I am measuring the CMRR of the LT1167 inamp on the board. The source imbalance is about 1 Ohm, the test frequency is 50 Hz, gain = 32, room temperature. I am getting over 130dB CMRR reliably, however some parts reach 150dB !!! How could it be that accurate?
Vladimir Vassilevsky DSP and Mixed Signal Design Consultant
At higher gain levels, topologies like that tend to have better CMRR numbers. The input transistors are well isolated from seeing the rails. I imagine that the current sources are made with nearly identical transistors. The CMRR is most likely largely set by the second stage amplifiers. These would be matched to 100dB which is still very impressive.
As you pass a few KHz, the stray capacitances start to become the determining factor. Keeping all copper away from the gain setting resistor and making sure that the distances to the ground and trace lengths are all nearly identical lets you have a product with a 120dB spec at 4KHz.
That's what made me wonder. I checked and rechecked everything; the setup looks all right.
On the actual board, at the higher frequencies the numbers seem to be determined by the direct feedthrough of the test signal rather then by CMRR. It is still about 120dB, as you mentioned.
BTW, there is the analog switch to set the gain of the stage; and the layout is not very symmetrical. The design goal was > 90dB CMRR; now I have to explain to the customer why the numbers are so good.
Vladimir Vassilevsky DSP and Mixed Signal Design Consultant
m
I expect that what you have measured is the truth. The CMRR really is that good. Now it is time to figure out what other spec it fails to meet. Does it distort a lot or oscillate etc. After all, the electronics gods do have a sense of humor.
In the cases I have worked on, another effect comes in also. The impedance you have to put in series to protect against ESD and the impedance to ground form a voltage divider that is never perfectly matched. I have resorted to things like:
Test for font: iiiii These should mmmmm line up
Ascii art:
-----[D1]------+---[D2]--- GND ! ! ! !\\ ! +----! >--/\\/\\-- ! !/U1 ! ---/\\/\\-----+-----!+\\ rest of dif amp
The unity gain buffer U1 drives the end of D1 and reduces the effect of its capacitance by a factor of about 100. A low cost high input impedance amplifier will do the job because ideally, it has no effect on the signal path.
I never explain it when I luck out. I just say "It meets the requirement" and let the boss believe what he will. I do try to figure it out so I can repeat the act later on demand.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.