Phase Shift Circuit Oddity

[dannybeckett]

Hi there,

I am building a lock-in amplifier, of which a quadrature phase shift section is required. I have constructed the circuit here:

https://www.tedpavlic.com/teaching/osu/ece209/lab1_intro/lab1_intro_phase_shifter.pdf

Which does work very nicely. I have chosen 5K1 resistors for the feedback, a 100K trimpot and a 3.3nF capacitor for the RC circuit. This gives me a range of about 500Hz to 100KHz at 90 degrees phase. The problem I am having is that at higher frequencies, the output waveform is somewhat distorted.

At 1KHz everything is OK:

https://snag.gy/yfcPO.jpg

At 80KHz the output waveform is somewhat distorted:

https://snag.gy/0wA5i.jpg

At each frequency I adjusted the trimpot to achieve a 90 degree phase shift. I am just wondering what could be the cause of this distortion? Could I be running into bandwidth issues with the op amp I'm using? (OPA4137)

Thanks for any help,

Dan

 

[godfreyl]

It looks like crossover distortion, not slew rate limiting.

- - - Updated - - -

The waveform in your picture (3.2V pk-pk @80KHz) nrrds a slew rate of less than 1V/us and OPA4137 is typically capable of 3.5V/us, so that shouldn't be a problem.

OTOH, OPA4137's GBW is only about 1MHz so in your circuit @80KHz, the loop gain is only about 6. I.e. there's very little feedback available to correct distortion.

You can probably get a much better waveform if you bias the opamp's output stage into class A by connecting a resistor between the output and one of the supply rails.

 

[dannybeckett]

Brill, thanks for the reply. What value of resistor would you recommend I start with? Does it matter which rail I connect it to?

 

[godfreyl]

Assuming the power supply is about +-12V, I'd try 4.7K connected to the negative rail to start with, but experiment. The distortion will probably be different depending on the value of the resistor, and which rail it's connected to.

 

[FvM]

Without seeing the complete setup, we can only guess. The OP can e.g. easily "run into bandwidth issues" by just connecting a 1:1 oscilloscope probe (as suggested by the screenshot) or similar capacitive load.

 

[dannybeckett]

That was awesome, it worked really well. Is this now eliminating crossover distortion from the output of the op amp? Thanks again for the help!

P.S. here's a picture of the biased op amp output **broken link removed**