhaps 2 channels is plenty, as people buying 2-channel scopes seems to believe. Am I wrong?
 
On the other hand, I'm still unable to clearly see how Siglent could do a fine work debugging SPI, which is more than MISO and MOSI, with just 2 channels. So i think they are perhaps stretching it as much as they can. Even worse, as per Dave's videos, Siglent's decoding isn't realtime. If it's not realtime,  I could as well go to my PC and put sigrok at work. Now, Siglent isn't in sigroks hardware list, but Rigol is.  Am I getting it wrong?

So I'm leaning to rigol, but I'm not in a hurry and will take my time to make this decission. Of course, to buy a Siglent DSD1204X-E would make things easy, and if I were to get into electronics as a profession, I would buy at least that. But I'm a hobbyst on a budget, and Rigol DS1054Z seems to have still more bang for the buck in my use case.

Still, many thanks to you all, and if I'm wrong in my reasoning, please let me know

[nctnico]

You need at least 10 points to get anywhere near an accurate sinewave.

Yeah, got it from Dave's videos. Other times he speaks about having at least four points, I think. But, excuse my ignorance: you are speaking about bandwidth, but this is about samplerate instead, it isn't?

As a rule of thumb: for a modern DSO the samplerate needs to be at least 2.5 times the bandwidth.  This is enough to reconstruct the signal. So 2.5 points per period for a sine wave with a frequency equal to the maximum bandwidth. Remember that according to Nyquist a sine wave can be sampled correctly up to (but not including) half of the samplerate. However a DSO needs some room in the frequency spectrum for the anti-aliasing filter and the length of the sine wave is finite; this is where the factor 2.5 comes from.

Also: take a good look at the GDS-1054B from GW Instek. It has many advantages versus the Rigol and Siglent (like individual channel controls). Especially if you are into decoding; unlike the RIgol and Siglent the GW Instek decodes the entire memory and not just what is on screen. For digital electronics having 4 channels is a must. Don't go for 2 channels.

[tautech]

On the other hand, I'm still unable to clearly see how Siglent could do a fine work debugging SPI, which is more than MISO and MOSI, with just 2 channels. So i think they are perhaps stretching it as much as they can. Even worse, as per Dave's videos, Siglent's decoding isn't realtime. If it's not realtime,  I could as well go to my PC and put sigrok at work. Now, Siglent isn't in sigroks hardware list, but Rigol is.  Am I getting it wrong?

With a setting called Clock Timeout.
With any 2ch DSO SPI decoding requires another way to go about it and decoding just one packet stream is the norm.

The 2ch SDS1202X-E also offers individual vertical controls unlike its 4ch brother SDS1104X-E however for little more than the 2ch X-E is the SDS1104X-U for $ 399.

[RoGeorge]

1- Which one works better with linux?

2- Can I do things as, say, send the data directly to my PC for storage? That could bypass memory limitations on the scope.

3- Can I load and execute shell scripts on these scopes?

4- Can I decode serial communications, not supported by the scopes, on my linux system, using data from the scope?

5- I do realize that four channels are better than two, but the siglent looks as being able to decode most of the mentioned serial protocols with just two channels. Should I buy the siglent, will I in the future have to repent about not getting four channels? Which would be the case for Atmel-related work? And for electronics in a more general way?

6- I do realize that 200 MHz BW is enough to work with Atmel micros working at say, 16-20 MHz, while it does look that doing it with an 100 Mhz hacked scope could be perhaps not good enough, because one has to get the seventh harmonic or more?  Excuse my ignorance, but it has somewhat shocked me. I was expecting 100 MHz should be more than enough. What is this 7th harmonic thing about? Does it matter as much that working at 100 MHz could be annoying?

7- Do you know of any other €400 -or less- scope that could fit my bill?

1.  Both.  Any instrument that has LXI can be controlled over LAN or over USB, the OS doesn't matter.  All you need is to send text SCPI commands.  There is already plenty of FOSS tools that can do that no matter the instrument model.  You can even telnet to the instruments if you want to.

2.  Yes if you want slow but indefinite data grabbing, for example one measurement per second.  No for real time acquisition, low range oscilloscopes can not be used for fast and continuous acquisition, they are not DAQs.  It is possible to dump to a PC the raw data that was acquired by the oscilloscope's ADC at full speed, but a transfer will take many seconds, so it will not be a continuous process, but you can save batches.

3.  Yes, but actually no.  That will be a hack, and the oscilloscope is busy anyway doing its tasks, it doesn't have resources to execute your scripts.  You won't be able to turn it into a 1Gsps DAQ if that's what you want.  However, LXI models can be controlled remotely, over the LAN or over the USB, so put your automation or data post- processing scripts in a PC that controls the oscilloscope remotely (using SCPI commands).  Once I made a thermostated soldering iron out of an oscilloscope, a power supply and a PC.  The oscilloscope was continuously polled to measure the thermocouple voltage, then the PC was controlling the power source that was heating the soldering iron.   
https://rogeorge.wordpress.com/2016/09/23/zero-parts-thermostated-soldering-station/

4.  Yes, with sigrok.  Most of the existing protocols are already decoded by sigrok, but can add your own custom decodings, too.  It's FOSS.

5.  You won't miss the extra 100MHz for Atmel work, but you will need 4 channel for protocols with most than 2 wires.  Buy a 4 channels Siglent.  Either Siglent or Rigol, buy the lowest model in its line and unlock it.  For the same line of models, the hardware is identical.  All the limitations are in software, and it is already known how to unlock that.  I have a 4 channels Rigol, and I would never want to go back to 2 channels.  100 or 200MHz is about the same range when it comes to frequency.  From 100 to 500MHz or to 1GHz, that would be hard to decide, but between 100 and 200 it almost doesn't matter.

6.  Bollocks, if 100MHz is not enough for your Arduino, than 200MHz wouldn't be enough either.  You don't care about the 7th harmonic, and harmonics doesn't stop to the 7th anyway, they go upto infinite.  FYI a square waveform (or any other shape as well, not only square) can be built by adding many synchronous sinusoidal signals.  Something like this.

Figure 2.1 of this books also has a more intuitive illustration, IMO less bamboozling than an animation.
https://www.analog.com/media/en/training-seminars/design-handbooks/Software-Defined-Radio-for-Engineers-2018/SDR4Engineers_CH02.pdf#page=14
(note that Gibbs phenomenon is something else, but the figure is the most intuitive I know for showing how the square wave in the sum of all those sinusoidal signals in the back)

Any shape can be constructed by adding many sinusoidal signals.  Some dudes were having fun drawing Bart out of sinusoidal oscillations.  If you want to learn more about this subject, search about Fourier.  Short story, you don't really care about higher order harmonics when working with an AVR.  100MHz is more than enough.

7.  I think both Rigol and Siglent have a 4 channel model in that price range.  I already have a 4 channels Rigol DS1054z unlocked like it would be a DS1104z, and it served me well for many years now.

If I would need to buy again today, I would buy the Siglent 4 channel one, because it's a newer model, with slightly bigger screen and some extra measuring modes (IIRC Siglent can do bode diagrams for example).  I won't go for anything less than 4 channels, and certainly I would not trade the 4 channels for those extra 100MHz bandwidth, no way!

[tatel]

So 2.5 points per period for a sine wave with a frequency equal to the maximum bandwidth

OK. So a 100 MHz , 4 channel oscilloscope that can do 250 Ksps on each channel, may be isn't the better thing in the universe, but is not "bad" in any case, and when using less bandwidth and/or less channels, gets better. Got it.

take a good look at the GDS-1054B from GW Instek. It has many advantages versus the Rigol and Siglent (like individual channel controls). Especially if you are into decoding; unlike the RIgol and Siglent the GW Instek decodes the entire memory and not just what is on screen.

Did it. After reading some scary thread about it getting locked, I did see Dave's teardown video. Well it could fit the bill. I would like it more if having some more metal, but I guess they know what they do. I don't know if that scary movie about becoming sluggish holds any water but I can't find any more about that. I guess other users would have something to say but the few I have found seem to be quite happy. Having a lightning fast UI and lots of wfps would be a plus. Is it linux-based?

Maybe I'm wrong, but I think Dave speaks in a video about Siglent decoding working with all memory data?

The flop could be hacking it to 100 MHz. Would I need other probes? The Rigol guys seem to know what they do, giving 150 MHz-rated probes  in a supposedly "50 MHz scope". So no need to buy new probes after "upgrading".

however for little more than the 2ch X-E is the SDS1104X-U for $ 399

Yeah, I find it really interesting. More after realizing than 2.5 points/wave is good enough. So It has the same basic specs than the Rigol, but no hacks needed, 4 channels for decoding and it does, among others, CAN. Probably no need to fall back to sigrok. It could have some use in my garage. 256 level screen vs 64. Linux based. Even if I can't touch the privative parts (which makes me grumpy), still I could perhaps put my knowledge at work for good. At about the same price, not some 25% more. Actually, I'm now leaning to this one. Tanks a lot to you, tautech.

@RoGeorge: Man, you did make my day. Answering my questions one by one, you have really helped me more than I hoped this forum could do. I'm giving to you even my feet's thumbs up

All your post is really helpful to me, but I specially appreciated how you begin the part about harmonics. Sometimes a clear word is what is really needed. Since I read that thing about the 7th harmonic, I was shocked and wondering about it. It reminded me about the seven circles of hell or the medieval monks having conciliums to discuss about how many dicks Satan has.

Luckily ntcnico rapidly said no need to go further than 3rd harmonic at maximum. Now you finish riveting it. Thanks you all, men. Really appreciated

[nctnico]

So 2.5 points per period for a sine wave with a frequency equal to the maximum bandwidth

OK. So a 100 MHz , 4 channel oscilloscope that can do 250 Ksps on each channel, may be isn't the better thing in the universe, but is not "bad" in any case, and when using less bandwidth and/or less channels, gets better. Got it.

take a good look at the GDS-1054B from GW Instek. It has many advantages versus the Rigol and Siglent (like individual channel controls). Especially if you are into decoding; unlike the RIgol and Siglent the GW Instek decodes the entire memory and not just what is on screen.

Did it. After reading some scary thread about it getting locked, I did see Dave's teardown video. Well it could fit the bill. I would like it more if having some more metal, but I guess they know what they do. I don't know if that scary movie about becoming sluggish holds any water but I can't find any more about that. I guess other users would have something to say but the few I have found seem to be quite happy. Having a lightning fast UI and lots of wfps would be a plus. Is it linux-based?

The flop could be hacking it to 100 MHz. Would I need other probes? The Rigol guys seem to know what they do, giving 150 MHz-rated probes  in a supposedly "50 MHz scope". So no need to buy new probes after "upgrading".

According to the specs it comes with 70MHz probes but don't worry about passive probes too much; these will do just fine for your purpose. Passive probes are a dime a dozen anyway (look on Aliexpress). Even very good ones from Testec are in the 30 to 40 euro price range.

Slowing down is an issue which occurs on rare occasions when switching a lot between functions like FFT and bus decoding but it is not something that gets in the way. I'm using the bigger brother (GDS-2000E series which is based on the same platform) regulary and in practise I have not seen the 'slowdown issue' happen for a long time.

And yes, the GDS-1054B runs on Linux. In the hacking thread you can find a application package which adds root access over ssh.

[tautech]

however for little more than the 2ch X-E is the SDS1104X-U for $ 399

Yeah, I find it really interesting. More after realizing than 2.5 points/wave is good enough. So It has the same basic specs than the Rigol, but no hacks needed, 4 channels for decoding and it does, among others, CAN. Probably no need to fall back to sigrok. It could have some use in my garage. 256 level screen vs 64. Linux based. Even if I can't touch the privative parts (which makes me grumpy), still I could perhaps put my knowledge at work for good. At about the same price, not some 25% more. Actually, I'm now leaning to this one. Tanks a lot to you, tautech.

Hunt out the reasonably recent vid that Defpom has done on the X-U and as he already has a SDS1104X-E he's gained a bit of experience on how to drive it yet in the X-U video comments I've added a few tips to how he can better use the feature set.

[tatel]

Well after quite a lot of reading (even the entire zoom out quirk thread) I think it could be worth to stretch my wallet a little more and go for a Siglent SDS1104X-E

I think I will get plenty of a scope for my purposes and does look to me as the best bang for the buck; able to do 200 MHz and having options for function generator and logic analyzer. Not that I'm going to get that outright expensive LA right now. I'm still thinking about a cheap USB LA but having the option on the scope looks to me as a kind of insurance should I need it in the future. I'm aware that need could never arise.

About the zoom out/history quirks: I guess I'll have to live with it until Siglent releases some FW update. Not that I'll be holding my breath. I realize they said it was on the works some 9 months ago, yet no FW update. But to have those specs in any other brand, it takes too much money for my wallet.

Should GW Instek have a similar offer, no doubt I would go that way.

I did also look at Rigol DS 1074Z+, but for the money, I'm going for 200MHz and newer, powerful hardware.

Anyway, thank you all for your help. It's really appreciated.

[tatel]

Now I'm here just to apologize 'cause I think I've been ridiculously arrogant on my previous post.

After reading some more, about-80-pages-long threads on this forum, I'm just realizing that not all shining things are gold. If you wonder about what I'm saying, please see:

https://www.eevblog.com/forum/testgear/siglent-sds1204x-e-released-for-domestic-markets-in-china/msg3527698/#msg3527698

the reply from @jan28

https://www.eevblog.com/forum/testgear/siglent-sds1204x-e-released-for-domestic-markets-in-china/msg3528422/#msg3528422

and follow up.

BTW, GDS1054B can be bought on tme.eu, €350+VAT, which I find much more reasonable than others are asking for that scope. Those guys must be crazy. 

@nctnico: I have read the open source scope thread. I think the people in that thread is doing it wonderfully. Liked what you said there. Please keep the good work on. I'm not planning to post there since the level is intimidating, and of course I couldn't help at all. But I'm just hoping  you all don't get mad with grazillions of MHz, Gsps and Wfmps. Should an open source scope, 4-channel, 200 Mhz, 1 Gsps/channel, 30.000 wfmps/64 levels  and the digital capabilities of a Saleae 16 ch clone be a thing, I would buy it on the spot, and would gladly pay about the same Siglent is asking for the SDS1014X-E + SLA1016, and quite probably even a little bit more. A wet dream, I know...

[rstofer]

Now I'm here just to apologize 'cause I think I've been ridiculously arrogant on my previous post.

After reading some more, about-80-pages-long threads on this forum, I'm just realizing that not all shining things are gold. If you wonder about what I'm saying, please see:

This topic runs just about weekly!

The hot setup a few years back was the 50 MHz Rigol DS1054Z unlocked for 100 MHz.  I have that setup.

Today the hot setup is the Siglent SDS1104X-E unlocked for 200 MHz.

Advanced Search for 'woman' and User CharlotteSwiss to get:

https://www.eevblog.com/forum/beginners/what-an-oscilloscope-recommended-for-a-woman-passionate-about-electronics/msg3100087/#msg3100087

This is an epic thread and even though she ultimately buys the 200 MHz 2 Channel variant (around Reply 150), read on.  We play with some very interesting tools as she worked through every single page of the User Manual.  This is a very long thread but it covers a lot of ground.

I see the SDS1104X-E on Amazon for $499 and the DS1054Z for $349.  Yes, the extra 100 MHz is worth the extra $150.

Only being able to display the 3rd harmonic implies that the rising and falling edges are a bit undetermined and this is often the exact thing you are looking for.  Setup and hold times relative to some clock.  I've always thought that the 9th harmonic was about the minimum so my 100 MHz scope can reasonably display a 10 MHz square wave.  My 350 MHz Tek 485 (an old analog scope) can reasonably display about a 38 MHz square wave.  If I care about the edges...

We can debate that stuff all day long and then somebody is going to point out that the only important number for the scope is rise time.  But risetime is tied to bandwidth by the constant 0.35

BandWidth=0.35/RiseTime

For a 100 MHz scope, the rise time will be 0.35/BandWidth = 3.5ns

It's pretty magic how the specs all come up with this constant but there is some math behind it.

[2N3055]

You need Digilent Analog Discovery USB scope and run it on Linux box.. It will do all you need. It also has API so write away.
And, being a computer person, you won't mind no buttons.

[nctnico]

We can debate that stuff all day long and then somebody is going to point out that the only important number for the scope is rise time.  But risetime is tied to bandwidth by the constant 0.35

Buzzz.... wrong answer! It isn't! The only way to measure bandwidth is by using a leveled sine wave (RF) generator. Rise time is set by the anti-aliasing filter and depending on the steepness of that filter the famous '0.35' can be anything else.

[TheUnnamedNewbie]

I'll try to directly answer your questions where possible.

1- Which one works better with linux?

I can't really answer that. I only tend to use SCPI/VISA interfacing with equipment, and that is OS-independent. If that is what you are looking to do, it doens't really matter which scope you get. Some these days have HTTP-virtual-front-panels, you just connect to port 80 on their IP address and get a mirror of their screen, sometimes with buttons.

2- Can I do things as, say, send the data directly to my PC for storage? That could bypass memory limitations on the scope.

You can't. Perhaps not technically on these lower end scopes, but the simple reason is that the bandwidth and latency on scope memory is very high/low respectively.

3- Can I load and execute shell scripts on these scopes?

4- Can I decode serial communications, not supported by the scopes, on my linux system, using data from the scope?

Yes! Look at things like https://github.com/azonenberg/scopehal and related projects from Andrew Zonenberg.

[tatel]

Well for me this is scope purchase is a one-off. This is my first scope, but it should be also the last one. I want to have my arduino-based things as cornered as -within reason- possible. I'm not interested in going further than that; even when playing with mightier hardware, I would be just looking at communications with DIY, 20-ish MHz boards. Period.

So I'll have a Saleae 16 ch clone no matter what, and could upgrade that in the future if better, cheap enough hardware is released.

Now for the scope, 200 MHz, 1 Gsps dual ADC, 4-analogic channels + working MSO option would fit that bill almost perfectly. But right now this isn't a thing unless I go to next level stuff. I could afford it, but really I do have better things to spend €1500-2000 in.

So problem is, I need to make a decision between 100 MHz, working MSO option, and 200 MHz, barely-working MSO.

Of course a experienced guy can work fine on arduino things with just a 100 MHz scope or even less. But me being a noob, and not having right now a clue about what the scope shows, being it a bug in my board or some artifact related to not having "enough" BW/sample rate, I'm thinking it could be worth to pay some 100-150 more to get this unknown variable filtered out and make my life easier. I do value my time, not just the money. And it would be into the golden standard for a scope to work with 16-20 MHz hardware, which is nice if it has to be a one-off buy

So  I want to have 200 MHz for every day work, but would really hate to spend €288 in that barely-working SLA1016 thing should I need it. I'm not holding my breath thinking Siglent people will get it working fine anytime soon.

On the other hand, to think about working every day at 100 MHz, perhaps making  just to be sure I will have a working MSO option should I need it "someday" is quite disappointing.

Could you guys give advice about this based on your experience?

 

[rstofer]

Could you guys give advice about this based on your experience?

That's what we have been doing.  You have reached the stage of 'analysis paralysis'.  If you want all the high end features, like the MSO thing, you're going to need to spend money.  A lot more money than just a simple entry level scope like the DS1054Z or SDS1104X-E.

For the difference in cost ($150), I would certainly go for the Siglent and unlock 200 MHz.  That is more than enough for Arduino projects and likely anything else I will do, even with my 100 MHz FPGA projects.

Back in the day when all I had was a 10 MHz Heathkit scope, I learned to simply slow things down.  There are all kinds of ways to troubleshoot digital systems.  Even today, I create a pushbutton single clock pulse component for my FPGA projects.  Single stepping isn't a good use of time but sometimes it is the only way.  Better choice for a CPU project:  Code in a breakpoint address arrangement that will stop the high speed CPU clock and revert to single stepping when some instruction address is reached.  Set the address on the dip switches or use an IO expander to fetch it from some offboard collection of switches.  It might be a million clocks in before something happens that is interesting.  I certainly don't want to push a button that many times!

You might research Dave's video on the MSO options.  He didn't recommend spending the money.  A logic analyzer does a better job.  Yes, you lose alignment with analog signals but Dave didn't consider that use as worth the cost.

BTW, the actual unlocked DS1054Z is closer to 130 MHz established by actually measuring the -3dB point.

At some point, these entry level scopes are going to come up short.  For the Teensy 4.1, the CPU is running at 600 MHz but SPI is limited to 30 MHz for master and slave modes.  Still, that's pretty fast for a 100 MHz scope.  Run it slower or buy a 200 MHz scope (minimum).

https://www.pjrc.com/store/teensy41.html
https://www.pjrc.com/teensy/IMXRT1060CEC_rev0_1.pdf

Apparently, the SPI clock on the Raspberry Pi 4 can run as high as 250 MHz.  I'm not sure the pins can wiggle that fast but 4 MHz seems achievable.  I haven't tried this and I haven't found a concrete answer - particularly for slave mode.  It may not even do slave mode.

The maximum SPI clock on the NXP LPC1768 (a 96 MHz ARM) is 12.5 MHz.  I am using it as a slave at a clock rate near this when driven by an FPGA.

Were it me, buying a scope today as opposed to a few years back, I would be all over the Siglent SDS1104X-E unlocked to 200 MHz.

You can chase bandwidth forever but at some point you need to drive a stake in the ground.

[Fungus]

On the other hand, to think about working every day at 100 MHz, perhaps making  just to be sure I will have a working MSO option should I need it "someday" is quite disappointing.

Could you guys give advice about this based on your experience?

It's an Arduino. 100MHz is plenty.

The MHz race is exponential. The step from 50Mhz to 100Mhz is twice as big as the step from 100Mhz to 200Mhz.

Real life electronics isn't some ideal, mathematical thing. As you go over 100MHz your limiting factor will start to become the probes and your probing techniques, not the 'scopes bandwidth. To see accurate 200Mhz waves on screen you need to be doing a lot more than just poking a probe at a PCB or clipping it onto a breadboard wire which is 10cm away from an Arduino.

Signals don't simply vanish once you go above the bandwidth limit, they attenuate. A 100MHz Rigol DS1054Z will show a 300Mhz signal on screen, it's just that the displayed peak-to-peak voltage won't be correct.

Short version: 200Mhz seems a lot more valuable on paper than it is in real life.

[nctnico]

You might research Dave's video on the MSO options.  He didn't recommend spending the money.  A logic analyzer does a better job.  Yes, you lose alignment with analog signals but Dave didn't consider that use as worth the cost.

You have to look at what usage that opinion is based on. When is the last time you have seen Dave work on a microcontroller or FPGA/CPLD project? From my own experience: If you work on microcontroller and/or FPGA/CPLD projects, having an MSO is almost is must. Being able to show more than 4 signals on a screen is really handy.

[rstofer]

You might research Dave's video on the MSO options.  He didn't recommend spending the money.  A logic analyzer does a better job.  Yes, you lose alignment with analog signals but Dave didn't consider that use as worth the cost.

You have to look at what usage that opinion is based on. When is the last time you have seen Dave work on a microcontroller or FPGA/CPLD project? From my own experience: If you work on microcontroller and/or FPGA/CPLD projects, having an MSO is almost is must. Being able to show more than 4 signals on a screen is really handy.

So is a logic analyzer with a state clock input.  For Arduino and Pi projects, a simple Saleae clone will work just fine.  I built up a 32 channel 200 MHz LA from an FPGA using the project at

https://www.sump.org/projects/analyzer/

At some point (16? 32?), displaying channels on a tiny scope screen is nonsense.  There's nothing like a 27" monitor for working with a LA.  Maybe one of the 34"...

ETA:  I have never tried the Web interface to a scope.  That may very well overcome the screen size concern.

At reasonable speeds, the Digilent Analog Discovery 2 blows the doors off of all of these tools.  Particularly when you're trying to do a Bode' Plot on a scope.  Can you get log-log output?  Labeled?

I have attached a more reasonable Bode' Plot