Hey guys, i've got this schematic going on on my board with a TLV2375ID opamp. When I turn on the opamp I hear a ringing. If I put my finger on (and push lightly) on the +5V regulator (LDO) the ringing subsides a lot.

You guys think it's the AGND vs DGND business? I don't have them isolated on the board.

Opamp is powered by +/-5V

Thank you,

 

Bottom left op amp appears to be configured as a relaxation oscillator so I expect one of the components is vibrating slightly with that oscillation, which is what you hear.

 

Yeah the first stage is a wein bridge oscillator. The oscillator saturates the opamp a bit ( I did this to ensure oscillation). I'm wondering if perhaps it shouldn't be saturating too much?I'm confused as to why the buzzing subsides when I put my finger on the 5V supply.


Thanks,

 

The ringing subsides when you touch the 5V PS? Have you checked with a scope to see if the 5V PS is itself oscillating? Do you have enough capacitance on your LDO? Check the LDO datasheet, they often require specific values of capacitance and sometimes even specify the minimum capacitor ESR.

 

Yes I have scoped it and it's pretty clean. Got input and output caps of various sizes on the 5V line as specified on the datasheet.

There is no buzzing when the opamp is off, it's quiet when only the digital components are on it, but when the opamp kicks on, the buzzing starts.

I've measured the opamp current, its 2mA on each supply pin, just as it should be, and the 10V it sees across the supply is well bellow the 16V limit for this opamp.

...ringing continues....

 

I don't know why you can stop the ringing by touching the regulator, but I would minimize stray capacitance on pin 13 of the op amp. Running it out to a probe point is just asking for oscillation of that section of the op amp.
I have heard ceramic disc capacitors "sing", but I don't know if they are even available nowadays.

You should also to add a bleed resistor from EC_READ to ground.

 

You should also to add a bleed resistor from EC_READ to ground.

Ron, can you clarify exactly why there absolutely has to be a bleed resistor? That signal goes to a micro's ADC input so I figured the pin's impedance will take care of that "bleeding". But what's wrong with letting the cap hold that voltage?

 

Ron, can you clarify exactly why there absolutely has to be a bleed resistor? That signal goes to a micro's ADC input so I figured the pin's impedance will take care of that "bleeding". But what's wrong with letting the cap hold that voltage?

Maybe I jumped the gun. What is the purpose of the peak rectifier?

 

Maybe I jumped the gun. What is the purpose of the peak rectifier?

First stage sets up an oscillation. Second stage is a variable gain. Third stage is a precision rectifier that senses the amplitude of the output of the gain stage. Then a micro reads off the value of "EC_READ" a bunch of times, and the opamp is turned off until the next measuring time.

The variable gain doesn't change during a reading period so the multiple micro reads are more to filter out noise.

 

First stage sets up an oscillation. Second stage is a variable gain. Third stage is a precision rectifier that senses the amplitude of the output of the gain stage. Then a micro reads off the value of "EC_READ" a bunch of times, and the opamp is turned off until the next measuring time.

The variable gain doesn't change during a reading period so the multiple micro reads are more to filter out noise.

I wouldn't count on the capacitors discharging when the op amps are turned off. How long will they be off between readings?

 

I wouldn't count on the capacitors discharging when the op amps are turned off. How long will they be off between readings?

minimum 1 hour, and i'm using a tiny .1uF ceramic cap

 

minimum 1 hour, and i'm using a tiny .1uF ceramic cap

See the attached schematic.
That cap could stay charged for hours when the power is turned off (or left on).
You must not make assumptions about ICs. The datasheet is there for a reason.

 

Do not rule out a simple solution such as hot glue.

I once made repairs to switching supply and removed large amounts of hot glue around several of the transistors and a few large inductors.

The repair was successful, but the supply started to 'sing'. Adding large globs of hot glue to the places where it was removed, stopped the noise.

The physical size and density of the components allowed for 'sympathetic vibrations' to occur due to the freq of the oscillations.

It would be a simple thing to try such a solution in your case as well.

 

See the attached schematic.
That cap could stay charged for hours when the power is turned off (or left on).
You must not make assumptions about ICs. The datasheet is there for a reason.

Thanks Ron, I'll add in a giant res in parallel with the cap.

And Kermit2, i'll see about the hot glue. It's a mass produced part so I'm trying to wager using hot glue on every pcb vs having the ringing be present for 1 second out of every hour.


Thanks guys