So how do you measure power amplifier PSRR then?

[GK]

Here is how I do it.

Show in the pic attached is a basic diagram of a typical test set up.

The amplifier under test is connected to a 4-ohm dummy load and the output monitored on an oscilloscope and an AC voltmeter. The amplifiers audio input is terminated/shorted to ground.

The earth return for the power supply is broken with a series 4 ohm power resistor / dummy load, which is driven by another audio power amplifier connected to a signal generator.

This allows a large audio frequency (and beyond) signal to be superimposed onto the power supply rails.

The same method can be used to measure the PSRR of each rail individually, but then independent floating supplies have to be provided for the positive and negative rail voltages.

[MJL21193]

Originally posted by G.Kleinschmidt
Here is how I do it.

Show in the pic attached is a basic diagram of a typical test set up.

Ok, now I get it. Very good.
The problem I have is that the readings on a voltmeter or scope aren't as "tactile" as the method I used, where I could actually listen to for the "ripple". I guess I could use a speaker instead of the scope or meter?

[KLe]

Thanks Glenn

Can the PSRR be simmed via LTSpice or Multisim, also?

[h_a]

Sure you can. Just put voltage sources into the corresponding places.

Have fun, Hannes

[PB2]

Don't forget that the supply voltage becomes the sum of the AC voltage plus the DC component. I'd suggest using a small signal injection, say .1 to 1V, perhaps 5V max.

Also, one could inject white noise and look at the output on a spectrum analyzer.

Small power resistors say .1 to 1 ohm could be put in series with the supply caps and the signal injected there to do the single ended test as shown in the attached diagram. The cap then acts as both the supply filter cap and the signal coupling cap:

http://baselaudiolabs.googlepages.com/psrr_PLB.JPG

Pete B.

[GK]

Originally posted by PB2
Don't forget that the supply voltage becomes the sum of the AC voltage plus the DC component. I'd suggest using a small signal injection, say .1 to 1V, perhaps 5V max.

Also, one could inject white noise and look at the output on a spectrum analyzer.

Small power resistors say .1 to 1 ohm could be put in series with the supply caps and the signal injected there to do the single ended test as shown in the attached diagram. The cap then acts as both the supply filter cap and the signal coupling cap:

http://baselaudiolabs.googlepages.com/psrr_PLB.JPG

Pete B.

Hi Pete.

Problem with a signal of only 0.1 - 1V is that you'd need a higher test frequency before you'd be able to pull anything out of the noise at the output.

By driving the ground of the power supply the total supply voltage stays fixed, just the relation to ground changes.
Most amps should be able to take at least a +/-10% variation here with ease (and should be designed to, to account for variations in mains voltage).

Cheers,
Glen

[andy_c]

Hi Glen,

Would this work without the 4 Ohm resistor (the horizontal one)?

[GK]

Originally posted by andy_c
Hi Glen,

Would this work without the 4 Ohm resistor (the horizontal one)?

Hi Andy.

I just put the resistor there because my test amplifier is happier driving a load.

Cheers,
Glen

[lineup]

The same basic setup will work for simulator PSSR test.

Glen,
are there any special things you recommend for SPICE setup?

I use now like 0.1 Vrms for preamplifiers and 1Vrms for poweramps.
100 Hz. Because of 50Hz full bridge gives 100 Hertz ripple.

[ostripper]

By line up..The same basic setup will work for simulator PSSR test.

3 voltage sources.. [SINE(60 2 120)],[SINE(60 2 500)]-for US..,and a pulse at 2V(HF harmonics) .. all fed to .1R resistors.. gives a realistic
representation of real world conditions..do the same for
negetive rail.
OS