Hi I'm Alan
I am using the LM837 for signal processing, for amplification and Sallen Key highpass and lowpass filters. I have found that a low level oscillation occurrs in some of the amplifiers, about 200mV 2MHz. I can get rid of it with a small capacitor across the feedback resistor from the ouput to the negative input, 10pF is sufficient, so my operation is not affected. However this should not be necessary according to the data sheet. Also the only amplifiers where I have encountered this are the 'A' amplifiers on the chip. Has anyone had a similar experience?
Thanks
Alan
I am using the LM837 for signal processing, for amplification and Sallen Key highpass and lowpass filters. I have found that a low level oscillation occurrs in some of the amplifiers, about 200mV 2MHz. I can get rid of it with a small capacitor across the feedback resistor from the ouput to the negative input, 10pF is sufficient, so my operation is not affected. However this should not be necessary according to the data sheet. Also the only amplifiers where I have encountered this are the 'A' amplifiers on the chip. Has anyone had a similar experience?
Thanks
Alan
I do not know, the circuit is on a prototype pcb and not easy to change. Since we can get operational with the 10pF capacitor additions I am reluctant to make alterations at this point. Certainly according to the data sheet the LM837 should be unity gain stable.
Alan
Alan
We need a detailed schematic.
If the feedback resistor is larger than a few kOhm, you will need a feedback capacitor for stability: http://www.ti.com/lit/an/slyt087/slyt087.pdf
Samuel
If the feedback resistor is larger than a few kOhm, you will need a feedback capacitor for stability: http://www.ti.com/lit/an/slyt087/slyt087.pdf
Samuel
Allegedly, but it is fast (GBW = 25MHz), so will probably be fussy about resistor values, layout, decoupling etc. TI's description and basic specs are here: Precision Amplifier - Low Noise - LM837 - TI.com
just had a look.
~40degrees of phase margin at gain0db crossing.
Bad overshoot and ripple with gain set to 1 on the inverting small signal plot. No comparable non-inverting to be seen.
~40degrees of phase margin at gain0db crossing.
Bad overshoot and ripple with gain set to 1 on the inverting small signal plot. No comparable non-inverting to be seen.
I recall an (experienced) analog designer colleague at a pro-audio company I worked for years ago saying he couldn't get the LM837 reliably stable, so he didn't use it.
There's 40 degrees of phase margin at the UGF. If you are hanging a capacitive load onto the output, you may run into problems.
Have you decoupled at the PSU pins?
As noted above, if the feedback resistors are high in value, you are also likely to have issues, requiring some extra comp. Post a circuit then you will get better advice. Just slapping a cap across the feedback resistor without understanding he exact mechanism causing your problem only get you a B.
Have you decoupled at the PSU pins?
As noted above, if the feedback resistors are high in value, you are also likely to have issues, requiring some extra comp. Post a circuit then you will get better advice. Just slapping a cap across the feedback resistor without understanding he exact mechanism causing your problem only get you a B.
Last edited:
As was pointed out, there's not a lot of phase margin in the part. So, if you're using an offset current cancelling resistor in the feedback, it creates a pole at the inverting input feedback node that pushes you over the edge.
I guess you're doing so because the bias current is kind of high for this part, and you're trying to minimize offset voltage due to bias current. If your resistors are low enough that this isn't a concern, and if I have guessed your configuration correctly, then you could just short out the resistor that goes from the output to the inverting input.
I guess you're doing so because the bias current is kind of high for this part, and you're trying to minimize offset voltage due to bias current. If your resistors are low enough that this isn't a concern, and if I have guessed your configuration correctly, then you could just short out the resistor that goes from the output to the inverting input.
Yes, in one case the feedback resistor is high (36K) in a simple inverting amplidfier and I was I was not surprised at the necessity of a capacitor accross the feedback resistor to provide some phase advance in this case. In the other case the configuration is a standard Sallen Key high pass filter fc of 3.4kHz in which the damping resistors are 3k48 (o/p to inverting i/p) and 51k1 (inverting i/p to 0V), I was a bit surprised to encounter the problem here. I think I do understand the mechanism (phase lag due to a high value feedback resistor) but to find this with a 3k48 feedback resistor is a bit of a puzzle. I am not using any offset current cancelling since the response is not required to extend down to dc.
What's your layout like? If you happen to have quite a bit of capacitive coupling at the inverting i/p to ground, this will exacerbate the problem.
- Status
- Not open for further replies.