common mode rejection ratio

If I read what you say correctly, the voltage at both ends of the sense resistor can be 15 Volts; i.e. as much as 15 volts of common mode with respect to ground. Is that correct? If the CMR is 40dB or one part per 100, the offset refered to the input is as much as 15/100 or 150mV. With a gain of 10, 1.5 volts of offset can occur on the output. It has nothing to do with the variation except to say that a steady, known offset can be nulled out. If you have that much common mode voltage and that poor CMR, you're in trouble. Furthermore, 15 volts of CM is beyond what most op-amps can handle. Bob

I had something like this in mind, common mode error caused by resistor error:

12-15v o---+----/\\/\\/\\----+------------o | Rs=.01 | | | R3 R1 | | | +---R2---, | | |\\ | | '--|-\\ | | | >--+---Vo +-----------------|+/ | |/ | R4 | | o----+---------------------------o

for (R2/R1)=(R4/R3) , Vo=RsIR2/R1

so for R1=R3=1K and R2=R4=10K and I=10A, Vo=1v

Now let's say R1, R3 and R4 are exact but you have an error of 10% in R2.

How much error does that result in at vo...

Now I get 136 mV error at vo when the input is at 15 volt

is that correct?

line

100,

gain

nulled

in

handle.

You know I really have a brain freeze when it comes to these text based schematics I see from time to time. I can't tell one thing from another, for example, what is all of that gibberish between R2 and R4??? I just don't get it, sorry. How about supplying a simple net list like you might supply to a spice program, I can read that. Or, post a schematic to alt.binaries.schematics.electronic. I'll be glad to comment. Thanks Bob

You have to set your newsreader to a fixed width (per character) font, like courier (or copy and past the post into Notepad and change the font, there). Then the characters line up into a drawn schematic.

for

get

to a

Sorry, I have more important things to do than 'eff around with goofy graphics. What's wrong with a net list?

Let me get this straight: Rather than you making a single click on a newsreader option (fixed font, in the upper right corner of the display in Google Groups) you want everyone in the world to switch from a graphic representation of schematics, that is archivable in Google, to a net list that must either be drawn into a circuit, manually, or pasted to a schematic drawing program, before we can see what the circuit looks like. Is that right?

Good luck with that.

By the way, if a single click constitutes 'effing around, I pity any woman you 'eff. ;-)

That would be the error for 1% deviation in R2. For 10%, you'd get

1.36V.

-- john

Looks like you're right. CMRR=20log(R1+R2)/(R1 X error) so the error in Vo is common mode voltage X (R1 X error)/(R1+R2)

15 X 1K X 10% / 11K = 1.36 v I keep dropping factors of ten.

No, that's still wrong. If that expression for CMRR is correct, with R1=R3=1k and R2=R4=10k for 10% resistor error CMRR=40.8 dB or a ratio of 110:1 I'm still getting 136 mV

Look at it in DC terms for a moment. Let the differential voltage be zero, so that the output should be zero, and the common mode voltage be

1. The resistors are the values you set, except R2 is 11 kohm (+10%). Then, the voltage directly on the amplifier inputs is set by R4/(R3+R4)

• 15 V = 13.636 V. The drop across R1 is then 15 - 13.636 = 1.364 V. The current through R1 and so through R2 is then 1.364 mA. But the voltage drop across R2 is 1.364 mA * 11k = 15V. The output voltage is then the NI input voltage minus 15V, or -1.364V. If R2 is 10% low instead, you get a drop of 1.364 mA * 9k = 12.2727 V, which, subtracted from the NI voltage, gives +1.363 V. This is the actual error voltage for a 10% deviation in R2 with 15V CM input, so the CMR is

20*log(15/1.363) = 20.83 db.

The CMRR is different from the CMR. The latter is common mode rejection, the ratio of output voltage due to the CM input that caused it, usually expressed as an unsigned db value. The CMRR is that value in db added to the db value of the gain of the circuit. Since your circuit's gain is about 10 (as modified by the resistor tolerance), the CMRR is actually about 20.8 + 20 = 40.8 db; but the CMR is what you want to use to find the size of the error output for a given CM input.

If you are looking for the comparison of a small change in signal input to the same size change in CM input, then the CMRR is the one to use. If you are looking for the size of output due to a CM input, then it's CMR.

-- john

Thanks for the explanation.

Thank you for pointing that out. I must admit that I looked at those schematics with a question and did not know how to view them. Now I do, thanks. There's no need for any pitty.

Glad to hear that.

The standard ways to presents circuits for discussion, around here, are:

1. If a very few components, just use words.
2. If not too complicated and no signal waveforms are needed, use an ascii drawing, like this one. By the way, someone has designed a very handy drawing program that does most of the work, at:
   formatting link
3. If details of operation are being discussed, the text *.ASC schematic file from the free spice simulator, LTspice, are often posted. This allows anyone to paste that snip of text into Notepad and save it with an ASC suffix, and then they can open it in LTspice and not only see the schematic, but see it operate.
   formatting link
   Far superior to net lists.