Is differential amplifier badly inaccurate in this circuit?

Hello,

I need the following differential amplifier to provide a gain of 18 on the voltage across the series current sense resistor.
(the diff amp resistors are 1K’s and 18K’s)

In steady state, the voltage across the (0.0876R) current sense resistor will be 0.1V.

Schematic with differential amplifier….
https://i50.tinypic.com/15ciz4i.jpg

The opamp’s supply voltage is 10V, and the voltage source at the left hand side of the schematic is 3.6V DC….(with battery internal resistance of 0.06R)


Do you think this will be badly inaccurate?

Will I suffer from common mode effects

Do I need to take measures to fight the input bias current of the differential ampifier

The Opamp in the schematic is LT1006, but in the real circuit I will uise the cheap LM358.

LM358 DATASHEET
https://www.ti.com/lit/ds/symlink/lm358.pdf

The accuracy is mainly determined by the resistor tolerances and the circuit offset voltage.

The common-mode rejection is determined by the resistor tolerance. Thus for 1% resistors the worst-case CM rejection is 40dB. For 3.6V and a gain of 18 that would give a worst-case zero current offset of 650mV at the output, which is about 36% of the 1.8V signal output. You could put a pot in series with one of the resistors to adjust this offset to zero (say a 1k pot in series with a 17.5k resistor).

The input bias current is not a problem as long as the resistor values are kept low (20k ohm or less).

If you don't want have to adjust for offset then you need to use more accurate resistors, or use a instrumentation amp which has a much higher common mode rejection, or use a high-side dedicated current-monitor IC. These are made by several IC vendors.

Crutschow

For 3.6V and a gain of 18 that would give a worst-case zero current offset of 650mV at the output, which is about 36% of the 1.8V signal output

Click to expand...

....Thanks, thats a dreadful error......i can't tolerate that......also, adjusting with pots isn't a possibility. An instumentation opamp sounds too expensive. High Side Current moniotrs dont tend to be able to measure currernts on a 3V6 rail, its just too low for them.

This is bad news, i doubt the project can handle the cost of an instrumentation opamp, have you an idea of the price of the cheapest ones?

Digikey shows several that start at less the $2. Search for amplifiers and then narrow that to instrumentation types.

As an alternate, can you put the shunt resistor in the ground side of the circuit? That would eliminate the common-mode problem.

Thanks but unfortunately i must keep the ground clear.

Its strange that the simulation, which uses the LT1006 opamp, doesnt show any of these inaccuracies....is this a simulator bug?

$2 is too expensive unless thats the single part cost.

i would be seeking something around $0.3 for 2000 pieces

Its strange that the simulation, which uses the LT1006 opamp, doesnt show any of these inaccuracies....is this a simulator bug?

Click to expand...

That's true as long as you don't introduce resistor tolerances to your simulation. But you can't blame the simulator for lack of realistim in your simulation setup.

A high side current shunt measurement either needs precision resistors (0.1% tolerance and low t.c.) or a dedicated high side current sense amplifier that uses a transistor current source to achieve high comon mode rejection.

..but the inaccuracy quoted was 36%...surely using 1% resistors can't result in a 36% error?

surely using 1% resistors can't result in a 36% error?

Click to expand...

1% of the common mode signal can easily make 36 % of the differential signal (or even more).

You can perform a sensitivity analysis. Vary each resistor by 1 % and record the output signal change.

That kind of accuracy requirement for resistors surely makes the differential opamp all but useless?.....its just not practical to have to use 0.1% resistors in a cheap application.

......there must be some reasonable way of making this workable?

....if i put the sense resistor with one terminal connected to ground...then that makes it all ok and only 1% resistors are OK to use?


Is it the case that the only way to make the setup workable (that in the top schematic) is to to use an instrumentation op amp?