'Scope display anomaly... any ideas please?

[Al (astral highway)]

The query du jour...

My scope is a GW Instek GDS-1022, a digital model. I am using Coline M12 x 10 probes.

I am using the MATH function using both scope probes to derive the output waveform from a gate drive transformer with a capacitative load.

Photo shows scope trace at 5uS/div. The red trace is the subject of interest.

I have two queries...


MATH function trace

Query 1) The 'scope doesn't allow me to just display the red MATH trace, which is what I want to see. It will allow me to display the additive MATH trace along with the blue trace (channel 2), overlaid. This is too chaotic to see what is happening...

The setting process goes...

select MATH = Math on/off
Operation type CH1+CH2. Press again for FFT/
Select addition/ subtraction CH1+/-Ch2
Select result position (UNRESPONSIVE)
Unit/Div (UNRESPONSIVE)

The red MATH trace that I want is partly concealed by the presence of a trace for channel 2, which I don't need, UNLESS I dial down the voltage per division for channel 2. This obviously stretches out the waveform from Ch2, leaving the red MATH trace visible, but I'd like to get rid of Ch2 altogether.

Normally, on this scope, you can select both channels to be displayed or just Ch1 or Ch2. So this seems to be an anomaly of the MATH function, unless I've missed something... which is what I'm asking in query 1.

(I also want to use the Fast Fourier Transform setting - not now but during this project - but this is even more chaotic.)


Query 2)

Scope trace 'jitter OR frequency instability?

If the 'scope trace wobbles about a bit and the frequency shown is a little unstable, is this always a reflection of what's happening with the signal that is input OR can it be an artefact of the timebase?


I can't see an obvious reason why the signal source would be in any way unstable - the logic side is all as it should be - so I'm wondering if it can be related to the timebase. This is obviously a pretty low frequency signal, with a period of about 120uS (actually it's variable between 100-300uS). It is possible that the gate drive chip could do with better more oomph with bigger low electrolytics on its power rail, but that's all I can think of.

It may just be that I'm stuck with this situation with this particular scope, but if I can get MATH to work, I'll be delighted as looking at the waveform purity up close will give me peace of mind before I start introducing power switching devices.

Thank you!

[cmjones01]

One of my clients has one of these GWInstek scopes, and I experienced very similar problems with it. The 'MATH' functions were more or less unusable because of issues like this. I remember trying to use Ch1-Ch2 to see the voltage across a current sense resistor, and it was hopeless: the result trace was obscured by the original signals, and there appeared to be no way to have a different sensitivity for the result than for Ch1 and Ch2. Moving either trace off the screen to get it out of the way resulted in terrible errors in the maths result, because the traces were getting clipped.

I'm used to analogue scopes in which the dynamic range of the inputs is considerably more than the screen height, so you can get sensible results when adding inputs even when their waveforms are too big for the screen, or completely off it. The GWInstek is not like that. The size of the screen is the limit of the output from the A-D converter and that's your lot. All processing and maths is downstream of that. In this instance, I gave up on it and started using an analogue scope instead.

Regarding the trigger jitter, I also experienced all sorts of strangeness with triggering. Sometimes it would just not bother to trigger even though everything was set up correctly, which was a nuisance when trying to capture intermittent signals.

With a lot of modern digital scopes, it seems that decades worth of knowledge about what makes a good instrument were just thrown away once somebody discovered cheap ADCs which could do tens or hundreds of megasamples per second. Sadly there's a lot more to analogue measurement than an ADC and some software. That's why, on my bench right now, the actual paid work I'm doing for a client on a thoroughly 21st-century project is using an ancient behemoth of a scope with 50+ valves in. It may be old and hot but it's breathtakingly well designed and reliably tells the truth. Rant over!

To do the measurement you want with a scope like the GWInstek, you really need a differential probe. I've heard good things about the EEVBlog HVP70. I haven't used one myself but (another) client has something similar and it works quite well. That's where I'd be looking.

Chris

[Al (astral highway)]

Hi Chris, and thanks for your great response. Really useful and much appreciated that you took the time to go into such detail.

Originally Posted by cmjones01

That's why, on my bench right now, the actual paid work I'm doing for a client on a thoroughly 21st-century project is using an ancient behemoth of a scope with 50+ valves in. It may be old and hot but it's breathtakingly well designed and reliably tells the truth.

That's wonderful! What make/era is it and can you post a photo?

To do the measurement you want with a scope like the GWInstek, you really need a differential probe. I've heard good things about the EEVBlog HVP70. I haven't used one myself but (another) client has something similar and it works quite well. That's where I'd be looking.

That's a great tip. I'll take a look.

Your post has saved me a great deal of time trying to get results that are unobtainable.

Thanks again, Chris. I'll post if I have any news on the direction you're suggesting.

[cmjones01]

Originally Posted by astral highway

That's wonderful! What make/era is it and can you post a photo?

Here you go. It's a Tektronix 535A and dates from about 1965. With its various plugins, it will do almost anything. It has dual timebases for zooming-in to the point of interest on a signal or making accurate time measurements. Plugins exist with up to 4 input channels, high sensitivity down to microvolts/cm, differential inputs with up to 500V common-mode, microwave spectrum analysis, and sampling measurements/TDR to 5GHz or so. Shown is the operational amplifier plugin, which allows you to do any processing on the signal you like with a couple of op-amps, like an analogue computer. Maths entirely in the analogue domain!

Its native 15MHz bandwidth turns out to be only a minor limitation in the real world.

Chris

[Al (astral highway)]

Originally Posted by cmjones01

It has dual timebases for zooming-in to the point of interest on a signal or making accurate time measurements. Plugins exist with up to 4 input channels, high sensitivity down to microvolts/cm, differential inputs with up to 500V common-mode, microwave spectrum analysis, and sampling measurements/TDR to 5GHz or so. Shown is the operational amplifier plugin, which allows you to do any processing on the signal you like with a couple of op-amps, like an analogue computer. Maths entirely in the analogue domain!

A wonderful bit of kit, Chris!

I have to admit to being a bit envious! But in a kind way...

I'm going to start a new thread to see if anyone magically is offering a pair of differential probes...

[Al (astral highway)]

Originally Posted by cmjones01

er!

To do the measurement you want with a scope like the GWInstek, you really need a differential probe.


I've heard good things about the EEVBlog HVP70. I haven't used one myself but (another) client has something similar and it works quite well. T

Hello Chris,

Initial impressions are that differential probes don't come cheap! I'm in a dilemma as I won't use them much beyond my current project.

Would you or any other members be able to recommend a pair that are 'good enough' for what I'm doing but not fancy enough to demand big £££?

Plus a source, ideally?

[G0HZU_JMR]

The other option is to dump out raw sample data from the scope for channels 1 and 2 and do the maths function and the FFT offline on a PC.

This will be slow and a bit clunky but you can do all kinds of post processing once you have the raw sample data. It could be done in excel or in a simple VB program or in Matlab or the free equivalent (Octave).

I assume the scope supports this via USB or some other interface.

The other option would be to buy something like the 100MSa/s 14 bit Analog Discovery 2 and do the maths functions and the FFT with the bundled software tools. This would be quite powerful and it should have enough bandwidth.

[emeritus]

Those old Tek scopes were certainly versatile. Lots of useful outputs. When we were developing a varactor tuner at Plessey we used a pot on the sawtooth output to carry out frequency sweep tests that required a max varactor diode voltage of about 90V. Another time when we needed to measure the frequency of a very low level signal, this was easiy done using two cascaded vertical amplifiers, connecting the output of the first to the input of the second, whose output was connected to the counter, a high impedance scope probe at the input of the first being used to tap into the circuit. Also good for warming up the lab (along with several HP 608-series signal generators) on cold frosty days!

[Herald1360]

Would something like the (relatively) old Tek THS700 series handheld 'scope do? These have floating isolated channel inputs and can run on internal batteries which makes them very useful for all sorts of "impossible" jobs.

Hmm, just looked at the secondhand cost of these, maybe not such a good idea. Seems the one I acquired a couple of years ago was a bit of a bargain!

There might be something with similar isolated inputs but cheaper from the big Chinese emporium, perhaps.

[Al (astral highway)]

Hi Chris, that would indeed have been ideal!

I’ll keep looking ! I can’t wait foe this all to be much simpler and more reliable

[G0HZU_JMR]

Another option would be to use something like the AD8130 to make a (high input impedance) summing amplifier. The inputs can be fed to +IN and REF and each input can be made to mimic a high impedance scope input. This would allow the use of a x10 scope probe or some other form of resistive divider if a large input voltage range is required.

Have a look at the AD8130 datasheet for the app circuit for the 'summer' amplifier configuration that can sum two high Z inputs.

I've not used the Eevblog differential probe but I'm not sure why it would be suitable for the equivalent of maths addition. Does it have this feature included?

[cmjones01]

Originally Posted by G0HZU_JMR

I've not used the Eevblog differential probe but I'm not sure why it would be suitable for the equivalent of maths addition. Does it have this feature included?

I believe astral highway wants to subtract channel 2 from channel 1, which in my mind is a differential measurement. Certainly the application (looking at the waveform across a MOSFET/IGBT gate at some arbitrary distance from ground) seems like a good one for a differential probe. The probe won't, I agree, get you arbitrary maths functions, though!

I like the idea of using something like an Analog Discovery 2 to capture good data and then do the post-processing maths to suit your own needs.

Chris

[G0HZU_JMR]

OK thanks. I may have misinterpreted the question if the aim is for CH1-CH2. The text in the OP does suggest the aim is to add the signals together and my interpretation of the scope screenshot is that CH1+CH2 is being displayed.


If the aim is to display CH1-CH2 then I agree this would be much more conventional and a differential probe would be fine for this. The AD8130 can also be configured for a differential measurement and this is the main function of this chip. The output of the AD8130 could then be fed to channel 1 of the GW Instek scope.

The Analog Discovery 2 already has differential inputs on both channels so this could be an advantage although the input voltage range is limited to something like +/- 25V. Otherwise, it could be used with the scope adaptor board and a pair of x10 scope probes to do the maths operation using channel 1 and 2.

I agree that the Analog Discovery 2 is a nice device. I don't use mine very often which is a shame really because it is a very powerful and versatile module and a great educational tool. Sadly, the sale price for them seems to be creeping upwards every year. They now cost about the same as a basic digital scope. I think mine cost about £200 with the scope adaptor plus a few other accessories. The price today seems to have nearly doubled.

[G0HZU_JMR]

One risk with the AD2 is that it might not be robust enough to withstand the kind of high voltage (high energy?) experimentation that Al does. So it might get zapped at some point. I'd be nervous to use mine like this and would probably try and use something 'cheap and disposable' instead.

The little AD8130 chip costs about £4 and it has good specs for bandwidth and common mode rejection. See the datasheet below.

https://www.analog.com/media/en/tech...D8129_8130.pdf

The application circuit for the 'summing' circuit is given on page 35 out of 40.

Otherwise, this chip could form the basis of a simple differential probe if used with the basic circuit at the start of the datasheet. I think the voltage range of the differential input is quite limited so it would probably need to be used with x10 scope probes and I'm not sure how well this would work in practice.

[Al (astral highway)]

Originally Posted by G0HZU_JMR

OK thanks. I may have misinterpreted the question if the aim is for CH1-CH2. The text in the OP does suggest the aim is to add the signals together and my interpretation of the scope screenshot is that CH1+CH2 is being displayed.

Hi Jeremy, well-spotted and my slip. As Chris somehow inferred, I did indeed mean CH1-CH2.

Originally Posted by G0HZU_JMR

If the aim is to display CH1-CH2 then I agree this would be much more conventional and a differential probe would be fine for this.

Thanks for confirming this.

I'm starting to feel more general dissatisfaction towards my 'scope. I frequently have jitter/trigger issues which vastly complicate measurements that I absolutely need to rely on before moving on a stage in a project.