Differential measurements using isolated oscilloscope

[Aners]

How much difference in measurment accuracy can I expect from using an oscilloscope with isolated channels and single-ended probes versus using a non isolated oscilloscope but with differential probes?

I understand that it depends on the scope and the probes, but from a general perspective, what affects the measurment outcome?

[Benta]

What exactly do you mean by "isolated channels"?
If you mean isolating the 'scope power supply by using an isolation transformer: don't do it. It's the most certain way to kill yourself.
There are numerous threads on this forum on this subject, as well as videos.

[Aners]

Scopes like these:

https://www.tek.com/oscilloscope/tps2024b
https://cleverscope.com/products/CS448

[David Hess]

There are two major considerations:

1. Differential probes have high input impedance on both connections but with common mode rejection limited both inherently and because of mismatched source impedances.
2. Isolated probes have low input impedance on the common side, chiefly because of common mode capacitance to ground, but essentially unlimited common mode rejection.

The result is that at higher frequencies where the common mode rejection of the differential probe falls, an isolated probe has an advantage in accuracy.  But an isolated probe can only be used where one of the connection points has a low impedance capable of driving its common mode capacitance to ground which can be 100 picofarads or more.

Note that common high voltage differential probes can have horrendous common mode rejection after a time because of a poor choice of substrate for the high impedance dividers and there are no provisions for the user to adjust it.  Isolated probes do not have this issue so might be considered more reliable.

[Wimberleytech]

Not an answer but a comment related.

Not too long ago, I had a case where I needed to make a differential measurement on an off-line SMPS.  I decided to build my own differential amplifier using the AD624.  The output went to a single-ended probe to a TDS2022B.

[David Hess]

Not too long ago, I had a case where I needed to make a differential measurement on an off-line SMPS.  I decided to build my own differential amplifier using the AD624.  The output went to a single-ended probe to a TDS2022B.

I have done that also when I did not have a suitable differential or isolated probe available.  A test circuit can be designed to drive a 50 ohm transmission line connected directly to the oscilloscope or coaxially drive a common high impedance probe.  But for off-line designs, the test circuit can be non-trivial and will not provide galvanic isolation so a special probe is a better choice.

[GeorgeOfTheJungle]

It's the most certain way to kill yourself.

There are portable scopes that run on batteries and float... and there are signals to be 'scoped that are NOT at lethal potentials... and many other reasons why your answer (even if he floats a mains powered scope) isn't accurate and IMHO is a bit much fear mongering-ish.

[tggzzz]

It's the most certain way to kill yourself.

There are portable scopes that run on batteries and float... and there are signals to be 'scoped that are NOT at lethal potentials... and many other reasons why your answer (even if he floats a mains powered scope) isn't accurate and IMHO is a bit much fear mongering-ish.

Sigh.

The OP, who is also new to this forum, would be well advised to read, learn, and inwardly digest this: https://www.eevblog.com/forum/projects/isolation-transformer-for-scope/msg2259465/#msg2259465 Especially the anecdote including "This guy was WELL AWARE of the danger and took a lot of precautions to prevent shock . . . Bottom Line: He died anyway."

[Wimberleytech]

Not too long ago, I had a case where I needed to make a differential measurement on an off-line SMPS.  I decided to build my own differential amplifier using the AD624.  The output went to a single-ended probe to a TDS2022B.

I have done that also when I did not have a suitable differential or isolated probe available.  A test circuit can be designed to drive a 50 ohm transmission line connected directly to the oscilloscope or coaxially drive a common high impedance probe.  But for off-line designs, the test circuit can be non-trivial and will not provide galvanic isolation so a special probe is a better choice.

I floated the SMPS with a transformer.  Then I eliminated the galvanic isolation (made the grounds common). Thought through it very very carefully before flipping the switch.  It worked.

[David Hess]

Not too long ago, I had a case where I needed to make a differential measurement on an off-line SMPS.  I decided to build my own differential amplifier using the AD624.  The output went to a single-ended probe to a TDS2022B.

I have done that also when I did not have a suitable differential or isolated probe available.  A test circuit can be designed to drive a 50 ohm transmission line connected directly to the oscilloscope or coaxially drive a common high impedance probe.  But for off-line designs, the test circuit can be non-trivial and will not provide galvanic isolation so a special probe is a better choice.

I floated the SMPS with a transformer.  Then I eliminated the galvanic isolation (made the grounds common). Thought through it very very carefully before flipping the switch.  It worked.

Sure, I have done the same when I had no better option.  However the issue for the test circuit goes beyond providing galvanic isolation; when a high side measurement is needed, it is difficult to do a 340 volt level shift with high bandwidth.  So instead the low impedance high voltage node needs to be grounded which impresses a 340 volt high voltage switching waveform across the common mode capacitance of the isolation transformer.

The whole situation is not impossible however things can quickly get out of hand for certain measurements which a high voltage differential probe or isolated probe make trivially easy.

Another way to handle it is to use a high bandwidth current probe.  The test circuit converts the measurement to a current and the current probe provides common mode rejection and galvanic isolation.  This is not ideal but it can work if there are no better options.

[Aners]

There are two major considerations:

1. Differential probes have high input impedance on both connections but with common mode rejection limited both inherently and because of mismatched source impedances.
2. Isolated probes have low input impedance on the common side, chiefly because of common mode capacitance to ground, but essentially unlimited common mode rejection.

The result is that at higher frequencies where the common mode rejection of the differential probe falls, an isolated probe has an advantage in accuracy.  But an isolated probe can only be used where one of the connection points has a low impedance capable of driving its common mode capacitance to ground which can be 100 picofarads or more.

Note that common high voltage differential probes can have horrendous common mode rejection after a time because of a poor choice of substrate for the high impedance dividers and there are no provisions for the user to adjust it.  Isolated probes do not have this issue so might be considered more reliable.

I am still a little new to some of this stuff but I use it daily and I feel that I need to better my understanding.

Could you give me an example of a meansurment with two high impedance signals?

[David Hess]

Could you give me an example of a measurement with two high impedance signals?

The signals only have to be high impedance compared to the probe impedance.  And since an isolated probe is shunted by 100 picofarads or more to ground on its common side, that is not all that much.  Even a low impedance high speed balanced transmission line would qualify in that case.

High side gate drive circuits are the common measurement case however their output is very low impedance so isolated probes work well for them; they are almost the ideal case.  But if you are measuring the difference between two nodes inside the driver itself, there could be problems with an isolated probe.

[Aners]

The signals only have to be high impedance compared to the probe impedance.  And since an isolated probe is shunted by 100 picofarads or more to ground on its common side, that is not all that much.  Even a low impedance high speed balanced transmission line would qualify in that case.

High side gate drive circuits are the common measurement case however their output is very low impedance so isolated probes work well for them; they are almost the ideal case.  But if you are measuring the difference between two nodes inside the driver itself, there could be problems with an isolated probe.

Ok, when you say it is shunted by 100pF what do you mean then? Because I have been looking at simplified probe schematics for single ended probes like the one below, and I dont see what you mean.

[GeorgeOfTheJungle]

Sigh.

The OP, who is also new to this forum, would be well advised to read, learn, and inwardly digest this: https://www.eevblog.com/forum/projects/isolation-transformer-for-scope/msg2259465/#msg2259465 Especially the anecdote including "This guy was WELL AWARE of the danger and took a lot of precautions to prevent shock . . . Bottom Line: He died anyway."

(Never Ask A Question If You Don’t Know The Answer...)

@tggzzz, have you never ever floated your 'scope? Everybody else: raise your hand if you've never done it.

[tggzzz]

Sigh.

The OP, who is also new to this forum, would be well advised to read, learn, and inwardly digest this: https://www.eevblog.com/forum/projects/isolation-transformer-for-scope/msg2259465/#msg2259465 Especially the anecdote including "This guy was WELL AWARE of the danger and took a lot of precautions to prevent shock . . . Bottom Line: He died anyway."

(Never Ask A Question If You Don’t Know The Answer...)

@tggzzz, have you never ever floated your 'scope? Everybody else: raise your hand if you've never done it.

No, I have never floated a scope. I've used my skill and imagination to find ways not to float a scope.

How many people have you watched die in front of you from avoidable accidents? That changes your perspective.

[capt bullshot]

@tggzzz, have you never ever floated your 'scope? Everybody else: raise your hand if you've never done it.

No, I have never floated a scope. I've used my skill and imagination to find ways not to float a scope.

Not that difficult. Just keep your scopes away from trees and rivers.

[magic]

Love those scope floating threads

Ok, when you say it is shunted by 100pF what do you mean then? Because I have been looking at simplified probe schematics for single ended probes like the one below, and I dont see what you mean.

Your picture is wrong, it shows a conventional scope. Cut the connection from input probe ground to oscilloscope ground. Replace with a bit of parasitic capacitance. That's a scope with isolated input channels.

[Aners]

Your picture is wrong, it shows a conventional scope. Cut the connection from input probe ground to oscilloscope ground. Replace with a bit of parasitic capacitance. That's a scope with isolated input channels.

Oh yeah, I just assumed the ground is floating so the symbol in it self could be ignored, but still the same idea. I was more wondering about the 100pf shunt, and what that is.

[David Hess]

Ok, when you say it is shunted by 100pF what do you mean then? Because I have been looking at simplified probe schematics for single ended probes like the one below, and I dont see what you mean.

In your example, an isolated probe has the ground connection on the right hand side cut.  The shunt capacitance is then the capacitance from the probe "ground", more properly referred to as common, and the environment including the isolation barriers it uses for power and signal.  This also includes the entire length of the probe cable.

If an isolation transformer is used, then it has its own capacitance from the primary to the secondary or from the secondary to an electrostatic shield and it will likely be considerably more than 100 picofarads although they do make expensive "low capacitance high isolation" transformers.  An isolated probe has a lot less "surface area" contributing to its common mode capacitance.

[not1xor1]

Sigh.

The OP, who is also new to this forum, would be well advised to read, learn, and inwardly digest this: https://www.eevblog.com/forum/projects/isolation-transformer-for-scope/msg2259465/#msg2259465 Especially the anecdote including "This guy was WELL AWARE of the danger and took a lot of precautions to prevent shock . . . Bottom Line: He died anyway."

I agree about the dangers of high voltage, but I want to stress that a random quote from a post forum has absolutely no value as a proof.... that smells so much of urban legend...

[tggzzz]

Sigh.

The OP, who is also new to this forum, would be well advised to read, learn, and inwardly digest this: https://www.eevblog.com/forum/projects/isolation-transformer-for-scope/msg2259465/#msg2259465 Especially the anecdote including "This guy was WELL AWARE of the danger and took a lot of precautions to prevent shock . . . Bottom Line: He died anyway."

I agree about the dangers of high voltage, but I want to stress that a random quote from a post forum has absolutely no value as a proof.... that smells so much of urban legend...

If you read the reference, that anecdote was written by the engineer that investigated the incident as part of his professional duties. It is much more credible than urban legend.

Yes, of course it is an anecdote - but one that is a concrete example of sound proven theory and practice. It is much more credible than urban legend.

It is worth repeating anecdotes, because some people - particularly those that distrust "theory" and "experts" - tend to place more reliance on personal stories than impersonal theory. Strange, but true, and worth understanding.

And I would still be interested to hear if "GeorgeOfTheJungle" has watched anybody die in an avoidable accident. Somehow I doubt it

[magic]

I have almost been killed myself in an avoidable accident and still don't see it as a reason to become a cowardly wimp

The best way to avoid high voltage accidents is to become a web developer Hate to say it, but once you open the cover of some mains powered appliance you already need to be careful around it, not matter how you probe it.

[Aners]

Guys, I really appretiate the warning about floating my oscilloscope but I was not planning on doing it and also I would rather this thread contain information about the question I asked. Even if it does not specifically state a by-design isolated oscilloscope I did post which types of scopes I meant in the reply I made. I still have some followup questions, I just need to write them down.

I just bet there are many many threads already about how dangerous it is to float a scope and another one is not needed.

[tggzzz]

I have almost been killed myself in an avoidable accident and still don't see it as a reason to become a cowardly wimp

That's a stupid non-sequiteur.

I don't avoid risks and have deliberately placed my daughter in risky situations, e.g. backpacking around India when 13yo, climbing and skiing, and being a solo pilot in aircraft that have to make forced landings (before she could start to drive a car in her case).

The point is to learn to understand what's risky, how to recognise incipient risks, and how to deal with them when they occur.

People that advocate floating a scope don't understand, can't recognise and hence can't avoid the risk.

I really don't care if such people injure themselves. I do care if they ignorantly teach bad pactices to other people.

The best way to avoid high voltage accidents is to become a web developer Hate to say it, but once you open the cover of some mains powered appliance you already need to be careful around it, not matter how you probe it.

True but trite and unhelpful.

Maybe you think statements such as "do not be on fire" are useful? Yes, I've seen/heard that statement too often - usually by people embracing the Dunning-Kruger syndrome.

[not1xor1]

I agree about the dangers of high voltage, but I want to stress that a random quote from a post forum has absolutely no value as a proof.... that smells so much of urban legend...

If you read the reference, that anecdote was written by the engineer that investigated the incident as part of his professional duties. It is much more credible than urban legend.

OK... I had missed that... you're right... in that case that's a reliable source