Has anyone here had success with using REGULATED power supplies on a gainclone?
On his website (http://www.geocities.com/react_71/audio/gainclone/gainclone.htm), Pedja says that improvements are heard using regulators, but I have seen suggestions that regulation kills the sound.
On his website (http://www.geocities.com/react_71/audio/gainclone/gainclone.htm), Pedja says that improvements are heard using regulators, but I have seen suggestions that regulation kills the sound.
As with everything else it will greatly depend upon implementation. If the regualtors are zero feedback, have a constantly low impedance way above audio and don't starve the amp for current they may well aid the sound. Building such is not trivial though.
I believe a regulated supply for a gainclone is unnecessary if you have nice big caps on the rails, but it can't hurt.
Do it. Enjoy it. Endure the comments from people that tell you it will/does sound awful. 🙄
/Circlotron - has built a gainclone too!
Do it. Enjoy it. Endure the comments from people that tell you it will/does sound awful. 🙄
/Circlotron - has built a gainclone too!
No intention to teach anyone what is good and what is bad, I just want to point out that if there are reasons for and benefits of usage of regulated supplies in line level devices, the same should be applied in power amps. By the other words, if it’s wrong to use regulated supply in power amps, then regulators should be banished from CD players, preamps and similar devices. However, there are not a lot of examples of such consistency.
And of course, there are many ways to sort a regulation. For the one I suggested, typical DIYer most likely already has necessary parts, only LM338 are needed and it is done (assuming that a double bridge is already used).
Pedja
And of course, there are many ways to sort a regulation. For the one I suggested, typical DIYer most likely already has necessary parts, only LM338 are needed and it is done (assuming that a double bridge is already used).
Pedja
I'm just about to slap some simple discrete voltage regulators on my chip amp, should see how it goes. The only thing is I need to buy some power resistors to use as current limiting resistors so that I don't blow my little MSR860 diodes when charging big caps, I'd rather be safe than sorry...
But with a regulator there should be a cleaner power suply, big caps and the benefits of smaller caps after the regulator. So technically it should be an improvement, now will it really? Joe said he uses regulated suplies in his amp and pedja as well reported an improvement.
So I guess all I have to do is try it and see...
But with a regulator there should be a cleaner power suply, big caps and the benefits of smaller caps after the regulator. So technically it should be an improvement, now will it really? Joe said he uses regulated suplies in his amp and pedja as well reported an improvement.
So I guess all I have to do is try it and see...
OK I have not built a gainclone but it sounds intriguing. However, I have built three power amplifiers with power between 100 and 300w per channel with fully regulated supplies and the difference is not subtle at all when implemeted properly.
You don't have to have monster caps in the prereg section and you don't need them after the reg either. Some good fast caps will generally do. The 1,000 uf that many use looks about right for the power range we are dealing with.
Make sure you have a reg that supplies more current than you think you'll ever need and make sure it has low output impedance into audio frequencies.
It is important that your sense line/eg output generally be very close to the device/3875. With a properly implemented reg I would not at all be surprised that the gainclones totally dispel their weak bottom end.
Be sure that the ic amp is attached to the heatsink with the lowest possible thermal resistance - that could require making sure the mating surfaces are perfectly flat. This is very important as the chip itself has very little thermal inertia/mass and will heat and current limit very quickly on bass transients. A poorly mounted device will negate the regulator effects.
This regulated gainclone thing was something I have been dreaming up for over ten years! The time might be right.
You don't have to have monster caps in the prereg section and you don't need them after the reg either. Some good fast caps will generally do. The 1,000 uf that many use looks about right for the power range we are dealing with.
Make sure you have a reg that supplies more current than you think you'll ever need and make sure it has low output impedance into audio frequencies.
It is important that your sense line/eg output generally be very close to the device/3875. With a properly implemented reg I would not at all be surprised that the gainclones totally dispel their weak bottom end.
Be sure that the ic amp is attached to the heatsink with the lowest possible thermal resistance - that could require making sure the mating surfaces are perfectly flat. This is very important as the chip itself has very little thermal inertia/mass and will heat and current limit very quickly on bass transients. A poorly mounted device will negate the regulator effects.
This regulated gainclone thing was something I have been dreaming up for over ten years! The time might be right.
The LM3875 has 120 dB power suppy rejection ratio (at 50 Hz), which in my opinion means there is no need for regulators.
For power rail noise rejection that's certainly true, but if you want to get maximum power output you have to give it a reasonably high rail voltage of course, and under low-signal conditions this rail voltage might drift uncomfortably high. And in some places the mains voltage moves up and down a bit too. With regulation on a higher powered gainclone you don't have to give it a second thought. All depends on what you want of course. Mine for example is not regulated. +/- 33v at no load, +/- 28v full blast on 1 channel. Good enough for the application.
I have been thinking that since the amp chips work so well as amps perhaps they also work well as regulators as the application notes also describe. So why not build a gainclone with the same chip as ps regulator? It is a simple and elegant solution.
There is something about PSRR. Many ICs claim high figures that I do believe is justifiable. However, whenever a wide bandwidth (e.g. Jung Didden super regulator) is utilized, they noticeably sound very different and always for the better. This leads one to suspect there are interactions that PSRR does not take into account.
From an engineering standpoint one questions the need, but practice deems otherwise. The key to regulated supplies in audio is bandwidth as was known since the 80's. Hang a poor PS on a better amp and it is for the worse.
By the same token, why does short wiring sound better? In theory, the few cms on a PCB should not make any perceivable difference. Judging by the gainclones being built it is obvious that many have discovered that it does and now seems the accepted norm.
Everyone is still learning, experimenting and listening .
From an engineering standpoint one questions the need, but practice deems otherwise. The key to regulated supplies in audio is bandwidth as was known since the 80's. Hang a poor PS on a better amp and it is for the worse.
By the same token, why does short wiring sound better? In theory, the few cms on a PCB should not make any perceivable difference. Judging by the gainclones being built it is obvious that many have discovered that it does and now seems the accepted norm.
Everyone is still learning, experimenting and listening .
when i built my gainclone, i had 2 15v regulated power supplies in my scrap bin, so i decided to use these. having not heard a non-regulated supply, i cannot comment of the differences.
Protos
I was thinking the same. For some reason though, all Jung-type regulators i've seen seem to have very low closed loop gain for dc and sometimes even lower at ac. As the 3875 is not stable at low gain it might not be suitable. Or maybe it can work with a lower reference voltage and higher gain?
I did some power testing for my dual LM3886 amp. I have non-regulated power supply. 8 parallel 33 ohm / 5W resistors were used as a load (= 4.125 ohms). I only tested one channel. I could only get 1 V p-p signal from my USB card. This was enough to get 18.9 V RMS or 55.2 p-p signal into load. This equals 86.6 Watts into one channel! The resistors soon got very hot, and i had to turn down power. The peak current was 6.6 A, still smaller than the 11.5 A current limit. There was no clipping. The ripple was 1.6 V. The supply voltage dropped from 33.0 V when idle to 31.4 V at full blast. I have 4*4700 uF capacitors on the PCB for power supply filtering and 2*24 V / 250 VA toroid transformer. The LM3886:s did not get very hot while testing.
See http://hepso/misc/Lm3886_amplifier.html for more details about the amp.
See http://hepso/misc/Lm3886_amplifier.html for more details about the amp.
Has anyone got report on how regulated supply sounds in comparison to no reg supply and if it worth buying parts & building
Pedja doesn't give much info - simply says it does all he expected it to do?
John
Pedja doesn't give much info - simply says it does all he expected it to do?
John
Correction: the input signal was 2V p-p (400 Hz sine wave).
And i agree that it might be a good idea to use regulator to make sure the amp chip is working at best possible voltage, to get maximum performance out of the chip. Just for this purpose a rather simple transistor circuit might be enough.
And i agree that it might be a good idea to use regulator to make sure the amp chip is working at best possible voltage, to get maximum performance out of the chip. Just for this purpose a rather simple transistor circuit might be enough.
ukram, that's exactly what I am using and why as well. So far it works well, although I'm still assembling it (tested it without a chasis) and I'll put it through some listening tests in the next week. I just put together a zener/emmiter follower type regulator and haven't tweaked it yet, but at first listen it looks promising.
- Status
- Not open for further replies.