Question on shunt regulation, JamesD's in particular

[ChrisV]

Allo
I need to build a very simple amp for a downsized system, for a small apartment far far away... I'm thinking a 2 stage SE, R120 triodes/Slagle iron driving Fertin field coils on open baffles. Needs to run cool but also without loads of iron, so I thought James shunt reg PS would be ideal.

Circuit will be either 27 or 76, LC coupled to the R120s. B+ 250V, 60mA per output valve. Very simple and sweet. The L's in this case are quite small, so a welcome tradeoff of weight vs heat. The driver stages will use them VR150 tubes fed from the 250V OP supply.

Now, regarding James' shunt reg circuit: I have no experience of any kind of shunt regs, having only used traditional LCs but will take forum members' word that these make wonderful supplies. The extra heat would be unwelcome but I could put up with another 40W or so to avoid a ton of low DCR chokes and big caps.

Therefore, I am wondering if one serving both channels will work well, or whether I really need 2. I should point out I am envisaging the BG version as I have a few 350V, 220u+220u BG WKZ to put to use.

Alas I won't have much time for experimentation; would like to get something decent first time, so any advice on the PSU or, indeed, other aspects of the plan is much appreciated.

The alternative to all this to get the 32 ohm version of the Fertins and build an OTL... and have a gainclone for the summer months...

Cheers and hope all's well with you fellas,
Chris

[NickG]

FWIW, I almost use a similar setup in my 211 driver stage. 76 (choke loaded) to 45. The supply is a Steve Bench type shunt reg, unlike James circuits its referenced to a VR tube and uses a error amplifier. But it was far better than a standard LC (black gate) supply, and also clearly better that a series reg.

I guess what I am trying to say is that shunt reg's certainly work, the next step is finding the best shunt reg.

Oh, and this is one reg feeding two chans, replacing a split LCLC supply with an half of a WKZ per chan.

[ChrisV]

Thanks Nick, exactly the kind of info I was after.

I was concerned about decoupling issues with both channels sharing the same supply, but it seems you've had no trouble and I guess the fact that the shunt maintains a low Z down to very low frequencies (unlike reasonably sized caps) helps prevent nasties like motorboating.

Cheers,
Chris

[James D]

Agree with everything said so far

I aim for something slightly different from my shunt design which is why I drop the reference and it really needs one per channel (although it works with one for both channels). My prime intention was to form a well defined return path for the output signal current loop hence I made mine ac only - it does reduce the noise on the output but it main purpose was to avoid caps in the rerun signal current loop... I haven't tried the two types side by side as my ac shunt made me smile so much I just stopped bothering

James

[ChrisV]

Thanks James. Do the channels share any part of the PSU? Ie both shunts feed off a common point with some series element (choke or R) to decouple them? Or are they entirely separate, monoblock style.

Actually, the biz about caps out of the signal path has given me an idea; I know you reckon the thing not working down to DC is a plus, but it's occurred to me that you could use a tetrode as shunt element, apply fix bias to g1 and use g2 to apply the modulating fluctuations from the supply rail. No time constants anywhere...
Trouble is you might need a few devices in parallel to get the required transconductance at g2. Need to research the numbers...
cheers
Chris

[IslandPink]

Hi Chris ,

Good subject . I want to this up the top of the list again, since I need to get to grips with how a shunt-regulated supply would look for my KR300B amp . As I remember it, James, we were left with the question of how best to improve the supply ripple of your shunt/BG hybrid supply . I suppose a conventionally smoothed supply with one or two big chokes is one option , but you had other ideas .. ?

Mark

[ChrisV]

Allo MJ,
I think they key thing here - if you want to improve shunt performance without recourse to big iron is the transconductance of the shunt valve.

So you could parallel up high gm valves (need a scheme to equalise currents presumably) or apply some kind of gain stage to the thing.

In the latter case, maybe a diff pair (to get phasing right) direct coupled to a screen driven shunt? Who knows how it would sound and what the stability issues might be though...

I defer further comment to the master...
cheers
Chris

[James D]

Chris,

Yes, I share the two channels psu before the shunt regs and use a choke on each channel into the shunt or I sometimes use seperate windings on the psu trannie and go seperate mono from there (if that makes sense)

The tetrode idea is interesting, I need to think it through a bit but I have two gut reactions to it.

1) The reason I removed the dc reference is that that this reference itself has some noise and it is injected in the high gain part of the regulator ensuring that the regulator is always moving up and down in response to the reference as well as the noise on the earth line and on the positive line - so it can never produce a truely quiet output and it is always using transconductance up in chasing the reference noise signal... The ac reg. acts as a differential amplifier acting against differential noise on the positive and earth lines directly to remove noise on either from the output...so in principle it can kill all differential noise (in practice it is limted by the transconductance of the regulator...)

2) Tetrode and pentode operation is predicated on G2 being held at a constant voltage wrt the cathode - and any signal applied to G2 moves the device away from pure tetrode or pentode operation into a less well defined and inherently more non-linear area of operation. It also generates intermodulation products between any noise signal present on the reference and the noise signal present on the power rails - which for SE operation would be an analogue of the output signal itself so the regulator is now a direct generator of intermodulation products rather than an indirect generator of such products.

more later if I think of anything

James