Need help with 600v+ PSU

[andrew_k]

Do you think ~200mA will be enough current for biasing? I'm considering a voltage quadrupler off the 24V tranny to ensure I can bias a pair of KT88.

[defaced]

A bias supply consumes very little current. I think it's because the bias supply is just providing a voltage reference and that any current draw is from the resistors in the circuit, but I could very likely be wrong. Enzo has elaborated on this a couple of times if you'd like to dig into it.

[andrew_k]

I will definitely dig into it (in 5 hours when I finish work )
Thanks

[ecc83]

200mA is WAY more than you need!

The grid leaks will be 220k? So you don't want to put more than another 50k to ground so that is how you calculate the current requirement.

The lower the bias source R the better because it keep Rg low and will be more stable.

Dave.

[Steve Conner]

The required bias voltage for a given idle current depends mostly on the screen voltage, and hardly on the plate voltage. The more screen voltage you put on, the more the screen works to lure those electrons towards the plate, and hence the more negative voltage you need to put on the control grid to hold them back.

So when you're designing your bias supply it's the screen voltage that you have to take into account, and if this is only 360V or so, you'll not need as much bias as you might think. You can read off the exact value from the curves on the tube datasheet.

I've tried doublers, triplers, LM337s and so on. But IMO, you only want to regulate the bias supply if you're also regulating the screen supply. One regulated and the other not is actually worse than having them both unregulated, in which case they both vary together with changes in mains voltage, and the variations cancel to a first order.

One bias supply trick that can get you out a tight corner, is to run a small transformer from a wall wart or the like, backwards off your heater winding.

[ecc83]

Hi Steve,

I take your point about screen volts and tracking. I might have missed it but I thought the OP was going for ultra linear working?

Dave.

[Steve Conner]

I thought we talked him out of it because 700V on the screens of KT88s was getting a bit hairy. The datasheet gives a screen rating of 600V, though, so it may work fine.

If he does use UL, he will indeed need a s**tload of bias voltage. I can't get KT88s to idle in my Crown amp at a current that wouldn't burn out the PTs. It's UL with 430V HT and up to -48V bias, 1k screen resistors in series with the UL taps, and works fine with EL34s.

[Enzo]

Hi, I have not been in this discussion, but a couple pages ago you made mention of finding a standby switch to handle the voltage. Wondering if we couldn't use a piece of semiconductor for the job and let the switch control that.

[ecc83]

Blame age and/or meds Steve! Yes I think I rememeber now!

BTW '34's need much less bias than these "old things but then they have a higher gm.

Keep up Enzo! We have settled on breaking the 230V ac side! I throw this in, probably a good reason no one has done it....

Why not just crank the bias up to -100V or so and shut'buggers down that way?

Dave.

[andrew_k]

Eeep, lots for me to read and attempt to digest.

I've attached a first attempt at the inclusion of a bias circuit. The limiting and range resistors will probably need adjusting but have I got the usage of the voltage tripler correct?

Yes, I've abandoned the idea of using the ultra-linear OT taps.

Thank you all for your time and advice so far

[andrew_k]

Attached is the completed PSU design as built.

The bias supply has been built and tested, the voltages listed are real. I might need to adjust the voltage, but hopefully I'll retain enough range to bias either KT77 (25W dissipation), 6550 (35W), or KT88 (42W).

The HT supply design produces different voltages in PSUD2 vs. an LTSpice sim, so I'm going to build it as drawn and adjust before sticking in the tubes. If changes to the power supply happen I will update the schem and post it here in case anyone uses this thread as a future point of reference. I'm still waiting on some parts so first fire-up won't be for a couple of weeks.

My sincere thanks for your expertise and patience guiding me through this design.

[ecc83]

Yes as a quick look (got to go back to work and fix some!) that looks ok.

Tiny point, you are not going to bias those valves to pull that max Pa are you?

Dave.

[andrew_k]

Cheers Dave.
re: Pa - 70% or lower. *maybe* up to 75% on the KT77 if it means I can avoid modifying the bias supply. I only have 6550 and KT77 pairs on hand so those will be what it gets fired up with.

[andrew_k]

Built.

The power supply is finished, everything turned out pretty much as expected.
The 230V secondaries both put out 246V due to the 240V mains. I've added the 820Ω resistors before the first cap on each to prevent the plate supply from ever being above the 6550's 680V maximum.

One strange thing; the voltages at the B+ points for PI and preamp are all the same. I assume this is because there is no load on them?







Thanks again for the invaluable guidance in designing this PSU

[tubeswell]

One strange thing; the voltages at the B+ points for PI and preamp are all the same. I assume this is because there is no load on them?

Yep - without a load everything is at the same potential.