If we do not have top quality output caps we could use some cheaper elco and solder good quality film cap in parallel from the bottom side of pcb. In that case we shall need 10mm (or more) spacers for pcb mount.
The quality of the capacitor affects the way current is delivered to a varying load, i.e. changes in current draw.
If the current is constant, the capacitor has virtually no effect on current delivery.
Now look downstream to the load.
Is the load properly supported by decoupling to achieve rapid changes in current delivery due to rapid changes in the loading?
The amplifier (or load) is where the biggest (by far) influence on changes in current delivery happen.
The PSU separated by many inches of complex impedance wiring is only expected to cope with constant current draw and slowly changing current draw. The PSU does not need decoupling to account for loading changes.
The PSU needs decoupling to attenuate ringing or oscillation generated within the PSU. This is usually done with a Zobel (a decoupling capacitor plus some intrinsic or added resistance for adequate damping) and other stability enhancing components, located in the correct posotion. It is not the job of the smoothing capacitors to enhance stability.
If the current is constant, the capacitor has virtually no effect on current delivery.
Now look downstream to the load.
Is the load properly supported by decoupling to achieve rapid changes in current delivery due to rapid changes in the loading?
The amplifier (or load) is where the biggest (by far) influence on changes in current delivery happen.
The PSU separated by many inches of complex impedance wiring is only expected to cope with constant current draw and slowly changing current draw. The PSU does not need decoupling to account for loading changes.
The PSU needs decoupling to attenuate ringing or oscillation generated within the PSU. This is usually done with a Zobel (a decoupling capacitor plus some intrinsic or added resistance for adequate damping) and other stability enhancing components, located in the correct posotion. It is not the job of the smoothing capacitors to enhance stability.
I believe this is correct, based on past experience with much better equipment than I have access to now.
I posted some screen pics of tests done using an ordinary 100MHz Tek scope with memory, and I could not find any evidence of ringing, when using Schottky diodes in the bridge. There is evidence of a small amount of overshoot at diode turn-off, but only on the unregulated side of the supply, not on the output.
Do posts 302 and 303 get any feedback?
Or are you all agreed that the suggestions in post301 are unlikely to make a better PSU?
Or are you all agreed that the suggestions in post301 are unlikely to make a better PSU?
Hi Guys,
This question ended up a page back and didn't get answered. Could one of you address this?
Thanks
http://www.diyaudio.com/forums/powe...hed-capacitance-multiplier-6.html#post3614738
This question ended up a page back and didn't get answered. Could one of you address this?
Thanks
http://www.diyaudio.com/forums/powe...hed-capacitance-multiplier-6.html#post3614738
Yes, we do. I will try to post something useful later this weekend. In the meantime, you can use the information in post 52, and scale it to your needs.
Test results @ different ripple voltages
At a minimum, Vdrop should cover 50% of the expected maximum ripple voltage (pk-to-pk value). I usually set mine to 100%. So, mine is set around 1.1~1.2V. (This is not like setting dropout voltage with a conventional regulated supply which would be much higher, and which would require dissipating much more power in the heatsinks.)
you don't want "drop out" during periods of high power output.
Try a variety of settings and see which drop out and which don't.
Then set to slightly higher than the ones that don't drop out.
Or just guess.
The Vdrop can be very small, because the circuit Miles used is more efficient than a normal Darlington Cap Multiplier regulator.
The minimum useful Vdrop should cover the ripple at low load, and some extra margin. With VSSA, or a class A-B amplifier with a similar idle current, this is about 0.75V. The maximum useful Vdrop with VSSA at 45V rails is about 2~2.2 V.
There is an explanation on Miles' home page, but this may help:
It is a linear power supply with a CM filter. The whole supply will rise and fall with line and load changes, and the Vdrop will stay about the same (tested).
When I adjust the AC line voltage with a Variac +/-10% to simulate high and low line conditions (during testing), the difference between the unregulated input and filtered output stays where I set it.
It is not a regulator with a fixed dropout voltage. It cannot drop out due to heavy load, or because of low line condition. There is little or no wasted power, unless you make the Vdrop too high to be useful.
It can only drop out if there is a sudden change in load, like a transient demand during heavy bass. Even then, the regulation is very good, the recovery is smooth, and the result is better than an unfiltered supply (tested).
Unless the Vdrop is set intentionally higher than needed for the filter to operate correctly, no useful voltage is "lost", unless you consider clipping power useful. The reduced voltage output is roughly equivalent to the range where the maximum ripple starts to distort the audio signal as it approaches clipping. The net result is to achieve a "cleaner" (smoother) clipping, at a voltage which is about 1-2V lower than the maximum unregulated voltage before the filter.
Terry,
Judging by the results from post 52, I concluded that if we use two channels (stereo) ESP DOZ class A amp, with recommended bias of 1,5A (each channel, 3A total current continuously drawn) it will be enough to set output of Mr Evil's cap multiplier some 2V lower than input voltage. Am I right? And that's probably hardest work (worst case condition) one can imagine for this PSU.
For class AB amps much lower difference is needed on average, but if you want to use AB class amps real loud than the amp will momentarily draw more than 3 amps (above case), but only for a short time (miliseconds). PMI proved that during loud passages this PSU can cope with that without significant output drop. Am I right?
Neglecting the power dissipation in the heatsinks, that is correct. In practice, you would have to increase the size of the reservoir cap and the heatsinks for the pass transistors. Even then, it would not be an optimum configuration for Class-A at that power level, because the average current is so high. The simple answer is to use one supply per channel, but I am working on a higher power version. There are pics of the board with some larger heatsinks, in another post above, to give you an idea where this is going... 🙄 JameshillJ is working on another angle, which may be better, not sure yet.
Correct. There will be an output drop, but it will be smoothed out in proportion to the RC constant of the cap multiplier, and the resulting disturbance will be below hearing range.
Hi Pete,
I guess I still don't know what I should be looking for to know if I have it set right. I hooked it up to an old symasym today. First I set it at no voltage drop, then 1.4V, then .75V drop. To be honest I can't hear any difference. Well, that's completely true. It did seem to have a little less punch with the 1.4V drop but that could just be a placebo affect. It's hard to tell sometimes when time has passed between listening. So what is it exactly that I should be looking for?
Thanks, Terry
I guess I still don't know what I should be looking for to know if I have it set right. I hooked it up to an old symasym today. First I set it at no voltage drop, then 1.4V, then .75V drop. To be honest I can't hear any difference. Well, that's completely true. It did seem to have a little less punch with the 1.4V drop but that could just be a placebo affect. It's hard to tell sometimes when time has passed between listening. So what is it exactly that I should be looking for?
Thanks, Terry
Terry,
Just my opinion here, but if the amplifier already has good PSRR then the capacitance multiplier may not even be audible at all. I have been working on a simplified version of the circuit and seem to get about -60dB ripple at the output under full load. The capacitance multiplier certainly can't hurt in any event but is likely to be of most benefit when powering a circuit with a poor PSRR. Just for reference PSRR, or Power Supply Rejection Ratio, is the amplifier's ability to reject ripple on the power supply rails so that the ripple doesn't appear at the output.
That said, PMI's suggestion of about 1V will likely do for most applications. My $0.02.
Just my opinion here, but if the amplifier already has good PSRR then the capacitance multiplier may not even be audible at all. I have been working on a simplified version of the circuit and seem to get about -60dB ripple at the output under full load. The capacitance multiplier certainly can't hurt in any event but is likely to be of most benefit when powering a circuit with a poor PSRR. Just for reference PSRR, or Power Supply Rejection Ratio, is the amplifier's ability to reject ripple on the power supply rails so that the ripple doesn't appear at the output.
That said, PMI's suggestion of about 1V will likely do for most applications. My $0.02.
So is this just a listening thing or is there something I can actually measure? Does the VSSA have poor rejection? Is that the reason something like this is needed?
Thanks, Terry
Thanks, Terry
No, the VSSA is actually not bad at all as far as its power supply rejection is concerned. You will notice it already had supplementary filtering in place for the front end and voltage amplifier stages, so in my personal experience it already works quote well with a plain 'ol linear bulk supply. As mentioned, the capacitance multiplier certainly can't hurt but may not make a subjective difference.
Hum, I wonder why LC pushes the SMPS so strongly then. I guess I don't understand the purpose for going to all this work and expense if a few aluminum caps will do just as well. I suppose it is a smaller footprint.
Hi Terry
VSSA has better PSRR that expected seen by varius measurements, but CFA loose against VFA in this point, thats why care for better PSU is taking place.
Think it has been mentionen and demonstrated earlyer how much better electrical measures will benefit from this circuit, but different to hear for many.
I have one suggestion to listening trials with possibility to better hear the difference when you set the varius dropout voltages:
Put good headphones on outputs instead (mayby with resistive load also), in this setup you get much closer to soundstage/studio and the possibility to hear the difference.
If it does not help then wait till you get a VSSA finished, it is a speedy/wideband beast with poorer PSRR than VFA, then you mayby be able to hear varius different voltage dropout settings on CM Circuit.
Ricky
PS Just got 4 pcb's from Pete myself, but will take some time before i get to point with trying them.
You can do a rough measurement with any multimeter which has an AC setting. Not accurately, but you can get a rough idea what the ripple voltage is before and after the filter. After the filter, AC should read zero or close to zero. That would take a lot of electrolytic caps to accomplish.
The power supplies LC recommends are small, compared to a linear supply which uses a toroid. If you look at his chassis, you can see the toroid would be a tight fit.
To get a relative comparison of the PSRR, you can look at the simulations posted by Esperado in the VSSA thread (scroll down to the section on Power Supply Rejection Ratio):
http://www.diyaudio.com/forums/vend...lateral-mosfet-amplifier-228.html#post3612161
post #2280 in case the link does not work
1) In general, when I compare the sound, it is a little better with a cleaner power supply, like a bench supply, or a Cap Multiplier.
2) With a conventional supply, I can also see some noise on the output on my scope. Not much, but it is there. All three versions of VSSA/PeeCeeBee act the same in this regard. With sensitive speakers, you may be able to hear that, not sure.
edit:
I should say that I am not normally too concerned about the output of a diy amplifier being "dead silent", like some people, when idle. I am more concerned about the power supply noise modulating the signal, which leads to degraded sound, often without being able to say why.
Last edited:
LC felt the amp sounded better with the SMPS, a subjective opinion but I won't argue with it.
The advantage I see is that the SMPS had its ripple frequency above the audio band and thus shouldn't have an audible effect on the VSSA's performance as far as ripple is concerned and the higher frequency also 'tops off' the reservoir capacitors more often the make the rails seem stiffer. The capacitance multiplier should allow a basic linear supply to approach, match or exceed the performance of the SMPS without the issues of a fully regulated linear supply.
All this said I will give you strong encouragement to get a set of the PeeCeeBee version amplifiers built, pick a flavour as they are all based on the basic VSSA framework and I'm sure won't disappoint. I'm listening to mine as I type, and it is sitting on top of a Bryston 3B 😉.
The advantage I see is that the SMPS had its ripple frequency above the audio band and thus shouldn't have an audible effect on the VSSA's performance as far as ripple is concerned and the higher frequency also 'tops off' the reservoir capacitors more often the make the rails seem stiffer. The capacitance multiplier should allow a basic linear supply to approach, match or exceed the performance of the SMPS without the issues of a fully regulated linear supply.
All this said I will give you strong encouragement to get a set of the PeeCeeBee version amplifiers built, pick a flavour as they are all based on the basic VSSA framework and I'm sure won't disappoint. I'm listening to mine as I type, and it is sitting on top of a Bryston 3B 😉.
OK, I just measured mine and at the fuses it measures approximately 4.5mVAC though it is jumping around a little. At the outputs it measures about 1.3mVAC, again jumping around a little. Of course that is with no load.
Don't get me wrong, I'm not against this. I enjoyed building them. I would like to see what it would take to increase the voltage they will do. I'm doing a Honey Badger right now and I will be about 65-0-65VDC rails on that.
Hi Ricky,
I canceled my VSSA order. I plan to build one off the Peeceebee thread if I can ever make up my mind which design to go with. I've been waiting for Pete to finish he latest design for it. I'll probably go that way if he like it.
Ha, Like I just told Ricky, haven't made up my mind which to build.
Funny, I'm listening to my X-BosoZ through a Perreaux 9000B as I type this. 😀
- Status
- Not open for further replies.