Trying to measure a noisy signal with scope and differential probe

[canadaboy25]

Hey guys,
I have a bit of a newbie question here.

I am troubleshooting a plasma TV which is missing half of the display due to a problem on the upper driver board.  There are 4 data signals coming into the driver board.  I am trying to test these 4 signals and compare them to the lower board which is working.  This sounds trivial but is proving to be difficult.

The driver boards utilize a floating ground to which all voltages are referenced.  The signals make their way to this floating ground domain by use of opto-couplers.  I had originally tried floating my scope (dangerous and not recommended, I know) in order to probe the signals but that did not go well (big surprise).  I bit the bullet and bought a high voltage differential probe for the scope thinking this would solve my problems.

The problem I am facing now is that the plasma TV by nature, emits a lot of electrical noise and my differential probe is picking all of this up.  If I move the probes anywhere near the TV, the noise I am picking up is larger than the signal I am trying to measure.  Shorting the probes of the differential probe still gives me significant noise on the scope.  I tried attaching wires to my test points and probing them further away form the TV, but the noise is just as bad.

How is this type of measurement normally done?  How can I probe these (5V I think) logic signals on a board with a floating ground, in this very noisy environment?

[alm]

If it's radiated noise picked up by the leads from the differential probe, try twisting them until close to where they connect to the device under test to reduce the loop area.

[canadaboy25]

If it's radiated noise picked up by the leads from the differential probe, try twisting them until close to where they connect to the device under test to reduce the loop area.

I tried that and it still picks up an unusable amount of noise.  The leads on my differential probe are quite short and thick which makes it difficult to make a tight twist and still be able to probe the pins I need to look at.

[IDEngineer]

It's probable your diffprobe's enclosure is not shielded, and it's picking up the ambient EMI internally. This would explain why twisting the leads didn't eliminate the noise. You may need to get creative in shielding the diffprobe itself.

[tautech]

How is this type of measurement normally done? 

Depends on the scope you're using, you haven't said so we can't really give any relevant advice.

[canadaboy25]

I tried using a length of coax attached to the points I wanted to test so that I could move the actual differential probe enclosure further from the TV.  While this improved the noise level with the leads shorted, the signal was still just as noisy.

Sorry, I should have mentioned I am using a Rigol DS2072A with the firmware hack.  I have tried playing around with the sampling rate, memory depth, acquisition options, and bandwidth limits.  Nothing helped very much.

[tautech]

I tried using a length of coax attached to the points I wanted to test so that I could move the actual differential probe enclosure further from the TV.  While this improved the noise level with the leads shorted, the signal was still just as noisy.

Sorry, I should have mentioned I am using a Rigol DS2072A with the firmware hack.  I have tried playing around with the sampling rate, memory depth, acquisition options, and bandwidth limits.  Nothing helped very much.

Use a 20 MHz BW limiter and ERES/HIRES

[canadaboy25]

I tried using a length of coax attached to the points I wanted to test so that I could move the actual differential probe enclosure further from the TV.  While this improved the noise level with the leads shorted, the signal was still just as noisy.

Sorry, I should have mentioned I am using a Rigol DS2072A with the firmware hack.  I have tried playing around with the sampling rate, memory depth, acquisition options, and bandwidth limits.  Nothing helped very much.

Use a 20 MHz BW limiter and ERES/HIRES

I tried the 20MHz bandwidth limit option with no results.  What is ERES/HIRES?

[2N3055]

Always start with all relevant data. So you have Rigol DS2072A, but what diff probe you have?
What kind of noise? What is amplitude, did you try to FFT it to see if it white noise or some signals?

HV diff probes are active devices, that have input attenuator (resistors make noise) that first make signal smaller than it is, and then it uses amplifier (more noise) to amplify two signals (for each probe end), subtract them (to calculate difference between them) and send it to scope.
Because of that, it will make noise on it's own. How much is normal, depends on probe.

So to make this efficient you need to:
- report exact probe model and type
- take a snapshot or two of scope screen showing data in question, with explanation of setup.
- Put all the data in one post so it's easy to analyze situation.

Otherwise, there will be many words and very little useful ones..

[tautech]

I tried using a length of coax attached to the points I wanted to test so that I could move the actual differential probe enclosure further from the TV.  While this improved the noise level with the leads shorted, the signal was still just as noisy.

Sorry, I should have mentioned I am using a Rigol DS2072A with the firmware hack.  I have tried playing around with the sampling rate, memory depth, acquisition options, and bandwidth limits.  Nothing helped very much.

Use a 20 MHz BW limiter and ERES/HIRES

I tried the 20MHz bandwidth limit option with no results.  What is ERES/HIRES?

Not only do you : I hate waiting! you also hate reading.
RTFM

[jonpaul]

Bonjour monsieur,


We are just speculating and rambling, you post has little engineering information to respond to a  vague and wide question.

To save everyone's time, can you PLEASE attach:

1/ Scope shot of noise signals with all scope settings

2/ Exact brand/model of TV under repair

3/ Exact brand model of HV diff probe

4/ Sketch (even by hand) of the entire test setup, TV, probe, scope.

Avec tous mes amenities,

Jon

[nctnico]

To the OP:
If the driver boards are the same I'd swap them so the broken driver board no longer has a floating ground. I'm also not sure whether probing signals is very productive. Did you already check all the power supply voltages? And if the driver boards are the same then take them out and check for differences using a DMM in Ohm mode (a thermal camera is also handy to look for hot spots and compare temperatures between parts). A pin or spot which shows a significant difference compared to similar pins is a tell tale sign you are getting close to the problem. You can also use this trick using a single board if it has the same circuit copied a few times. But since the driver board seems to do nothing at all it is likely power related.