Also,
When listening , I like to unplug and amplifer should continue playing for at least 5 seconds Without obvious noise and distortion ....
I have no doubt 80 000 uf is fine . The irony is that 2 KVA and 10 000 uF is cheaper than 1 KVA and lets say 88 000 uF if talking high quality ( it is in my world ) . I have a small hunch the 2 KVA / 10 000 will sound better and especially in therms of punch . Maybe even better than 700 VA and 80 000 uF . That one might be capacitor heavy ?
A well designed PSU with respect to all the usual things ( which I never seem to get right if my oscilloscope isn't lying to me ) . Would deliver massive current even if 10 000 uF . The wires to the transistors should be massive ( half inch I don't doubt , direct to collectors usually or drain ) . That causes problem number one . As massive cables cause HF reduction . The decoupling caps can rescue that problem ( usually about 220 uF ) . If the PSU voltage is about 5 V above the driver PSU voltage there never is a problem with ripple . If class A it gets more complicated . Caps on the PCB isn't such a bad idea with copper reinforcement . Almost design the PSU first if doing that .
The 5 seconds is not required ( I like it also ) . 0. 1 will do . However a bank of lets say 22 000 uf charged every one second has merit ( charge is mS in reality ) . The amp never seeing the rectifier by using a flip flop switch ( MOS FET's in saturation ) . Some guy has a patent on that . What a nerve as it is so obvious . A commutator or distributor . One could even have something very simple like a PSU that switched in caps simply as the volume was advanced . It wouldn't need a fancy computer for that . It might then have a more delicate sound when it mattered ( 1000 uF + 10 000 uF + 22 000 uF .... ). Equally a very simple feed-forward signal switching idea could work . Not class H , just part of the general commutation . If hammering the speakers switch the commutation off and go conventional . I had the idea years ago but didn't do anything .
The LC resonance of the PSU is interesting . Nearest I can give you is the inductance of a big VOX output transformer I have somewhere at 500 H .
A well designed PSU with respect to all the usual things ( which I never seem to get right if my oscilloscope isn't lying to me ) . Would deliver massive current even if 10 000 uF . The wires to the transistors should be massive ( half inch I don't doubt , direct to collectors usually or drain ) . That causes problem number one . As massive cables cause HF reduction . The decoupling caps can rescue that problem ( usually about 220 uF ) . If the PSU voltage is about 5 V above the driver PSU voltage there never is a problem with ripple . If class A it gets more complicated . Caps on the PCB isn't such a bad idea with copper reinforcement . Almost design the PSU first if doing that .
The 5 seconds is not required ( I like it also ) . 0. 1 will do . However a bank of lets say 22 000 uf charged every one second has merit ( charge is mS in reality ) . The amp never seeing the rectifier by using a flip flop switch ( MOS FET's in saturation ) . Some guy has a patent on that . What a nerve as it is so obvious . A commutator or distributor . One could even have something very simple like a PSU that switched in caps simply as the volume was advanced . It wouldn't need a fancy computer for that . It might then have a more delicate sound when it mattered ( 1000 uF + 10 000 uF + 22 000 uF .... ). Equally a very simple feed-forward signal switching idea could work . Not class H , just part of the general commutation . If hammering the speakers switch the commutation off and go conventional . I had the idea years ago but didn't do anything .
The LC resonance of the PSU is interesting . Nearest I can give you is the inductance of a big VOX output transformer I have somewhere at 500 H .
How true!
I just received the new price list of my transformer manufacturer:
500 VA: 135*60 mm, 4.3 kg (9.5 lbs), V. drop 4%, US$ 64.50
1000 VA: 170*80 mm, 6.7 kg (14.8 lbs), V. drop 2%, US$ 103
1500 VA: 210*70 mm, 9 kg (20 lbs), V. drop 2%, US$ 162
2000 VA: 210*90 mm, 11 kg (24.2 lbs), V. drop 2%, US$ 198.
"V. drop" refers to the voltage drop from load off to full load on voltage.
Now, I have to pay €24 app. US$ 30) per BC Components 22,000 uF/63V capacitor. So, assuming that each channel will need 2 such caps, the caps alone will cost me €96, or about US$ 120.
A good quality 10,000 uF/63V cap will set me back €12...15, depending on which one I choose and its quality, let's say € 15. That's about US$ 18.8.
So, a 2000 VA transformer with 4 10,000 uF caps, as Nigel suggested, would cost me (198 + (4*18.8)) $ 273.20.
A 1000 VA transformer with 4 22,000 uF caps would cost me (103 + (4*30) $ 220.
Therefore, Nigel's reasoning is at least for me locally is absolutely true.
Of course, things may change somewhat if one needs caps of 80 or 100 V, because their prices quickly escalate above the popular 63V limit, I don't know by how much because I never need them.
On the other hand, it all depends at what kind of power are you looking at. A 200 WPC amp will necessarily cost more than a 100 WPC amp. The way Nigel is going, he'll end up asking his amp to do something like 1,500 Watts into 1 Ohm, in which case he's in for a whopping power supply bill.
I just received the new price list of my transformer manufacturer:
500 VA: 135*60 mm, 4.3 kg (9.5 lbs), V. drop 4%, US$ 64.50
1000 VA: 170*80 mm, 6.7 kg (14.8 lbs), V. drop 2%, US$ 103
1500 VA: 210*70 mm, 9 kg (20 lbs), V. drop 2%, US$ 162
2000 VA: 210*90 mm, 11 kg (24.2 lbs), V. drop 2%, US$ 198.
"V. drop" refers to the voltage drop from load off to full load on voltage.
Now, I have to pay €24 app. US$ 30) per BC Components 22,000 uF/63V capacitor. So, assuming that each channel will need 2 such caps, the caps alone will cost me €96, or about US$ 120.
A good quality 10,000 uF/63V cap will set me back €12...15, depending on which one I choose and its quality, let's say € 15. That's about US$ 18.8.
So, a 2000 VA transformer with 4 10,000 uF caps, as Nigel suggested, would cost me (198 + (4*18.8)) $ 273.20.
A 1000 VA transformer with 4 22,000 uF caps would cost me (103 + (4*30) $ 220.
Therefore, Nigel's reasoning is at least for me locally is absolutely true.
Of course, things may change somewhat if one needs caps of 80 or 100 V, because their prices quickly escalate above the popular 63V limit, I don't know by how much because I never need them.
On the other hand, it all depends at what kind of power are you looking at. A 200 WPC amp will necessarily cost more than a 100 WPC amp. The way Nigel is going, he'll end up asking his amp to do something like 1,500 Watts into 1 Ohm, in which case he's in for a whopping power supply bill.
Yes, sorry, I swapped theses.
As for how each might sound, well, there's only one way to find out.
China's import of copper has driven the price of raw copper up over the last ten years.
A decent 1kVA toroidal does over $225 overhere.
For standard series transformers above 300VA, transformer price goes up almost linearily with VA/weight number, so 2kVA would be roughly twice the amount of a 1kVA.
Unfortunately, 2kVA is not standard for most toroidal winders, so $/VA goes up instead of down.
(raw copper is expected to take a dive in a couple of years, hopefully tranny prices go down as well)
Not so for electrolytic capacitors, the large the cap value, the lower $/uF (for sane voltage ratings).
But for small time DIY folks there's also price breaks : 10 times 10.000uF may be cheaper than buying 22.000uF lytics.
A comparison doesn't hold water anyway, the number of uF will depend more on the secondary voltage of the transformer than the VA size.
(1kVA with 2% regulation is amazing, a low weight one normally does 5%, 7.5kg/16.5lb and up do 4%)
A decent 1kVA toroidal does over $225 overhere.
For standard series transformers above 300VA, transformer price goes up almost linearily with VA/weight number, so 2kVA would be roughly twice the amount of a 1kVA.
Unfortunately, 2kVA is not standard for most toroidal winders, so $/VA goes up instead of down.
(raw copper is expected to take a dive in a couple of years, hopefully tranny prices go down as well)
Not so for electrolytic capacitors, the large the cap value, the lower $/uF (for sane voltage ratings).
But for small time DIY folks there's also price breaks : 10 times 10.000uF may be cheaper than buying 22.000uF lytics.
A comparison doesn't hold water anyway, the number of uF will depend more on the secondary voltage of the transformer than the VA size.
(1kVA with 2% regulation is amazing, a low weight one normally does 5%, 7.5kg/16.5lb and up do 4%)
I just asked my supplier the same and got the reverse answer . I chose DNM ( BH ) slit foil caps at £20.42 ( 10 000 uF 80 V ) for an example . Mundorf was too complicated to look at ( too many choices ) . BBH would be fine .
1KVA + 160 000 uF = £407.72 inc VAT 20% ( £ in the old Latin for pound ).
2KVA + 20 000 uF = £250.84 inc VAT
2KVA + 40 000 uF = £291.68 ( Best ? )
I am placing the voltage at 70 +/- max which seems about right if using these VA's
Those prices change with quantity . Doubtless you can buy cheaper caps . As JC would say we should always do our best .
I recently bough loads of 4700 uF 100 V at £1.84 . they are great to do experiments with ( Sun Tan , yes , China , Rapid UK ) . They even have decent ripple current !
1KVA + 160 000 uF = £407.72 inc VAT 20% ( £ in the old Latin for pound ).
2KVA + 20 000 uF = £250.84 inc VAT
2KVA + 40 000 uF = £291.68 ( Best ? )
I am placing the voltage at 70 +/- max which seems about right if using these VA's
Those prices change with quantity . Doubtless you can buy cheaper caps . As JC would say we should always do our best .
I recently bough loads of 4700 uF 100 V at £1.84 . they are great to do experiments with ( Sun Tan , yes , China , Rapid UK ) . They even have decent ripple current !
Last edited:
Prices were £ 175 and £120 ( 2KVA 1KVA ) + VAT 20% . In UK pounds . The important fact is UKP £60 for 1 KVA difference . Less than 4 capacitors value .
I was assuming +/- 70 V which allows 10% to go to +/- 77V ( 80 V caps ) .
Most go for +/- 55V which I suspect doesn't require more than 1KVA ( if that ) .
I suspect as said before 2KVA and 4 x 10 000 uF is optimum . One should realize that it is in fact 10 000 uf if comparing with old single supplies ( 10 000 uF at 160V max ) .
I have an EL34 amp on my bench at 440V and 2 x 680uF ( 340 uf as in series ,generous for such an amp ) . That would be roughly equal to 2 x 2000 uF if at the lower voltage . Admittedly a 24 watt amp class A ( 48 pentode 14 triode ).
Years ago someone said over doing capacitors makes an amp muscle bound . The charge rate thing sort of says the same thing . I suppose we should use switch-mode . As with digital I don't like to . Transistors is my only attempt at being modern . I use CD and DigiTV . No great fan of CD , I like MP3 as music on the move is great . The Sony MP3 player I had which naturally went wrong was very good ( not much worse than CD ) .
I was assuming +/- 70 V which allows 10% to go to +/- 77V ( 80 V caps ) .
Most go for +/- 55V which I suspect doesn't require more than 1KVA ( if that ) .
I suspect as said before 2KVA and 4 x 10 000 uF is optimum . One should realize that it is in fact 10 000 uf if comparing with old single supplies ( 10 000 uF at 160V max ) .
I have an EL34 amp on my bench at 440V and 2 x 680uF ( 340 uf as in series ,generous for such an amp ) . That would be roughly equal to 2 x 2000 uF if at the lower voltage . Admittedly a 24 watt amp class A ( 48 pentode 14 triode ).
Years ago someone said over doing capacitors makes an amp muscle bound . The charge rate thing sort of says the same thing . I suppose we should use switch-mode . As with digital I don't like to . Transistors is my only attempt at being modern . I use CD and DigiTV . No great fan of CD , I like MP3 as music on the move is great . The Sony MP3 player I had which naturally went wrong was very good ( not much worse than CD ) .
Something came to mind when an earlier post said an amplifier should "play for 5 seconds" after power is removed.
I recall back in the late '70's using the big mainframe Cyber 74 computer (okay, using terminals connected to it). There was a lot of lore about the machine. Someone mentioned there was a motor-flywheel-generator setup feeding it power, that would keep the thing running when mains power went out for up to three seconds.
A similar arrangement (with or without the flywheel, which would effectively be a mechanical capacitor) could be done for a high power amplifier. The generator could be set up to give the proper voltage output to power the rectifiers and reservoir capacitors directly, bypassing the amp's power transformer. The generator output could also be three-phase, with the reduction in ripple that gives.
The "secondary" of this setup can pull any desired current waveshape without affecting the power factor. The only power factor correction needed would be a capacitor across the line to compensate for the motor's inductance.
While this shouldn't generate a lot of acoustical noise, it will generate some, and the generator-motor would of course best be set up outside the listening room.
- Status
- Not open for further replies.