ultra low noise like 50uV from the psu for 80V 15A possibility?

[rhythmsandy]

Is there anyway to reduce the psu noise to substantial level yet providing the 80V 15A current?

 

[sreten]

Hi,

Why ? The PSU is dominated by ripple and noise is not an issue.

You can regulate or use a capacitance multiplier to reduce ripple.
Both are generally not needed except for simple class A,
which you won't be doing at 80V and 15A.

rgds, sreten.

If you don't ask the right questions, most answers are useless.

 

[rhythmsandy]

Im more concerned with noise than ripple. For ripple no problem can use CLC circuits but for the noise thats the problem.

 

[peranders]

Why the fear for noise? What is the application?

 

[infinia]

the short answer yes , but I doubt you need it for the whole power output section.
look at using a separate low power supply for the front end regulator using discrete transistors and LN references. several ways to do this, and still keep some decent efficiency rather than brute force.

 

[sreten]

Hi,

What noise ? Caused by what component ? What do
think the noise of the final capacitor is determined by ?

rgds, sreten.

 

[rhythmsandy]

I found some regulator from TI for VAS stage which can give noise performances as much low as 50uV but for the output stage we dont have any choices as the voltage and current requirements are high.

 

[abraxalito]

Low noise PSUs tend to be only such when a very constant current is drawn. Otherwise the load dependent current draw * the Zout gives rise to much more noise than 50uV. Output impedance is king in my experience.

 

[kouiky]

I wonder if the regulators could be paralleled. Is your load dynamic or constant?

 

[infinia]

yes the idea is to split the supplies, clean up all gain stages. With the normal supply to the OP stage (gain <1). Then re measure the output noise into a very light load at O V ( see CMRR / PSRR measurement for power op amps ) compared to the unclipped Vo .
20 LOG (Vo RMS / 50uv ) is an unreasonable S/N

 

[abraxalito]

All gain stages need to run in SE classA (i.e. CCS load) to get minimal load-dependent noise. Once that's addressed it makes sense to go for a lower noise regulator.

 

[infinia]

All gain stages need to run in SE classA (i.e. CCS load) to get minimal load-dependent noise. Once that's addressed it makes sense to go for a lower noise regulator.

yes probably a balance of both depending on what he has to work with.