In the datasheet rev D, CMRR is specified as 155dB typical @ 5V and 160dB for 30V operation, yet figure 30 shows it to be only 110dB from 10Hz to 100Hz, declining thereafter. Presumably the graph is in error?
Is CMRR normally specified at DC?
The CMRR performance in the specs table is DC CMRR. If you are going to extend figure 30 down to DC using ideal setup, it should yield to the same DC CMRR performance. Therefore I would say that both measurements are correct.
Depending on the common mode voltage (CMV) present on your circuit, you can refer to specs table if it is DC CMV or in figure 30 if it is AC CMV to compute for its error contribution on the system
I hope this helps.
Hi, Emman, thanks for the prompt response.
I'm sorry but I find that explanation very hard to accept. It seems to me that it would be a very unusual op-amp where the CMRR drops by 50dB from DC to 10Hz - that is a huge amount, especially given that isn't a very slow micro-power amplifier. It would be even stranger for the CMRR to then level out and remain steady up to nearly 100Hz before it starts to decline again.
I have looked at various datasheets since and whilst most seem to be consistent, the same problem seems to apply to the ADA4528 (160dB DC, 110dB @ 100Hz in fig. 35), AD8538 (150dB DC, 108dB @ 100Hz, fig. 15)
The AD8531 seems to suffer the opposite problem - DC CMMR spec is 47dB typical whereas fig. 22 shows it to be 86dB from 1kHz to 10kHz before it starts to drop off. Which is the correct figure?
Just a few comments on CMRR. Very difficult to measure. See:
For DC CMRR, we move the supplies.
For AC CMRR, there are several ways, the most common is to put the op amp in the classic
four resistor difference amp circuit. But you need super matching to get more than 80 dB.
The R and input C will affect the AC CMRR at higher frequencies.
Because different op amps were done by different application or product or design engineers
in different groups over different time periods, you can't really compare two op amps without
knowing how it was measured.
Spec table min/max numbers are determined by mean/sd of several hundred parts along with
tester repeatibility. Graphs are usually one or two parts from the frist fab run at 25C.
Retrieving data ...