Need help with 600v+ PSU

[andrew_k]

Following Steve Conner's suggestion here, I'm trying to design a PSU with a 600-650V supply for the plates of a pair of 6550C, using a toroidal PT with two 230v windings.

Originally I planned to take the screen/PI/pre voltages from one winding and the plate voltage from both windings in series, but since I have an ultra-linear transformer I realized I only need to take the PI and pre from one winding and he screen comes from the UL taps.

Attached is my first attempt. Is this on the right path? Have I at least got the theory of combining the two supplies correct?

[black_labb]

seems on track. you will need to put in the current drain from each section to get the right values for the preamp voltage. not sure how final the values fro the components are, but 47uf for the first stage(s) should really be higher. because they are in series the actual capacitance from ground to the doubled supply will be 1/2 (23uf) which is not going to be enough.

as for the ultralinear, you dont need to do it that way. i would probably suggest against it. the screens have a max voltage of 500v from memory, so you would have to drop a fair bit of voltage (which would have to be between the ot and the screens). either going from the 1/2 voltage, or dropping the voltage another way. having the screens a bit higher could give you more power if thats the aim, but it will be harder on the tubes (screens in particuar). im not sure about kt88's/6550's, but el34's have issues with their screens dissipating too much power which sends them to an early death

[andrew_k]

seems on track. you will need to put in the current drain from each section to get the right values for the preamp voltage. not sure how final the values fro the components are, but 47uf for the first stage(s) should really be higher. because they are in series the actual capacitance from ground to the doubled supply will be 1/2 (23uf) which is not going to be enough.

Doh! I stuffed up the cap values as I was in a rush to get it done before going out. The PI and preamp target values are the "as built" (no load) values from the AX84 October Stage, which is what the rest of this amp is. I know I should allow 5mA per triode in the 12AX7s, but I'm quite confused about what mA I should use in the model for the plates and screens.

as for the ultralinear, you dont need to do it that way. i would probably suggest against it. the screens have a max voltage of 500v from memory, so you would have to drop a fair bit of voltage (which would have to be between the ot and the screens).

You've confused me Don't the UL taps deliver 43% of the voltage fed to the OT's centre tap? - 646 X 43% = 277.78
For reference, the tubes I'm designing around are the Winged C 6550C, which have a max of 680V on the plate and 400V on the screen.

Thanks for your help

[black_labb]

Doh! I stuffed up the cap values as I was in a rush to get it done before going out. The PI and preamp target values are the "as built" (no load) values from the AX84 October Stage, which is what the rest of this amp is. I know I should allow 5mA per triode in the 12AX7s, but I'm quite confused about what mA I should use in the model for the plates and screens.



You've confused me Don't the UL taps deliver 43% of the voltage fed to the OT's centre tap? - 646 X 43% = 277.78
For reference, the tubes I'm designing around are the Winged C 6550C, which have a max of 680V on the plate and 400V on the screen.

Thanks for your help

the screen taps are on the same winding as the plates, so they get the same voltage, +- a few volts. the 43% refers to how many winds from the centre tap to the screens related to the number of winds before the plate taps (or the other way around)

see the diagram of the trannies here http://www.hammondmfg.com/1608.htm the screens are between the CT and the plates on the same winding.

with a max screen voltage of 400v, i'd suggest just running the screens off the 1/2 voltage point.

for the current needs of the plate and the screen, use the datasheet with the closest voltages to what you will have

[sasquatchito]

Howdy All-
I realize this isn't a toroid, but this is a PS design that I've used to power 2x6550A; with an OT primary of 4.5-5K, the result is right around 100 watts.
(from a UNIVOX 1061 Bass head; originally ran 4x6L6GC at these voltages!!)
Hope this helps...
Tom

[ecc83]

If I might interject!

Whenever you put caps in series in power stages it is very good practice to shunt each one with identical value resistors. This ensures equal voltages on the caps since electrolytics have very wide tolerances and the charge can give uneven volts (not it this particular case I know) but also to act as drains when the kit is off, how I have cussed amp makers who dont' DO this!

68K 2W will be about right and pull a smell under 5mA for 323V.

Dave.

[andrew_k]

black_labb: I'd mixed up 43% imp. vs. 43% voltage. That was dumb. Thank you! I've attached a revision of the design, this time with loads emulated. The voltages used for load calculations are the unloaded voltages; I couldn't think of any other way to calculate it.

Tom: It's good to see a commercial PS similar to what I'm planning.

Dave: Thanks for your suggestion, I will add the resistors. To emulate their affect on the circuit would I add 5mA load to each of the two current taps that follow the series caps?

[black_labb]

seems good so far. for the resistor moddeling, you should be able to put the resistors there which should give the correct current drain and such, without putting a separate supply there.

[ecc83]

Hi again.

The extra juice pulled by the bleed R's should not be important, the vaguaries of cap values and mains voltage make this a "cut and try" art to some degree!

Whilst I accept that this circuit is in gestation can I make a few comments?
Fusing: If both anode and G2 supplies (i.e. you don't go ULin) are fused you could have the situation of no Va but keeping VG2, the screens will melt. Some low Ohm fusibles to feed each screen would be advisable.

Is there yet another transformer for the filaments? If so you could have 2 mains switches and get the heaters cooking before HT slamms on. Less stress all round and the HT switch becomes "standby".
Or! You could have a delay circuit to feed the torroids, they are a bit notorious for innrush current and primary fusing can be a headache.

Just my 2 p's worth!

Dave.

[Steve Conner]

The PS under discussion doesn't need the resistors for balancing: the capacitor voltages are forced equal because each one is fed by its own rectifier. Same applies for the full-wave doubler, and the center tapped bridge circuit from the Univox. (This latter is the same as the split-rail supply in a solid-state amp, but with the ground reference moved to the negative rail.)

However, it's still a very good idea to put the resistors in as bleeders in all of these circuits. If you forget them, you can get into a nasty situation the first time you fire the thing up for testing with no tubes: you turn it off and all the capacitors stay charged for days. Don't be tempted to use that screwdriver! If the capacitors are any good they'll blow the end clean off it.

UL taps on an OT provide 43% of the audio signal voltage, but the same DC voltage as the plate taps. I'm not sure it's a good idea to run UL at such a high supply voltage. (My only experience with UL is my recently-restored Crown amp, and I can see the screen wires on that glowing at idle.)

A standby switch would be nice, if you can find one capable of breaking 600V DC. I like to get the heaviest toggle switch I can find in DPST, and use the two poles in series to spread the arc out a bit. But even so, you're probably using the switch outside of its ratings.

BTW, I think this PS circuit with the two bridge rectifiers in series was originally used by someone like Sound City or Orange.

[ecc83]

Hi Steve,
I did mention that this circuit did not need balancing R's but I am glad you agree they should be there to save blue words!

Yes 600dc can be a pain so I suggest breaking the ac feed to the bridge, no arcing with ac. But if the op is just using the torroids for HT he can switch the primaries.

Dave.

[black_labb]

that explains why my old holden/wasp amps leave the plate voltage (700v) on during standby, and the high voltage isnt fused. i always wondered why. still dont understand why they use 2 fuses on the mains, a 3A on the live, and a 5A on the neutral. dont think that one will ever be explained.

[ecc83]

Black lab,

Fuses in neutral lines were a plot to kill off amp technicians!

Dave.

[Steve Conner]

Indeed. Fuses in the neutral line are probably illegal in the UK. They're only really useful in countries with non-polarised plugs, where either line could be "live" depending on which way round the plug was inserted.

I think Australian plugs only go in one way round, though, so it still doesn't explain why Holden did it.

[andrew_k]

I've attached the latest revision, this time including switching and the complete transformer. The only part left is to modify the bias circuit to feed off 230v instead of 310 as in the original October Stage schem.

Questions
- Will the bleed resistor on the top winding see the full 646 volts? If so, it's going to be expected to dissipate 6.137W!

- Is 5A an adequate fuse or perhaps even overkill? I've calculated the power consumption as 197VA, but I don't know how to calculate what fuse to use.

Thanks!