470uF for amps psu bypass?

[martinbls]

Hi all!

The filter bank of my single power supply for two amp modules consists of 2x10.000uF/50V Panasonic TS-HA per rail, so it's 40.000uF in total. Rail voltage is about 35V.

For testing purpose (and for the fun of it) I recently solderd in some bypass caps at the output of the psu (so these bypass caps would be the caps the amp "sees"). These are ELNA Silmic II, 470uF/50V each. I guess this is quite uncommon because most people use 0,1uF up to, say, 10uF here.

I'm not exactly sure about the result. I like the ELNA's very much in other applications, and I think there is a noticeable increase of musicality with these caps in the psu. On the other hand, I have the feeling that at very high volume levels the sound becomes slightly harder than it used to be before I soldered in these caps.

Unfortunately I don't own a o'scope to check how the output voltage looks like...

Any ideas? Has anyone experiences with such high-valued bypass caps in an amps psu?

Here's a pic...

Thanks a lot!
Martin

[AndrewT]

Hi,
these bigger value bypass caps are more usually mounted on the amplifier PCB very close to the main current consumers. Here they are referred to as decoupling.

You have used two different values of electrolytic in your PSU. Due to significant ESR in the electrolytics there is much less likelyhood of incipient oscillation when very low ESR caps are used. Some low ESR electrolytics can cause ringing on the supply lines, but I don't have a rule for identifying which may cause problems.

Does your amplifier have HF and MF decoupling on board?

[martinbls]

Hi Andrew,

yes, there are decoupling caps on the amp boards near the output devices, rated 47uF/100V, Nichicon Muse.

I'm aware of the fact that these bypass caps are usually mounted near the power consumers, but nevertheless I think of the amp boards and psu as two "separated" units. Even a completed amp board with decoupling caps near the output devices does react (sonically) to different psu topologies. Of course one might say there is no need for such psu bypass caps if they are on the amp boards already. But I think the psu does still "sound" different with different additional caps.

Recently I read an artikle about the idea that small psu bypass caps (like 0,01uF or 0,1uF) do actually degrade the performance of the psu. They performed a lot of measurements to proof their claims. Sorry, can't remember where I read it...

But in the end I'm left a bit puzzled about the possibilities of NO bypass, SMALL bypass or LARGE bypass values....

Martin

[DF96]

Bypass capacitors can cause a rise in impedance at high frequencies if the main capacitor has very low ESR. Sometimes a bypass can help, sometimes not. This is a complex situation, made worse by myth propagation and fashion. If used, a bypass could be perhaps 100-1000 times smaller than the main cap. Alternatively, replace the big cap with two of half the capacitance each.

[AndrewT]

The unseen problem is the complex ESL and ESR of the capacitances interacting with the C & L & R of the connecting traces and/or cables when a fairly fast step change in current demand arrives.

Multiple caps increase this complexity and sometimes one hears the resultant SQ change.

Oscilloscope investigation is the only method I know, to go from heard to seen voltages

[martinbls]

You guys,

when we are talking about this bypass-thing I think the result is that it makes things not necessarily better or worse, but certainly much more complex and unpredictable.

I will rip the caps off. The simpler, the better! :-)

Thanks a lot for sharing your thoughts!

Martin

[LostForWords]

Ironically, I was about to post a similar question, but am reluctant to expose my ignorance. Nevertheless, here goes. While Mr Pass seems to recommend an 4.7uF MKP capacitor as the final component in the power supply. I have also seen 4.7uF capacitors used across the secondaries of a transformer although it has also been suggested in an old post that they might be better positioned across the +/- terminals of the bridge rectifier. While I accept that the use of such capacitors at the amplifier end might lower ESR I am puzzled by their use across the bridge or secondaries. Can anybody enlighten me as I believe it has something to do with inhibiting ringing in the circuit rather than an attempt to simply lower ESR of the first capacitor in the circuit. What effect does it have?

[DF96]

4.7uF seems much too high for the AC side of the rectifier - it could encourage ringing. 4.7nF may help with switch-off spikes.

[LostForWords]

"From simulations it seems that 4,7uF across the plus and minus of the bridge is more effective than across the AC inputs (= the secondary of the transformer). Smaller values (100n-1u) do work, but often require another bigger capacitor (470n-10u) with series resistor in parallel as a damper. Ratio of C's approx 5-10, resistor value depending on the inductance of the transformer secondary and the value of the C, but in the range of 5-15 Ohm."

Quoted from 'Choke for Pi Filter# in Solid State without permission - my sincerest apologies to Steven

[martinbls]

Hi guys,

quick update: I did it, I removed the 470uF ELNA caps from the psu, and the result really is like I said before. The overall sound of the amp seems to be just slightly less musical (yet VERY neutral!), but its main sonic character now remains absolutely stable, even if I crank the volume up to really high levels. No more sign of "getting harder" or something like that at high listening levels.

Maybe this kind of bypass works for minimum current requirements, but fails at high current demands.

Anyway, great! I consider this part-removal a serious upgrade for my amp, that's why I want to thank you all for your participation on this topic!

Regards!
Martin