Dual rail PSU

[beatnik]

I am making a linear power supply with two outputs, see attached schematic

top part is 32V for audio circuit
bottom part 12V for relays and LEDs

I am trying to make this from a single 12V AC transformer

I have noticed that when the 12V section is connected there is an increase in hum on the audio circuitry powered by the 32V rail

If I remove R7 the noise drops considerably.

Do you have any suggestions ? Perhaps the rectification of the 12V section needs changing ?

 

[Matador]

My guess is IC1 doesn't have enough dropout when the 12V rail is sourcing current (dragging down the 12VAC slightly).

What is the voltage at the top of C9, and V2+, both with and without R7 populated?

 

[beatnik]

The margin is tight but I don't think this is the cause of the problem

There is 35.4V at the input and 32.2V at the output of the regulator

It makes no difference whether R7 is in or out of the circuit, however the noise further increases when the load is connected to the 12V rail

To give you an idea, the noise level without R7 is -74dBu, with R7 in circuit but no load it's -66dBu, with load -56dB

Throughout these variations the input to IC1 remains constant at 35.5V and the output at 32.2V

 

[beatnik]

I have also tried the 12V rail without LM317 section and leaving it unregulated but the same problem occurs

Could the problem be caused by an interaction between the rectification of the 12V rail and the voltage tripler for the 32V rail ?

 

[Matador]

It's drawn strangely to me, now that I look closer.

For example: you've declared AC2 as the common line for both supplies, so I don't see how it's possible for D9 to ever turn on, once C10 charges.

 

[JohnRoberts]

A common problem in power supply design is dealing with reference ground integrity. PCB traces have finite resistance so current flowing in those traces will cause voltage drops that can corrupt clean references.

It helps to visualize PCB traces as resistors so the voltage drops are easier to conceptualize.

A common area of problems is around reservoir capacitor grounds.

JR

 

[Matador]

It's typical to see a bridge rectifier for the 'bottom' supply: see this (which is a doubler version, so would need a third stage added for a tripler):

 

[beatnik]

I have tried the solution you proposed with the bridge rectifier but I get basically the same problem, there is a 10db difference in noise with the 12V supply line connected.

 

[ccaudle]

There is 35.4V at the input and 32.2V at the output of the regulator

What kind of regulator? Unless the regulator is specifically a low-dropout design it is probably not actually regulating with only 3V drop across the regulator.

 

[beatnik]

LM317 I believe these have a minimum 3V input to output differential ?

When the 12V rail is connected the voltage before the regulator decreases only 0.2V, and the output voltage remains stable at 32.2V

I have tried adding a multiplier stage, so now there is 48V on the input of the LM317. No difference

Just another thought, since I am using a 12V transformer with dual secondaries connected in parallel, I'm wondering if I should try using one secondary for each rail, instead of powering both rail from the same two windings in parallel ?

 

[scott2000]

a lot of those printer brick supplies are dual voltage.... 32 and 16v for one example
Just saying... not that it's relevant..just reminded me...