Tim sent me a couple of PSUD screenshots by email confirming the issue he described.
What happens is that depending on the DC load current, C1 can discharge faster than the secondary voltage after the peak. Without the choke, the diodes will just stay on past the peak and the voltage on C1 will follow the secondary voltage down for a while. At some point, the secondary voltage will be dropping faster than C1 discharges and the diodes will turn off.
With the input choke, apparently there is a delay during which the diodes are off but then later, on the same cycle, the voltages reverse and the diodes turn on again exactly as Tim described. This problem will be aggravated by: 1) lower line frequency, 2) smaller C1, 3) higher DC current. So for people experimenting with the flywheel, it's something to watch out for.
Interesting behavior that I hadn't seen or thought of. Thanks, Tim, for pointing it out.
-Henry
What happens is that depending on the DC load current, C1 can discharge faster than the secondary voltage after the peak. Without the choke, the diodes will just stay on past the peak and the voltage on C1 will follow the secondary voltage down for a while. At some point, the secondary voltage will be dropping faster than C1 discharges and the diodes will turn off.
With the input choke, apparently there is a delay during which the diodes are off but then later, on the same cycle, the voltages reverse and the diodes turn on again exactly as Tim described. This problem will be aggravated by: 1) lower line frequency, 2) smaller C1, 3) higher DC current. So for people experimenting with the flywheel, it's something to watch out for.
Interesting behavior that I hadn't seen or thought of. Thanks, Tim, for pointing it out.
-Henry
Of course you don't "see" a resonance because the load will be enough to damp it. Unfortunately your load is the 2A3 which should not be supposed to do that in order to work linearly. The consenquence of that wimpy filter will be an increased intermodulation distortion. Measure the IMD and see what an effect box. Do it again with a proper filter and compare.
Happy New Year!
Last edited:
When I said there was no resonance, I meant the "flywheel effect" is not due to the L1/C1 filter section peaking at the power line frequency or some harmonic. Figuring out why the "flywheel" behaves the way it does was interesting for me and once every couple years or so I enjoy talking about it with people. Beyond that, I don't claim the "LSES" supply is superior, or even a good idea at all.
-Henry
I went back and had a look at the first few pages of this thread. Drlowmu has been posting the same comments on AudioAsylum for years. Your replies to him are spot-on.
Ok I didn't quote properly and I didn't want to discount your findings. Do you get those results with L1=0.48H, C1=20 uF, L2=1.6H and C2=40 uF?
P.S.
Total DCR is 16.1 R and total current draw is about 90 mA.
Last edited:
Edit: L2=1.4H and total DC current is 45 mA (it's only one channel).
More precisely DCR=10 ohm for 0.48H and 6.1 ohm for 1.4H.
More precisely DCR=10 ohm for 0.48H and 6.1 ohm for 1.4H.
Sorry, I was off because of the days. I just want to thank Henry! I' ve been following this thread hoping for an explanation like that! Happy New Year everyone!
Last edited by a moderator:
I don't have a Windows system currently running to install PSUD. Perhaps you can try it yourself? It's not hard.
As a general rule, an input choke of 480mH is much too large for "flywheel" operation. I don't know for sure how the filter you describe will work. Minus the input choke it seems typical of a classic tube-era filter. With the relatively low DCR, small caps, and medium size second choke it'll probably have a bouncy step response.
You know, absent careful analysis and listening tests, so much of this is really just arbitrary instinct and going with what concepts appeal to you aesthetically. I find this time around my enthusiasm for the flywheel is diminishing. A big-old 10H choke of intermediate DCR and several hundred microfarads of capacitance in a single-stage filter has worked great for me in the past and has a lot of appeal. In a low-current application make sure to add enough bleeder resistance to keep the choke in regulation.
I have some really great parts sitting in the attic. I'm in the middle of a major motorcycle project, but after that perhaps this will be the year I finally build my Magnequest battleship amps. Stranger things have happened.
-Henry
Member
Joined 2009
Paid Member
Once you decide to 'go modern' you realize that the better solution may indeed be a solid state bridge rectifier with CRC filter.
I know it's not hard but I only use Mac's.😀
That's what should be the LSES of the latest "ultimate" amplifier by mr. drlowmu. I had already bumped in one of your posts and guessed it was too big and for that reason criticised his choice that was merely aiming to low DCR but possibly not necessarily low impedance. Of course all he could reply was the usual story of compilations but no proof.
Besides all this story about "flywheel" that wimpy filter is still a good way to get more IMD. As I have eliminated AC filmanent supply for DHT's I not going to compromise distortion again for some little benefit when the amp is running at full power for sustained periods. An amplifier should not be supposed to be used that way.
Or a proper capacitance multiplier where you don't need large electrolityc capacitors but can still get rather small ripple.
Maybe it could replace the second LC filter in the LSES. In that case, if it works, it could be interesting....
Member
Joined 2009
Paid Member
cap multipliers are not without their own warts and generally need some decent capacitance after them.
100 uF or so is not a problem anymore even for film caps if one wants to avoid electrolitics. You only need one and don't need to buy a choke.
Yes, this is a characteristic with the flywheel approach. You need to adjust for mains frequency to get the optimal response.
Of course adequate filtering is always useful too.
Ian
Last edited:
IMHO everything posted by dr.lowmu (now banned) should simply be ignored. He does not understand this approach and from everything I have seen from him, has never actually properly designed or built one.
I however, have actually tried Henry's flywheel approach. So have a few others . It appears to work just as he has suggested it would (mains frequency adjustment aside).
I do not think it is worth redesigning a power supply for though. Especially not if you are happy with your current set up. I agree that it is useful to consider, and the he parts to build it need not be expensive.
Ian
Last edited:
At work i have opportunity to Hioki expensive LCR. Any e-caps, even low esr, did go down quickly with rated µF at higher freq. Foils had very stable value from bass to treble freq. Old metal-paper cap did quite well.
"reservoir" is exact name for them, that´s what are they good at. Filtering, not much.
Since ideal supply has zero ohms impedance, to achieve that across entire hifi range- foil caps must be added. Otherwise this will be valid in bass range only
Reservoir caps are required to act as a reservoir of charge. Filtering is done later. I can just about understand people who want to use film caps for PSU filtering, but they are a waste of money for the reservoir cap.
'Bypassing' is a good way to increase the impedance of a capacitor.
'Bypassing' is a good way to increase the impedance of a capacitor.
RE: Hasquin and Swenson?
Well at least one of them has no EE degree. If they did, wouldn't it be listed in that person's bio in his many hack audiophile articles? No, can't do that because somebody may pay the $0.99 fee to check on it.
Also Swenson makes his living these days by selling useless USB repeaters based on some app note and linear power supplies for Mac computers. He goes on and on how his linear power supply reduces noise with digital audio streams over the stock switching wall wort supply. Yet he fails to acknowledge there is another switch-mode regulator right inside the MAC! Classic audiophile hogwash. "We can't fix that problem so we'll just pretend it doesn't exist".
I'm so glad to see Drlowmu banned from this otherwise fine forum. He wrecked TubeDiy with his junk science. I guess the moderators here are in fact good engineers and see right through the BS.
What are the shortcomings of capacitance multipliers?
Please could somebody explain in what way capacitance multipliers have shortcomings.
My view to date is that they provide the same order of mains & signal ripple filtering as a voltage regulator / stabilizer with a less complex circuit and with a lower voltage drop from input to output; thus they are more efficient than a voltage regulator (less heating and fewer components).
Nonetheless, so far as I have read, they get no mention by the great authorities (e.g. Nelson Pass & Ben Duncan et al).
Who need a voltage stabilizer for an audio amplifier (<100Watt) in a domestic situation?
(OK for a high power public address system or in places with wildly varying mains voltage, but I am referring to the usual domestic situation in Europe.)
Please could somebody explain in what way capacitance multipliers have shortcomings.
My view to date is that they provide the same order of mains & signal ripple filtering as a voltage regulator / stabilizer with a less complex circuit and with a lower voltage drop from input to output; thus they are more efficient than a voltage regulator (less heating and fewer components).
Nonetheless, so far as I have read, they get no mention by the great authorities (e.g. Nelson Pass & Ben Duncan et al).
Who need a voltage stabilizer for an audio amplifier (<100Watt) in a domestic situation?
(OK for a high power public address system or in places with wildly varying mains voltage, but I am referring to the usual domestic situation in Europe.)
- Status
- Not open for further replies.