OPA4314 instability

Hi Andrew,

Our audio expert is away on travel right now, but I wanted to try and get a prompt response back to you.

According to OPA4314 data sheet Figure 1, Open-Loop Gain and Phase vs Frequency, the OPA4314 should have an inherent phase margin of about 90 degrees at 123 kHz. If it is oscillating, then the external feedback circuit connected between its output and back to its input is contributing delay that reduces the phase margin to zero. Whe that occurs positive feedback results and the circuit can oscillate. That result indicates that certain component settings of the Baxandall network are resulting in the positive feedback condition and leading to the instability.

This Baxandall network is complex and analysis of it would be a time consuming task; especially, because of the mulitple feedback paths and variable components. There are analysis on line of the original two-channel network including the work done by Baxandall, published in Wireless World, in October 1952:

http://home.comcast.net/~stphkeri/NegativeFeedbackTone.pdf

His circuit didn't include the Mid control. When I search the web for 3-channel Baxandall circuits in Google images a number of variations on the circuit are shown:

https://www.google.com/search?q=3-channel+Baxandall&espv=2&biw=1280&bih=923&site=webhp&source=lnms&tbm=isch&sa=X&ved=0CAcQ_AUoAmoVChMInNSvpKvCxwIVRhSSCh1YmQ16

The 3-channel networks shown in the first several images do use different component values than yours. You might find that one of the variations provides a stable solution. You should be able to verify that with the TINA simulation.

Regards, Thomas

PA - Linear Applications Engineering 

Hi Thomas,

Many thanks for your prompt reply - much appreciated.

The circuit that I am using comes from Douglas Self's book "Small Signal Audio Design" as shown below:

My component values are different because I am using 50K digital pots not 20K analog pots. The op amp in Self's circuit is an NE5532.

Changing the component values in order to find a stable configuration as you suggest is a difficult since there is little room for manoeuvre once you have pinned down the -3dB cut-off points and gain(s) that you want. The other difficulty is that I can't create instability under simulation so I wouldn't know what set of values has fixed it.

FYI - I made a mistake in the measurement of the oscillation frequency - it is in the high 500K - 600KHz range under both load, no load and no input conditions - NOT 123KHz as previously stated. Also, the instability only occurs when the MID pot is at its 50% setting +/- a bit (i.e. 0dB gain). Setting the MID pot to either 0% or 100% does not cause oscillation.

I will try and get a better model for the digital pots - right now I am modeling them as purely resistive - and see if I can get any further insights into the problem. From your perspective, do you know of any recurring problems that your customers have had with digital pots?

Regards,

Andrew

Hi Thomas,

A reply to myself as I think I have fixed it. I modelled the digital pots using the input capacitances given in the datasheet. The circuit and frequency response curves are as follows:

A nice little pole right around the frequency I'm seeing. Modifying the circuit with a 22pF feedback cap fixes it under simulation as follows:

The value of 22pF was trial and error - nothing scientific I'm afraid. Anyway, I'll try this out on the prototype and get back to you to confirm if it fixes the problem or not.

Kind regards,

Andrew

Hi Andrew,

I have been out of the office on vacation a couple of weeks. Thank you for all the updates on your OPA4314 application and resolution of the stability issue. It is great that you were able to find a solution. There is always much more to be gained from resolving an issue on our own! Others will certainly benefit from your findings.

Best Regards, Thomas

PA - Linear Applications Engineering