Optimal supply design for UCD and Zappulse modules

[ghemink]

C'mon guys,
will nobody criticize my wiring nor my, in my view, indispensable power filter?

Equi=tech ask big money for its transformers, though you can buy or wire one for much less.
This "technical balanced power" is said to allow even addition of Y caps, from secondaries to ground, without noise contamination of gound, against wich Bruno also warned us.

Just see post #26 from Jan Didden, wich states the extreme PS dependency of UCD, to achieve superior sound.

Regards
Mauricio

PS looks good to me. I also like the balanced power and filter. Have no balanced power yet myself but I'm planning to do it for my audio equipment.

Gertjan

 

[Bgt]

just testing 4x Vishay102 10.000uF/63V caps.(replaces the 154 series) in my PSU for the UCD400's.
Sounds very nice. Although the specs(ESR) look a bit worse I cannot notice a degradation in bass, high tones are superb.
Comparing 3 stereo amps. 1 UCD with 154's another with BCC AL30's and 1 with the Vishay. All UCD 400's.
The BC/Vishay family sounds the best.

 

[Portlandmike]

Re: Thread resurrection:

I thought about resurrect this interesting thread. It has some good info that passed overlooked to my eyes.

After my initial success with my UCD180 I had bad HF noise with my UCD400 monoblocks, wich led me to ask our nice big brothers here at the wonderful forums.

Eminent EE's don't agree about the best PS wiring scheme for UCD. The picture is what I tried last night: PS is floating, or I least think so, with only T-piece bolts attaching to chassis. I added tween rectifiers (no snubers). My fear of lack of VDC symetry was absurd: perfect symetry, even without grounding. I have common type 35A diode bridges on one monoblock and Ixiz, 68A, on the other.
Wiring/solder are still burning-in but I find this is the least noisy wiring that I tried so far.
I still use copper plate for PS caps ground because of mechanical reason and easiness of wiring.

An externally hosted image should be here but it was not working when we last tested it.

[/URL][/IMG]

I will next add a Power filter, "technical balanced power", wich is represented on the next drawing. Now it is serving the secondary UCD180 based system . All I can say is "enhanced clarity". I'll report back when burn-in proceeds.
Also I will try XLR's Pin 1 lifted from chassis,

[URL=http://imageshack.us][IMG][URL=http://imageshack.us]

Please refer to:
[url]http://www.equitech.com/articles/articles.html[/url]
For technical explanation about iso-Tx.

Any comments, regading snubbers, wiring or other topics are welcome
Regards
Mauricio :) [/B][/QUOTE]

Mauricio,

You are a busy guy. You must not have kids :)

This all looks good, but I'd be concerned about the inductors your adding. Not that they may not help, but they may hurt too.

That will increase the inductance to the bridge, and you will lower the ring of that snubbers are suppose to cure. You should try adding yet another cap accross the rails, perhaps to ground as you allude to, after the inductors. Thus a CLC filter.
that second C ( or two if grounded) could be on the secondary side of the transformer too. AND, you will likely want to add a RC in addition to the second set of C's to damp out bridge ringing.
I know I promised some snubber advice for you, but don't have it yet.

You need to loose the plate, or maybe get out a hack saw and cut some slots in it. Can you post a picture of your chassis layout again?

Best Regards,

Mike

 

[classd4sure]

Re: Re: Thread resurrection:

Mauricio,

You are a busy guy. You must not have kids

This all looks good, but I'd be concerned about the inductors your adding. Not that they may not help, but they may hurt too.

That will increase the inductance to the bridge, and you will lower the ring of that snubbers are suppose to cure. You should try adding yet another cap accross the rails, perhaps to ground as you allude to, after the inductors. Thus a CLC filter.
that second C ( or two if grounded) could be on the secondary side of the transformer too. AND, you will likely want to add a RC in addition to the second set of C's to damp out bridge ringing.
I know I promised some snubber advice for you, but don't have it yet.

You need to loose the plate, or maybe get out a hack saw and cut some slots in it. Can you post a picture of your chassis layout again?

Best Regards,

Mike

Hi,

Mike I think that's just a power conditioner off an isolation xformer in addition to the above supply.


Mauricio, I can't comment much on the plate aspect, will leave that to Mike, I haven't experimented with it. I think slotting it may make sense though, to direct the current further? Maybe a healthy plastering of fine holes or slots for damping eddy currents too? Who knows.

One area to perhaps experiment with is how you tap the common point from the bridge rectifiers.

For instance it may be advantageous to connect the common point of the bridges with a short thick lead, and tap into that in the middle of that wire, with another one going over to your caps.

 

[Portlandmike]

Re: Re: Re: Thread resurrection:

Hi,

Mike I think that's just a power conditioner off an isolation xformer in addition to the above supply.


Mauricio, I can't comment much on the plate aspect, will leave that to Mike, I haven't experimented with it. I think slotting it may make sense though, to direct the current further? Maybe a healthy plastering of fine holes or slots for damping eddy currents too? Who knows.

One area to perhaps experiment with is how you tap the common point from the bridge rectifiers.

For instance it may be advantageous to connect the common point of the bridges with a short thick lead, and tap into that in the middle of that wire, with another one going over to your caps.

Chris,

You get at my point about the plate. I don't like how it will have the big diode currents flowing through it, and modulating the potential of ground. It goes along way to defeat this ideal approach to power supplies. I guarantee it will be better if the cap terminals are treated as 4 ports. That is, rectifier connections come in one side, amp rails go out the other. He's got monoblocks, so he doesn't have the compromise you have. He could tie the cap grounds at the UCD if he wanted to.

Oh, as for the power conditioner.

http://www.zero-distortion.com/techno/powersupply/powersi_05.htm

This link you are framiliar with deals with it. I just don't see any cap accross the secondaries that will stiffen them up at high frequencies. This can cause the ring to go quite low in frequency I think. I have yet to bench the snubber thing, but EVA suggests, and I concure, that the best way to deal with ringing in the diodes is to add a RC accross the secondaries rather than accross the diodes. Across the diodes is connecting the mains with great HF caps, to the main large electrolytics, with lots of L.
Just seems like a bad idea. Nothing to back it up with yet, listenign or measurement.

Regards,

Mike

 

[classd4sure]

Re: Re: Re: Re: Thread resurrection:

Chris,

You get at my point about the plate. I don't like how it will have the big diode currents flowing through it, and modulating the potential of ground. It goes along way to defeat this ideal approach to power supplies. I guarantee it will be better if the cap terminals are treated as 4 ports. That is, rectifier connections come in one side, amp rails go out the other. He's got monoblocks, so he doesn't have the compromise you have. He could tie the cap grounds at the UCD if he wanted to.

Oh, as for the power conditioner.

http://www.zero-distortion.com/techno/powersupply/powersi_05.htm

This link you are framiliar with deals with it. I just don't see any cap accross the secondaries that will stiffen them up at high frequencies. This can cause the ring to go quite low in frequency I think. I have yet to bench the snubber thing, but EVA suggests, and I concure, that the best way to deal with ringing in the diodes is to add a RC accross the secondaries rather than accross the diodes. Across the diodes is connecting the mains with great HF caps, to the main large electrolytics, with lots of L.
Just seems like a bad idea. Nothing to back it up with yet, listenign or measurement.

Regards,

Mike

Experimentation is the main thing n a little research.

Far as ringing in the diodes he's using FRED rectifiers like me so it shouldn't be too bad. However I'd like to try much the same snubber you're recommending because I'm not inclined to bypass in the conventional way.

Also I believe analogspiceman had a similar solution.
What values would you recommend as a good "stiff" snubber?

I have a bunch of 22nF and 47nF caps I could use.

Regards

 

[maxlorenz]

Thanks everybody for your interest and help

You are a busy guy. You must not have kids

On the contrary, I have four (4) , I think they grow up better with her mom taking care...his father has a terrible case of DIY fever

Sorry to answer too fast but I'm building something here...

About UCD's PS, I will try separate grounds ("four poles").
Now I have a 12AWG, solid OCC copper short wire joining the ( +) and (-) of bridges, then I run a wide OFC speaker cable to the copper plate.

About Power Filter, I will make some test and then decide what is best: for example I thought about putting the X1 cap on TX's primaries instead and no cap or a pair of Y2 to ground.
I could also try the TX alone without any other filtering.

As per the supposed reactive currents cancelation (see equi=tech's documents) at the grounded connection, I fear that my present PS wiring (floating, only grounded through T-piece and bolts?) could fail to take advantage of it

Anyway, the filter is now powering my UCD400 monoblocks and it's sounding good: as I said music is more "clear", less grain; the voices are really believable. Burning-in proceeds with a CD from "the Doors". I though previously that the original recording was not so good or that the late singer was experimenting the effects of some nights of excess, but now the voice is very clear and warm, kind of "suspended disbelief" with lights off
After that will go some "Van der Graaf Generator" CD's

Ciao
M

PS: later I will try to answer other issues commented

 

[Portlandmike]

Re: Re: Re: Re: Re: Thread resurrection:

Experimentation is the main thing n a little research.

Far as ringing in the diodes he's using FRED rectifiers like me so it shouldn't be too bad. However I'd like to try much the same snubber you're recommending because I'm not inclined to bypass in the conventional way.

Also I believe analogspiceman had a similar solution.
What values would you recommend as a good "stiff" snubber?

I have a bunch of 22nF and 47nF caps I could use.

Regards

Chris,

I have done some guessing simulations, and that points to a cap, rather large, and larger really won't hurt, of about 470nF to 1uF with a resistor of something in the range of 10 to 100ohms, 33 and 47ohms seeming to work out best given my sim results. These are accross the AC secondaries. The big issue is what is the transformer inductance seen looking back to the line. The line has a good share of inductance as you might guess, 20-100's of feet. The secondary inductance really goes up if the primary isn't a stiff source at high frequencies.
That all said, I'd really like to bench it. I tried a quick test, but its like a needle in a hay stack. That, and I think I was mostly looking at hash generated from the electronic balast in my lab. I really didn't give it due dilligence. I think the best way is to use a current probe. I'll keep you posted.

Mike

Thanks everybody for your interest and help



On the contrary, I have four (4) , I think they grow up better with her mom taking care...his father has a terrible case of DIY fever

Sorry to answer too fast but I'm building something here...

About UCD's PS, I will try separate grounds ("four poles").
Now I have a 12AWG, solid OCC copper short wire joining the ( +) and (-) of bridges, then I run a wide OFC speaker cable to the copper plate.

About Power Filter, I will make some test and then decide what is best: for example I thought about putting the X1 cap on TX's primaries instead and no cap or a pair of Y2 to ground.
I could also try the TX alone without any other filtering.

As per the supposed reactive currents cancelation (see equi=tech's documents) at the grounded connection, I fear that my present PS wiring (floating, only grounded through T-piece and bolts?) could fail to take advantage of it

Anyway, the filter is now powering my UCD400 monoblocks and it's sounding good: as I said music is more "clear", less grain; the voices are really believable. Burning-in proceeds with a CD from "the Doors". I though previously that the original recording was not so good or that the late singer was experimenting the effects of some nights of excess, but now the voice is very clear and warm, kind of "suspended disbelief" with lights off
After that will go some "Van der Graaf Generator" CD's

Ciao
M

PS: later I will try to answer other issues commented

M,

I was thinking the same thing about caps to ground. I really don't see any huge advantage and the potential downside is big. But if you've got more time then me...

Mike

 

[maxlorenz]

Mike:

All this transformer quest was enhanced in my spirit because of a post from a french coleague that swares that his C-core (or R-core) transformer based UCD was way better than a toroidal based UCD. Since sourcing BIG C or R-core TX's isn't easy I supposed an isolation TX would approach (or better) the former's performance.
Later, I was taught about "technical balanced power" on another, almost frustrated thread of mine, by dear Paul, AKA pmkap. See, for instance:

With this configuration, I'd also suggest you consider using Y capacitors to ground at the output on each of the output legs. With unbalanced power, using Y caps to ground. (this does not preclude your use of the ferrites). With typical 'hot' - 'neutral' mains, the use of Y caps can be problematic, often tripping leakage fault devices and/or shunting '****' and contaminating the ground line. But realize that differential(transverse) noise becomes balanced differential noise at the outputs, and because the shunted noise is nominally antiphase, they will 'cancel'.

Today I will make some experiments without caps (wich must have been done in the first place, anyway; we loose objectivity when we behave as fans ).

Enjoy
M

 

[Robert F]

Hi,

I must admit I have always loaded the secondaries as opposed to using RC networks accross each diode as per suggestions from Allen Wright and Ben Duncan. Their suggestions in relation to low voltage secondaries at least is to use a Resistor only. In higher voltage applications the cap really only prevents the resistor dissipating too much power and robbing too much "headroom" from the power transformer.

Allen's explanation is that the resistor (or the RC) keeps the secondary loaded thereby preventing or suppressing the ringing excited by the diode snapping off and the secondary suddenly becoming unloaded.

I have no means of measuring the leakage inductance of power transformers, so it has just been a matter of trial and error for me.

The last power amp I built (UCD 180) ended up with a 0.1uf cap and from memory a 47 ohm R accross each secondary. It may not have been optimised but it certainly was an improvement sonically, and that was with using those IXYS hexfred bridge packages.


Rob.

 

[Portlandmike]

Hi,

I must admit I have always loaded the secondaries as opposed to using RC networks accross each diode as per suggestions from Allen Wright and Ben Duncan. Their suggestions in relation to low voltage secondaries at least is to use a Resistor only. In higher voltage applications the cap really only prevents the resistor dissipating too much power and robbing too much "headroom" from the power transformer.

Allen's explanation is that the resistor (or the RC) keeps the secondary loaded thereby preventing or suppressing the ringing excited by the diode snapping off and the secondary suddenly becoming unloaded.

I have no means of measuring the leakage inductance of power transformers, so it has just been a matter of trial and error for me.

The last power amp I built (UCD 180) ended up with a 0.1uf cap and from memory a 47 ohm R accross each secondary. It may not have been optimised but it certainly was an improvement sonically, and that was with using those IXYS hexfred bridge packages.


Rob.

Rob,

thanks for your comments. I tend to agree that a resistor is best, but they sure get crazy power wise in a big rail. I see no possible downside to loading the secondaries with a RC, but putting caps accross the diodes seems like a step in the wrong direction.
Why would one want to put the energy from the secondary, 100 of uH or more depending on wiring and line treatment, into the main supply caps when the diode turns off?
I guess it works.... or does it, but it just seems like the secondaries is the right place to do it suck up this current discontinuity.
I think the ideal solution is a Cap accross the secondaries, say 220nF, with a snubber, say 1uF, 47 ohms. The resistance is a guess based on simulations, but its tough to nail given the incomming line impedance. I think a X cap on the transformer input would define the HF inductance of the secondaries for these purposes.


By the way, I tried to look at this ringing everyone talked about, and could not see it on a ordinary 50A bridge. I had a current probe on the line too. Anyone with with a tip on how to measure, and see, this would be greatly appreciated.
Mike

 

[Portlandmike]

Rob,
discontinuity.
I think the ideal solution is a Cap accross the secondaries, say 220nF, with a snubber, say 1uF, 47 ohms. The resistance is a guess based on simulations, but its tough to nail given the incomming line impedance. I think a X cap on the transformer input would define the HF inductance of the secondaries for these purposes.
Mike

Just a note of correction. After simulation, I would NOT use a cap on the primaries or secondaries with the suggested 1uF, 47 ohms.
Turns out 470nF, 47 ohm seems to be pretty good.

Not saying this is ideal, but its obvious from the simulations that adding a cap on the secondaries can cause the ringing to drop in frequency enough that the snubber will need to be retuned, bigger, hotter, more $$$....
It is worth noting that adding a cap there does slow the slew of the voltage alot, so its likely the best solution, but as in many thing, it starts to get more complicated and for sure more expensive.

Rob, Your values are good. There is benifit to up that cap value to 470nF in my sims though. No ring vs one complete cycle.

Mike

 

[maxlorenz]

I found this interesting and maybe well known article about snubbers. I post it here in case someone missed it. I can't say I fully understand it but I sniff its consequences. I liked the TX part of it

http://www.hagtech.com/pdf/snubber.pdf

Enjoy
M

 

[Portlandmike]

I found this interesting and maybe well known article about snubbers. I post it here in case someone missed it. I can't say I fully understand it but I sniff its consequences. I liked the TX part of it

http://www.hagtech.com/pdf/snubber.pdf

Enjoy
M

It seems he agrees about putting them accross the secondaries!
Good article.


Mike

 

[Robert F]

Just a note of correction. After simulation, I would NOT use a cap on the primaries or secondaries with the suggested 1uF, 47 ohms.
Turns out 470nF, 47 ohm seems to be pretty good.

Not saying this is ideal, but its obvious from the simulations that adding a cap on the secondaries can cause the ringing to drop in frequency enough that the snubber will need to be retuned, bigger, hotter, more $$$....
It is worth noting that adding a cap there does slow the slew of the voltage alot, so its likely the best solution, but as in many thing, it starts to get more complicated and for sure more expensive.

Rob, Your values are good. There is benifit to up that cap value to 470nF in my sims though. No ring vs one complete cycle.

Mike

Thanks Mike I will give your values a go. I tried a few from my parts box, I am pleased to see I got close!

Regards,
Rob.

 

[classd4sure]

Bit of an arrogant question I suppose but how true to life is that sim?

I'm about to try 47nF //100R across the bridge... just to see what it sounds like. It's what I have and no scope to measure so it's just a listening test anyway. Could also try a 22nF // to that... not really enough difference to make it worthwhile though. Let you know what I hear. Could be an interesting case in mine given the Jensen 4 poles.

 

[classd4sure]

Bit of an arrogant question I suppose but how true to life is that sim?

I'm about to try 47nF //100R across the bridge... just to see what it sounds like. It's what I have and no scope to measure so it's just a listening test anyway. Could also try a 22nF // to that... not really enough difference to make it worthwhile though. Let you know what I hear. Could be an interesting case in mine given the Jensen 4 poles.

I meant to say 47nF + 100R and maybe // to 22nF.

I skipped the 22nF.

Could be imagination but I've played a several rather complex pieces (Telarc/Hans Zimmer junk) along with a few old favorites so far, and who knows how long it'll take the new caps to break in... but it's more dynamic than ever. Micro-detail seems to have gotten a finely honed edge, to the point that each and every instrument in full orchestral pieces have a clearly defined voice and distinct presence.

Then there's the unexpected.

On top of that it seems everything else that was good before got even better, with a whole new sense of power and dynamic range... could that be? I wish I could measure it and see what happened, but this can't have been a bad thing, seems I lucked out with this "what I had kicking around" combo!

BTW I installed them right at the rectifiers as close to the pins as possible. The caps are Digikey # 495-1614-ND
47nF 275Vac X2 rated EMI suppression cap... (might not need a resistor??)

The resistor is a standard ceramic 100R .5 watt.

Anyway,

!!!!!!!!!

 

[maxlorenz]

Hi Chris:
That was with your soft recovery diode bridge???
Or a common type DB?

Thanx
M

 

[Robert F]

I meant to say 47nF + 100R and maybe // to 22nF.

I skipped the 22nF.

Could be imagination but I've played a several rather complex pieces (Telarc/Hans Zimmer junk) along with a few old favorites so far, and who knows how long it'll take the new caps to break in... but it's more dynamic than ever. Micro-detail seems to have gotten a finely honed edge, to the point that each and every instrument in full orchestral pieces have a clearly defined voice and distinct presence.

Then there's the unexpected.

On top of that it seems everything else that was good before got even better, with a whole new sense of power and dynamic range... could that be? I wish I could measure it and see what happened, but this can't have been a bad thing, seems I lucked out with this "what I had kicking around" combo!

Chris,


BTW I installed them right at the rectifiers as close to the pins as possible. The caps are Digikey # 495-1614-ND
47nF 275Vac X2 rated EMI suppression cap... (might not need a resistor??)

The resistor is a standard ceramic 100R .5 watt.

Anyway,

!!!!!!!!!

Chris,

Your subjective experience appears to parallel mine exactly. However, a few years back I tried the cap alone and did not like the effect at all. Different amp- rectifiers, tx, filter caps of course...
BTW I think it would be worthwhie you trying a non-inductive resistor in place of your wirewound. I used some metal-oxide 2 watt jobs.

Rob.

 

[classd4sure]

Hi Chris:
That was with your soft recovery diode bridge???
Or a common type DB?

Thanx
M

That's the IXYS monster FRED.

Chris,

Your subjective experience appears to parallel mine exactly. However, a few years back I tried the cap alone and did not like the effect at all. Different amp- rectifiers, tx, filter caps of course...
BTW I think it would be worthwhie you trying a non-inductive resistor in place of your wirewound. I used some metal-oxide 2 watt jobs.

Rob.

Cool maybe I didn't imagine it all then. Thanks for the non inductive reminder, will put that on my to-do list