my take, the right amount of sag is when you are able to hit your FTC power rating targets for class AB amps, for class A sag is almost non existent...
Thanks a lot for the very valuable advice. I had this feeling of the importance to use a transformer with very low impedance secondaries.Over sizing the caps does not necessarily prevent sagging rails. You *need* about 20,000 uF per rail with normal speaker impedances, and anything beyond that does very little other than let it play longer after you pull the plug. The power supply will still draw down based on the current demand - and the transformer’s impedance determines that more than anything else. It may hold up at full voltage for a few more milliseconds, but not put out one watt more sustained power. Sustained sine wave capability (for a short term) power determines how loud it will play, period.
I hope that a resistance measurement should be enough. Impedance meters are not exactly cheap. I guess that thickness of secondary wires can be telling.
I will take 20mF per rail per channel as a optimal value. Thanks a lot !
I’ve found that “the right amount of sag” is such that the DC voltage drops down to the unloaded RMS secondary voltage when the amp is driven with sine waves at 2 ohms. It’s about a 4X overload to its VA capacity, but will run long enough to make the measurement. Any idea amp that can’t run at a sine wave at 2 ohms long enough to take a power reading isn’t good enough for my use.
Doesn’t have to run that way much longer, but anything that shuts down or blows up before I can read the instruments isn’t going to like being bridged into 4 ohms playing music at clipping.
Very very interesting. 2 ohm is not that rare with some exotic speakers.
But going a little OT ... i read often about people bridging amps ... but i do not understand the reason for bridging.
Is it not better to get a more powerful amp ? 🙄
Also because i read of the risk of doing that.
my take, the right amount of sag is when you are able to hit your FTC power rating targets for class AB amps, for class A sag is almost non existent...
Very interesting. Could this be a reason why people like class A sound ? 🙄
Changing the Vsupply the working point of output transistors change ? 😱
Very interesting. Could this be a reason why people like class A sound ? 🙄
Changing the Vsupply the working point of output transistors change ? 😱
May as well play your HiFi through a marshal or fender stack..
If you want zero sag.. then the cryogenic fridge for superconducting may make more noise 😉 we'd all be moaning about the cost of liquid helium. I wonder if a valve would work given there's plenty of research papers for superconducting thermionic emissions.. but more todo with quantum tunnelling, reducing qubit noise and extending cohesion.

Very interesting. Could this be a reason why people like class A sound ? 🙄
Changing the Vsupply the working point of output transistors change ? 😱
class A is a constant power load.....i happen to like the PASS mosfets at this time....
What is wrong with oversizing ? (cost aside) .... of course always within a reasonable limit. My limit would be like 15-20mF per rail per channel.
i use over 100mf per rail in my amps psu for two channels...😀
I use a 572VA transformer for a 110W RMS amp. From idle to full blast I get about 15V of drop (idle 320V@480mA, full blast 305V@~2.5A (It's a PPP triode connected sweep tube monoblock amp with about 4mF on the B+).
I was also lucky to find 5 of these transformers for 25$ each!
For comparison, my Yorkville AP1200 power amp (rated at 1200W into 2R in bridged mode, class AB) uses a 720VA transformer with a 12.5A circuit breaker.
My tube amp will put out 30Hz sine at full power all day without overheating the power transformer. I dunno if the Yorkville would do the same.
I build a 10Wx2 tube amp with a 100VA power transformer.
I build a 30Wx2 tube amp with a 250-300VA power transformer.
I built a 15Wx2 tube amp using an old Hammond 100VA transformer. It worked for 2 years before the transformer smoked, replaced with a 160VA coil and it's worked for years.
Like others have said. The bigger, the more efficiency, lower DCR, less drop.
And generally 40uF per Watt (for a tube amp, more like 400uF per Watt for an SS amp).
I was also lucky to find 5 of these transformers for 25$ each!
For comparison, my Yorkville AP1200 power amp (rated at 1200W into 2R in bridged mode, class AB) uses a 720VA transformer with a 12.5A circuit breaker.
My tube amp will put out 30Hz sine at full power all day without overheating the power transformer. I dunno if the Yorkville would do the same.
I build a 10Wx2 tube amp with a 100VA power transformer.
I build a 30Wx2 tube amp with a 250-300VA power transformer.
I built a 15Wx2 tube amp using an old Hammond 100VA transformer. It worked for 2 years before the transformer smoked, replaced with a 160VA coil and it's worked for years.
Like others have said. The bigger, the more efficiency, lower DCR, less drop.
And generally 40uF per Watt (for a tube amp, more like 400uF per Watt for an SS amp).
Last edited:
Very very interesting. 2 ohm is not that rare with some exotic speakers.
But going a little OT ... i read often about people bridging amps ... but i do not understand the reason for bridging.
Is it not better to get a more powerful amp ? 🙄
Also because i read of the risk of doing that.
“2 ohm loading” in my world is driving 4 pairs of speakers or two subs in bridged mode. Not the same as an exotic speaker load.
A more powerful amplifier is always more preferable than bridging - but power supply voltages over +/-150V are impractical, even if you’re pulling every trick in the book to get efficiency. If you need more than 1000 watts per 8 ohm subwoofer driver, you’re stuck bridging.
I
For comparison, my Yorkville AP1200 power amp (rated at 1200W into 2R in bridged mode, class AB) uses a 720VA transformer with a 12.5A circuit breaker.
My tube amp will put out 30Hz sine at full power all day without overheating the power transformer. I dunno if the Yorkville would do the same.
And generally 40uF per Watt (for a tube amp, more like 400uF per Watt for an SS amp).
Almost all tube amplifiers are designed to put out a full power sine wave indefinitely. Usually the idle dissipation is set high enough that it is comparable to the maximum possible with sine wave drive, even at the 60% power level where dissipation is maximized. You also can NOT tolerate regulation figures in the power supply as poor as most solid state amps. That would put the B+ without a load just too high, so your power transformer is sized appropriately for the regulation required. That means a lot more VA than you would use for a transistor amp, and then the amp can run at high levels a lot longer without overheating.
Your Yorkie might put out full power at 8 ohms all day, and almost assuredly for an hour or two at a time. But don’t expect that at 4 or 2. You could probably run it at 2 ohms all day with the clip lights flickering with no issue. Average power would be on the order of 100-200 watts per channel. But turn it up to 11 when the roof is on fire, and it might shut down in a few minutes.
About cap size and tube amps - it’s all about the corner frequency between the cap and the load. You want the impedance of the capacitor to be significantly lower than the load impedance. 20,000 uF and 4 ohms is the SAME as 20 uF and 4000 ohms. Most tube amps don’t NEED that much capacitance to get equivalent power supply performance. Use 1000 uF and it’s like using a farad on a 200 wpc solid state. That’s a LOT of capacitance and a VERY low impedance supply. . Try putting a screwdriver across that and see what happens.
Hi ! well to be honest i do not think that guitar speakers are wideband enoughMay as well play your HiFi through a marshal or fender stack..
Maybe the power amps could be used for music ? i do not know
want ... let's say that it would be nice to see a "steady" ampIf you want zero sag.. then the cryogenic fridge for superconducting may make more noise 😉 we'd all be moaning about the cost of liquid helium. I wonder if a valve would work given there's plenty of research papers for superconducting thermionic emissions.. but more todo with quantum tunnelling, reducing qubit noise and extending cohesion.I may have deviated from the post.. 😀
When the voltage supply varies also the working points of the transistors vary
So the amp does not behave as per design ? distortion could increase ?
Interesting. I have a question about class A ... how do you set the bias ?class A is a constant power load.....i happen to like the PASS mosfets at this time....
how you know when it is optimum ? using a scope ?
Moreover ... have you tried reducing it and listening for differences in sound ?
😱 nice 😱 i like big caps as i hate many smaller ones ... maybe the best is in the middle ... many medium size caps ?i use over 100mf per rail in my amps psu for two channels...😀
Contrary to what some state you simply want as much power reserves as you can get/fit/afford according to your specific use. Reading through some of it most have a commercial/economic perspective meaning good enough to meet technical specs but I am not sure that has ever been what DIY audiophiles are about.
For home audiophile type domestic use at relatively limited spl levels you want a dead quiet noise floor and explosive dynamics which works with a smaller trafo and a BIG capacitor bank, preferably with a lot of filtering or other clever ways keeping noise/ripple out.
For a party or PA set up you need the big trafo and enough capacitance.
If you decide on a smaller trafo/home type approach don't get all worked up by all the scare of inrush currents etc that some argue, remember a smaller trafo will limit the current in itself protecting the diodes etc and as long as the trafo live through the charge up second or so it and the system will be fine.
If you go super crazy on the trafo and/or capacitor bank it's easy to add a soft start using a NTC and a relay, potentially a fuse on the rails before the amp stage if you don't want to stay in the danger zone.
For home audiophile type domestic use at relatively limited spl levels you want a dead quiet noise floor and explosive dynamics which works with a smaller trafo and a BIG capacitor bank, preferably with a lot of filtering or other clever ways keeping noise/ripple out.
For a party or PA set up you need the big trafo and enough capacitance.
If you decide on a smaller trafo/home type approach don't get all worked up by all the scare of inrush currents etc that some argue, remember a smaller trafo will limit the current in itself protecting the diodes etc and as long as the trafo live through the charge up second or so it and the system will be fine.
If you go super crazy on the trafo and/or capacitor bank it's easy to add a soft start using a NTC and a relay, potentially a fuse on the rails before the amp stage if you don't want to stay in the danger zone.
Interesting. I have a question about class A ... how do you set the bias ?
how you know when it is optimum ? using a scope ?
Moreover ... have you tried reducing it and listening for differences in sound ?
bias is pretty much determined by the design, and depending on your heatsink the temperature rises and then stabalise, i am not fond of playing around with bias that much.....
😱 nice 😱 i like big caps as i hate many smaller ones ... maybe the best is in the middle ... many medium size caps ?
i have 4 each of 68kufd/100vdc and 47kufd/100vdc to play around...
as a matter of practice, i also put in smaller valued caps like 4700uf/100vdc into the mix..
Contrary to what some state you simply want as much power reserves as you can get/fit/afford according to your specific use. Reading through some of it most have a commercial/economic perspective meaning good enough to meet technical specs but I am not sure that has ever been what DIY audiophiles are about.
Hi ! thank you very much for the very kind and valuable advice. You are very right. The commercial units are designed and built at a budget. I am not sure but the MSRP could be 5 time the parts cost ? then they have labor and distribution costs that i do not have
So i see DIY like a challenge ... to take something basically sane but limited for commercial reasons and improve it with some mods aimed at specific strategic points.
As an example the connections to the external speakers ... in some cases the traces are thin 😱 or the connections are realized with some aluminum sheets 😡 I am sure that some piece of good section wire and decent binding posts could help the flow of current to the speakers.
Then some bypass ... caps replacement ... the mains transformer ..
I am looking more to power amps these days. Line preamps will be my long term project ... i have already a good solution in mind. Very minimalist and very promising. I have to beat my friend's preamp for less than 100 USD ... he spent 1200 USD I think
thanks a lot ! this is what i was looking for. There is a candidate incoming ... an old Rotel ra-960bxFor home audiophile type domestic use at relatively limited spl levels you want a dead quiet noise floor and explosive dynamics which works with a smaller trafo and a BIG capacitor bank, preferably with a lot of filtering or other clever ways keeping noise/ripple out.
I will try to fit in 2 big 30-40mF caps in place of the old stock ones
I am even thinking to remove completely the front controls and make it a little but robust power amp. I prefer to have a separate preamp.
In understand the preamp section in many commercial integrateds is the weak part. It can shrink the soundstage and ruin the midrange.
not my case ... unfortunately I love parties 🙂For a party or PA set up you need the big trafo and enough capacitance.
to put in place a very robust integrated diodes bridge i think i can do thatIf you decide on a smaller trafo/home type approach don't get all worked up by all the scare of inrush currents etc that some argue, remember a smaller trafo will limit the current in itself protecting the diodes etc and as long as the trafo live through the charge up second or so it and the system will be fine.
If you go super crazy on the trafo and/or capacitor bank it's easy to add a soft start using a NTC and a relay, potentially a fuse on the rails before the amp stage if you don't want to stay in the danger zone
I like the very big ones. With big wires in and out. Debottlenecking as much as possible.
But it has been very important for me to understand that caps F is what is needed in my case I have already a suspicion but now i am convinced
thank you very much again.
The only issue is that it is easier to place big caps horizontal... and i have the strong feeling that this is really not the best solution ? if i remember well some Rifa caps can be only mounted vertically ... if not they could have issues. I do not understand why only Rifa ... for me it is a very general issue. Caps must be vertical.
Thanks again i think that this by far the best way. To check the actual T on the output devices heatsink ... if it increases too much that is not good for reliability.bias is pretty much determined by the design, and depending on your heatsink the temperature rises and then stabalise, i am not fond of playing around with bias that much.....
I understand that some nice sounding cheap class A amp have reliability issues due to the too high temperature on the heatsink. The Musical Fidelity A1 being one case ?
I was shocked to see how good and large heatsinks can cost ... someone use fans to keep them smaller.
This is good capacitance indeed 😀i have 4 each of 68kufd/100vdc and 47kufd/100vdc to play around...
as a matter of practice, i also put in smaller valued caps like 4700uf/100vdc into the mix..
I think i have understood that caps are the key for my needs.
I did an experiment years ago with big caps. The feeling was of lower noise and more dynamics ... and they were old big Mallory very abused i am sure
Last edited:
ripple voltage is directly proportional to load current and inversely proportional to C, therefore it is very easy to see from relations how each affects ripple voltage...
ripple voltage is best minimized......but beware having too much C will require soft starting circuits...
ripple voltage is best minimized......but beware having too much C will require soft starting circuits...
I wonder how increasing C might alter PRaT. It's been made clear that increasing it benefits bottom end impact/extension. But is there a consequence to overall subjective speed/agility?
I've been here a decade but here's a "noob" question...
Which PRAT?
PRAT - Definition by AcronymFinder
Which PRAT?
PRAT - Definition by AcronymFinder
- Home
- Amplifiers
- Power Supplies
- Mains transformer VA rating and musical power.