Differential probe for a scope

[Corvus]

I'm looking for an affordable manner to get differential measurements with a scope. The dedicated diffrential probes do exist, but I'm having difficulties in paying hundreds for an accessory that is not used all the time. Most of the time a normal ground referenced measurement or using the two channels simultaneously will get the job done.

I'm interested in audio frequencies, so 20-20000Hz with high integrity is enough. It would be nice to accommodate differential and common mode voltages up to 300V for measuring power supply related voltages without fear of breaking anything, but what I really am interested in are signal level voltages, so <20V is probably enough.

Some alternatives considered so far

1. Dedicated differential amp circuit like INA134 with a separate high impedance input stage with OPA2134 as proposed by the BB data sheet
Good
* Wide enough bandwith
* Reasonable cost
Doubts
* Needs +-18V floating power supply (4x 9V batteries for instance)
* Only for signal voltages -> need overvoltage protection of some sort

2. High quality transformer isolation for the signal (with Lundahl LL1684 for instance)
Good
* Totally passive - no need for power supply
* Could be used for the full input voltage range of the scope
* Bandwith should be adequate
Doubts
* Not cheap even if clearly cheaper than a dedicated differential probe
* How's the trafo behaving when used with very high input and output impedance?

3 Dirty tricks involving the safety grounds
* Not going to even try

4. AC power line isolation transformer for the scope or the unit being measured
* Not willing to try due to compromized safety grounding

5. Biting the bullet - find a used differential probe somewhere
Good
* Works
* Bandwidth beyond what I'll ever need
Doubts
* Very expensive
* Most differential probes are for high voltages - how are they working with small signal voltages?

I'm finding the signal transformer solution (#2) tempting but I'm not sure I got the facts right or missed something totally.

Older scopes (at least) have their signal ground (the BNC connector) connected to the safety ground. I've understood that newer ones may have the signal ground floating. Does this mean that you really do not need a differential probe unless the voltages are too high for the regular probe?

All comments appreciated

-martti-

 

[Samuel Groner]

You are aware that you can flip the phase on a two channel scope and get a differential measurement?

I don't like the transformer idea. It's input impedance is way low and it will show horrible time domain aberrations unless it is a wide-band Jensen properly terminated.

I'm interested in audio frequencies, so 20-20000Hz with high integrity is enough.

So what do you need your scope for? I believe that most of the time I use my scope to check stability and then I want to see up to 100 MHz to be sure.

I'd build an instrumentation amp with JFET opamps in front and a good BJT differential amplifier afterwards. There are tricks to make this work for high CM voltages by subtracting the CM voltage from the input. Check H & H if you have a copy at hand.

Samuel

 

[bcarso]

I pretty much agree with Samuel on the transformer, especially---you really will want to see d.c. offsets, at least optionally, and have small loading as conferred by the high Z buffers ahead of the instrumentation amp. The battery power is not always going to be needed but it is nice at high gains to eliminate one of the common-mode corruption paths.

I've managed for many years without one of the Tek differential probes, but they are very nice for looking at what is really going on in class D stuff. They have clip-on attentuators I believe for higher voltages.

Scopes with battery power have other issues, usually, with the internal switching supplies getting into the measurement somehow. All line-operated scopes I have seen have high conductivity between the safety ground and the BNC shell. Their ability to do differential measurements with two probes are constrained by the limited input range---but you can usually get a reality check on what you see by changing the two channels' gains together---if the A-B result changes in character a lot then you can pretty well say that the higher sensitivity setting is bogus. If the overload recovery is good some parts of the A-B waveform may not be lying at least.

 

[Corvus]

Thanks guys, you are indeed right about the transformer impedance. You can get to 40k input impedance with some step down Jensens and that is not a lot for measurements with high impedance circuits.

Differential measurements with two channels are of course available but the digital scope I have does not support automatic calculations for p-to-p voltages, RMS etc. For sin waves this of course needs only some math.

In addition to check for oscillations I'd like to use the scope for troubleshooting. If two channels are already used for the differential measurement, there's none left for monitoring some other section of the circuit.

Sorry for my ignorance but what is H & H?

regards,

-martti-

 

[Samuel Groner]

Sorry for my ignorance but what is H & H?

www.artofelectronics.com

Samuel

 

[Corvus]

Boy, do I feel stupid. Horowitz & Hill is indeed something I have in my bookshelf. A book that is my primary reference and source of learning. Thanks for reminding.

-martti-