Low noise HT

[James D]

There has been some discussion of using BG WKZ in the psu and shunt regulation to get a better supply. This is my take on the design. To my ears it is the best psu I have built by quite a large margin.

The shunt reg valve is chosen and biased to pass at least 100% of the power stage current. It can be a pentode triode connected. For instance for a 2A3 power stage passing 50mA I use an EL34 shunt valve triode connected and biased for 50 - 65 mA.

The other point to note is the use of a balanced power connection on the transformer input - this removes most input common mode noise from the transformer provided the earth line is low impedance and clean itself.

Hope its useful. Comments, critisims and rude remarks all welcome.

James

[NickG]

Ahh, very interesting, I meant to ask you about your comment about shunt regs without dc references. I see now. Almost like a bouncy load resistor.

[James D]

Almost like a bouncy load resistor.

Exactly! It copes with all programme driven variations and any differential noise between the two power rails. I find any dc reference introduces some noise into the shunt reg and this avoids that. The BGs and resistor only in the passive part ensure very low noise from the psu and using BG in the shunt circuit continues that. This circuit in practice is capable of rendering the psu noise at less than 10uV peak to peak. In theory it can acheive less than 1uV...

The shunt reg and BGs allows the full glory of SE to come through

When playing around to get different B+ from standard power trannies I sometime replace the first cap after the rectifier with a WKZ and make an RC element with this - tweaking the R to acheive the required voltage.

James

[IslandPink]

James,

Sorry to be a pain -
could you e-mail any pics you've posted today , since I can't see them ?
No rush . Sounds v. relevant .
Thanks for the last e-mail . I forwarded it to Thom Mackris ...

Mark

[James D]

No probs. On there way.

Part two of the email coming up :-) They should form the basis for the tonearm piece here (eventually ...)

I've ordered some neo magnetics today for a Schroeder style beastie - except all in wood except for the magnets of course... and the fixing bolts... and the wire ... and the thread... and the...

James

[IslandPink]

Wow, Nice !
Lots to take in .
Is the valve shunt bit able to (a) lower the noise and (b) take the PS impedance down to quite low levels down into the single Hz ?

What are the options for the shunt for a 300B or super-300B ? KT88 or something cheaper ? I suppose the shunt valve does not in itself have to be as fine or low-distortion as the DHT ?

I'll be musing over the circuit as I listen to the system tonight .

MJ

[James D]

a) The shunt valve is very effective at removing differential noise from the supply rails.

b) Shunt reg doesn't lower the psy impedance down to very low levels like a series reg does... but the psu impedance is only one measurement of psu goodness and just like THD it doesn't tell anything about how a psu will sound. To my mind it is much more important to have a consistent and well behaved impedance characteristic than a miximally low one - for intance I'd raterh listen to a psu that has a 10ohm output impedance but is a flat ten ohms from 1Hz to 100mHz than one that is 10ohms at dc, 100milliohms at 500Hz, 500milliohms at 10KHz and 100ohms at 1MHz... I don't mind if the impedance gently shelves down with frequency either.

I guess ideally we use the same valve for shunt as for opt but run the shunt at 125% of the opts current. I have built then like that and they do sound better but only just better than with say EL34 as shunt for a 2A3. So for a Super 300B I would use an 813 or equivalent or a cheap KT88 - whatever provided the current and volts rating and was cheapish!

Remember that the shunt is in the output current loop as is the output BG so you can hear them.

Hope that helps

James

[IslandPink]

OK
(b) but it's lower than a typical passive supply , which from Thorsten's article (and calcs ) on the Verdier one, would typically reach several-100's ohms at 10-15 Hz unless silly cap sizes were used ?


"Remember that the shunt is in the output current loop as is the output BG so you can hear them"

..but is the shunt valve more significant only at low frequency, or does it have a part to play at all parts of the frequency spectrum ?

Thanks for the info .
Mark

[James D]

Yes the shunt provides a lower output impedance than a typical passive supply

The shunt is active from lower than audio to higher than audio and makes an audiable difference and improvement across the whole range - partly because it lowers the noise in the system by 30-50dBs (with good design) as so lets more of the sound through, partly 'cos the lower noise floor lets more of the BG goodness show, partly because it provides 'in time' current delivery to the opt when needed at whatever frequencies it is needed (programme dependent) and partly because it lowers and evens up the psu impedance in the output current loop...

There may well be more reasons too

James

[dave dove]

hi james

that looks lovely
have you got a triode
can pass @250mA?

dave