Better diff probe to avoid inaccurate scopes

[Faringdon]

Hi,
The attached is the scope of the secondary coil of a isolated 240VAC , 150w  one transistor forward with HV txformer reset.
Its got an oscillation in it of T=10ms. As can be seen
This is not real.
Its due to the leakage current to earth in the diff probe.

Do you agree, we must float the scope to avoid this and get a proper waveform?

We cannot afford a better probe, and an isolation transformer and 100:1 probe gives the same problem.

TA041
https://uk.farnell.com/pico-technology/ta041/probe-active-differential-powered/dp/1667343?gclid=CjwKCAjwq-WgBhBMEiwAzKSH6BTgsoaHaPZhLv3tN57SIsaMBnTLFUtTgh5geS-qpZcRgl3VZM1zuRoCTxgQAvD_BwE&mckv=s_dc|pcrid|605279997276|kword|ta041|match|p|plid||slid||product||pgrid|143566925168|ptaid|kwd-314725982045|&CMP=KNC-GUK-GEN-SKU-MDC-Test-Tools-PS

[nctnico]

Never float a scope. And a scope still has at least tens of nf of leakage to mains so it will be worse.

Show us the schematic,test points you want to measure and tell us precisely why you want to measure at those points.

[Faringdon]

Hi,
Thanks, its just a 150W, 240VAC one transistor forward, with high voltage transformer reset.
Its isolated.
We wish to  measure across the secondary coil.
Also, we wish to measure the primary, post rectifier  DC Bus.
Both these measurements are totally ruined by the TA041.
The SMPS is fed from an osolated AC source.
But its the same if we feed it from the raw mains.
The Diff probe is lousy. (TA041)
--------------------------------------------
Here is another scoping problem with TA041......The blue trace in the attached two scopes is the post rectifier DC bus capacitor (5u6 Film) of a 240VAC 150W One transistor forward with HV transformer reset. (with 150W load). Its taken with a TA041 Diff probe.

In one, taken up to 5 seconds after switch ON , the blue trace is always running down to a “valley” of some 165V.
In the other, the blue trace is sometimes going down to some 40V "valley".
The reason for the 160V "valley" (or any “valley” voltage above 40V) is the leakage current in the TA041 Diff probe.

Do you agree, we must float the scope to avoid this and get a proper waveform?

We cannot afford a better probe, and an isolation transformer and 100:1 probe gives the same problem.

[bdunham7]

If the DUT is isolated what exactly do you gain by floating the scope??

[Weston]

By what mechanism do you think "leakage current" is causing this? The differential probe has a 4M input impedance, if you are probing a low impedance node such as the DC bus voltage it seems unlike that any current though the input impedance of the probe would cause issues.

If the waveform is distorted is it possible that you are exceeding the common mode input voltage of the probe? That would distort the waveform produced by the probe.

If the DUT is floating and has no ground reference the common mode voltage is poorly defined and could reach some high DC potential. Floating the scope would not necessary fix this, no would a better differential probe. You need to provide some sort of ground return for the DUT to limit the voltage it could float to.

[jonpaul]

No need for diffa probes.


scope ch 1 5 or 50V/ div>>10x probe>>DUT B
scope ch 2 5 or,50.V/ div>> 10x probe >>DUT B

Set ADD, CH2 I V

scope creates,Differential!

Use,this,all the time,,on 120,240, 360V.  eg H and,half hbridges, transformers, UPS

j

[ogden]

The Diff probe is lousy. (TA041)

It's decent enough. When using diff probe, you better interconnect scope ground to ground of the DUT. Obviously use isolation transformer for DUT. It is said like million times: never float a scope. Float DUT instead.

[Martin72]

Show us the schematic,test points you want to measure and tell us precisely why you want to measure at those points.

+1

[Faringdon]

Thanks, but none of the measures kindly depicted here, will prevent the faulty scope shots which are being shown in the top post, and in post #2 above.

Or are you suggesting that we need to connect the DUT ground to earth with eg a 1MEG resistor?
The DUT is being supplied by an isolated AC source.
But the problem is also seen when DUT supplied by  raw mains.

[tggzzz]

Show us the schematic,test points you want to measure and tell us precisely why you want to measure at those points.

+1

Agreed.

But this is a Faringdon/treez topic.

[gf]

If the DUT is floating and has no ground reference the common mode voltage is poorly defined and could reach some high DC potential. Floating the scope would not necessary fix this, no would a better differential probe. You need to provide some sort of ground return for the DUT to limit the voltage it could float to.

The probe input impedance is "4 MΩ || 5.5 pF each side to ground", according to the datasheet. Isn't the resulting 4M+4M divider sufficient in order that the mean voltage between the two test points becomes the DUT's common mode floating voltage if the DUT itself is fully isolated (say 1GOhm) from ground?

[Weston]

The probe input impedance is "4 MΩ || 5.5 pF each side to ground", according to the datasheet. Isn't the resulting 4M+4M divider sufficient in order that the mean voltage between the two test points becomes the DUT's common mode floating voltage if the DUT itself is fully isolated (say 1GOhm) from ground?

Just trying to make the most of Faringdon's babble. I can't think of any other reasons the diff probe would be not giving the expected waveforms, other than the power supply to it browning out. Actually, he should check that too. Is it battery powered? Batteries could be draining over the course of several minutes and distorting the waveforms.


The DUT is DC isolated but there are capacitively coupled leakage currents. It's not uncommon to get a bit of a tingle from capacitively coupled voltages on laptop chargers if they use an ungrounded plug. The capacitively coupled voltage can also be rectified and cause a DC offset. I have killed a few devices due to the output of a lab power supply floating to some voltage above ground and then zapping the DUT when connecting a grounded cable.

[Kean]

Just trying to make the most of Faringdon's babble.

Unfortunately this has proven to be a lost cause time and time again.

[2N3055]

Just trying to make the most of Faringdon's babble.

Unfortunately this has proven to be a lost cause time and time again.

Of course it has. I don't know why people even bother anymore..
If you carefully read his first two posts, he claims that passive 100x probe and isolation transformer and TA041 show same waveform..

And both must be wrong. Because he expected to see something and it showed something else...
Instead of saying "oh I have oscillation here let me look in to it" he says it is not real. How does he know it is not real?
Statement could have been "It shouldn't oscillate like this". Not "it's not real" how does he know "it is not real"? Because he measured twice using two different techniques and got same result... Hmmmm.... Yep, not real, I need better probe. Definition of better here being "one that will show same waveform like Ltspice predicted"..... So it's probe that is not real. LTSpice simulation is...

Hence, waste of time... Because this is persistent pattern of behaviour, sadly...

[tggzzz]

Just trying to make the most of Faringdon's babble.

Unfortunately this has proven to be a lost cause time and time again.

Of course it has. I don't know why people even bother anymore..
If you carefully read his first two posts, he claims that passive 100x probe and isolation transformer and TA041 show same waveform..

And both must be wrong. Because he expected to see something and it showed something else...
Instead of saying "oh I have oscillation here let me look in to it" he says it is not real. How does he know it is not real?
Statement could have been "It shouldn't oscillate like this". Not "it's not real" how does he know "it is not real"? Because he measured twice using two different techniques and got same result... Hmmmm.... Yep, not real, I need better probe. Definition of better here being "one that will show same waveform like Ltspice predicted"..... So it's probe that is not real. LTSpice simulation is...

Hence, waste of time... Because this is persistent pattern of behaviour, sadly...

I think people are getting the idea, and acting (IMHO) appropriately.

Here are some of his recent threads with negligible responses...
https://www.eevblog.com/forum/manufacture/smps-transformer-hipot-test-without-any-damage-whatsoever/msg4775711/#msg4775711
https://www.eevblog.com/forum/reviews/what-mains-harmonic-class-is-hair-removal-product/msg4773638/#msg4773638
https://www.eevblog.com/forum/renewable-energy/sense-resistor-in-switching-node/msg4772918/#msg4772918
https://www.eevblog.com/forum/renewable-energy/two-current-sense-transformers-reveal-the-current-best/msg4768889/#msg4768889

In order to give people the information that will help them choose whether or not to respond to his posts, I suspect it will be necessary to repeatedly point out his behaviour on new posts that he makes.

It is a shame that is the case.

[David Hess]

By what mechanism do you think "leakage current" is causing this? The differential probe has a 4M input impedance, if you are probing a low impedance node such as the DC bus voltage it seems unlike that any current though the input impedance of the probe would cause issues.

If the waveform is distorted is it possible that you are exceeding the common mode input voltage of the probe? That would distort the waveform produced by the probe.

If the DUT is floating and has no ground reference the common mode voltage is poorly defined and could reach some high DC potential. Floating the scope would not necessary fix this, no would a better differential probe. You need to provide some sort of ground return for the DUT to limit the voltage it could float to.

What you describe could be the problem here.  Differential probes must be used with a ground return to control the common mode voltage.  For instance they cannot be used with an isolated oscilloscope input, or with an isolation transformer on the device under test, without the risk of the common mode leakage causing the common mode voltage to go out of range.  Tektronix specifically mentions this limitation in the documentation for their differential probes or maybe isolated inputs.

[gf]

If an ordinary differential probe is not good enough, what about optically isolated probes?
Seems they support common mode voltages of tens of kV, and CMRR of 150dB and more
Price is scaring, though

[David Hess]

If an ordinary differential probe is not good enough, what about optically isolated probes?
Seems they support common mode voltages of tens of kV, and CMRR of 150dB and more
Price is scaring, though

They are a higher performance solution *if* their high common mode capacitance can be tolerated, which is the case if the measurement is made with respect to the emitter or source of the output transistor.

[bson]

I'd suggest finding the root cause of the oscillation rather than adjusting your measurement method to make it go away.  If it's so close to oscillation that just putting a probe on it triggers it you have a problem, one which will eventually happen in some production setup.  Be happy you found it now rather than later, and use the probe to reproduce it - and focus on tracking down the root cause.  Then design it out of existence.