Dear community,

I built a circuit to switch gain of a 20kHz to 40kHz signal between 1, 10 and 100 with an TL082 and two times two 2N7002 switches [1]. I got horrible ringing / osccilations of about 400kHz. reducing Rx_1 and Rx_amp_out1 to 10kOhm and adding a 100pF cap parallel to Rx_amp_out1 did only slightly improve the behavior. Simulating in LTspice [2] I found that my original circuit phase margin was too low but with the 100pF cap should have been sufficient.
Any guidance who do debug / improve?
I already ordered a PGA103 which I hope makes the job for me but maybe someone can guiode me how to fix my circuit

BR,
BastelBaus

[1]

[2]

 

Could it be the drain capacitance?

 

Additionally.... I don't think the TL08x opamps have the gain-bandwidth product to amplify a 40 Khz signal 100 times.

 

Better would be to use CMOS analog switch, such as the quad CD4066, since it has much lower capacitance than the 2N7002.
(That CD40066 maximum voltage is 20V so the sum of Vdda+and Vdda- must be less than that, otherwise you would need a higher volage switch.)

To also minimize the effect of the switch on-resistance, you can use the non-inverting gain configuration (below), since now the gain resistor current does not go through the switch.

The nominal gains for the 1% resistor values below would be:
U2a Closed -- Av=1
U2b Closed -- Av=10.0
U2c Closed -- Av=100.3

 

@schmitt trigger thanks a lot
- you are right, GBP with 4 MHz (max) is at the limit, actually I am tresting with 23kHz which should be even below min specification of 2.5MHz but yes, on the limit
- 2N7002 capacitances should be belkow 20pF which I hoped is small enough. LTspice did not show a big difference
@crutschow also thank for your proposal. Looks good would be an elegant option. I am not realy sure if the drain capacitance is the reason for the oscillation but I would give this circuit a try if the PGA103 does not work. Are you sure the high on resistances of the CD4066 (<400ohm) do not influence the amplification factors ?

BR, BastelBaus

 

Are you sure the high on resistances of the CD4066 (<400ohm) do not influence the amplification factors ?

Yes, the purpose of that configuration is that no significant current (just the op amp input bias and stray capacitance currents) goes through the CD4066 resistance, so there is negligible voltage across the resistance to effect the gain.
Note that all gain resistor currents go directly to ground, not through the CD4066.
The CD4066 just taps the voltage off the voltage divider generated by the resistors.

 

I already ordered a PGA103

You probably will not get past 250khz with a gain of 100x. (750khz at 10x) It is a real good part.

 

Do you understand how the gain is calculated for the Post #4 circuit?

 

@ronsimpson : as said, 40khz is max, probably I could live with 30. the only thing is that pga103 isn’t as cheap as the tl082 combi

@crutschow: I think so, if u2a is closed, it’s a voltage follower with gain 1. if u2b is closed it’s a voltage devider with 11.23 over 113.33 giving a gain of ~10 and if u2c is closed it is 1.13 over 113.33 with a gain of ~100.

Is there a possibility to simulate the ringing in ltspice? Would help me to debug.
BR
BastelBaus

 

Is there a possibility to simulate the ringing in ltspice?

Adding a few tens of pF stray capacitance from the inverting input to ground may give more ringing.