Re: Differential voltage probes

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: "Jerry Chamkis by way of Terry Fritz <teslalist-at-qwest-dot-net>" <jchamkis-at-bga-dot-com>

On Saturday 28 June 2003 02:25 pm, Tesla list wrote:
 > Original poster: "Terry Fritz" <teslalist-at-qwest-dot-net>
 >

 > >*A trimpot on one of the 1K resistors for CMRR trim (maybe T-network type
 > >thing)
 >
 > I would like to not have any adjustments.  Have to really try it and see if
 > it needs any additional parts to make it work right.

In that case, it will be well worth your while to match the resistances in the
two arms as closely as you reasonably can.  It's a good way to go- pots are
expensive, unstable, bulky and capacitive.
 >
 > >*A variable capacitor somewhere for HF CMRR trim (maybe put a small 10pF
 > >fixed capacitor across one 1K resistor and a 2-20pF trimmer capacitor
 > >across the other)
 >
 > A few caps may indeed be needed.  have to try and see.

This will make a huge difference in HF rejection and really should be done as
suggested.  Use an NPO ceramic for the fixed one and Mouser has tiny cheap
ceramic trimmers that are pretty good  (Digi-Key too).
 >
 > >*Metal film resistors because carbon ones have a voltage coefficient of
 > >resistance.
 >
 > Should not matter too much.  I was thinking the probe would be like 5% if
 > one selected the resistors a bit.  Even at 10% it would still do
 > practically everything we need.

The whole point here is to reject the common-mode signal.  It works by
canceling the possibly huge signals coming in identically on the two probes.
The degree of matching has a big influence on the CMRR.  The difference
between 100 dB CMRR and 60 dB- which I wouldn't want to guarantee with 5%
parts,  is 100x more of the undesired interference in your trace.  Metal film
resistors are much quieter and more stable. The probe is a great idea, a very
useful tool, and you can build a really good one with just a little more
effort.

The opto-isolator is a good way to go too for galvanic isolation- but of
course it's a little more trouble.  At that point you might consider using a
much better op-amp  like an OPA-627 and you'd have a very nice piece of test
equipment.  Then you could sell a bunch of them on ebay and build a REALLY
big coil...   :-))

Jerry
 >
 > >I have all the parts lying around and I'll give it a try
 >
 > Cool!  I am not sure I will have time this weekend.
 >
 > >I just remembered hewlett-packard (I think) make a 'linear optocoupler'
 > >which is like an ordinary optocoupler but with 1 LED and 2 isolated
 > >photodiodes. You use one photodiode for feedback stabilization and the
 > >other for output.
 > >
 > >You can use one of these plus two op-amps to make an opto-isolated scope
 > >probe with about 1MHz bandwidth. This approach would probably give better
 > >performance when looking at small signals like gate voltages etc.
 > >
 > >Agilent part# HCNR200
 > >http://literature.agilent-dot-com/litweb/pdf/5988-4104EN.pdf
 >
 > Sounds like it is getting too complicated ;-))  If we could just figure out
 > the "basics", then folks could always add fancy stuff as they please.  More
 > worried about phase shift, frequency response, safety... right now.  Steve
 > Ward blew a channel on his scope this morning :-((  A probe like this would
 > have easily prevented that...  "Normal" scope probes and stuff just do not
 > like to work with voltages over about 100V.  In the case of AC circuits,
 > the grounding and all can easily blow things up unless you are super
 > careful.  A simple probe like this would fix all those problems.  In most
 > cases, extreme accuracy and bandwidth are simply not needed.  Steve was
 > just trying to check AC voltage but the probe got left in the 1X
 > position...  Hopefully just a few blown scale resistors...
 >
 > A very simple device like this would allow people to check AC stuff and
 > <1000V stuff very easily and safely.  Floating scope cases and using low
 > signal level probes is just too risky when a simple thing like this would
 > solve all that.
 >
 > Cheers,
 >
 >          Terry
 >
 > >Steve C.

-- 
Jerry Chamkis
jchamkis-at-bga-dot-com