I'm going to rebuild my headphones amp and I'd need some advice on the power supply.
Currently, the amp is using a very simple supply : a 250VAC xformer and a 220uF cap. That's it, and that's for both channel. No need to say that ripple measures are awful in psud.
The amp is class A and its total consumption is 40W. B+ is around 320VDC. Am I right assuming I've to work on psud with a current of 150mA (both channels, 150mA * 320 = 48W) ?
As it is class A, is it necessary to use seperate power supplies, for each channel ?
Here are the two options giving me the best results in psud :
For a single supply :
For seperate supllies :
If someone has better ideas, I'll gladly take them
Just don't ask for more iron please. Costs are already high enough 
.
Currently, the amp is using a very simple supply : a 250VAC xformer and a 220uF cap. That's it, and that's for both channel. No need to say that ripple measures are awful in psud.
The amp is class A and its total consumption is 40W. B+ is around 320VDC. Am I right assuming I've to work on psud with a current of 150mA (both channels, 150mA * 320 = 48W) ?
As it is class A, is it necessary to use seperate power supplies, for each channel ?
Here are the two options giving me the best results in psud :
For a single supply :
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
For seperate supllies :
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
If someone has better ideas, I'll gladly take them
Just don't ask for more iron please. Costs are already high enough .
A couple of comments:
First, and most important, you have to consider the circuit being driven as well as the supply. What's the PSRR of the amplifier? Can that be improved? Remember, every 20dB increase in PSRR is like reducing the ripple by an order of magnitude.
Second, the supply can be reconfigured to give you a cap input, then a pi section. This should smooth things out quite well. Careful bypassing at the output will take care of higher frequency crosstalk and coupling between stages.
Third, if you really want the supply rail to be DC and nothing but DC, an active regulator is absolutely the best way to go.
First, and most important, you have to consider the circuit being driven as well as the supply. What's the PSRR of the amplifier? Can that be improved? Remember, every 20dB increase in PSRR is like reducing the ripple by an order of magnitude.
Second, the supply can be reconfigured to give you a cap input, then a pi section. This should smooth things out quite well. Careful bypassing at the output will take care of higher frequency crosstalk and coupling between stages.
Third, if you really want the supply rail to be DC and nothing but DC, an active regulator is absolutely the best way to go.
Thx for the comments.
The amp is an old kit, which evolved into a commercial product, the Audiovalve RKV mkII. According to the website, the circuit is strictly the same. Evolutions are at first sight : use of a toroidal xformer, dual mono power supply and a bunch of bypass caps. However, they still use a basic power supply, with only one big cap and perhaps a resistor. So I suppose the PSRR is quite good, considering the amp is 1000$ and is among the best of its category.
A pi filter is ruled out : no money for a big choke sadly.
Is a 0.006V ripple as I got in the first case not good enough? Would the advantages of a dual supply justify the increase to a 0.26V ripple ?
The amp is an old kit, which evolved into a commercial product, the Audiovalve RKV mkII. According to the website, the circuit is strictly the same. Evolutions are at first sight : use of a toroidal xformer, dual mono power supply and a bunch of bypass caps. However, they still use a basic power supply, with only one big cap and perhaps a resistor. So I suppose the PSRR is quite good, considering the amp is 1000$ and is among the best of its category.
A pi filter is ruled out : no money for a big choke sadly.
Is a 0.006V ripple as I got in the first case not good enough? Would the advantages of a dual supply justify the increase to a 0.26V ripple ?
there's a low background noise but in an amp with : a shitty old pot, 20years olds big output caps and a noisy xformer located at 2 inch of the circuit, without shielded wires, it'd be a miracle if I didn't have noise Actually, the background noise is really low.
All of the above issues will be fixed of course.
Actually, the background noise is really low.All of the above issues will be fixed of course.
Honestly, it sounds like you've got some other issues to address before you go crazy with ripple reduction.
Nonetheless, I'd strongly consider some regulation- you can use a simple three-pin adjustable floated with a HV pass transistor in this application. If you're nervous about using a chip, use the reg as a preregulator to get rid of the ripple, then use a simple R->C to your circuit.
Nonetheless, I'd strongly consider some regulation- you can use a simple three-pin adjustable floated with a HV pass transistor in this application. If you're nervous about using a chip, use the reg as a preregulator to get rid of the ripple, then use a simple R->C to your circuit.
More than you thinkSY said:
but I've already ordered a stepped attenuator, shielded wires, I've a nice big transformer and lots of caps to change.
In fact, I've now to order metal sheets to make a new box (I didn't tell you that it was needed too ?
) and the layout and thus the size of the box depends a bit of what I take for PS. As i have those 4 big caps, I'd just want to know the best way to do it.It's not just a lifting, more like a rebuild as you can see.
Sorry to put cold water over ideas.....OF Using all those caps.
Why not dispense with most except one/two and use a cap multiplier or basic mosfet stabiliser which will give you easily 30dB ripple rejection, a far lower output impedance and noise and you can connect A+B together without intermodulation effects between channels. I use it in my amps.
Look under Steve Bench's pages in Triode electronics site to get a crack what's involved.. There is a whole section about stabilised PSu's. It's a big site and the simplest circuit using SS works wonders on tube circuits.
You shouldn't have to use sep supplies for each channel If you you a SS regie-.
As SY mentions.......it's all up to your ability.
rich
Why not dispense with most except one/two and use a cap multiplier or basic mosfet stabiliser which will give you easily 30dB ripple rejection, a far lower output impedance and noise and you can connect A+B together without intermodulation effects between channels. I use it in my amps.
Look under Steve Bench's pages in Triode electronics site to get a crack what's involved.. There is a whole section about stabilised PSu's. It's a big site and the simplest circuit using SS works wonders on tube circuits.
You shouldn't have to use sep supplies for each channel If you you a SS regie-.
As SY mentions.......it's all up to your ability.
rich
It isn't more complicated........The thing to remember about using chip reggies on H.V applications above their intended application voltage i.e lifting the reference term way higher than ground potentials is to put plenty of protection between adj and o/p and IN terminals....With MOSFEts, a15V zener between gate and source and feed... gate via resistor of a few hundred ohms. Often I fit a low ohm in the drain as well.
We aren't looking for perfect regulation as these SS devices have a much faster slew rate when compared to the old circuits using OA2's etc.
Linear technology appls note AN2 <<performance enhancing techniques for three terminal regulators>> gives a practical overview of using 317 types with lifted adj term.
I regulary use IRF840's in 450V supplies, but don't ignore gate protection w.r.t source. No protection = No working device.
hope this helps
rich
We aren't looking for perfect regulation as these SS devices have a much faster slew rate when compared to the old circuits using OA2's etc.
Linear technology appls note AN2 <<performance enhancing techniques for three terminal regulators>> gives a practical overview of using 317 types with lifted adj term.
I regulary use IRF840's in 450V supplies, but don't ignore gate protection w.r.t source. No protection = No working device.
hope this helps
rich
If you are not hearing hum/buzz now:
1) Fix other problems
2) double-up the existing caps (470µFd in place of 220µFd, etc) just for fun.
In some amps, microvolts of ripple is too much, in others a few volts is no problem. But since you have it working, you don't even have to wonder about it. If it is OK now, leave it alone or double-up just-in-case.
> class A, is it necessary to use seperate power supplies, for each channel?
Depends again on PSRR. If there is crosstalk, you can usually make most of the bulk-filtering common to both channels, then have one RC stage for each channel. If your main filter terminates in, say, 100µFd, and you add a 100Ω 100µFd fither for each channel, then any crosstalk has to get through three caps and two resistors to reach the other side, dozens of dB rejection across the audio band.
1) Fix other problems
2) double-up the existing caps (470µFd in place of 220µFd, etc) just for fun.
In some amps, microvolts of ripple is too much, in others a few volts is no problem. But since you have it working, you don't even have to wonder about it. If it is OK now, leave it alone or double-up just-in-case.
> class A, is it necessary to use seperate power supplies, for each channel?
Depends again on PSRR. If there is crosstalk, you can usually make most of the bulk-filtering common to both channels, then have one RC stage for each channel. If your main filter terminates in, say, 100µFd, and you add a 100Ω 100µFd fither for each channel, then any crosstalk has to get through three caps and two resistors to reach the other side, dozens of dB rejection across the audio band.
I know you didn't ask for this ...
... but I simply have to suggest adding a choke (if not two).
OK, cost is an issue; but have a look at the
www.audiokit.it site and there are many good value items there.
A Digitex 10H 50mA choke of 190 ohms costs EUR33 and is probably a pretty good unit. I think it would significantly improve the sound, by removing all those dropper resistors and allowing the use of *much* less capacitance (which can then be higher quality at the same price).
If you're looking at a lower price point, there are Hammond, 10H, 195 ohm at EUR 12 !
Possibly better still; Hammond 9H, 300 ohm at a mre EUR 9; this would allow you to split the PSU by channel for high isolation.
I'd suggest you look at say:
10uF foil cap; 10H choke of 190 ohms dcr; 20uF poly/oil motor run cap.
Or say:
10uF foil cap; then two legs each 9H, 20uF
Or even better:
10uF foil; then 2 legs each 15H, 20uF motor run.
Note the Hammond choke, 15H 100mA 256 ohms, EUR 17.
I'd go for this as the DCR of the two chokes in parallel is only 128 ohm; less than the dropper resistors you are considering.
Low resistance and small fast caps will give you speed, natural flowing dynamics.
Poly/oil motor run caps will give you a sweetness and liquidity that you won't get from big electros.
Results should be fast, dynamic, highly musical.
I am sure the result will far exceed the RC filtering with large electros, whatever the PSUD results might show. I'd expect the RC filters above to make the music seem a bit slow and compressed.
Note that the ripple results of these suggestions, analyzed by PSUD, will not be as good. But chokes and fast caps do a good job of removing higher order, more objectionable, ripple and hence I suggest they sound 'better' than their figures would suggest.
The more cost and effort I put into the PSU, and the more chokes I use, the better the results.
Good luck
... but I simply have to suggest adding a choke (if not two).
OK, cost is an issue; but have a look at the
www.audiokit.it site and there are many good value items there.
A Digitex 10H 50mA choke of 190 ohms costs EUR33 and is probably a pretty good unit. I think it would significantly improve the sound, by removing all those dropper resistors and allowing the use of *much* less capacitance (which can then be higher quality at the same price).
If you're looking at a lower price point, there are Hammond, 10H, 195 ohm at EUR 12 !
Possibly better still; Hammond 9H, 300 ohm at a mre EUR 9; this would allow you to split the PSU by channel for high isolation.
I'd suggest you look at say:
10uF foil cap; 10H choke of 190 ohms dcr; 20uF poly/oil motor run cap.
Or say:
10uF foil cap; then two legs each 9H, 20uF
Or even better:
10uF foil; then 2 legs each 15H, 20uF motor run.
Note the Hammond choke, 15H 100mA 256 ohms, EUR 17.
I'd go for this as the DCR of the two chokes in parallel is only 128 ohm; less than the dropper resistors you are considering.
Low resistance and small fast caps will give you speed, natural flowing dynamics.
Poly/oil motor run caps will give you a sweetness and liquidity that you won't get from big electros.
Results should be fast, dynamic, highly musical.
I am sure the result will far exceed the RC filtering with large electros, whatever the PSUD results might show. I'd expect the RC filters above to make the music seem a bit slow and compressed.
Note that the ripple results of these suggestions, analyzed by PSUD, will not be as good. But chokes and fast caps do a good job of removing higher order, more objectionable, ripple and hence I suggest they sound 'better' than their figures would suggest.
The more cost and effort I put into the PSU, and the more chokes I use, the better the results.
Good luck
Yes Frank...Many years ago When I worked in the console industry we had this <thing> about screening E&I cores for lower fields and also using thinner stampings. It worked great and nowadays this idea has been lost for good except for those manufacturers who know about the problems. More weight..sometimes larger core.....higher costs.....yes. When I'm looking for quality, the proper solution is to lower the flux density (as I do when drawing up a spec) and find a manufacturer who is going to listen when I want more copper added. However applications vary from one extreme to another. I know it's not cost effective but thank heavens in the UK there are manufacturers still around.
The toroid would seem to many as being the ideal tranny solution.......but NO.....at 1.6T flux density and badly wound can be a horrific offender of spewing out stray flux. Unfortunately I see this seldom done, the result that the input tube has to be a mile away from the iron and also the iron correctly orientated. Chokes as you say are excellent dB rejectors,esp of line frequency ripple provided Bmax is constrained.
I do like chokes but as my amp already weighs 50kg....enough is enough! and have to use higher efficency SMPS technology.
Keep the silicon up!
rich
The toroid would seem to many as being the ideal tranny solution.......but NO.....at 1.6T flux density and badly wound can be a horrific offender of spewing out stray flux. Unfortunately I see this seldom done, the result that the input tube has to be a mile away from the iron and also the iron correctly orientated. Chokes as you say are excellent dB rejectors,esp of line frequency ripple provided Bmax is constrained.
I do like chokes but as my amp already weighs 50kg....enough is enough! and have to use higher efficency SMPS technology.
Keep the silicon up!
rich
Hello,
I am interested in your headphone amplifier. I read that the amp (AudioValve RKV II) will be a perfect match for my AKG K1000 headphones.
Since you mentioned that you built the device from a kit I want to ask you for posting more information about the source or details of the circuit.
Thanks,
Michael
I am interested in your headphone amplifier. I read that the amp (AudioValve RKV II) will be a perfect match for my AKG K1000 headphones.
Since you mentioned that you built the device from a kit I want to ask you for posting more information about the source or details of the circuit.
Thanks,
Michael
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.