Kofi Annan in: "Push and Pull with Me"

[Kofi Annan]

Originally posted by Miles Prower


I did this voltage regulator for the screen supply of my current project (E-Mail for zip archive of design details). This is a smaller regulator, but the principle is the same. Starting with the requirements: Vo= 270Vdc; Io= 5.0mA -- 16.0mA. Since the main DC rail is at 365Vdc (nominal) that gives a Vpk= 95Vdc for the series pass device: a 6AQ5A operating as a triode. So you enter that on the plate characteristic graph. The extremes of the current lines intersect the main load line at a little less than -5.0Vdc and somewhere around -10.0Vdc, so use -10.0V. Knowing these bias voltages gives you the voltage across the DC voltage amp's Rp. Since this regulator uses an 0C2 for a reference, that gives a Vpp for the 6CB6A of: 270 - 105= 165Vdc. That gives the zero plate current point on the 6CB6A plate characteristic. You can then just play with possible load lines until you get one that looks viable. In this case, at the Vpk= 157.5Vdc (estimated bias for the 6AQ5A) and a plate current of 1.2mA looked good. This gave a value for Rp= 6.72K (select 6.8K). I picked off a couple of points for extremes of bias and current.

As for the voltage feedback divider, simply try to come up with a pair of resistors that'll divide 270Vdc down to 105Vdc (zero biasing the 6CB6A). These "next closest" values actually give +4.3Vdc, however, this is not a problem. Before starting this, simply set the 100K pot (used 10-turn cermet pot for this, since it's basically a "set it and forget it" control) to halfway. In practice, this actually gave too much voltage when tuned into a 50K load. NBD to adjust to 270Vdc. When operating with the 807's, the final measured screen voltage was 268.8Vdc. This doesn't vary under different operating conditions, nor when the line voltage shifts a bit. This circuit works, and gives dead quiet DC.

Of course, for more current, you won't be able to use a 6AQ5A. In those cases, use one of the triodes designed for series pass regulation: 6AS7A's, 2A3's, depending on what the plate dissipation will be. If paralleling series pass VTs is necessary, be sure to include a current balance resistor (~100 Ohms) in each cathode path. Grid stoppers, though not really required, aren't a bad idea and don't cost much anyway. So you might as well include it. Even more important is the Cdom to enhance Miller capacitance of the voltage amp tube (one of those rare cases where Miller capacitance is a good thing) since this is supposed to be a DC, not AC, amp. Also, the less AC passband you have, the less noise you have.

The 6CB6A is really good for this in that it has a heater-to-cathode voltage rating of 200Vdc. It's unusually high for a small signal VT.

Hope this helps.

Ay yay yay yay YAY!!!

Holy cow. I know that it was a lot of work to post this, but there's no way I'll get this. Let me explain:

As a child my mother would say, "Kofi, make sure you use a current balance resistor for this project". I'd look at her blankly and she'd say, "Kofi, you can identify the cathode path, can't you?".

Then she'd pause and say, "Kofi, your fly is open again".

I love my Mom and all, she's a great girl. Possibly the greatest representation of her love for me was and is her realization of the concept that I'm kind of a dumbass.

Please, please don't take this the wrong way. Its not you, its me. Well, actually, it may be you. You just haven't had enough experience with my magnitude of dumb-assedness.

Honestly, and without invoking any sarcasm, I know that this is helpful. Its just that my level of understanding could not possibly fit on the same graph as yours. On the other hand, if we tried to graph a representation of the chasm that exists between my comprehension and yours, we may actually wind up creating a mathematical concept of the infinite.

Maybe that would be bad, though. We may inadvertently deprive aspiring scientists in their junior year at MIT a chance at solving a quantum puzzle akin to Fermat's Last Theorem, a potential "A" in Quantum Mechanics III and a shot at a lucrative defense contract with the US gover... Uhh... BAD! BAD! BAD!!!

IMHO, you should replace Mr Jones's "285V regulator" with a choke and a cap. Current draw of the first two stages is 16.6mA, so a Hammond 154E 20H@20mA, followed by a 100uF-350V cap will fit the bill nicely. You'll thank me later.

Easy. Easy does it. Getting it... Getting it...

OK, so should the choke be in series with the voltage and the cap be grounded?

Me like easy.

Kofi

[Sheldon]

Originally posted by Kofi Annan
OK, so should the choke be in series with the voltage and the cap be grounded?
Kofi

Yes. But a question for Joel. Why not just a couple of RC sections? Easy to adjust to get the right voltage drop and very low ripple and more compact. Sonic considerations?

Sheldon

[Joel]

But a question for Joel. Why not just a couple of RC sections?

Actually just one RC filter would be needed, and that's perfectly acceptable - especially if the input tube and split load inverter are set to equal currents.

Joel

[SY]

There's really no need to regulate the HV for that design- it's nearly in class A. If I were to build a regulator, I would use a heavy-duty version of the Maida or, at the expense of more expense, the Curcio reg. But Mr. Jones seems quite content with a plain vanilla LC filter for the output stage. The input stage regulator will provide superior decoupling compared with the passive that Joel suggests, but his passive would no doubt work fine too.

[Kofi Annan]

Originally posted by SY
There's really no need to regulate the HV for that design- it's nearly in class A. If I were to build a regulator, I would use a heavy-duty version of the Maida or, at the expense of more expense, the Curcio reg. But Mr. Jones seems quite content with a plain vanilla LC filter for the output stage. The input stage regulator will provide superior decoupling compared with the passive that Joel suggests, but his passive would no doubt work fine too.

Yeah! Thanks Joel, Sy, Miles and Sheldon! I'm ready to get my parts list together, so I have a few things to think about which of course means that as these issues arise, you'll have these things to think about as well.

Woo hoo!

Kofi

[Kofi Annan]

Resurrecting this old post...

I was on the path toward a 300b (see here), but budget constraints will likely push this project into 1Q 2007.

Given that, I think I can still whip up this little neato Bevois Valley push-pull job, but I'm looking for a simple solution for the HV regulator that Morgan Jones recommends in his circuit.

I think I need 285V regulated for HT and the screen supply and I thought that maybe the Supertex LR8 would do. It looks like you can set it up just like an LM317, but it will take an input from 12V to 450V and output 1.2V to 440V. Plus it's called a Supertex.

So, thoughts on a Supertex-Regulated Bevois Valley EL84?

Hurry up and respond-- I'm going to be tied up with this North Korea missile test thingee soon.

Kofi

[Sherman]

Originally posted by Kofi Annan
...
I think I need 285V regulated for HT and the screen supply and I thought that maybe the Supertex LR8 would do. It looks like you can set it up just like an LM317, but it will take an input from 12V to 450V and output 1.2V to 440V. Plus it's called a Supertex.
...
Kofi

Kofi,

Never used this one but I have a question; is 20mA enough current? It appears that is the output current limit.

[Kofi Annan]

Never used this one but I have a question; is 20mA enough current? It appears that is the output current limit.

I knew there'd be a catch. Story of my life.

Well, I'm back to searching for an easy-to-implement regulator solution. Any thoughts one this thingee?

Any other easy HT regulator would be OK-- I just need to be taken a bit further down the path here. Or.. hey, remember this old tune?:

IMHO, you should replace Mr Jones's "285V regulator" with a choke and a cap

Is this still a popular solution to the regulator issue?

Kofi

[Andrewbee]

How about a regulator per tube in true diy overkill?

If I had easy access to those regs I would get a few for screen regs myself.

[Sheldon]

Originally posted by Kofi Annan


I knew there'd be a catch. Story of my life.

Well, I'm back to searching for an easy-to-implement regulator solution. Any thoughts one this thingee?

Any other easy HT regulator would be OK-- I just need to be taken a bit further down the path here. Or.. hey, remember this old tune?:



Is this still a popular solution to the regulator issue?

Kofi

Any of those things would be fine.

Cheapest and easy is a couple of RC sections - just use PSUD (FREE) to play with values to get the right voltage and low ripple.

Next cheapest and also easy is Joel's suggestion - again play with PSUD (see the pattern?).

Or try Jone's two transistor regulator. Pretty cheap. Trim voltage by adjusting the divider resistors to the base of the bottom transistor (you do have the book, right?).

Or buy the Welborne regulator. Not as cheap, and you don't have to read anything except how to hook it up, but you won't learn as much. Easy though, and will probably work quite well.

Or one of SY's favorites the Maida (look it up on the National Semi site, or see this: TIP50 + LM317 High Voltage Regulator). Or?

Your call.

Sheldon

edit: You only need about 17mA per side, so one of the Supertex jobs per side should do it, if they are otherwise suitable (I didn't check). I also did a quick run on PSUD. The HT supply has only about 10mV ripple to the power tubes. One RC section with a resistor of about 1.5-2k and a 47uf cap would give about the right voltage drop and ripple of well under 1mV. You could easily split this into two sections of 3-4k with 22uf cap. Many other combos will work too.