Low noise voltage regs

[Extreme_Boky]

I’ve been planning to make few of these and start using them in all my modifications. The original design did not have voltage reference HF by-pass cap and resistor combination (499 + 0.1uF). The gate stopper resistors (2 X 750) were also missing and I added NE5532 power supply decoupling caps (2 X 10uF).
The max current with IRF’s on small heatsinks, is around 1.5A – just what I need for CD players, DAC’s and low voltage sections in amps and pre-amps.

Has anyone built this regulator before? It seems easy to build at very reasonable price, giving good output noise performance. The trim-pot can be omitted and various output voltages can be obtained by proper ref voltage selection.

I invite all to comment this circuit and the proposed extras I have added.

Any comments are welcome.

Regards,
Boky

[janneman]

Originally posted by Extreme_Boky
I’ve been planning to make few of these and start using them in all my modifications. The original design did not have voltage reference HF by-pass cap and resistor combination (499 + 0.1uF). The gate stopper resistors (2 X 750) were also missing and I added NE5532 power supply decoupling caps (2 X 10uF).
The max current with IRF’s on small heatsinks, is around 1.5A – just what I need for CD players, DAC’s and low voltage sections in amps and pre-amps.

Has anyone built this regulator before? It seems easy to build at very reasonable price, giving good output noise performance. The trim-pot can be omitted and various output voltages can be obtained by proper ref voltage selection.

I invite all to comment this circuit and the proposed extras I have added.

Any comments are welcome.

Regards,
Boky

Hi Boky,

A few comments:

- the 10uF caps across the opamp are in parallel with the 2500uF caps (C23,25), so I doubt if they will make a difference. If what you are after is local supply decoupling for the opamp you probably get more out of a good small cap directly across (or as close as possible to) the opamp pins;

- an important one: the return for your two polarities comes back on a single line from the output connector. Presumably, the common return comes also in as a single line from the load. That means that any AC (noise, ripple, signal) current in the common line impedance (R, L) from one side comes back as an error voltage on the other side. Put the star ground as close as possible to the load, and run all wires separate (also ground and return) as much as possible. Use a 4-pin output connector;

- I'm not sure of that 499+0.1 ref filter. Normally we would take the filtered ref voltage from the R-C junction and go to the opamp from there;

- depending on input and output voltages and opamp selection, you may be able to run the opamp from the output voltage rather than the raw voltage, which gives lower noise and output ripple. You may want to include that option on your pcb.

Jan Didden

[Extreme_Boky]

Originally posted by janneman


Hi Boky,

A few comments:

- the 10uF caps across the opamp are in parallel with the 2500uF caps (C23,25), so I doubt if they will make a difference. If what you are after is local supply decoupling for the opamp you probably get more out of a good small cap directly across (or as close as possible to) the opamp pins

Yes, I will use .1uF/50V Black Gate - one of my all time dirt-cheap favourites. Thanks.

- an important one: the return for your two polarities comes back on a single line from the output connector. Presumably, the common return comes also in as a single line from the load. That means that any AC (noise, ripple, signal) current in the common line impedance (R, L) from one side comes back as an error voltage on the other side. Put the star ground as close as possible to the load, and run all wires separate (also ground and return) as much as possible. Use a 4-pin output connector;

The returns should go to common ground connection where the PS decoupling caps (2 X 2200uF) are connected to transformer centre tap. This is a DC current loop:” from capacitors via IRF’s via load back to common ground point via separate ground returns for + side, and for - side. I have PCB ready for experiment – one plane is ground plane-only, with “cut-outs” to route different returns – but I was thinking that the “star” point should be where two 2200uF capacitors and centre tap are connected together? I think that you have suggested star point to be at the “load end”?

- I'm not sure of that 499+0.1 ref filter. Normally we would take the filtered ref voltage from the R-C junction and go to the opamp from there;

I’ll leave provision for R and C series connection on the PCB – and see what combination(s) give lowest noise on the power supply rails…. I’ll start with 0.1uF cap and see what that does to HF noise (above 100kHz)

- depending on input and output voltages and opamp selection, you may be able to run the opamp from the output voltage rather than the raw voltage, which gives lower noise and output ripple. You may want to include that option on your pcb.;

Yeah, I was thinking about the same thing – but in case that I need to adjust the voltage on-the–spot, it would be better to leave the OP’s supply off the “raw” DC. Making two different PCB’s could be an option, though.

Regards,
Boky

[banana]

Did something similar, but used BJT instead.

Yet another Opamp based regulator