At 04:07 AM 7/29/01 -0700, you wrote:
>I have an application where I need to monitor and
>digitize the current waveform output of an audio power
>amplifier into a subwoofer (speaker).
>Previously, I used a hall-effect based sensor, but for
>cost purposes (consumer product), it is impractical.
>So instead, I have decided to use a 1 milli-ohm
>current sense resistor.

Nothing like making things difficult...

>The problem now is creating an
>electrical interface to convert the instantaneous
>current (0 - 50 amps) into a proportional
>instantaneous voltage (0-2.5v). Since the audio output
>is not referrenced to ground, some type of
>differential circuit must be used. But although the
>voltage across the resistor will always be small, the
>voltage at each of the 2 resistor taps, with respect
>to ground can swing as high as much as +/-100 volts.

You've created an extrememly difficult problem for
yourself. While this is possible (for example, isolate
the circuitry to the front end, build it inside of a
Faraday cage that is connected to one side of the
resistor and use transformers and/or optocouplers to
transfer power and signal across the barrier), it
may not be the best way to tackle this.

>Unfortunately most instrumentation op-amps that can
>handle such a high common-mode input voltage, are also
>quite expensive. In addition, a textbook differential
>op-amp circuit doesn't provide the needed accuracy,
>even with .1% resistors, since any component
>tollerance is multiplied by the gain of the circuit
>(50). I though it would be easy finding an interface
>circuit that would solve my problem, especially since
>almost every DMM must use something similar for DC
>current measurement.

Most of them float the front end (the whole meter on
cheap ones) at the signal voltage. Ones that don't tend
to have the box within a box arrangement if they have
decent AC performance (as does my nice 6 digit DMM).
You can handle the higher voltage by dividing down the
voltage, but as you have found, the resistor ratios
(and parasitic capacitance ratios..) have to be within
very narrow limits. If you want accuracy of +/-500mA,
and you divide the inputs down to +/-10V, that is
an accuracy of 50 microvolts DC (you could AC-couple
it to get rid of this), and 5ppm in the
resistor ratios (good luck). The B-B IN148 is designed
for similar applications and is probably close to
state-of-art for monolithic designs
http://www-s.ti.com/sc/psheets/sbos123/sbos123.pdf
but CMRR is only 70dB min, and you need more like
106dB just to get a lousy half an amp of accuracy.
Other commercial inamps have the CMRR but not the
input range, and are costly ($6/1K), and you'd have
to add some fancy (eg. $$ Caddock) resistor sets and
prolly some HF compensation to get the input range
and CMRR.

>But I have had no luck. I would
>appreciate some advice, even better a schematic.

Maybe you could use a CT  (limited high frequency
performance due to the typical tape-wound core) or an
active current balancing scheme using a ferrite CT
and a feedback winding. If you only need to work
up to a few hundred Hz (subwoofer), then the CT
may be the easiest way. Only a couple of $USD.

Best regards,
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Spehro Pefhany --"it's the network..."            "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com
Contributions invited->The AVR-gcc FAQ is at: http://www.bluecollarlinux.com
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listserv@mitvma.mit.edu with SET PICList DIGEST in the body