The difference amplifier circuit seemed to work well; it was nice and linear in fact. But there was something not quite right: There would be attenuation at certain frequencies. Not very much, but troublesome nonetheless. It had something to do with balancing the resistors in the compensation loop, Bill Klein concluded. Watch him go to work on "A Call We Took Last Week."
My caller last week had a design on his bench that seemed to work, but he was nervous. It seemed to be an obvious solution but it did not appear in any of the applications sections of the data sheets. There were no application notes that discussed this solution. Why? What was he missing? Any offer that looks too good to be true usually is just that.
He started out with one of our difference amplifiers and modified it with input resistors and capacitors to get a low frequency roll off. His schematic diagram is shown here.
Click to Enlarge
He had taken care to match the added resistors and capacitors. The prototype seemed to work. There was one small issue in that the gain was not quite as accurate as expected. He had based the calculation on the internal resistor values given in the data sheet.
I pointed out that he had missed a note in the data sheet that stated the absolute accuracy of the resistors was ±25 percent. The gain accuracy and common-mode rejection specifications were obtained by laser trimming the resistor ratios but not the absolute values.
We continued to chat and he arrived at the next stage in his design. He could add two more resistors as shown here that would make up for the difference in absolute value. By proper selection the ratios could be set equal.
Click to Enlarge
With RA = RB and RC = RD the gain accuracy would be restored as well as the CMRR. While this could be done at DC the balance in the filter action cannot be maintained. The balanced filter attenuation between the two sides required that RA = RB. That condition is not satisfied now.
To allow a differential input filter a full instrumentation amplifier is required. With the high input impedance the added input resistors cause a minimum of disruption. It is necessary to match the input elements however. If the input filters are not matched then the attenuation at any frequency will be different. An AC common-mode signal will be attenuated and phase shifted differently causing a differential signal to appear at the amplifier input. Rather than being rejected, this signal will be amplified.
Thanks for reading…………Comments Please.
And be sure to copy, you-know, firstname.lastname@example.org
*Applications guru Bill Klein in host the Analog eLab web seminars sponsored by Texas Instruments and CMP Media.