Linear regulator noise

[Potato Cakes]

Hello, Everyone,

I have a capacitance multiplier circuit that I have been using with great success in conjunction with off the shelf switch mode power supplies. In one of my designs, I am running into a scenario where adding a linear regulator (LM337) to the audio PCB after the initial filtering would be advantageous. However, when doing so, the audio circuit generates a high amount of white noise/static. I am using values around the LM337 that have been successfully used in other builds, but not so much in this instance. I have added additional capacitance to the output section which attenuates the higher frequencies of the noise, but not nearly enough to be useful. My thought process was that since the DC going to the regulator is properly filtered, then using the recommended values from the LM337 data sheet should be sufficient but clearly that is not the case. Putting the capacitance multiplier after the regulators makes everything quiet again and in the long term that is more than likely what will happen, but I thought I would take this opportunity learn a little bit more about working with linear regulators in an audio circuit.

Do I still need the 1000-2000uF capacitance on the input section of the LM337 that I normally see in linear power supplies?

Thanks!

Paul

 

[gyraf]

No, you shouldn't need those bigg'ish ripple-reducing capacitors when dealing with smps.

Excessive noise here could very-well be a by-product of some out-of-band oscillation somewhere.

/Jakob E.

 

[Potato Cakes]

It seems to be a consistent issue with various audio circuits when arranging the signal flow as SMPS - Filter - Regulator regardless of the SMPS used (at least the ones I've tested) so I'm focused on the regulator itself at this point. I'll try a some more things later this week to see if anything is discovered.

Thanks!

Paul

 

[merlin]

Schematic would help. Since you think it's a low frequency problem, try adding a small resistor before a big input cap (i.e. an RC filter). This can be effective for pre-cleaning the input to the reg, with minimal voltage loss.

 

[Arni]

I always use ferrite beads and a good HF capacitor (NPO ceramic) directly on the output of switched mode power supplies. They can often have narrow spikes in the order of 100 mV or more, which can create havoc downstream.

 

[Cqwet Dbdfte]

Good advice above, multiturn ferrite beads work OK, directly on SMPS, (both positive and negative). Capacitive loads on SMPS may force it to draw out more HF energy making it harder to tame.
Depending on noise spectrum a "differential line filter" may be useful. Look at TDK the iDQ48 circuit for ideas. This part has ~60dB suppression at around 300KHz.

 

[Potato Cakes]

With this capacitance multiplier circuit it seems to filter out noise to incredibly low levels of noise with anything needed to be added to the output of the SMPS. I does not matter the brand or type, the audio circuits downstream of it have remarkably improved noise specs. I used it with bipolar and single source powered circuits with the same results. However, anytime I have a regulator after the capacitance multiplier I always get a hiss like a high noise floor or impedance mismatch. I have configured the regulator several different ways using diodes, ceramic caps, etc., but I still get noise. I'm fine with this being the case but there have been instances where it would have been simpler to use one capacitance multiplier and then regulate to the different desired voltages.

Thanks!

Paul

 

[Walter66]

Could you please provide a schemo for your circuit?
Why not placing the cap multiplier after the regulator?
A regulator should regulate input and output (load) voltage, so a cap multiplier should always filter that last noise and this means to put it after the regulator, just in the output section of a PSU, right?

 

[Cqwet Dbdfte]

There is no spec for cap multiplier loss, it is current dependant,
so regulation is not its job. The regulator (317) chip needs the FET to handle the input voltage, adding an RC network to the FET gate makes it into a cap multiplier and can be set to make whatever ramp up time needed, and the FET absorbs most of the noise from the rectifiers, making the 317's job easier.
So no.
The voltage regulator works in series with the load, rejecting voltage variations from the input and the load. As it can only source voltage it cannot absorb any rising voltage on the output. The bypass diodes direct higher than set voltage to the rectifiers. A shunt regulator would be needed to deal with reactive loads, usually not a problem.

 

[Newmarket]

Trying to be helpful - imho this thread isn't making a great deal of sense ? As wizard @merlin said, a schematic (or several) would help a lot

 

[Cqwet Dbdfte]

https://www.ti.com/lit/pdf/snva583Also works with a HV MOSFET in place of the Q1/Q2 Darlington, with a cap to ground on D1 Cathode. 10uF is OK, R1 can be much higher value.
I prefer using higher value for R5/R6 and C1, as the RC constant is longer.

The 317 only regulates the gate-to-source voltage.
No capacitor on the 317 input.

I have posted schematic in other threads on this forum. I claim no originality on this circuit.
Simulate to your hearts content in LTspice, then turn on the soldering iron.

 

[Potato Cakes]

Could you please provide a schemo for your circuit?
Why not placing the cap multiplier after the regulator?
A regulator should regulate input and output (load) voltage, so a cap multiplier should always filter that last noise and this means to put it after the regulator, just in the output section of a PSU, right?

Correct. However there have been a few instances where it would be helpful for the capacitance multiplier to filter SMPS first and then break out from there to hit several regulators for varying voltage needs in an audio circuit. This cuts down on the components.

Thanks!

Paul

 

[Potato Cakes]

Schematic here:

https://groupdiy.com/threads/smps-class-a-microphone-preamp-discoveries.82639/#post-1068415