DSO life expectancy

[bd139]

If course it is. However if I was going to do that, I'd probably buy an Analog Discovery 2. That has 14-bit 100MS/s which is better price/performance ratio than standalone digital scopes of any class:

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

[Sylvi]

The AD2 was referred to previously: it is a toy with no proper connectivity. Electrically it has merit, but only 30MHz bandwidth and 14-bit. Were I going to get a USB scope and deal with its latency issues I would get the Picoscope 16-bit unit, which also has limited bandwidth and is a lot more expensive but is aimed more at audio work.

Alternatively, there is Dr.Jordan-MLS and similar that use 24-bit sound cards, but then you have uncalibrated inputs and outputs to deal with and the silly 3mm jacks to adapt to.

The stand-alone DSO as it is today seems to have a very quick probe-to-display time, which used to be much worse and was one of the things that kept me from going with a DSO when my first CRO died and I replaced it with an inferior CRO still in production last year. That one died at 18-months of age and will be fixed once the OWON arrives and I get it working for me.

Whether you have or get a USB scope, a DSO, or a used scope of any kind, it will be a novelty and allow you to explore and do new things. So of course it is great! If you get serious about whatever you are doing, then you will move towards the "real deal" for whatever that niche demands. For example, the digital channels on all these USB scopes and DSOs seem to only be able to handle TTL voltages at best, and some are only 3V or so. That is actually fine if all you look at is recent production digital. I'm old school and use CMOS at 9-15V in hard-ware-only circuits when I need something "digital". Even though the chips are capable of high speed, my applications are distinctly ultra-low speed inasmuch as they are logic for channel selection, on/off muting and similar things in audio equipment. So, the digital inputs on modern scopes are useless for what I do.

I'm not an RF guy, nor do I have a complete handle on all the DSO specs, but I believe that for those very high bandwidth apps, like GHz+ scopes, the low-bit converters are the only ones fast enough to do a capture AND the sample rate must be extremely high AND the memory depth must be higher than for a mere 1-200Mhz scope. All of that costs money. Lucky for me I don't need such a thing since I don't have that kind of budget.

[Fungus]

I think some of the comments about CRO vs DSO are naive or unsubstantiated.

My CRO was 50MHz, so a lot of high-frequency stuff is visible even if the very highest of it was being attenuated by the scope's roll-off. Others here had and have CROs with much higher bandwidth into 100s of MHz and more

Sure, but at "normal" brightness levels they don't show the high frequency noise in the same way that a DSO does. It's just a faint glow around the trace due to the high horizontal scan rate and your eye ignores it.

To see the noise properly on a CRO you have to turn the brightness way up, but nobody does that because it looks wrong.

[Sylvi]

Um... you can't say no one turns the brightness up on a CRO to see the noise because that is what they do to see the noise.

Maybe it's been a while since you've used an analogue scope? Analogue is my reference and I have not yet used a digital scope, so I have a lot to learn yet.

I will be getting my CRO fixed after I have the DSO running as I can't imagine never needing the CRO for audio. DSOs are getting better at mimicking CRO appearance, with DPO techniques and increasing bit-resolution, but what does that tell you? It says that there is still something missing in the digital offerings that the analogue offerings can still deliver. Yes, absolutely there are things you can do with a DSO that are extremely difficult or even impossible with a CRO, so I see that there is room for both on my test bench.

[David Hess]

The stand-alone DSO as it is today seems to have a very quick probe-to-display time, which used to be much worse and was one of the things that kept me from going with a DSO when my first CRO died and I replaced it with an inferior CRO still in production last year.

The newest oscilloscopes which I really like are the almost 40 year old 7000 mainframe and 22xx portable oscilloscopes from Tektronix simply because they have the absolutely fastest probe to display time under all conditions.  Even the later Tektronix oscilloscopes are slower.

Give me peak-to-peak automatic triggering, or give me death!  - Patrick Henry

[james_s]

I was an analog scope holdout for a long time, citing most of the reasons that have been discussed here. After a couple years with a good DSO though I'd have a hard time going back. Using one requires a slightly different technique, it's not the same instrument, but once you get used to it they're really powerful. You can do things easily with a DSO that would be very challenging to do with analog. I like the "real" feel of analog, but it sure is nice to be able to capture a single shot several screen widths long, then zoom in and take measurements of various parts of the waveform at my leisure, or take screen captures at any point to review later.

[precaud]

I don't agree. I have a 10 bit scope on my bench and the FFT is much more useful compared to an 8 bit oscilloscope.

I totally agree, 10 bits is much more useful for FFT than 8, especially for sinewave test signals. In a modern 10-bit DSO, an FFT'd data record of multiple sine cycles gives the similar SNR improvement as time domain averaging of multiple records combined with numeric accuracy improvement of freq domain averaging. With solid triggering and an ADC with low DNL, averaging as few as four of these records will easily give 80dB of useful, repeatable dynamic range.

Early DSOs and FFT analyzers didn't give this benefit due to data processing limitations (slow cpus, no math co-processors, small memory maps).
1. They used pre-weighted integer math in their averaging routines. Ex: for 8 averages, each integer ADC sample was right-shifted 3 bits and then added to an integer accumulator. This is fast but results in averaged data that is no better than the ADC's inherent bit depth.
2. Their FFT routines also used integer data throughout, using lookup tables for things like trig functions, log conversions, etc.

One of the 10-bit DSO's I still use (a Lecroy 9430) has some of these limitations in its FFT. No matter how many time domain averages you do before FFTing the data, you'll never get even 2-bits of dynamic range improvement from it. But if I pull its individual time records into the computer, average and FFT them using real numbers, I get 80 dB easily. This shows that its ADC is very good but it is crippled by the scope's data processing.

Despite these limitations, they sold the FFT package as a $1500 option! My, how times have changed...

[David Hess]

You can do things easily with a DSO that would be very challenging to do with analog.

That is why I like the Tektronix 2232.  Only that series of DSOs had automatic peak-to-peak triggering (that I know of) so I get the best of both worlds although you would not use it for single shot captures or at slow time/div but that applies to any automatic triggering.

Tektronix lost that feature in the following generation of both digital storage and analog oscilloscopes which sort of makes sense.  (1) Later when digital triggering was implemented on DSOs, it could have been reintroduced but that was 20 later and it had been forgotten by designers and new users.

(1) The last Tektronix analog oscilloscopes had a different technique which involved using the trigger itself to make amplitude measurements and this allowed inferior automatic level triggering for essentially free.  Modern DSOs duplicate this in one way or another;  there might be some exceptions but I have not found them.

[Old Printer]

The AD2 was referred to previously: it is a toy with no proper connectivity.

Dismissing the AD 1 or 2 as a toy is nonsense, it has earned it's reputation as an excellent educational tool that fulfills several T&M jobs. The "proper connectivity" is a $25 option, hardly a barrier. It's input voltage range is +_ 25V @ 1M ohm with a 1X probe, neither TTL or CMOS limited.

[Sylvi]

Yes, the AD1 can have  some dangling BNCs tied to it while itself dangles from a computer. That is not exactly a proper format for serious test equipment. For a uni student on a budget in a dorm it will suffice.

I'm used to "real" test equipment that is unitised in a box with all the right connectors securely in place.

Obviously, those are two aesthetics that you either like one or the other, can tolerate, or just are used to. I'm used to the the Test-equipment aesthetic particularly for my bench. I have no desire to have a computer there as well. For many people, especially more computer-savvy people than me and/or younger than me, their aesthetic incorporates a computer for everything they do, so add-ons are not a nuisance; rather, the add-ons are just more stuff tied to the centre of their world. I guess in my aesthetic there is more compartmentalisation and separation.

I don't have anything wireless (I've opted out of that experiment), so all the links between potential devices or add-ons is by wire, which makes the AD2 and its accessories more messy to me.

Respectability of or for a given device is not the issue here and I don't generally choose things on that basis. Although, I do admit to having more respect for Keysight because of its HP heritage. I find their "get rid of the toys" slogan a bit offensive in their literature, where all the techniques they use are given a Keysight buzz-name that is meaningless rather than saying how they make 8-bits looks like a scope of yore. The images of displayed data don't actually look all that hot. In any case, Keysight is respected and so is the AD2 for what they do for the people who choose to use them, just in the same way that a Ferrari is respected for what it can do even though that is different than what the grocery-getter reliable Chevy can do, which is respected for its virtues.

[Old Printer]

Yes, the AD1 can have  some dangling BNCs tied to it while itself dangles from a computer. That is not exactly a proper format for serious test equipment. For a uni student on a budget in a dorm it will suffice.

I'm used to "real" test equipment that is unitised in a box with all the right connectors securely in place.

Digilent uses a pin and socket joint to connect the BNC board to the main board. If you pop the hood on much of the older "real" test equipment you are "used to" you will find that companies like Tektronix used the same type of joint for many years to connect a myriad of vaccume tubes and then transistors to the main boards. That's not even mentioning the inter-board connections of more modern equipment. Is it unsuitable now because it can be seen?

[Fungus]

I'm used to "real" test equipment that is unitised in a box with all the right connectors securely in place.

It wouldn't be difficult to make a box.

Much more difficult would to meet the performance specs of the AD2 with a piece of "real" equipment.

[bd139]

https://store.digilentinc.com/bnc-adapter-for-analog-discovery/ ?

And anyone who complains about this not being  “real test gear” and “real BNCs” hasn’t seen a default state fucked up 54600 or TDS scope. Alas any further discussion on that may make me dig out the photos of cracked boards, broken off connectors and my favourite, the ext trigger that Tek forgot to solder on...

The metric of success isn’t the lore around the equipment, nor the spec sheet but the ability to solve a problem and deliver a product or repair. AD2 is a compromise but a quite decent one. Plus the pricing is disposable territory.

[james_s]

I personally don't like PC based test gear. It originally promised lower cost, but today the display and interface subsystems have become so cheap that there is little savings. I have had issues getting it to work properly with some computers, the software is often lousy and gets abandoned at some point, and I don't like having to drag around a laptop and the test gear. I still have an older Bitscope and a couple different PC based logic analyzers and they're not nearly as nice to use as a standalone instrument.

[bd139]

All my big test gear gets plugged into the PC anyway. That's where the useful analysis is done, usually with numpy/scipy/sympy/matplotlib.

[nctnico]

I'm used to "real" test equipment that is unitised in a box with all the right connectors securely in place.

It wouldn't be difficult to make a box.

I think that is the whole point. The digilent AD2 would be much more interesting if it didn't had such a high DIY factor. There certainly is a market for high resolution oscilloscopes so it would make sense to have something like the AD2 with BNC inputs and some protection circuitry.

[MrW0lf]

There certainly is a market for high resolution oscilloscopes so it would make sense to have something like the AD2 with BNC inputs and some protection circuitry.

But it could not be knobbed box, that would be just silly. Usage scenario / target group is too different. Sometimes I wonder why Pico does not have "do it all" box to compete (with some digital outputs and 2x signal gen). There should be similar but dirt cheap ADALM2000, which I ordered last summer. Still have not received it, you see its not shipping yet...

[james_s]

All my big test gear gets plugged into the PC anyway. That's where the useful analysis is done, usually with numpy/scipy/sympy/matplotlib.

I guess if that's your use case then PC based makes sense. Personally for what I do with a scope I can count on one hand the times I've wanted to do any sort of analysis on a PC beyond storing a screen capture. Normally I'm testing something in real time and just want to see the waveform and measure things like frequency and voltage levels.

[Old Printer]

I did add a couple switches to my BNC adapter to control the AC-DC Coupling. Those fiddly little blue jumper blocks were a PITA.

[bd139]

All my big test gear gets plugged into the PC anyway. That's where the useful analysis is done, usually with numpy/scipy/sympy/matplotlib.

I guess if that's your use case then PC based makes sense. Personally for what I do with a scope I can count on one hand the times I've wanted to do any sort of analysis on a PC beyond storing a screen capture. Normally I'm testing something in real time and just want to see the waveform and measure things like frequency and voltage levels.

Display is much bigger on a PC though...



 

[Fungus]

Display is much bigger on a PC though...



 

And an extra digit! 

[bd139]

Yes that was interesting to find. Also there is an entirely undocumented SCPI interface hiding in it.

[james_s]

Again if one's use case requires a bigger display then that's another reason a PC interface could be useful. I have no trouble reading the display on my multimeter though and I don't have to find a spot on my workbench to put my laptop.

[mtdoc]

I did add a couple switches to my BNC adapter to control the AC-DC Coupling. Those fiddly little blue jumper blocks were a PITA.

Nice mod.  I like it!  How did you make those panels?  Do you have a laser cutter?

[Old Printer]

I did add a couple switches to my BNC adapter to control the AC-DC Coupling. Those fiddly little blue jumper blocks were a PITA.

Nice mod.  I like it!  How did you make those panels?  Do you have a laser cutter?

Yes, I have a laser, actually 2, an Epilog 50 & 75 watt. I am the production manager for a sign shop, lots of neat toys.