[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Differential voltage probes 1



Original poster: "Terry Fritz" <teslalist-at-qwest-dot-net>

Hi Richard,

At 03:41 PM 6/28/2003 -0400, you wrote:
>Terry,
>
>Excellent project.  Maybe later you can adapt it to measure differential
>electrostatic TC fields. They're quite a bit different than computer
>simulated ES fields.

Indeed.  We could easily make the input impedance very high as I am sure 
you are well aware.  This article by Niels Jonassen describes that can of 
worms one opens with such things ;-))

www.ce-mag-dot-com/archive/01/11/mrstatic.html

Once you place a probe in an E-field, it just screws everything up...


>I have designed and built dozens of these solid state differential
>amplifiers over the past several years.  I have a few suggestions.
>
>         1.      While the TL082 j-fet op amp is a venerable old veteran 
> and the over
>riding goal is to keep it simple, there are several later generations of
>cheap op amps that will produce better and more accurate differential
>amplifiers.  In your TL082 configuration there are several problems that
>need to be addressed.  On these j-fet high input impedance op amps it's
>almost impossible to make the battery +/- supply rails discharge evenly.
>The batteries will discharge at different rates.  This makes a big
>difference in biasing the op amp.  Even at a small 0.5 volt supply
>difference there is often a high input bias.  A rock solid "virtual ground"
>at zero volts has to be established.  This will get rid of this annoying
>supply bias problem.  The TL2426 "Rail Splitter" precision virtual ground
>can be obtained from Digikey for ~$1.50.  Alternately, you can build a
>virtual ground rail splitter with a single op amp.  It's just an op amp
>follower with about 100 ma output.  Put a 100k pot across the +/- rails.
>The pot wiper goes to the non-inverting input and tie the output back to
>the inverting input.  That's it.  Adjusting the pot voltage divider sets
>the output virtual ground.  Usually we want it to be 50-50 of the total
>rail supply.  Since the pot voltage divider is a set ratio, the voltage of
>the virtual ground always remains at this set ratio - even as the batteries
>discharge.
>
>(While it's not good practice, I have found that if there is a lot of
>supply bias, a common voltage signal can be applied to both the
>differential inputs and the virtual ground pot can be adjusted a little so
>the virtual ground output is zero volts.  The virtual ground won't quite be
>50-50, but it sure gets rid of the supply voltage bias.)

I was not going to worry much about CMRR or offset bias in such a simple 
design :o)  However check this out:

http://hot-streamer-dot-com/temp/TekP5205-01.gif

The impedance of the Tek5205 drops drastically from 4Meg to 200 ohms!! with 
frequency.  This manual is nice in that it has the calibration procedures 
in it to tell me how such things are tested...  There are no adjustments on 
the probe.  It either works or it does not.


>         2. While TL082 input impedance is 10^12 ohms, you've essentially 
> shunted
>its impedance to ground at 1k.  It's unlikely though that the non-inverting
>input impedance is the same as the 1k inverting input impedance.  Due to
>output impedance, transorb and output measuring device, the non-inverting
>impedance may differ greatly from the 1 k inverting impedance.  This
>likewise introduces input bias.  A single differential op amp design isn't
>the best.  The solution is to have two high input impedances from two
>individual op amps as separate front ends.  These two drive a third
>differential op amp.  If you prefer TL j-fets I suggest TL084 (quad) op
>amps.  Another advantage wound be to choose another op amp with built in
>shunting diodes across the inputs.  I haven't done it, but maybe simply
>putting a transorb across the inputs would work.  This would depend on the
>impedance of the nonconducting transorb though.  I have done quite a bit of
>work with mosfet op amps in the past and for the most part they seem a
>little easier to use than the j-fet versions.   There are a lot of high
>impedance differential precision instrument op amps with most of these
>features built in.  Digikey has them for ~$5.00, give or take.  Finally, I
>have been using chopper op amps almost exclusively for a year now.  They
>are cheap, accurate and do away with almost all the biasing problems
>associated with high input impedance op amps.

I am thinking we need two probe designs.  One super simple one like I 
suggested.  It would be great for many things like Steve's test this 
morning (Steve says the scope just lost a few ranges on one channel, so it 
should only be minor resistor damage :-))  A very simple probe would 
eliminate floating scopes and clipping the ground lead to the hot line 
errors so common when scoping AC stuff...  Very easy for anyone to make.

But maybe we also need a higher performance design too?  We could maybe 
spend a few more bucks on a true higher bandwidth design that had the CMRR, 
offsets, opto-isolation, good impedances and all that fixed for more 
serious work.  PCBs and parts kits re no problem for those that want 
something "better".  Probably be surface mount...

Could you suggest "good" op-amps to use?  I have not worked with op-amps 
recently and I am not sure what the modern good ones are these days.  I 
could go then look them up and work on it.  The supply rails and all could 
just be voltage regulated.  Given the prices of "real" diff probes, we have 
a lot of room to work with!!

BTW - The TekProbe power supply #1103 (low current +-15 volts) "only" runs 
$1550!!  I assume, at that price, it is fusion powered :o))))

Cheers,

         Terry