OK, spoke too soon. The 1% resistor is just one place. The differential amp does better with matched resistors Mostly in CMMR. There is going to be a gain error here too.
If you talk about U1B CMMR, datasheet says, at Vcm = 0 V to (V + −1.5 V), its about 65 to 85 dB. Not sure, why you are saying about gain error relating CMMR.
In practice, a CMRR in excess of 80dB to 100dB will be needed for high accuracy measuring system ,
http://www.ntu.edu.sg/home/aschvun/FAQ/DiffAmp.html
This idea is telling that,
To eliminate the effect of the common mode component, we can either
(i) make the input common mode component equal to zero, i.e. make V2 = -V1
such that the average value of the two input signals equal to zero
or
(ii) choose the resistor values of R4 to R7 in such a way that (
Common Mode Gain ) Acm is zero.
If the device is going to be manufactured on a large scale then one of the first things I would want to know is whether the trim would work for all resistors. Some being too high and some being too low. Yep, 0.1% resistors in the differential amplifier would give better CMMR, but is that necessary? Are these differential amp resistors 10%, 5%, 1%, 0.1%, 0.01%?
What maximum resistor value do you think it can measure? Can you give me an idea how could I implement a wide range of differential amp resistor?
At low ohms, you really want a KELVIN type measurement. measure the current through the unknown and the voltage across the unknown. This meter won;t do that.
It also doesn't use a 4-terminal resistor. e.g.
http://www.vishay.com/docs/30217/cpsl.pdf
Good point, but I dont think its a temperature dependent measurement. I saw other measurement system like "bio impedance " measurement.
People are using, smart MCU or DSP, generating high accurate constant current source. I started a PhD there, but did not continue.
LT,TI are offering many KIT to measure accurately, I have used those in 2015. Here, just I want to use simple circuit those components I have in my stock.
Is the ZERO adjust a trim or is it a user adjustable part of the meter. Call it lead compensation?
Yes, well said.
How low of a resistance do you expect to measure? What are the limitations? What are the temperature effects?
Its also arise in my mind, not sure how much I can measure in this design.
Should some of these resistors be in a network package where the temperature coefficient is better?
There are 2, 4, 5 and 6 wire ohm measurements. See:
http://www.tek.com/document/whitepa...measurements-using-6-wire-ohms-measurement-te
What's the low end performance? With a single supply, it's difficult to null Vos completely and it won't go to zero, but it will get close.
it's not meant to be a 5 digit milli-ohmeter.
Yes, these points are important, but as long as this design is poor we cant think all facilities.
==
One OP-amp is used as a buffer. What can I say, They are dual OP-amp packages. You can elect not to use the buffer and configure it as one amplifying ground, just so it doesn't oscillate and warmup the IC package. The buffer can negative affect the temperature performance. but is the temperature performance necessary.
Are you indicating not to use, U2B? Not need to set gain 1? not to shift from high output to low input impedance ?
First order approximation, yes. What if Vcc changes? Vcc can easily change with temperature. If the reference of the A/D is Vcc. then those effects go away, but a voltmeter can get the wrong answer.
Here
http://cds.linear.com/docs/en/datasheet/1009ff.pdf is a reference. Here
http://www.analog.com/en/design-cen...ge-reference-selection-evaluation-wizard.html are other references.
Any particular design?