Radical (insane) PSU ideas

[Fuling]

Good afternoon fellows.

I´m working on a pair of 813 SE monoblocks and I´ve run into some PSU problems.
I´ve made a pair of mains transformers from kits but looking at them I get some doubts about using them, the 750VAC they are going to put out requires serious insulation and I don´t know if I can trust my own winding capabilities...

Let´s put those monsters aside for a while and look at some alternative solutions.
In my amps I´ll need about +800-900V @ 90mA for the power section and +450-500V @ 50mA for the driver/input section.

My idea is to build a 500V 140mA supply (I have the necessary parts for that) and then stack a 300V 90mA supply on top of it to get the full B+.
Something similar can be seen here: http://www.svetlana.com/graphics/TB/No.26fig2.jpg

For various reasons I plan to use SS rectifier diodes, but I want a slow startup sequence and I have an idea for that: Tube diodes in series with the rails. I have a good supply of 6AU4GTA´s that should be fine here.

The major reason for not using them as rectifiers is that simulations shows that ripple current can and probably will be a problem.
If the diodes are placed after the first capacitor it should only see quite clean DC so ripple can´t be an issue anymore.

I´ve read about people who has used TV damper diodes in a similar fashion but most seems to use a relay that shunts the diode aftre a while, but is that really necessary? At 90mA a 6AU4GTA seems to drop less than 10V and I don´t care about 10V more or less in a 800V supply.

I´ll post a schematic later today when I get access to a scanner so stay put!

Cheers!

[jan.didden]

Originally posted by Fuling
[snip]My idea is to build a 500V 140mA supply (I have the necessary parts for that) and then stack a 300V 90mA supply on top of it to get the full B+.
Something similar can be seen here: http://www.svetlana.com/graphics/TB/No.26fig2.jpg
[snip]

That'll work, but it isn't optimal. The load current for the output stage will also run through the supply for the input stages, so the lower supply will have much more ripple because of this (about three times as much). I don't know how significant that is.

Why don't you try out those xformers you wound? If you have a variac, slowly power the primary through a separate fuse and watch for signs of shorts and/or arcing. You may worry unneccesarily. If all seems well, let them sit a few hours to get them used to the work

And oh yes, do this outside, with a healthy distance between you/the variac and the transformers.

Jan Didden

[SY]

Stacking supplies can be very tricky from a safety point of view. Insulation failures that in other circumstances would be harmless can now kill you. I'd be inclined to go find an old ham transmitter and rescue some proper PS parts from it.

Regarding series diodes, think about the effect on dynamic regulation.

[Fuling]

Oh, my optimism suddenly got shot down

I don´t have a variac, but I have afriend that does. Maybe I should try them out after all, but I´m not too optimistic about them.

I´ve thought about the ripple in the driver rail and my cure for that would be to put some heavy CLC filtering after the point where the two supplies connect to each other. A bit hard to explain without a picture, but that CLC filter would be exclusive for the driver stage.

The HV rail would also have CLC filtering with the last cap referenced to ground instead of to the 500V rail. I have a pair of 30H 200mA chokes that should be able to quiet things down quite good

Safety issues? Well, we´re talking 800-900 V that connects to a top can on the output tubes, need I say more?
The "bottom" transformers puts out about 380V and are referenced to ground, so they shouldn´t have any problems.
They are actually 380-0-380V types made for tube rectified supplies, but that doesn´t matter here.

The problem that I can see would be the "top" transformers, but I´m going to use regular mains toroids here that usually are insulated for 4kV between prim and sec.

The diode regulation shouldn´t be too much of a problem, since they would work with way more constant currents and voltages that when they do their usual job as rectifiers, right?
Their sins should be small compared to the chokes, for example.

[Colt45]

why not just put a FWB on a 380-0-380 trafo?

that should do.

[Fuling]

Colt45:
Two reasons, 1: They can´t handle the current. It would work with a choke input filter but that would only give just over 600V.

2: Since it´s designed for use with a grounded CT I doubt that it´s insulated for FW bridge use. Since it has two filament windings a breakdown would end in a mighty soup

[Eli Duttman]

Fuling,

Consider the following Pete Millett idea. Use a CT plate power trafo, like those Hammond offers. Wire a hybrid bridge rectifier to the ends of the trafo. Stacked UF4007s and 6AU4s make a good bridge. Wire both anodes of a 7Y4 or similar rectifier to the CT of the power trafo. You take the high B+ from the damper cathodes and the low B+ from the 7Y4 cathode. Both B+ rails are full wave rectified; the SS diodes form the path to ground for both rails.

Obviously, the VA rating of the power trafo has to be adequate for the total draw. FWIW, I'd use choke I/P filters on both rails. A 20 H. inductor, with its low critical current, is practical on the low rail.

[Fuling]

Eli Duttman:

Not a bad idea, but the problem is that the range of Hammond transformers that allow 230V primary voltage are specified for FW CT use, not FW bridge use. Their 115V series can handle both. A bit disturbing if you ask me...

Well, I´ve run some more simulations and it seems that the ripple current problem isn´t as bad as I thought so so a 6AU4GTA/1N4007 bridge might work fine after all.

The transformers I´ve made are 300VA types with the following windings: 750V , 350V, 12,6V, 6,3V and 4V.
The problem is that they are not especially well made (my windings skills are a bit lacking...) and I´m not sure about the insulation between the windings.

The amps does not necessarily have to be monoblocks, so
one way to fix this would be to rip the transformers apart and use one bobbin to wind a single 750V 400mA winding for the B+.
Maybe 725 + 25 +25V to allow a bit of fine tuning.
A single winding would be way less troublesome to get right.

Heaters and driver stage can be fed from separate transformers, that´s not a problem.

[slowmotion]

Hi Fuling

If you don't feel good about your transformers
then maybe use them for something else.
Peace of mind is very important,
don't you think?

You can order custom made transformers from
many suppliers, I got mine from a supplier in England,
I use 980v/600mA transformers with 3B28 FWB,
one for each channel.


cheers

[Fuling]

Jan:

Yes, peace of mind is important. I don´t want any arcing or flames to disturb my listening sessions

I guess I could custom order a transformer from Sowter in England or askjanfirst.de in Germany, but it would be far from cheap I guess.

Demostenes in Norway has a massive 10kg monster with 2x700V secondaries plus several filament windings that would be nice, but someone told me that it costs a fortune to order stuff from them.

I think that rewinding one of my transformers and putting my best effort into that single HV winding might be the best solution.
That would cost lesst than 1/10 of a custom ordered transformer from some foreign company, and price is important.

Wow, 980V and 3B28´s, what kind of monster amp is that??


Cheers