• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Sanity check? HV supply for 125wpc sweep tube amp

ericj

Member
2008-12-08 10:24 pm
I mentioned a while back that i think i would like to try and build a 125wpc stereo amp based on pete's (almost) universal driver board, some big sweep tubes, and some of the solutions applied to making big power versions of the Big Red Board amp.

I know I need a pair of 300-ish volt supplies, and tightly regulated 150v supply for the screens.

My plan is to etch a pair of power supply boards, one for each channel. At the moment i have a pair of big industrial control transformers which can supply a bunch of 300vac each. I'd be using those plus an antek toroid for the heaters, and a small dual 36v transformer for the bias supply. Bias supply is handled by the universal driver boards, so that isn't reflected in my schematic.

So, shamelessly copied from the big red board, here's what I have come up with so far:

Ybeea0R.png


Should the secondaries of my 300v trafos be tied together at pins 2 and 3 of P1? I suspect that they should be.

I am also curious if capacitor values should be increased, since the big red board was originally specified for much lower output.

Am i on the right track?

Thanks for any input.
 
Last edited:

ericj

Member
2008-12-08 10:24 pm
ZaDKRwb.png


Using 35mm footprint for the B+ capacitance so i can fit whatever i need to fit.

I've got a dozen or so 8"x4" sheets of copper clad FR4, thus the formfactor.

Managed to do this with just the one top side jumper.

I'll have to go through this again once or twice before i etch. There's got to be a better way to use wide traces with to220 parts, for example, and I might need to adjust the pads for the big caps, since I'm not sure that footprint really works for a snap-in cap.

I don't really like that long trace at the top. I might ditch the 5 position terminal block and go with a 2 position for B2 and screen supply, and a 3 position for ground, heater reference, and B1. I could delete my 0 ohm resistor at that point, too.
 
Last edited:

rayma

Member
2011-04-29 8:37 pm
Should the secondaries of my 300v trafos be tied together at pins 2 and 3 of P1?

The two transformer secondaries must NOT be connected together, because two of the supplies
are stacked in series. Both C1 and C2 will likely need to be several times larger for a 125W amplifier.

It may be better to use three identical circuit blocks side-by-side for this layout,
leaving some extra room for the Zeners in the screen supply block.

The two series supply blocks should be adjacent, with the grounded series supply in the center,
and with the screen supply on its other side, nearer the output common.
 
Last edited:
Why only one screen supply? Are you aiming for individual monoblock type channel separation, or happy to have some interaction?

Have you done a PSUD2 simulation of say the max continuous loading on the basic rectifier/filter to determine the ripple level, and then tried to settle on a fixed output regulated voltage that is suitably lower than the min raw B+ (due to ripple) and including mains voltage tolerance?

Have you then worked out the max power dissipation in your regulator FET when the raw B+ is at max due to mains tolerance, and the loading is max?

That would at least indicate if your FET choice and heatsinking is practical, or as rayma highlights you may want to do some redesign of the capacitance needed (and hence pcb layout).
 
Last edited:

ericj

Member
2008-12-08 10:24 pm
The two transformer secondaries must NOT be connected together, because two of the supplies
are stacked in series. Both C1 and C2 will likely need to be several times larger for a 125W amplifier.

It may be better to use three identical circuit blocks side-by-side for this layout,
leaving some extra room for the Zeners in the screen supply block.

The two series supply blocks should be adjacent, with the grounded series supply in the center,
and with the screen supply on its other side, nearer the output common.

Whether or not the secondaries should be in series was not clear. It's an easy change to make.

I know that the caps may need to be much larger than the original 47uf in the schematic. That's why the footprint is big enough to accommodate 470uf caps.

I'm unsure if i can put the screen supply at the end of the board in a clean way. Putting it in the middle was relatively easy.

Why only one screen supply? Are you aiming for individual monoblock type channel separation, or happy to have some interaction?

Have you done a PSUD2 simulation of say the max continuous loading on the basic rectifier/filter to determine the ripple level, and then tried to settle on a fixed output regulated voltage that is suitably lower than the min raw B+ (due to ripple) and including mains voltage tolerance?

Have you then worked out the max power dissipation in your regulator FET when the raw B+ is at max due to mains tolerance, and the loading is max?

I'm attempting to go off of what people have done on big builds of the "big red board", but i want to use pete's universal driver boards.

The advice i got was to stick with the <400v supply for the driver board, and have a higher voltage supply for the output tubes.

I plan on two of these boards, in a semi-monoblock configuration. The channels will share power transformer iron but nothing else.

I have not worked out the max power dissipation in the regulator FET. Allegedly at least a dozen 125w versions of the big red board have been built, and this fet seemed to do the job.
 

rayma

Member
2011-04-29 8:37 pm
I know that the caps may need to be much larger than the original 47uf in the schematic.
That's why the footprint is big enough to accommodate 470uf caps. I have not worked out
the max power dissipation in the regulator FET. Allegedly at least a dozen 125w versions
of the big red board have been built, and this fet seemed to do the job.

A 470uF in place of the 47uF should be fine, but verify this. The FET's temperature depends
on the heatsink, current, and voltage drop, and should be designed to be within the margins.
The heat sink used is not likely to be very low in thermal resistance, though.
 
Last edited:

ericj

Member
2008-12-08 10:24 pm
A 470uF in place of the 47uF should be fine, but verify this. The FET's temperature depends
on the heatsink, current, and voltage drop, and should be designed to be within the margins.
The heat sink used is not likely to be very low in thermal resistance, though.


So far, this is the heatsink I'm looking at:

530002B02500G - Board level heatsinks

Also keep in mind that there is one of these boards per channel.
 

ericj

Member
2008-12-08 10:24 pm
If possible, the filter cap pad should be a star point for incoming and outgoing current flow - that is appropriate for the main caps on either side of regulator.

Tricky on a single-sided printed board.

The regulator circuitry should preferably reference the output capacitor negative terminal, as that is the interface.

The 1M resistor it references ground through is far more significant than a couple cm of copper.

The B1 and B2 supplies aren't technically regulated. They're just ripple filters.
 

ericj

Member
2008-12-08 10:24 pm
Keep each rectifier/input capacitor as a sub-circuit, and run +/- traces directly
from the capacitor's pads to the remainder of the circuit (a Kelvin connection)
to minimize Ohmic noise coupling from the charging pulses.

Not really following your logic.

Are you saying i should leave those traces off the board and run jumpers, or what?

Anyway, I admit i wasn't happy with the position of C2. So, shoved things around a bit.

And yeah, I don't love the trace that runs under R1, but the alternatives are also ugly.

sAXQBhe.png
 
Its good to appreciate where the pulsing currents in the rectifier circuit are mainly looping. The transformer winding causes a pulsed current to circulate through its winding, and through the rectifier diodes and NTC to the first main filter cap. Its best to not connect any other load circuitry to anywhere on that loop except for the capacitor pads. There are many forum posts on hum where that particular issue is the problem, especially for chassis/tag board style fixed wiring amps.

For most amps, that gives acceptable hum/noise performance. For hi-fi where reducing hum and noise floor is a key aim, then it is appropriate to continue that methodology for all bypass filter caps, such as after the regulator, and for each amplifier stage.

Although the 'ripple filter' circuit may not appear to be a regulator, the basic function is to maintain a constant voltage on the FET gate (in the sense of the long time-constant of the 10uF reference voltage). So if that reference voltage has noise/ripple on it then the FET will attempt to recreate that noise/ripple at its output as well. Making a connection along a pcb trace, rather than at a pad, may seem of no consequence, but is the basic tenet to why vintage style busbar grounding is now normally replaced by distributed star grounding in amps. This attention to detail is for pedantic hi-fi, so is often not considered a concern.
 
Last edited:

ericj

Member
2008-12-08 10:24 pm
Made a pass at rearranging things so that the B+ supplies can use a large off-board heatsink.

Aesthetically, I don't like it. Though these contortions do use space more efficiently, giving more room for capacitors. In retrospect, just adding another big can to B2 was probably not the wisest use.

I might move C7 to between the B+ fets, which would require me to use a top-side jumper.

6fSBdAm.png