Hybrid valve/SS supply help please

[lion]

For years I've been working on and off on a hydrid valve/SS circuit for a preamp/shaper in one of my projects. It started as the Shaka-Tube, but developed from there and very little of the intially circuit remains now. I'm finally pretty satisfied with the circuit so I'm starting a general clean up (of mess from the constantly modding and adding to the preamp/shaper circuit) and in particular considering how to optimize or improve the initial and very basic power supply section.

The schem shows the very basic PSU (mostly the remains from the Shaka-Tube)
- with +15V for the solidstate parts (actually 15.8V),
- a HT B+ for a 12AX7 (utilising a backwards 12/115V toriod)
- and 12VAC (13.6 actually) for the valve heaters running in series.
.


I don't have major noise/hum problems but there's certainly room for improvements. As you can see everything runs from one external PT (and I'd really prefer to keep it that way if possible), which makes me uncertain how to go about possible improvements.

I'm considering better rectification/filtering of the +15V branch.
A full wave bridge with a low dropout regulator might be able to give me the 15-16V I need with the 13.5V from the PT - and work happily parallel to the toriod. Would that be possible – and any hints to a suitable LD regulator/circuit?

Another issue is the valve heater setup – which has no ground reference at all.
I know how to do that the usual way, with a pair of resistors to ground or a elevated DC - but I don't think it'll work in this case with the two other sections sharing the same PT.

If needed I could opt for a separate heater voltage source/PT with a CT or 2x100R to ground - or even rectified to 12V DC.

Finally there could be better or alternative ways to optimize the PSU that I just can't see.

Any thoughts, comments or advised would be much appreciated. Thanks.

Erik

[PRR]

> the valve heater setup – which has no ground reference at all.

?? I see one side of the heaters hard-grounded. (top image)

> hum problems

AC heat "CAN" be done very cleanly (low hum); but as long as you HAVE a near-right DC supply IN there, why not run the heater on DC? Don't have AC anywhere near your audio parts.



Your present +15V DC affair is fine for a few mA. For 150mA you really should use full-wave for better transformer action. Also lower ripple. Even so, you want to up your first cap to like 1,000uFd because of the higher current you will be pulling with the heater load.

First-cap "DC" is not pulsating but has significant ripple. If this is run among sensitive audio nodes, the buzz can be "worse" than pure AC hum. Also you need to drop the raw +17V down close to 12.6V. So add a resistor and another big cap. Feed your 12V heater here.

> better rectification/filtering of the +15V branch. ....a low dropout regulator

Since the wall voltage (and the 13VAC and the 17V DC) will vary from day to day and place to place, trying to sharp-pencil an LDO to "just barely" have room to work will usually disappoint you some day. Maybe it works fine in your house with 242V mains power, but not when the oven is heating-up and your wall voltage drops to 233V, or when you go to an important job and the cheap stage wiring only gives 219V when all the stage-lights are on.

And most audio does NOT need regulation. What it needs is smoothing, no sudden or repetitive wobble of voltage. More R-C filtering will do this, simple, cheap, and will do it even with wide variations of wall voltage.

If you are happy now with 220uFd 1K 1,000uFd, I suggest 1,000u 470r 1,000u 470r 1,000u, or even 1,000u 330r 1,000u 330r 1,000u 330r 1,000u. 2-stage and 3-stage ripple filters are much more than 2X or 3X cleaner than a 1-stage filter. And since caps cost less each in 10-bags your cost does not go way up.

[lion]

Thanks Paul – this is all I was hoping for. Appreciate your time.

> First-cap "DC" is not pulsating but has significant ripple. If this is run among sensitive audio nodes, the buzz can be "worse" than pure AC hum<

So, a dedicated ground wire to the star ground – and tucked away in the farest corner of the enclosure I guess. Or are you hinting at something else to consider?

DC for the heaters looks good. I remember reading somewhere (a vintage radio forum IIRC) that in the days when battery supplies were normal valves where marked with a +/- polarity on the heaters for lowest hum – but since AC got the norm those specs have disappeared. Just a myth – and/or not relevant here?

Erik

[PRR]

> +/- polarity on the heaters for lowest hum

No. These tubes had no separate "cathode", the heater was the cathode.

In biasing a tube you want the heater-cathode voltage to be correct. But the "cathode voltage" of a naked filament varies from one end to the other. You need to get the polarity right relative to where you tie the grid.

I'm not sure polarity is critical on these things, but you are not using naked-filament tubes, you have the separate cathode. There's perhaps a slight preference for heater somewhat Positive of cathode; since your cathodes sit at 1V-2V and the midpoint of the lower cathode is 3.15V, this works out fine.

[lion]

Thanks Paul.

Erik

[lion]

This project keeps growing – added stages and features. As a consequence the unregulated voltage string is decreasing by each added load. Time to step up to a regulated supply I think.

Something like this:


As it happens the mains XMF I'm using has an unused secondary winding (2x12V), so I'm thinking I could  keep feeding the backwards toriod for the valve HT from one winding, AND to feed a LM317 by running both secondaries in series for a 24V supply (aprox 35V rectified).

Heater supply as 12VDC from a separate D-bridge also from the 12V secondary – parallel to the toriod.

I haven't looked at calculating R3 yet, and R1 should probably be 470R or more – but otherwise I think this could work. Not sure whether I have to heatsink the LM317 or can do without.

Comment or advise would be appreciated.

Erik

[PRR]

> unregulated voltage string is decreasing by each added load. Time to step up to a regulated supply I think.

Regulation only makes it lower, not higher.

I forget what you are doing, but the 22K dropper resistors stand out as voltage losers. With those cap values, 10K or 5K will still give filtering, at less drop.

> Not sure whether I have to heatsink the LM317

1) Build it. If the '317 is over-hot, it will shut-down (1.2V out).

2) Calculate. What is the '317 subjected to? Looks like 36V in 16V out or 20V drop. The drawing shows 10mA, which is 0.010A. 20V*0.010A is 0.2 Watts.

The naked TO-220 package will usually dissipate 1 Watt without harm.

Even the TO-92 package is rated 0.3W-0.6W, so "should" work if not in a hot-zone, though it will burn your finger.

Note that '317 is usually rated for maximum 35V drop. Yes, the running drop is 20V, but at start-up C5 is dead-zero so there is 36V across the regulator until C5 charges-up some. In this case with multiple transformers, it is likely you won't have 36V, or not while C5's charging current is flowing. Also I know experimentally that few-% over-specs rarely give trouble. But technically you are past-the-edge of rating until proven otherwise.

[lion]

Thanks Paul.
I should have been more precise. The challenge I'm having is just with the 16V supply string!

Previously the 1. version/prelim supply (right voltages/poor filtering and grounding) was reworks on the basis of your previous advise above – like this FYI:



The HT string for the valve works fine – the dropping/filtering gives the desired (for this project) voltages and they are steady. The upper part with the heater and 16 volt section barely worked with respect to the 16V, as I just managed to squeeze the needed 16V out by using efficient Schottky's and low (maybe too low) value dropping resistors. But it worked and the noise specs was greatly improved in the version 2 supply.

What I want to improve now is the 16V part to be able to add more circuitry/features WITHOUT pulling the supply voltage below 16V (which noticably disturbs carefully tweaked stages in the solid state part of the preamp/shaper circuit). To get enough voltage to supply a LM317 the single 12V secondary feed will obviously not do - hence the plan to series couple the secondaries for 24V AC, but leaving the HT section as is.

> The naked TO-220 package will usually dissipate 1 Watt without harm.
No problem here is see.

> Note that '317 is usually rated for maximum 35V drop.
I see your point – hence my comment about R1. Would it be bad practice to add some dropping ahead of the LM317 via R1 to lower the voltage to the regulator a bit. At 5mA 470R would give almost a 5V drop – maybe even use 1k for a 10V drop?

Erik

[PRR]

> couple the secondaries for 24V AC

Sounds like a plan to try.

[lion]

Will do - almost there, thanks.

Erik

[lion]

Almost done, and it works – sort of?
One issue -  is the DC heater supply supposed to float or not?



With the rectified heater supply floating (not connected to ground), the voltage is 12V. With the "O" side grounded (which I though it should be) the voltage into R11 rises from 18 to 35V! It's over my head why – probably showing my lack of knowledge again.

Did some research but couldn't seem to find a definite answer. There's mention of both "floating" and reference to ground in a couple of ways – via a 0.1 ceramic cap, elevated above ground, etc.

I have a book with preamp projects (mostly hi-fi), most of these shows used DC heater supplies as floating, but then again 1 or 2 is grounded. Confusing. What's up and down?

Is it correct/OK to run the DC heater supply totally floating?

Apart from this, all seems well. There's 18V into the LM317 and it barely gets warm (no heatsink), but I haven't done a real audio test yet, just measured voltages and checked a sine wave with the scope.

Erik

[PRR]

> is the DC heater supply supposed to float or not?

Without context.... it may work floating, but invites buzz and insulation trouble.

In context developed below: you would expect some buzz bleed from the heaters (not truly floating, but riding with raw DC). If it lives, and doesn't buzz, I'd go with it.

> With the “O” side grounded (which I though it should be) the voltage into R11 rises from 18 to 35V! It’s over my head why –

You have *two* bridge rectifiers on different taps of the same winding.

Such connections often lead to dead-shorts (smoke-bomb).

I think you got lucky-- with the two "-" bridge ends connected, the "+" ends output the same voltage.

I had not foreseen that, and am glad you caught it before anything went wrong.

Asking for too many different voltages in one project is always a royal pain.

With your present inventory, _I_ would think about tying your heaters in series, see how close to 35V you can get, so it can feed from the upper bridge.

There may be a clever plan but I am full of turkey.

[lion]

>With your present inventory, _I_ would think about tying your heaters in series, see how close to 35V you can get, so it can feed from the upper bridge.

There's only one 12AX7 in the circuit!

How about simply adding an extra LM7812 (or even LM317) while I'm at it. Something like this?:



I'd assume R1-C1-C2-C3 could be shared by both regulators, but maybe the cap values should be upped?

Erik

EDIT: Alternatively I guess a 7812 circuit could be run after/in series with the LM317? I seem to remember reading the later will increase current draw.

[PRR]

> I'd assume R1-C1-C2-C3

0.150 Amps in 470r R1 gives 70.5V drop from your 35V point. (??)

Clearly R1 is too big to hang a heater from.

I'm not sure the LM317 even needs pre-filtering. A zillion commercial products run raw rectified power to the regulator.

Certainly the heater does not need super-clean.

I'd figure (35V-12.6V)/0.150A = 149 Ohms 5.06 Watts of simple resistor-drop to get heater power off your 35V point. (Use 150r 10 Watt part.)

[lion]

> 0.150 Amps in 470r R1 gives 70.5V drop from your 35V point. (??) Clearly R1 is too big to hang a heater from.

Huhhh! I know I know - THINK, don't assume.

Decided to go with the simpler alternative:



R11 as 135R gave a perfect 12V output. With this setup the supply voltage in the LM317 string dropped a little. Lowering R1 to 235R took care of this - now 25V into the LM317. The regulator barely gets warm even without heatsink.

I'll try this out for a couple of days, but unless something shows up I think I'll stick with it.

Thanks for the help Paul, much appreciated.

Erik