Regulated Voltage Supplies

[bob p]

Which brings up a delicate point. Almost all tube amps get powered up at one time or another with no tubes or no output tubes in them. That means that the load goes effectively to zero. The regulated option works fine then, but the L-C filtered version becomes a C-only filter unless you also install a minimum load bleeder. That would waste a lot of power in this setup.

Back when I was doing the initial modeling I ran three different scenarios for each PSU layout: zero load, idle load, and full load. I also examined the transitions from zero load to idle, and from idle to full load. As you'd expect, the zero load condition took the voltages quite a bit higher and we'd obviously have to accommodate that situation.

Regarding the totem-pole cap configuration, I had been calculating the optimal value of the bleeder resistors to balance the charge on the caps based upon their capacitance, the total voltage across the series pair, and the maximum permissible voltage per cap. I don't see any way around that; I'd rather rely on calculating the right value of bleeder for the application, rather than omitting the balancing resistor and hoping that caps from the same production lot have equivalent leakage currents.

In the totem-pole scenario the balancing resistors do bleed off some current, though I don't think its enough to waste very much power. Am I correct in thinking that the kind of bleeder that you had mentioned would have a lower R and dissipate a lot more power than a set of balancing resistors?

From a practical standpoint, I think that using totem-poled caps is a good idea from an equipment damage prevention perspective, though I have considered ways of getting by without them. Those kinds of corner cutting measures, though, are ones where you have do define a service protocol that has to be strictly followed in order to save a few dollars on caps. In that situation, I think its false economy to complicate the service protocols in the hopes of chasing down a couple of dollars in savings. Series parall totem-poles (where each series par has its own balancing resistor) work for me. It's the safest design as it's inevitable that one of the caps will eventually fail.

[R.G.]

Good. I was actually broaching the subject gently so I could talk you into it. I can see you're already there.

My approach would be to use 400-450V caps and use resistors to balance the voltage. The resistors don't actually balance the voltage on capacitance, they balance it against capacitor leakage imbalance. The voltage divides on capacitance only for the first few AC cycles. As the initial surge subsides, the final voltage division on the series caps is set by their leakage and any external resistors. I've had good luck with 100K resistors, period.

And as you suggest, I would recommend NOT to use these resistors to make a minimum load. I would in fact not do a minimum load for either class AB or A.

I would do the Class A with totem pole caps to withstand the initial voltage surge. The tubes themselves will pull the voltage down into the Inductor-regulated region when they warm up.

I would do the AB amp with just regulators. The caps would always be running at about 750Vdc.

Do you want to have the ability to turn the regulated plate voltage up and down? If so, over what range?

[bob p]

Good. I was actually broaching the subject gently so I could talk you into it. I can see you're already there.

Well, I'm glad to see that we're thinking along the same lines.

I've had good luck with 100K resistors, period.

The numbers always seem to be coming out in the range of 100K-220K. To some extent, repeated number crunching isn't much more than doing math for the sake of doing math.

Do you want to have the ability to turn the regulated plate voltage up and down? If so, over what range?

Yes and I don't know.

I realize that creating the ability to adjust the regulated voltage up and down complicates the design of a high current supply -- especially as the adjustment range increases -- you need the ability to dump a lot more power as heat. I remember thinking how great it would be to have a continuously variable bench supply that ranged from about 350-600 VDC. But that could be somewhat expensive and impractical to build.

As far as the adjustable regulation circuit for the amp goes, I guess that my best answer is yes, I think there should be some adjustability -- but I'm not sure about: a) how much variability is actually needed to assure that we hit the target, and b) how much more we can design into the circuit while keeping the cost and complexity under control. (that bench supply does seem so appealing)

So I guess I have to answer a question with a question: How much does it complicate the design to have an adjustment range of 100V compared to 200V? I only ask this because I have no idea what the expense curve might look like, and I'd like to avoid the situation where the design goals change after you've started doing the work.

Looking at tube charts, the applications use B+ voltages that run anywhere from 375 to 600 VDC. My original Class AB plan was to use a 560 VDC B+ rail to get to the 70-100W zone, but that turns out to be a high distortion operating point at 3%. The distortion figures get a lot better at as you move closer to 400 VDC. Granted, the difference between 0.6% and 1.5% and 3.0% THD may not matter much from an MI perspective, but that's pretty signficant in HiFi terms (and tweeter protection). Maybe adding pairs of tubes isn't a bad idea.

I just noticed something while looking at the tube charts that I hadn't noticed before: Looking at the GE 6550 tube data sheet, page 3, pentode mode:.

As you move down from the 600 VDC operating point down to 400 VDC, the max signal plate and screen currents stay pretty much the same. What I hadn't noticed before was that the quiescent plate currents increase markedly as you move to the lower voltages. At 400 VCD they're 175% of their values at 600 VDC.

At first glance this seems like it might be good news, because it looks like the total amount of power consumed is pretty similar in all cases. Am I correct in interpreting this to mean that the design of the adjustable regulation circuit might be less demanding than I had originally thought? I'm thinking that this case (where the total amount of power doesn't change very much) has to be better than the case I had been thinking about (where current remained constant, voltage changed, and lots of power had to be dissipated as heat as you dialed down to lower voltages).

At a casual glance, these extremes look like they only differ by about 10% in terms of the number of watts that the regulation circuit would have to dissipate as heat. Is this too good to be true?

[kg]

so bob, did you build this sucker yet or what???

[Gingertube]

Found this thread very late.

If running Pentode Mode Output Tubes then regulation of the screen supply is all you need. Just use a reasonably stiff supply for the anodes, say 470uF per 100 Watt.

I would only consider regulating the entire B+ if using Ultralinear connected output tubes.

I have a pair of 100W Monoblock Ultralinear (and Cathode Feedback) Amps on the drawing board.

I'm using going to use rectifier => 470uF => electronic choke => 470uF + 5uF Film Cap for the supply.

Here is the electronic choke or gyrator if you prefer (Use in the Time OFF mode for Class AB, adjust the 10 Ohm in the Source for higher current). You can even buy it as a totally overpriced kit but I would'nt (Search for Electronic Choke and it will probably lead you right to it).

The Time ON mode is recommended for Class A, it compromises the gyrator (simulated inductor) function but adds some capacitance multiplier function.

Cheers,
Ian