I have a customer using the 2021 in place of an old LM108A. Circuit has a gain of 30. The 2021 in circuit is outputting 30mv when the inputs are zero during a test. The 1mv is on the non-inverting input. The rails are at +/-13V. Measurement inputs are on the low side of a servo driving H bridge. They used a 0.15ohm to ground for current measurement. 3K input R and 102K feedback, with no compensation on the 2021. He wants to know why the 2021 has a 1mV Voffset on pin 3? He can use a 5k compensation on pin 1 to –Vcc and pull the pin 6 output down, but doesn't think it should be needed. Most of the time the op amp should be sitting at zero as it just moves a motor occasionally.
My first thought was that the way they were measuing the devices was causing the issue, but they assured me that wasn't the case. I also asked if something downstream was putting voltage back on the op amp. They told me Pin 6 output goes to a mux thru 110K then A/D for use to control the torque used on the servo. The inputs are seeing the FETs switch at 488Hz. The test engages the servo and provides a 20ohm and 1ohm load. The output is set to zero. Measurements for the current are taken across the 1ohm load and show 95uV. Measurement on Pin 6 output shows 34mV.
They use this part in 3 locations that measure voltage and three that measure current, and they all show the same issue. They have bread boarded the part, and it appears to work well. Shorting the inputs gives rail voltage out, although I would expect a zero output. When they lift pin 3 lead on the bread board the output goes to 30mv and 1mv appears on pin 3. When they ground the .15ohm input on pin 3 the part seems to work fine. When they ground pin 1 thru .15ohms and lift open pin 3 to simulate a zero command they get 2mV on pin 3 and 60mV output due to my gain of 30.
Is there some leakage? Anyone know what could be causing this.
Thanks
