Converting voltage across terminals to voltage from ground

[Sigmoid]

Hey,

I've come across a very interesting problem. I'm sure I'm overlooking something and the solution is quite simple, but I'm really intrigued right now.

So say I want to measure current somewhere in a circuit, and act on it in a microcontroller, or use it as an analog feedback signal. Now this somewhere is anywhere but close to the ground rail, maybe it's an audio output or a driver or something. The load is either non-ohmic or of unknown resistance.
I put a resistor with a small known, precise value in the signal path, or create a measurement shunt if necessary, and get the voltage between the two terminals of our measurement resistor.

Now let's say for example's sake that I'm using a .1% 10Ohm resistor, and I have a 20mA current across the load. The positive rail is 5V from ground. The voltage at the hot terminal of the resistor is 2.1 Volts (right now - otherwise it's assumed to be variable and unpredictable, and potentially dependent on whatever control actions I'll perform based on this measurement), and due to the voltage drop at 20mA, voltage at the cold terminal is 1.9V.

What I am struggling to find out is how would I go ahead and create a voltage follower configuration with multi-megohm impedance on both inputs that would take these two voltages, and output a .2 Volts (for the 20mA) for me to feed into the ADC and drive some control logic off of.

[ejeffrey]

This is what an instrumentation amplifier is designed for.  There are lots of ICs that will do this, like the INA111.

[Paul Price]

Use an instrumentation amp.

[Sigmoid]

Lol, of course. "There's an IC for that!" XD

[planet12]

Linear Technology's AN105 app note would be a good read for you on this subject - title is "Current Sense Circuit Collection" and it covers many techniques for getting a high side sense ground referenced, including various specialist IC and op-amp based options.

http://cds.linear.com/docs/en/application-note/an105.pdf

[kizzap]

The other popular option is to use a differential amplifier

-kizzap

Edit:

Huh...just looked a little closer at the instrumentation amp. It literally is a differential amplifier with input buffers? I'm not to sure on the benefits.

Would someone take the time to explain the benefit of the instrumentation amplifier over the differential amplifier?

[baljemmett]

Would someone take the time to explain the benefit of the instrumentation amplifier over the differential amplifier?

Not really my field, but I believe the answer is "input impedance" --  and for an integrated device, better matching of the amp(s) and on-chip resistors.

[Marco]

I'm no expert but I have read enough experts use the terms difference amplifier and differential amplifier, unlike instrumentation amplifier those terms aren't very meaningful to refer to a specific type of circuit topology.

If with differential amplifier you mean a single opamp used as a differential amplifier, then I think input impedance is one of the problems. Because one of the inputs is connected to the feedback path you won't be able to get close to the input impedance of the opamp itself, which an instrumentation amplifier can do.

[ejeffrey]

Instrumentation amplifiers have high input impedance, which is important for achieving high common mode rejection when you have lead resistance or source impedance mismatch.  The standard 3 opamp circuit has the gain set by a single resistor, so it is much easier to change the gain without screwing up the circuit.  An IC based instrumentation amp will have the critical resistors laser trimmed on die for better matching.

[kizzap]

Well I guess the important question to ask is, which would be better for high-side pre-voltage regulation monitoring? Would the input impedance really play a big deal in this case, especially if you have excess current to use?

-kizzap