Sans Choke Smoothing

[lowell]

Hey all,

I'm curious what options there are for achieving little ripple on my supply for a single ended preamp. Interested in this to reduce size, cost issues etc... Was reading a bit about gyrators but not sure thats the right direction. Looking for small size and low cost with minimal parts.

Must be able to supply 320vDC and current for 2 12ax7 stages (1 tube) and 1 6V6.

Thanks!

[Enzo]

Two or three caps, a resistor or two, just make a conventional supply. They have been doing that for decades.

[pdf64]

Yes, easy with class A, just start out with more HT than needed so that dc and ripple can be dropped across an additional stage of filtering, eg http://bmamps.com/Schematics/vox/Vox...0Schematic.pdf

[uneumann]

Hey all,

I'm curious what options there are for achieving little ripple on my supply for a single ended preamp. Interested in this to reduce size, cost issues etc... Was reading a bit about gyrators but not sure thats the right direction. Looking for small size and low cost with minimal parts.

Must be able to supply 320vDC and current for 2 12ax7 stages (1 tube) and 1 6V6.

Thanks!

You can do the conventional RC networks or try something different ...
https://en.wikipedia.org/wiki/Capacitance_multiplier

The version below smooths a voltage-tripler output (see below) to run 3 12ax7s. It will work for any transformer/rectifier setup. You want a small RC at the end for decoupling, but this circuit has the effect of a much larger cap so it may save you some space (or cap cost). Since it's a preamp (low current) - the transistor can be cheap too since it's power dissipation is negligible.

[Enzo]

But you will need an isolated 120vAC winding.

[uneumann]

But you will need an isolated 120vAC winding.

Yes - good point - isolation is needed for a doubler or tripler supply - but the cap multiplier circuit itself can also be used with any conventional rectifier setup. That little circuit just filters ripple as if there were a bigger cap. You also don't necessarily need the totem-pole cap arrangement. A single cap works fine. I used the totem-pole since the open circuit voltage went up to ~500v. A conventional full-wave bridge, for example, would probably not see such large variations in open circuit vs loaded voltage, making a single cap the better way to go.

The circuit I show takes the rectifier/tripler output with about 40v of ripple to about 40 millivolts of ripple using only the 2x20uf caps - effectively a 10uf cap - for a load current of ~10ma.

It's something to consider when PS size/cost are issues. The pic below shows a 3x5.5" card with a PT, tripler, and filter circuits on the left side (with the ground plane). The multiplier circuit itself is just the transistor (TO220 case) in the upper left and the two smaller caps next to it. That's about all there is to it.

The other half of the board is a 60w power amp (another story).
The bigger 3 caps are used in the tripler circuit. I wouldn't have needed those if the PT had the right voltage to begin with.

[R.G.]

At the risk of further offending the vacuum deities, have you considered using a MOSFET a couple of resistors and zeners in a source-follower regulator? Yeah, there are several parts there, but they're simple ones, and smaller than the resistor/caps that would give an equivalent ripple reduction. This is not strictly correct, but it is close to using a MOSFET as an impedance before the stage. A 1H inductor has an impedance of 754 ohms at 120Hz. The MOSFET drain is a couple of megs in the simple follower case. As a practical matter, the ripple reduction is equal to the reduction of ripple to the reference voltage on the gate.

Yes, all the issues with using MOSFETs apply, including having to wire them right and protect the gate, but in a supply for a preamp stage, the power expended in the MOSFET will be pretty small, as preamp stages run only a couple of ma per stage, so heatsinking may not be needed. A good high voltage MOSFET will be well under $2.00 for this kind of app.

[Dave H]

Must be able to supply 320vDC and current for 2 12ax7 stages (1 tube) and 1 6V6.

You said it's a preamp but what current is the 6V6 running?

[lowell]

RG... yes, that's what I'm looking for. Exactly. I'm aware I can go the tradish route, but looking to reduce real estate requirements.

Dave H - It's running resistive loaded with a 5k Ra. It's biased at 14W. I can't remember the exact Va... but somewhere around 250v if I remember correctly. Rk is somewhere around 150ohms. Again I don't have it open right now and have been using my prototype for gigs etc... But that's about where it's at. I would like to improve hum rejection in the next version. I guess at 14W that means square root of (14*5k). So 264v Va? This seems high. But I do remember biasing it as close to 14w as I could.

RG, like this? http://www.one-electron.com/Series-reg.png. Can you link to something close to what I'm going for?

[R.G.]

That one will probably work just fine. It fooled me at first with the schemo drawn input-right, output-left and the time delay transistor, but once I looked at it, should be fine. The important points are:
- input voltage generates a reference voltage on the two 82V zeners, with current through the 100K/2W resistor
- reference voltage smoothed and filtered with the 470K/0.22 RC network
- smoothed reference fed to the MOSFET gate
- input voltage fed to the MOSFET drain
- output from the MOSFET source
- 100 ohm "gate stopper" resistor right at the MOSFET gate
- 12V gate protector zener

Those are all things I'd have put into a design of my own.

As long as you only need ~~14ma, you're good. Beyond that, calculate the power needed very carefully. The MOSFET as sits is dissipating only about 55V*0.014A = 0.77W, so a TO-220 in free air will be fine. If you have more input voltage - the 250V you mentioned? - and/or more current, recalculate the MOSFET power and heatsink to keep it cool. A bare to-220 mounted vertically can get rid of maybe 2W. More than that and it needs a sink. The voltage the MOSFET has across it is the input voltage minus the output voltage. The output voltage is the reference voltage minus about 5-6V.

[lowell]

RG how do I provide reference voltage for the gate? The schematic I posted has some sort of time delay input. Do I need that transistor q106?

Im guessing not and that I could connect the 100k to R102

Finally, what kind of gate voltage am I looking for? Just enough to turn it on?

[R.G.]

Sorry for the delay. I didn't get back to this thread til now.

RG how do I provide reference voltage for the gate?

The reference voltage for the gate is provided by the two zeners, D104 and D105 with the current supplied through R106 and Q106.

The schematic I posted has some sort of time delay input. Do I need that transistor q106?
Im guessing not and that I could connect the 100k to R102

The time delay is done by leaving the base of Q106 open, which starves current out of the zeners, and leaves the gate of Q100 floating, mostly. That's a bad design practice. But I'm quibbling.

You can, as you guessed, just connect R102 and R106 to the 213V supply.

Finally, what kind of gate voltage am I looking for? Just enough to turn it on?

I'm not sure what your question is. The gate voltage of Q100 is determined by the two zeners D104 and D105, and the leakage resistances of the circuit board and such. I would have put a 10M or so across the two zeners just to force the gate of Q100 to go down to zero volts if the 213V supply was off. So the gate voltage on Q100 is about +161 when the +213V supply is on, and zero when the "+213V" supply is really zero volts.

Remember that Q100 is drawn backwards, with the input voltage on the right connecting to Q100's drain on the right, and the output +158V connecting to Q100's source on the left. So as the input 213V rises, the zeners do not conduct until the "213V" gets higher than 160-odd volts. At that time, the zeners stop the voltage let through to the gate from rising any more.

Q100 is an enhancement mode MOSFET, so nothing much happens till its gate gets about 5-10V higher than its source. In this case, when the "+213V" supply rises to, lets say 20V, cap C103 will rise to that same 20V by charging through R104. The gate is an open circuit to DC, so the gate voltage rises to that same 20V. Presumably there is a load of some kind on the "+158V" output, and that load holds the voltage down on the output until the gate-to-source voltage on Q100 hits the 5-to-10V threshold that lets it conduct. If the gate voltage is 20V, as we were thinking, then Q100 conducts enough current to make its source rise back into the 5-to-10V range again.

Another way to look at it is that Q100 is a DC source follower. R106, D104 and D105 set a voltage on the gate. The output on the source follows that voltage, but a few volts lower as is needed to keep Q100 on.

Am I making this muddier or clearer?

I guess maybe a shorter answer is that the circuit makes the gate voltage be right, all by itself, so you don't need to supply a gate voltage other than what it makes for itself.

[lowell]

RG first off thanks so much for your time and commitment to answering these questions for me.

I get it now...I think. Calling it a source follower clears a lot up for me. I realize this is what you called it from the first post. That and your detailed response helped a lot. So whatever voltage is on the Gate will be on the Source minus the 5-10v "on" voltage. The less ripple on the Gate = less ripple on the source. I mostly understand a tube cathode follower, and emitter follower...this is the same but with a FET.

Ok so the Gate needs to be 5-10v higher than the Source for the FET to conduct. In the circuit I posted, the Source is only 3v lower than the Gate. I guess there is a RANGE and not so much a solid figure for Vgs.

So I mostly get it. And the harder the load pulls on the supply the more ON the FET becomes, and therefore counteracts the potential sag that would otherwise occur with a higher impedance supply. This is because as the Source wants to sag, the Gate does not, and this increases Vgs therefore turning it ON more.

[lowell]

Interesting document falling right inside the subject of this thread.

http://www.next-tube.com/articles/hvs/hvrEn.pdf

[lowell]

If my 6v6 MAX current is 320/5k =64ma. The 12ax7 is 150/100k = 1.5ma x2 = 3ma. So my HT rail neednt supply more than 70ma max. Can leave some room and go with 100ma. Which is way over spec.

What is a suitable enhancement mode MOSFET? And I believe it is N channel? I'm thinking the IRF820 from your follies page is about 4 times over spec for this application. So that should be fine, and cheap.