Keantoken's CFP cap multiplier

I've looked at the output of a 350VA toroid I have. Yes, there are flat tops to the sine wave but the flat tops always have the same geometry when I increase the voltage, it can't be a saturation issue. Rather it must have to do with line impedances earlier on which resist the charging pulses. I got up to 6.66A and couldn't test further because my graphite resistor submerged in water (affordable power resistor!) wouldn't go less than 3.3R for mysterious reasons.

There are actually two kinds of saturation which confuses the issue. There is flux saturation, which occurs when input changes fast enough, so it happens at the zero-crossings. I observe this with my toroid. It happens worst when the transformer is not loaded, because loading it relieves the magnetic stress. Because the rectifiers don't conduct during zero-crossings, toroids must avoid this form of saturation because the primary acts like a short as it happens. The second kind is the kind usually discussed concerning transformer power ratings. This is current saturation where inductance falls along with output, and this also causes an input current spike I think. I read that current saturation in a toroid makes flux saturation much worse, which is why toroids can have monstrous inrush current.

All of my measurements suggest the issue of perceived power in amps has little to do with transformer rating, but everything to do with supply filtering and transformer impedance. I have a very simple Lfet power buffer which sounds like it's lost it's bass when I don't use a line filter. A large trafo can sound very weak if you mess up the amplifier circuitry.

So this leads me to think that someone who knows what they're doing does not need to pay for an oversized transformer. My 350VA 55-0-55 toroid gives more than 6.66A peak without saturating, which is at least 518W. I don't think dynamic power can be an issue.

The secondary only supplies the same average current as is drawn by the load, so any saturation losses must be caused by charging pulse width, which does not seem to change much depending on reservoir capacitance except at very low currents. In a sane simulation model, line impedance and transformer leakage inductance act as the current limit, so the charging pulses across 4 octaves of capacitance is almost identical. You would need to severely undersize the capacitance to change the pulse shape.

So, impedance and filtering are the only things that add up here.
 
There are numerous resonances in any power supply unless one takes the effort to measure each one and damp it. This is not as straightforward as it seems because all the impedances change when the rectifier conducts, from very high impedance to very low impedance.

If you damp all resonances in one condition you may get even worse resonances in the other condition.
 
Member
Joined 2006
Paid Member
From your findings I take it that my subjective listening impression results have more to do with the type of transformer than it's VA rating (IMO, in my application, a Rcore outputting 300mA at 45vdc sounds much larger and refined than the same using a toroid.)

Would you please comment on that ?

Also, why low output currents do cause higher saturation ?

PS:; With Kmultiplier on my paradise psu TX heats much less than before.... why should this happen ?
 
Last edited:
The trafos in your pictures have split bobbin windings which means that capacitance from mains to amp will be very small. This means very little ground noise. Furthermore non-toroid transformers have more leakage inductance and core losses which means they filter more RF inherently.

Low output currents usually don't cause more saturation because flux saturation occurs when the rectifiers are not on. If you added a load before the rectifiers you may reduce flux saturation but it certainly wouldn't be efficient. Flux saturation occurs when the voltage crosses zero; if the trafo has enough loading during this time, which it will not because the rectifiers aren't conducting, then there will be no flux saturation. It would be interesting to experiment with to see what the affect on sound is.

I seem to recall looking at the Paradise PSU that I worried the regulators would be unstable. If so adding the Kmultiplier may change impedances and stop it by random luck. But I don't know why a Kmultiplier would cause a transformer to run cooler. That's a new one to me. If total capacitance had anything to do with charge pulse current, than I would suspect that the Kmultipliers were blocking ripple so that the transformer isn't charging all the capacitors after them. But that does not seem to be how it works.
 
Member
Joined 2006
Paid Member
The trafos in your pictures have split bobbin windings which means that capacitance from mains to amp will be very small. This means very little ground noise. Furthermore non-toroid transformers have more leakage inductance and core losses which means they filter more RF inherently. Maybe that is why I like R-Core so much :)

Low output currents usually don't cause more saturation because flux saturation occurs when the rectifiers are not on. If you added a load before the rectifiers you may reduce flux saturation but it certainly wouldn't be efficient. Flux saturation occurs when the voltage crosses zero; if the trafo has enough loading during this time, which it will not because the rectifiers aren't conducting, then there will be no flux saturation. It would be interesting to experiment with to see what the affect on sound is. Maybe I can place a small RC between the TX and rectifier to load the TX during the zero cross :yummy:

I seem to recall looking at the Paradise PSU that I worried the regulators would be unstable. Not mine because I used special jap power transistors as per Salas advise... If so adding the Kmultiplier may change impedances and stop it by random luck. But I don't know why a Kmultiplier would cause a transformer to run cooler. That's a new one to me. If total capacitance had anything to do with charge pulse current, than I would suspect that the Kmultipliers were blocking ripple so that the transformer isn't charging all the capacitors after them. But that does not seem to be how it works.
Cooling might have more to do with my new layout after all.
 
Try a large cap like 4.7u, and try it with and without a 22R series resistor (or use a 50R trimmer if you want). I'd be interested in hearing your results. This is what I've done actually, but it was in the interest of damping winding resonances rather than eliminating flux saturation.
 
Low output currents usually don't cause more saturation because flux saturation occurs when the rectifiers are not on. If you added a load before the rectifiers you may reduce flux saturation but it certainly wouldn't be efficient. Flux saturation occurs when the voltage crosses zero; if the trafo has enough loading during this time, which it will not because the rectifiers aren't conducting, then there will be no flux saturation. It would be interesting to experiment with to see what the affect on sound is.

I didn`t read the whole thread, but would like to add...
When a rectifire starts conducting the charging current lowers the secondery voltage output , So the diodes start bouncing ON/OFF riding on the regulation margin of the tranny. this happens till the charging cycle voltage increases above the transformer regulation margin, Then the diode remains open till the closing cycle occures..
This usually happens with ultra fast diodes..
To eliminate or minimize the regulation margin of the transformer a simple load resistor/s across the secondery output will do the job quite effectivly.
 
Last edited:
Thanks for the info. I've seen diode bounce on the scope when there was a major resonance in or around the transformer, but I haven't seen that once the resonance was damped properly.

Yes,
Dealing with this by dampening the diodes is very good.. but still a partial solution.. The regulation margin of a transformer introduces non linearty that has an impact on the sound.
 
Last edited:
Years ago Mr Erno Borbely suggested to me a 10% resistive load on the seconderies as a rule of thumb.
If you want to custom tailor a load for your specific tranny.. Say if your transformer is 50VA , test it with several loadings starting from zero load to about 15% and plot the results into a graph, This way you will find accuratly to what point your particular transformer should be loaded.
Transformers voltage output sag faster for the first 5 to 10% of their power rating.

But as Mr Borbely said "You can`t go wrong with a 10% load".
 
Last edited:
Thanks a ton for this information! I wasn't sure myself what you meant because transformer regulation is at full load, whereas according to you the resistor needs to be sized at around 10% the VA of the trafo, and I don't see a connection between that and regulation.

So if you have a 50VA transformer, you need a resistor across the secondary drawing 5W. Since secondary output is roughly a sine, you'll need to convert to RMS. So if you have a DC output of 24V, 24*.707=17Vrms. 5W/17V=294mA. 17V/294mA=58R. That will be a lot of heat.

However if we use a cap here, it will provide the necessary load without dissipating any heat inside the chassis. For this we choose a capacitor with a reactance of 58R at 60Hz. However, since saturation losses are mainly 3rd harmonic and above, I think we could use a cap set for 58R reactance at 180Hz, the 3rd harmonic. This would reduce idle power consumption and importantly, the size of the cap.

For a 50VA transformer with 24VDC output, I get 15uF for the right size secondary cap. This would draw about 1.7W idle. For a 500VA 24VDC trafo then, you'd want 150uF which would draw about 17W idle. Technically, the capacitor isn't drawing any power (thus no heat), it's just shunting current, but the power company will charge you anyway because they leave phase out of the calculations.

At 15uF you may find a suitable lytic cap but at 150uF I don't know if a suitable lytic would be less expensive than a film cap, or last long enough to be worth considering. I've looked for film caps at this size and they go for about $20. An amp will get the same measured performance with or without the cap, so there's no need for it per se but it may reduce inrush for toroids.

The right size cap may actually reduce idle consumption from saturation losses but this would require experimentation.
 
Hi ben, should this test be done before rectification or after ?

Should I measure voltage drop due to increased current in AC mode or after rectification in dc mode ?

VAC readings can be misleading because of RMS. For instance a square wave and a sine wave with the same peak voltages actually have a different RMS voltage reading. So measure voltage after the rectifier.