Since there seems to be a lot of opinions on the best design for a UCD or Zappulse power supply, and the advice is rather scattered over a number of threads, I think it would be useful to document a decent basic design here. Hopefully others will also post tweaks as well.
In any design we need some parameters to optimise to, so I am going to loosely propose the following (which might hopefully be applicable to most builders)
- Need a good basic design. There is little point trying to add exotic parts for a 1% improvement if a better design exists that gives a much better improvement
- Stereo amps. Two modules in a box
- Modules placed physically apart (to reduce interference, and give more options to control heat)
- Case sizes need to be reasonable 80mm to 120mm maximum. For most purposes, this probably means that dual large torroids are not feasible, hence dual mono supplies are going to be difficult to implement. Note that LCAudio do have some very slim 200va torroids where you can pack 2 in a 1U case, and 4 in a 2U case!
- These digital modules appear not to need quite such high end supplies as a more traditional design. Therefore the design should be simplified appropriately.
Now, from reading here:
http://www.tnt-audio.com/clinica/ssps1_e.html
..and also from the notes on the UCD180/400 threads, and from comments from Lars and Jan-Peter, I believe the following design is suitable for a decent digital amp power supply:
However, there are still some unknowns that I don't quite understand:
Firstly, the small caps across each diode were suggested in the UCD180 thread.. Are they still appropriate to this design and have I drawn them in the correct place? In the UCD 180 thread it was suggested that they only be place on the AC to DC negative diodes, is that advice good, and how would it relate to the design above? The LC Audio design does not include these
Secondly, LCAudio, and the link above recommend a small cap placed (100-330nF) between the two ac inputs to each rectifier. Should we add this to the design here?
Thirdly, does the use of double rectifiers make much useful difference? Would, say, two torroids and single rectifiers per channel be more useful for these modules in practice?
Finally, I drew in the two pairs of caps to be physically closer to the modules. Is this sensible/necessary?
Does anyone have any comments on this design? Can we improve it (usefully)? Any suggestions on minimum sizes for the various components (given a certain application)
In my case, I have a choice of a single large 1000Va torroid + 2 rectifiers to feed two UCD400s. Or I have a pair of slimline 200Va torroids that I could use with one rectifier per channel (dual mono though). Any thoughts on how these two options would stack up?
Thanks
P.S. Anyone know why my images aren't appearing inline?
In any design we need some parameters to optimise to, so I am going to loosely propose the following (which might hopefully be applicable to most builders)
- Need a good basic design. There is little point trying to add exotic parts for a 1% improvement if a better design exists that gives a much better improvement
- Stereo amps. Two modules in a box
- Modules placed physically apart (to reduce interference, and give more options to control heat)
- Case sizes need to be reasonable 80mm to 120mm maximum. For most purposes, this probably means that dual large torroids are not feasible, hence dual mono supplies are going to be difficult to implement. Note that LCAudio do have some very slim 200va torroids where you can pack 2 in a 1U case, and 4 in a 2U case!
- These digital modules appear not to need quite such high end supplies as a more traditional design. Therefore the design should be simplified appropriately.
Now, from reading here:
http://www.tnt-audio.com/clinica/ssps1_e.html
..and also from the notes on the UCD180/400 threads, and from comments from Lars and Jan-Peter, I believe the following design is suitable for a decent digital amp power supply:
However, there are still some unknowns that I don't quite understand:
Firstly, the small caps across each diode were suggested in the UCD180 thread.. Are they still appropriate to this design and have I drawn them in the correct place? In the UCD 180 thread it was suggested that they only be place on the AC to DC negative diodes, is that advice good, and how would it relate to the design above? The LC Audio design does not include these
Secondly, LCAudio, and the link above recommend a small cap placed (100-330nF) between the two ac inputs to each rectifier. Should we add this to the design here?
Thirdly, does the use of double rectifiers make much useful difference? Would, say, two torroids and single rectifiers per channel be more useful for these modules in practice?
Finally, I drew in the two pairs of caps to be physically closer to the modules. Is this sensible/necessary?
Does anyone have any comments on this design? Can we improve it (usefully)? Any suggestions on minimum sizes for the various components (given a certain application)
In my case, I have a choice of a single large 1000Va torroid + 2 rectifiers to feed two UCD400s. Or I have a pair of slimline 200Va torroids that I could use with one rectifier per channel (dual mono though). Any thoughts on how these two options would stack up?
Thanks
P.S. Anyone know why my images aren't appearing inline?
Interesting topic.
I consider buying the ultrafast diode bridges from the group buy elsewere in this forum for my zappulses (2.2 SE) or alike
Probably one toroid per zap, so I'll need four single bridges.
Whether the capacity will be a singe big one per rail or a configuration as per TNT suggestion I do not no yet.
The 200VA seem to small. It has been regularly stated that the Zap modules benefit from sufficiently sized transformers to really make them sing.
What is unclear to me is whether that only relates to VA rating or also to the secundary voltage.
I consider buying the ultrafast diode bridges from the group buy elsewere in this forum for my zappulses (2.2 SE) or alike
Probably one toroid per zap, so I'll need four single bridges.
Whether the capacity will be a singe big one per rail or a configuration as per TNT suggestion I do not no yet.
The 200VA seem to small. It has been regularly stated that the Zap modules benefit from sufficiently sized transformers to really make them sing.
What is unclear to me is whether that only relates to VA rating or also to the secundary voltage.
I am aware of that, I'm still indoubt as I have 8* 68.000Uf in stock. Unfortunate they are 50V. I'm still considering to go to a 33V tranny as I do not need the full power of the zapp but would like to have the current (2*800Va tranny or alike)
I have some time before building starts. Tomorrow I'll get the key of a new house so I will be building
although the amp will have to wait. The zapps are on stock waiting to be used.
I have some time before building starts. Tomorrow I'll get the key of a new house so I will be building
although the amp will have to wait. The zapps are on stock waiting to be used.
Softstart circuit
Hi,
In the last days I experimented with a softstart circuit. The result is to be seen here. May be, this circuit was done before anywhere, but I did not find this version. The sum of all elements of the PSU's softstart part cost below 10 Euro. It is very simple too.
The SCR's gate is very sensitive, so high resistor values work well. It's extra supply with the two 1N4004 is not tested yet, I connected the circuit to the 10000µF caps directly until now. This version's disadvantage was the long switch-off time (appr. 10 seconds, dependend on the on-state duration). With the extra supply this will be shortened to below a second, hopefully.
The diac is necessary to turn on the SCR safely (another pulse generator may do that too). The two resistors at the relays' contact hold the inrush current below 2A. It takes around 0,7 seconds to load the main caps up to appr. 5V below the nominal supply voltage. After that the relays switches on and the caps will be charged fully. The circuit works fine.
Any suggestions for advancements?
Best regards, Timo
Hi,
In the last days I experimented with a softstart circuit. The result is to be seen here. May be, this circuit was done before anywhere, but I did not find this version. The sum of all elements of the PSU's softstart part cost below 10 Euro. It is very simple too.
The SCR's gate is very sensitive, so high resistor values work well. It's extra supply with the two 1N4004 is not tested yet, I connected the circuit to the 10000µF caps directly until now. This version's disadvantage was the long switch-off time (appr. 10 seconds, dependend on the on-state duration). With the extra supply this will be shortened to below a second, hopefully.
The diac is necessary to turn on the SCR safely (another pulse generator may do that too). The two resistors at the relays' contact hold the inrush current below 2A. It takes around 0,7 seconds to load the main caps up to appr. 5V below the nominal supply voltage. After that the relays switches on and the caps will be charged fully. The circuit works fine.
Any suggestions for advancements?
Best regards, Timo
Softstart addendum
Good Evening,
Some modifications of the circuit above:
C4: 100µF -> 22µF,
added diode (small signal) between C1+ (anode) and C4+ (cathode),
Resistor 15k between C4+ and GND.
And the diodes D5 and D6 were added, of course, while C4+ was disconnected from the positive rail.
This modification shortens the relay switch-off time to below 0.5s (after mains switch). This time is independent from the rail voltage now. That means, if the mains switch is off for a period longer than 0.5 seconds, the softstart is functioning, otherwise (well below 0.5s), the relay switches immediately on. That's not too bad, because the rail voltages are still near to their nominal value. So there cannot be an exceptional inrush current in this case.
The supply voltage of the relay has a ripple of about 3Vpp due to the small filter cap of 22µF. This is ok, because there is no connection to the rest of the amplifier circuit except ground and secondary transformer AC. And the currents are very low, therefore I do not expect any disturbances.
The delay is in the range between 0.5 and 1 second with the given values (on my bench).
The relay is a 16A-type for 48V nominal supply, consuming less than 10mA.
I would be happy about some criticism. Maybe, this circuit could get better.
Best regards, Timo
Good Evening,
Some modifications of the circuit above:
C4: 100µF -> 22µF,
added diode (small signal) between C1+ (anode) and C4+ (cathode),
Resistor 15k between C4+ and GND.
And the diodes D5 and D6 were added, of course, while C4+ was disconnected from the positive rail.
This modification shortens the relay switch-off time to below 0.5s (after mains switch). This time is independent from the rail voltage now. That means, if the mains switch is off for a period longer than 0.5 seconds, the softstart is functioning, otherwise (well below 0.5s), the relay switches immediately on. That's not too bad, because the rail voltages are still near to their nominal value. So there cannot be an exceptional inrush current in this case.
The supply voltage of the relay has a ripple of about 3Vpp due to the small filter cap of 22µF. This is ok, because there is no connection to the rest of the amplifier circuit except ground and secondary transformer AC. And the currents are very low, therefore I do not expect any disturbances.
The delay is in the range between 0.5 and 1 second with the given values (on my bench).
The relay is a 16A-type for 48V nominal supply, consuming less than 10mA.
I would be happy about some criticism. Maybe, this circuit could get better.
Best regards, Timo
What I am curious to know is, how important is the use of the ferrite beads? Also would using 2 10000uf capacitors per modules have any negative effect on the sound, over just using 1. Or how about 1 15000uf cap per module. If cost was absolutely no object, what would be the very best way to go for the absolute highest sound quality? Considering the efficency of the modules, and ability for them to operate with such a small power supply, compared to class a, or class a\b designs, cost wise a person can put alot better quality parts into the power supply, and still keep the project cost low. Hopefully achieving better over all sound quality then class a or a\b designs, costing the same price. At least thats what I'm hoping.
I've built several versions with zappulse 2.2SE modules.
Started with one stereo version with a kit power supply built 12 pcs 100v deawoo caps. 550VA 55V torroid. one bridge built of 4 fast skotchky diodes with polycap bypass.
This sounded terrible, especially the treble was very bad. sounded more like a first generation (well maybe second) soundblaster card.
the second step I did was shortening the cables between capacitors and modules form about 15cm to 8 cm. This gave a modest but good audible improvement.
I only swapped the torroid for with a 1000VA 42 volt version and the sound quality dramaticlly improved especially the treble.
After that I built my own power supply according this scheme: http://www.zero-distortion.com/techno/powersupply/powersi_05.htm
I basically used 2 1000VA 42 volt torroids, 4 BC components 33000uF capacitors per channel, 2 skotchky based bridges per channel including all bypass capacitors on the bridges and the capacitors before the bridge (this should filter HF rubbish from the mains) I've also installed 10 watt bleeder resitors
This lead in to a very good sounding amplifier with sweet treble, like a tube amp and very deep bass like an old Krell. (I actually was not aware that my speakers could go so deep and found that my REL subwoofer became obsolete when listening to music.
I also built 2 monoblocks and have NOT encountered any problems with unsynched switching frequencies so i had no background noise (until one module broke down and started humming and whistling a lot)
I used to have cabling between the poles of the capcitors but I've changed that to solid 5mm copper plate, this was more for mechanical reasons. I also added a lot of bitumen damping to the aluminium housing of my amps.
Apart for the defect (that is being repaired) I'm very happy with my current result but I wonder if it is not just a lot of overkill.
- LC audio suggested smaller and better quality capacitors: because I have low impedance and current hungry speakers I decided to go for the larger bank.
- According LC audio a 500va torroid should be sufficient: without any doubt I do not agree with that statement. The change of the transformer was by far the biggest improvement.
I believe (but can not explain why) that a current capability of the power supply is by far the most important part when designing a power supply for a zapp amplifier. This might be less important when speakers are used with good sensitivity.
BTW: I also have added a softstart circuit from ESP in each monoblock, it was impossible to start the amps without blowing the mains fuses and I had to change the timing because I need a few seconds to slow-charge the capacitors, instead of the suggested 1/2 second for these softstart units.
Started with one stereo version with a kit power supply built 12 pcs 100v deawoo caps. 550VA 55V torroid. one bridge built of 4 fast skotchky diodes with polycap bypass.
This sounded terrible, especially the treble was very bad. sounded more like a first generation (well maybe second) soundblaster card.
the second step I did was shortening the cables between capacitors and modules form about 15cm to 8 cm. This gave a modest but good audible improvement.
I only swapped the torroid for with a 1000VA 42 volt version and the sound quality dramaticlly improved especially the treble.
After that I built my own power supply according this scheme: http://www.zero-distortion.com/techno/powersupply/powersi_05.htm
I basically used 2 1000VA 42 volt torroids, 4 BC components 33000uF capacitors per channel, 2 skotchky based bridges per channel including all bypass capacitors on the bridges and the capacitors before the bridge (this should filter HF rubbish from the mains) I've also installed 10 watt bleeder resitors
This lead in to a very good sounding amplifier with sweet treble, like a tube amp and very deep bass like an old Krell. (I actually was not aware that my speakers could go so deep and found that my REL subwoofer became obsolete when listening to music.
I also built 2 monoblocks and have NOT encountered any problems with unsynched switching frequencies so i had no background noise (until one module broke down and started humming and whistling a lot)
I used to have cabling between the poles of the capcitors but I've changed that to solid 5mm copper plate, this was more for mechanical reasons. I also added a lot of bitumen damping to the aluminium housing of my amps.
Apart for the defect (that is being repaired) I'm very happy with my current result but I wonder if it is not just a lot of overkill.
- LC audio suggested smaller and better quality capacitors: because I have low impedance and current hungry speakers I decided to go for the larger bank.
- According LC audio a 500va torroid should be sufficient: without any doubt I do not agree with that statement. The change of the transformer was by far the biggest improvement.
I believe (but can not explain why) that a current capability of the power supply is by far the most important part when designing a power supply for a zapp amplifier. This might be less important when speakers are used with good sensitivity.
BTW: I also have added a softstart circuit from ESP in each monoblock, it was impossible to start the amps without blowing the mains fuses and I had to change the timing because I need a few seconds to slow-charge the capacitors, instead of the suggested 1/2 second for these softstart units.
What Ericpeters did, is almost identical to my psu with my Zappulse monoblock. I have not filtercaps over the diodebridges, and no bleeders. Otherwise, this is the best result I have made with the zappulse modules.
I use 2 pcs Elna Cerafine 18000uf/module (monoblock), and 1pcs 800VA trannie 42 v/channel (monoblock).
Internal cabling is pure silver, except the power cabling.
Best Regards
I use 2 pcs Elna Cerafine 18000uf/module (monoblock), and 1pcs 800VA trannie 42 v/channel (monoblock).
Internal cabling is pure silver, except the power cabling.
Best Regards
Perhaps a stupid question...
Given the efficiencies of these amps, is it at all feasible to use a bank of SLA batteries and a charger as a power supply? That is, if one was willing to ignore size and weight. What would it take in terms of amps-hours to run, say, a UcD180? I'm not very knowlegeable about this stuff. But, the idea intrigues me when I consider trying to find and afford good capacitors.
Are there other potential problems besides current capacity and/or recharge time, to using batteries?
I seldom listen to my system for more than a few hours, giving the rest of the day for the batteries to recharge.
Is this a ridiculous notion?
Thanks.
Paul Ebert
Given the efficiencies of these amps, is it at all feasible to use a bank of SLA batteries and a charger as a power supply? That is, if one was willing to ignore size and weight. What would it take in terms of amps-hours to run, say, a UcD180? I'm not very knowlegeable about this stuff. But, the idea intrigues me when I consider trying to find and afford good capacitors.
Are there other potential problems besides current capacity and/or recharge time, to using batteries?
I seldom listen to my system for more than a few hours, giving the rest of the day for the batteries to recharge.
Is this a ridiculous notion?
Thanks.
Paul Ebert
ericpeters said:
Hi, this is a really interesting comment, because it pretty much confirms something that I am seeing here. My second amp with single 1,000va 45V torroid is heaps better than the dual 200va amp. However, there are other differences as well like spacing out the two modules, and removing the Y caps on the power inlet, also triple the size in main caps. However, I suspect that the Y caps in the main filter might be a big part of the problem...?
I have that whistling going on one of my modules. It hadn't really occured to me that the module might simply be broken though..
Just a quick question though. How big a difference did you find your final design was from the original version with the 1,000va torroid and shorter wires? Just curious as to the benefits of all that extra work (was it a big change or just a little better..?)
ewildgoose said:
Very difficult to say because I had about 2 days without listening between dismantling the first design and starting to listen with the new ones. My feeling is that bass improved and also imaging and treble.
I would probably built the same monoblocks again but use a normal heavy duty rectifier bridge if I would built them again.
Because one module is being repaired I'm now looking to change the layout to make the wires as short as possible. I will also change the low-level signal wires to single strand pure silver. (I currently use teflon isolated twisted pair copper (CAT 5) which is quite good but I do have plenty of silver wire laying around and can't use the amps anyway until my module get's returned.
My Zapp module PROBABLY broke down after feeding source into the input of a module without power supply, with disconnected speakers (but no short circuit, open circuit) and then while this was the case it was switched on. The result was a high pitch whistle and humming and other background noise. This noise is also there when the other monoblock is switched off. I tried several scenarios but the noises were alwas related to that one single zap module.
Before that problem both amps were absolutely silent.
My next project will be using the old power supply from the first design to drive 2 Ice modules, I want to test how small the power supply for a class-d amp can be built without loosing quality but when using power hungry speakers. I already tried this with the stock evaluation kit (400VA transformer) but this was unsatisfying. I will first put another transformer on the kit and after that I will move to a bigger capcitor bank and test both transformers.
So you guys are saying a larger transformer would be more of an improvement than a large bank of caps? My amp case is pretty limited in size, so I could only go really big on one area.
I am currently using a 500va toroid, and the stock predator PS.
would it be better to swap out the 500va with a 1000va. or just add another 500va? If I do that, I think I have enough room to add 4 more 10k caps.
And, what kind of speakers are you guys running?
I had a chance to try my amp on a pair of B&W 703's, and the sound was absolutely stunning. It beat all the other amps they had there. On a pair of Sonus Faber's, it sounded downright horrible.
I am currently using a 500va toroid, and the stock predator PS.
would it be better to swap out the 500va with a 1000va. or just add another 500va? If I do that, I think I have enough room to add 4 more 10k caps.
And, what kind of speakers are you guys running?
I had a chance to try my amp on a pair of B&W 703's, and the sound was absolutely stunning. It beat all the other amps they had there. On a pair of Sonus Faber's, it sounded downright horrible.
mattjk said:
I running on Apogees; my main speakers ar Duetta signatures but I also got Calipers. They do require a lot of current and the D-sigs got an almost flat impedance of 3 ohm, the calipers smaller but even less efficient. That's one of the things I don't know: is the size of the transformer directly related to sound quality or is there also a strong relationship with the efficiency of the loudspeaker?
I would start with the transformer. (Try Farnell i found their prices for Torroids are 1/2 the price I paid for mine.)
I hooked up a 100V 82,000uF capacitor to only one channel tonight. I set my preamp to mono and I listened to one channel at a time. It definately made an improvement in the high frequency. Cymbals sound a bit more detailed and forward. Before it was rolled off and too laid back sounding.
Still too soft sounding though.
Still too soft sounding though.
Now, I hooked up 3 additional 10,000uf caps to the other channel, for a total of 40,000uf.
I can't tell the difference between the two, but the amp certainly sounds more lively. There is a certain song that is all acoustical guitar. It was unlistenable before, but now it sounds very very good, but not perfect. Time to try a larger tranny now.
The overall sound seems more effortless as well.
I am running acoustat 2+2 speakers, BTW.
still too rolled off though.
I can't tell the difference between the two, but the amp certainly sounds more lively. There is a certain song that is all acoustical guitar. It was unlistenable before, but now it sounds very very good, but not perfect. Time to try a larger tranny now.
The overall sound seems more effortless as well.
I am running acoustat 2+2 speakers, BTW.
still too rolled off though.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.