EEVblog #983 - A Shocking Oscilloscope Problem ! aka Whack Triggering ...

[RGB255_0_0]

Dave, you've been called out by Keysight_Daniel: https://youtu.be/OgDuL-or12c?t=440

[Fungus]

Who wants to see Dave whacking his new $300k dumpster find?

[FrankBuss]

Dave, you've been called out by Keysight_Daniel: https://youtu.be/OgDuL-or12c?t=440

Lol, I'm pretty sure Dave knows how to use single mode and the auto/normal trigger modes. Sometimes single is just better, if you want to catch the first trigger and not mucking around with holdoff, or accidentally catch another trigger, but you wanted to analyze the first triggered signal.

And the Keysight guy shows the piezoelectric effect of the probes, which is well known, while Dave demonstrated it inside the scope, which might be not so well known and varies greatly between scopes.

[mikerj]

I discovered my Owon SDS7102 was horribly microphonic a few years back when I bought it.  This is the result if I flick the Channel 1 BNC socket with a finger nail, about 700mv pk-pk!

[flodins]

I recently got the R&S RTO2024 2GHz 10GS/s 1Mwfms/s 16bit ADC and it seems to be the best. BNCs and frontend are built like a tank. RTM-ZP10 probe is made mainly from a rubber. Only the very tip is made from a plastic and it is susceptible to whack test.

[Damianos]

I wonder what will be the result if someone performs the auto-calibration while listening loud music! 
The manufacturers say "disconnect everything from the inputs". So no damping things will be present there.


Edit: correctness of my stupid English!

[pascal_sweden]

R&S should have used capacitors from Murata for reduced microphonics:

https://books.google.no/books?id=y0kGRzbh-owC&pg=PA43&lpg=PA43&dq=microphonics+murata&source=bl&ots=-R_bm7FLG2&sig=zc2FjzQ2JzfsqAG_ZueYseqFH6s&hl=no&sa=X&ved=0ahUKEwiutsCil4TTAhWqK5oKHUq7AaIQ6AEIITAB#v=onepage&q=microphonics%20murata&f=false

[KedasProbe]

I tried it with a 50 ohm termination on it and then it's too low to detect it on that channel.
So I assume the problem is located before the amplifier. (if it would be after it my to measure circuit output load wouldn't have any influence on it)
This also means it's less likely to happen during measurements since you are not measuring with an open input.

[jonovid]

try tapping your partner? ...jokes aside. 

But seriously-
expensive top of the line oscilloscopes should not have this microphonics problem IMO. 
say your useing the oscilloscope in a industrial environment,  mechanical vibrations from
running industrial plant and equipment can get on the test bench. also what if the test bench is a moving vehicle?
as example, commercial fishing boat at sea. another example aircraft have mechanical vibrations from engines into the fuselage.

[Petter]

I was expecting someone to have replaced input caps on his/hers scope already ...

[mikeselectricstuff]

I was expecting someone to have replaced input caps on his/hers scope already ...

Bit busy at the moment..

[David Hess]

The only way I can do this on my old oscilloscopes is to tap the BNCs directly but what is going on with them has nothing to do with ceramic capacitors; the BNCs in this case are mechanically isolated from the printed circuit boards.

The BNC coaxial insulator between the body and inner conductor is piezoelectric and/or triboelectric and I verified this by attaching a short RG-316 (1) coaxial cable.  The BNC on the end of the cable has roughly the same sensitivity as the oscilloscope BNCs and the cable itself is even more sensitive.  If I connect the inputs together with the cable, then both channels show exactly the same thing no matter where on the BNCs or cable I tap.  The high impedance vertical inputs are operating as charge amplifiers and when the input are shorted together, they see exactly the same signal produced by charge in the BNCs and cable.

If I tap the bodies of the oscilloscopes, I have to get close to the BNCs to have any effect unlike most of the examples in Dave's video.

They make specially rated connectors including BNCs and cable for applications where this is a problem.

I agree with Dave; it is insane that tapping the touch screen on the RTB2004 as lightly as he did has the effect he showed.  Even tapping the body should not do anything unless it is coupling into the BNCs themselves.

(1) RG-174 sized but with a Teflon dielectric and silver plated shield and center conductor.  It is just what I had in front of me.

[Fungus]

But seriously-
expensive top of the line oscilloscopes should not have this microphonics problem IMO. 
say your useing the oscilloscope in a industrial environment,  mechanical vibrations from
running industrial plant and equipment can get on the test bench. also what if the test bench is a moving vehicle?
as example, commercial fishing boat at sea. another example aircraft have mechanical vibrations from engines into the fuselage.

Chill.... it's only down in the millivolt range, unlikely to affect most things.

[mtdoc]

But seriously-
expensive top of the line oscilloscopes should not have this microphonics problem IMO. 
say your useing the oscilloscope in a industrial environment,  mechanical vibrations from
running industrial plant and equipment can get on the test bench. also what if the test bench is a moving vehicle?
as example, commercial fishing boat at sea. another example aircraft have mechanical vibrations from engines into the fuselage.

Any environment where vibration levels are high enough to trigger this issue when a probe is attached is going to cause bigger issues for your measurements IMO.

If you were in such an extreme environment, a vibration isolation table would be needed.  These are pretty much standard fare for any sensitive electrophysiology measurements where vibration from someone just walking by your measurement set-up can cause problems.

I think this issue is being overblown.

[joeqsmith]

In case you're not following the R&S thread, the issue pretty much vanishes as soon as there is any kind of probe connected, so pretty much a non-issue.
 
What a scope does with an unconnected input just doesn't matter, as you'll never use it like that.

Yep.  Repeat the test with the probes attached, then tap the BNC end of the probe..  Even that is not something I do.  I don't even tap the chassis.  Tapping the touchscreen on my scope is no problem but I run it with the mouse.   Now fire up my ignition tester in the same room, the LCD will start to act up and the mouse will hang and need to be unplugged to recover.   

[EEVblog]

Dave, you've been called out by Keysight_Daniel: https://youtu.be/OgDuL-or12c?t=440

LOL!
The impulse he's getting is from the MLCC cap in the x10 probe, not inside the scope.
Not that his rant had anything to do with the microphonic problem anyway.

[EEVblog]

In case you're not following the R&S thread, the issue pretty much vanishes as soon as there is any kind of probe connected, so pretty much a non-issue.
 What a scope does with an unconnected input just doesn't matter, as you'll never use it like that.

Yep.  Repeat the test with the probes attached, then tap the BNC end of the probe..  Even that is not something I do.  I don't even tap the chassis.  Tapping the touchscreen on my scope is no problem but I run it with the mouse.   Now fire up my ignition tester in the same room, the LCD will start to act up and the mouse will hang and need to be unplugged to recover.   

I'll repeat this again. I have tried using a shorted x10 scope probe and I can still get the problem.
Yes it's diminished in amplitude but it's still possible.
Do I have to make a video showing this?

Also, be fully aware that x10 probes contain microphonic MLCC's too, sometimes in the probe tip, other times in the base of the cable in the BNC box.
I have done a video on this years ago and part of it was included my Shocking video.
I showed this in my videos, and it's 10 times worse than tapping the BNC.

[KedasProbe]

Maybe you can make a video 'fixing' an input channel of one of your many scopes and then with the '300k' scope you can check if the BW is still ok. 

[pascal_sweden]

Dave mentioned Murata in his follow-up video about this issue. I wonder if that was based on my below comment

R&S should have used capacitors from Murata for reduced microphonics:

https://books.google.no/books?id=y0kGRzbh-owC&pg=PA43&lpg=PA43&dq=microphonics+murata&source=bl&ots=-R_bm7FLG2&sig=zc2FjzQ2JzfsqAG_ZueYseqFH6s&hl=no&sa=X&ved=0ahUKEwiutsCil4TTAhWqK5oKHUq7AaIQ6AEIITAB#v=onepage&q=microphonics%20murata&f=false

[Petter]

I am a little curious as to which caps are causing the problem in the first place. It seems strange to me that you would need a lot of capacitance in the signal circuit proper, hence C0G or whatever capacitors might be usable and they should be very stable wrt shock.

If the caps are in the decoupling of any power rails that could explain large values and the desire to use capacitors with many layers. The obvious solution in that case would be to use tantals + smaller ceramics in tandem. Now this probably introduces a cost problem which the manufacturers don't care to pay for.

I challenge Dave to fix this problem and do a cost analysis of the fix. That would enable the industry to move forward on this issue and help out with edge cases and touch screen usage .

[joeqsmith]

In case you're not following the R&S thread, the issue pretty much vanishes as soon as there is any kind of probe connected, so pretty much a non-issue.
 What a scope does with an unconnected input just doesn't matter, as you'll never use it like that.

Yep.  Repeat the test with the probes attached, then tap the BNC end of the probe..  Even that is not something I do.  I don't even tap the chassis.  Tapping the touchscreen on my scope is no problem but I run it with the mouse.   Now fire up my ignition tester in the same room, the LCD will start to act up and the mouse will hang and need to be unplugged to recover.   

I'll repeat this again. I have tried using a shorted x10 scope probe and I can still get the problem.
Yes it's diminished in amplitude but it's still possible.
Do I have to make a video showing this?

Also, be fully aware that x10 probes contain microphonic MLCC's too, sometimes in the probe tip, other times in the base of the cable in the BNC box.
I have done a video on this years ago and part of it was included my Shocking video.
I showed this in my videos, and it's 10 times worse than tapping the BNC.

No need to make a new video on my account.  Pick what you consider is a sensitive scope and just run the various probes across it, tapping just the body.  Skip the tip.   Take what you consider is the least sensitive 10X probe and try it on each of the remaining scopes.  Do you see different results?   This could be a huge matrix..  You need a controlled tapper as well.  Sorry, I tend to hyper focus if you have not yet sorted that out....

[bktemp]

I am a little curious as to which caps are causing the problem in the first place. It seems strange to me that you would need a lot of capacitance in the signal circuit proper, hence C0G or whatever capacitors might be usable and they should be very stable wrt shock.

If the caps are in the decoupling of any power rails that could explain large values and the desire to use capacitors with many layers. The obvious solution in that case would be to use tantals + smaller ceramics in tandem. Now this probably introduces a cost problem which the manufacturers don't care to pay for.

I challenge Dave to fix this problem and do a cost analysis of the fix. That would enable the industry to move forward on this issue and help out with edge cases and touch screen usage .

Maybe the capacitor for AC coupling the signal (probably the big one next to the optocoupler).
There seem to be quite a lot of different capacitors in the signal path. The grey ones are typically C0G while the brown ones often have a higher capacitance.



It is probably difficult to solve the problem without reverse engineering at least the complete input section (probably coarse input range selection switch) of the frontend.

Image from Mike's teardown video:

[Petter]

Probably right about that.
So it should be quite easy to desolder that part and measure it - then figure out if if falls into the C0G or whatever range, perhaps stacked 3 high or whatever.

If it is easy to get to this while operating the scope, it should be a decent test to tap it directly. I can't remember how hard it was to get to this point in the teardown.

I would think that the smaller parts are safe. Since this scope operates at only 300MHz, serious reverse engineering is unlikely to be very important. Smack a replacement in, test scope for linearity and shock sensitivity, then done if it works.

[Keysight DanielBogdanoff]

Dave, you've been called out by Keysight_Daniel: https://youtu.be/OgDuL-or12c?t=440

LOL!
The impulse he's getting is from the MLCC cap in the x10 probe, not inside the scope.
Not that his rant had anything to do with the microphonic problem anyway.

Haha, the point was definitely about triggering and not about the piezoelectric thing - I saw a teachable moment and couldn't pass it up

[mikeselectricstuff]

I have a couple of suspicions about other possible factors - quite possible there are multiple contributung factors.
The plastic in the BNC connector (triboelectric)- banging a bnc plug on the end of a cable does cause some effect.
Another possibility- if there is a sensitive node that has DC bias on it, then variations in stray capacitance to that node will produce a signal. I'm specifically thinking about the shielding cans - even a small vibration of the can could have a significant effect. I wonder if this might be a reason for the use of copper cans, though I'd think there would be cheaper and better ways to damp vibrations, like sticking heavy tape on.