audio-talk

Owston was a eye opener this time for me,

Thomas's and. Dave's amps both sounded really nice, with good detail but it was the cleanness of the sound that impressed me , those power supplies are impressive.. When each valve has a separate supply it clearly make a difference.

So a question .. Does the panel think simple cap decoupling a stages power supply is 100% effective, its seems not to my ears

My second point is regarding choke loaded stages, the choke would provide another level of decoupling, that would partly explain their improved sound ?

Now taking the above , which is likely to be based on my misconception

So my next task is test that misconception and to build a single power supply with totally separate smoothing to each valve. I have a few smaller maplins 10h chokes that would be fine and still mount in the amp, and some lower resistance types for the outputs,
I have a new power supply transformer coming so i was planning to rebuild the px25 amp anyway...will keep you all posted.

PSU and B+ design is both simple and complex, depending on the view one takes of it. If you view as just voltage supply to the stage then one just sizes the cap to provide enough peak current under load and Bob's your Uncle...

Unfortunately that ignores that the B+ line is part of the signal output current loop of the stage and so one hears the quality of the PSU in the signal - this also means that the power stage currents speak back to the input stages louder than they talk to the output stage, decoupling effectiveness ameliorates this but doesn't stop it.

This is why shunt regulation PSU are so effective - they provide a cleaner and better path for signal currents through the PSU and in doing so provide better interstage isolation.

DTBs amd Thomas's amps at Owston showed that even this is a compromise and that separate low noise psu for each stage are better yet... It is the way to go for best sound...

Now we just have to work out how to design cheap but effective psus so we can afford to give each stage its own psu!

J

My KT120 PP pentode amp has separate power supplies for the driver and output stages.
Apart from the teachings of Dr Gizmo, my take on it is that the power stage and the input stages have different requirements in terms of what they are being asked to do.
The power stage needs to be able to produce clean power into a speaker without sagging under load, so needs the brute force passive solution of serious storage capacity, preferably followed by a hefty series regulator, as in push-pull amps like mine, the standing to peak current requirement of four big class A/B pentodes excludes shunt regulation from the spec.

The input stage needs an ultra low noise supply, which can be provided from a choke input setup, with shunt regulation, making sure the supply always draws critical current. The current requirements of this supply are obviously nowhere near what the power stages want.

Even without regulation, this twin power supply solution, works superbly in my amp. The only thing you have to watch out for, is to make sure that one supply does not inject hum into the other. I had that problem at first, with the present amplifier, until I sorted the grounding.

I have three supplies in my amp, as the output stage screens are fed from another well filtered power supply, with shunt regulation by gas tubes.

I wouldn't do things any other way now I've heard the benefits of this sort of setup.

Something to also be aware of is the effect of regulation on the supply before it. Shunt supplies will tend to draw a constant current from the supply, the supply pre reg will be isolated from the signal. Series supplies will make the pre supply see a exaggerated version of the load. So for series regulators, the supply that feeds the regulators can be just as important.

That would argue that a shunt feeding a series feeding a shunt might be a good solution.

That may make a shunt feeding a series, feeding individual shunts for each stage a good compromise.

Troublemaker

Series regulators turn a single ended output stage into what is in effect a push pull one.

Troublemaker

What me?

I know they have their disadvantages, but are chip CCS (as in 10M45s) good at isolating a driver stage from PSU vagaries ?

Seems to work OK on the driver of my GK-71 amp.

A good CCS can do a good job of appearing to isolate a signal stage from the power supply - it seems to work well for preamp and some driver stages.

A CCS feeding a shunt reg is generally better but that is getting into Nick territory

DTBs solution is probably optimum i.e. having separate fully regulated psus for each stage although following the Nick philosophy, they would be better if his solid state regulated psu fed a valve shunt stage followed by the signal stage

Its much less efficient so it must be better

J

JamesD wrote:
A CCS feeding a shunt reg is generally better but that is getting into Nick territory

It doesn't have to be too complicated if the current swing on the driver is not too great . The simplest version is the CCS into a couple of VR tubes ( eg. OD3's to give you 300V ) . If the VR tubes are by-passed, it's clean & quiet enough for a driver stage & a lot better than an RC or even an LC decoupling stage .
I wouldn't want to use a CCS direct as an anode load - never found any CCS that sounded clean in this regard - I'm sure Steve would pick up the downsides ; but CCS into VR shunt is a different game .

Thanks to all for your input. I may revert to bread boarding some power supplies and see how i go. I have this un definable urge to make things as simple as i can.,, some of these suggestions challenges that, but what I'm proposing in my 1st post involves quite a lot of hardware, I could easily make a separate chassis and have a double decker amp... But its not my preference

Cheers..

Just thinking aloud Steve.

If your HT transformer provided a suitable voltage for all levels and channels in the amplifier, you could specify sufficient over capacity to ensure no sag in voltage from the power stage, so you could then work from rectifier level. Drivers get an ez80 or 81 output stages 5r4 5u4 gz or whatever else (mercury if you will).

Or you can use low resistance solid state rectifiers and work from that point on et al.

A big transformer is a great place to start.

Been having a play about with my px 4 amp this morning,,i selected some tracks that sound clean and some other tracks that dont sound so clean.
The last couple of tracks on madonna's hard candy turned out to be good in that respect.. Ie not so clean

I did what i've done in the past, put the fluke measuring the dc power supply at the split to both chanels outputs.
depending on the tracks there was slight drop in voltage, from .1 to around 3 volts showing on the fluke, but constant larger variations seem to be down to mains fluctuations., which i was measuring simultaneously. No big changes with the music signall but enough to note.
Then i switched to measuring AC the normal .009 ac on the ht changed once the music was playing, 3.9 volts rms was the worst i saw But to be fair the average was less than a volt , with the amp flat out
Now as far as music signals are concerned i need to multiply these voltages by1.4 to get the peak volts..
Now i would guess this voltage ac voltage is the power supply being modulated by the load, or i suppose it could be the music signal not being decoupled.. but the results are the same.
What was interesting, and something that many of you had noticed with my amps that the confused sound on some tracks related to the amout of ac being measured, those madonna tracks were way above simpler music. With an average of 1.5 volts,
That explains alot to me.
The old valves and transformers have high resolution, and replays this confusion very well,

I was going to go with split power supplies for the drivers and outputs. I now cant see the point unless there is one hell of alot of smoothing betweeen the valves.i was going to split the supply and put 2-3 chokes on each leg
Regulation seems the easy choice, but i think i havve no choice except to try 4 x power supplies, hope fully i'll get them into the case Ive been making. Picture below...

Nice experiment Steve. The ripple levels are not great, but I suppose it's the effect of that ripple feeding back into the driver stage that causes the trouble ?

Mark, I'm not quite sure how it works, the interstage inductance and the smoothing / voltage drop before the point i measured, would reduce the inter channel and driver/ output mix between each valve to a degree,
but say theres up to a couple of peak volts on the ht, im not sure how it relates to the couple of peak volts music on the grid from the source. Which is the level i measured at

I also checked the interchannel, i disconnected the input to one channel and disconnected the speaker on the other, the cross talk was at a low level, but still at the level of the ac heater hum on my old amp,
I've also done this before and thought its not worth worring about,
But this time when i checked that the ac on the power supply, its less than two thirds of when both channels have a signal to them, so in reality the cross talk is actually 3 times worse when both chanels are playing
scary....

Nick wrote:Something to also be aware of is the effect of regulation on the supply before it. Shunt supplies will tend to draw a constant current from the supply, the supply pre reg will be isolated from the signal. Series supplies will make the pre supply see a exaggerated version of the load. So for series regulators, the supply that feeds the regulators can be just as important.

That would argue that a shunt feeding a series feeding a shunt might be a good solution.

That may make a shunt feeding a series, feeding individual shunts for each stage a good compromise.

Hmmm, I was thinking a series reg feeding a shunt reg.
I need to learn more, obviously. And think more.
An excuse to make everything more complicated!