Cheap high quality filtering?

[50AE]

Ok, I have the transformer and rectifiers, I don't have the electrolytic caps. My transformer outputs 33V, ~45V rectified. I don't necessary need to use 45V in my case, it's preferrable to use 25 to 35V.
I want to drive two class AB amps.
One amp has an output power of 150W at 45V and because the speakers I want to drive are rated 30W continous each, I can sacrifice some voltage and let the amp work at 60-100W.
I'm a poor student and I'm looking for a shortcut to obtain a nice output like with an Elna, nichicon or panasonic cap.

My question is, can I use this capacitance multiplier circuit and get away with low value elna silmics 2, say 100uf, when I need 10 000?
Capacitance Multiplier Power Supply Filter

Or again using a low valued cap and a voltage regulator to regulate the voltage to 25-35V? Because I am able to sacrifice a lot of voltage, I could get away with a bad ripple.

[DF96]

You can use a cap multiplier for the smoother, but not for the reservoir cap.

[agdr]

And here is what DF96 said, in pictures. Kind of interesting. Since the cap multiplier can't store any significant energy like a reservoir cap can, the input 120 Hz waveform goes almost all the way to the 0V turf between cycles without the 4700uF reservoir (first plot). With the 4700uF cap to store energy between the cycles (second plot) all is fine.

On the second question of a small reservoir cap, with subsequent larger ripple followed by just a regulator chip, probably not. I tried simulating a few and at the voltages and currents you are using the regulators exceed their thermal limits and shut down, due to the large voltage swing of the ripple. You need to tame the ripple a bit and decrease the input voltage a bit before feeding it to the IC regulator to keep it from going poof. The cap multiplier reduces the voltage a bit going into the regulator due to the 8.2k resistor. A small R value power resistor in series with the regulator input (4700uF etc reservoir cap with no cap multiplier) could also burn off some excess power and drop the input voltage to the regulator IC to acceptable levels.

[DF96]

Another option is to use a choke input PSU, then the choke does the energy storage. You might get away with a cap multiplier for the first cap (after the choke). Snag is that a decent choke is bigger, heavier and more expensive than a decent big cap.

[simon7000]

Use a CRC filter. The first C should be big and can be cheap. The R should be under an ohm and the final C only needs to be 470 uF. If you want to tune the amp to your particular loudspeakers you can play with the R value.

[50AE]

Thanks for the help. I searched a little further about these.
I put a CRC filter for now. It consists of 14 000uF cheap as first C, 0,15ohms resistor and 1100uf second C(2200/2 panasonic FCs). It gives me around 1V ripple at 0,5A, 85 volts. I say "around", because my oscilloscope display sucks so much, it's just a pain in the ***.

[simon7000]

CV=Q by definition,

C x dV/dT + V x dC/dT = dQ/dT = i,

allow dC to be close to 0

C x dV/dT = i or dV = i x dT / C = .5 x 1/100 / 14000 e-6 = .357 volts.

Bulgaria having a 50 hertz line frequency and allowing for full wave rectification.

[AndrewT]

That prediction of 0.357Vpp of ripple is for the the first C of the CRC.
The ripple at the second C is likely to be ~ -6dB ref 357mVpp. i.e. ~200mVpp.

A multimeter that reads Vac, while a high DC voltage is present, will read ~ 1/3 of that pp value, i.e. ~70mVac.

If you are seeing much more than that, then suspect a wiring fault or a component fault.

[50AE]

I'm an idiot, the oscilloscope was in AC mode.
Now with 0,7A of load and 20V per division, there's almost no visible ripple. I measure nothing with the multimeter on Vac on 200V calibre

[AndrewT]

The scope whether on DC mode or AC mode should show rail ripple.
Switched to AC mode allows a more sensitive vertical scale and thus to expand the ripple so that you can approximate a peak to peak measurement.

The same applies to a multi-meter.
Reading 0.0Vac on a 200.0Vac scale, indicates it is safe to switch to a more sensitive scale.
Switch to 20.00Vac scale, what is the reading?
If it is still at or near zero then switch to 2000mVac scale. What is the reading? If it is near zero, then switch to 200.0mVac scale.