Mosfet B+ reducer in Voltage Doubler

[Astronomicum]

With the current transformer, which most likely came out of one of the 1200 series Univoxs, you can expect to see a voltage in the high 600s. I have seen a few 1200s just over 700V. I attached a U1221 for reference (which eccentrically is what this modified U320 is now).

To hit a target of 450V, you need to shed over 200V. If you target the max of 550V for 7591s, then over 100V. U320s with stock transformers typically run about 500V.

https://el34world.com/charts/Schematics ... ox1221.pdf

[R.G.]

I thought the reason folks were recommending the Mosfet/zener was because a dropping resistor will react badly to fluctuating current and sag and suck tone and inevitably melt?

The resistor will do what resistors do: V = I * R. Resistors simply can't react badly to fluctuating current.

Resistors only melt if you use a resistor that can't dissipate enough power. For 10W of power in a resistor, a 25W resistor will get noticeably warm, a 10W resistor will heat to a surface temp between 75C and 100C, and a 1/4W resistor will go up in smoke.

Resistors can cause voltage sag by that V = I * R thing. If the current changes through a dropping resistor, the voltage dropped by the resistor changes too. If it's in series with B+, then the amp's B+ sags as the current changes. Some people love this, some don't. A resistor, by itself, generally does not suck tone. The circuit it's in may change its frequency response in reaction to its power supply voltage sagging, but the resistor only changes the voltage.

Using internet "folks" for technical information is a tricky business.

[martin manning]

RG, I was thinking as long as there is a cap of some size after the resistor there will be a ready source of current, and there wouldn’t be much “sag” effect.

What if the resistor is placed in the return, in the same place as proposed for the MOSFET? Any difference in the response?

[R.G.]

RG, I was thinking as long as there is a cap of some size after the resistor there will be a ready source of current, and there wouldn’t be much “sag” effect.

What if the resistor is placed in the return, in the same place as proposed for the MOSFET? Any difference in the response?

A series resistor and a cap to ground make a low pass filter. At frequencies above the corner frequency (F = 1/(2*pi*R*C) ) there is little if any sag because the capacitor is integrating out the current movements. For frequencies below the RC corner frequency, there is always sag because the resistor is always doing V = I * R. The capacitor is always doing dV = I(t)*dt/C.

Power supply caps are chosen so that the cap value only allows delta-V (the ripple run-down voltage) to be thus-and-such based on the needs of the circuit. For tube amps, this is often 20V to 50V or more, depending on how much current the output stage pulls, and whether it's Class A (that is, nearly constant power supply current) or Class AB (about half max current to max current). We get a sawtooth at 120hz of tens of volts. The output stage is set up to cancel this sawtooth out in push-pull stages, and we use chokes and many stages of R-C filtering to get rid of the rest for preamp tubes.

So for signal frequencies, yeah, effectively no sag. For the time between notes, little if any sag because the cap can hold the B+ up mostly. For loud passages where the amp circuit changes from its low-signal power use to a high-power, high-current operation, the cap can't hold the DC up that long and the "DC" level sags. It's time dependent. The ripple voltage increases during times of high current, and the average DC in the middle of the ripple sags. This has effects on the tonal response of the output stage and to some degree the preamp.

Most of the time, people are looking for "sag" over loud passages or notes. The amount of sag on a note depends on the size of the cap, just like rectifier ripple does.

Tossing in a series resistor and a second "first" filter cap changes things because if the resistor is big-ish, it relieves the issues of maximum capacitor size for vacuum rectifier tubes, and the second "first" cap can be made huge. If you do that, you can run the R-C time constant for the second "first" filter cap up to BIG, longer than loud passages if you can afford enough caps. For the example 390 ohm series resistor, and a second "first" cap of, say 47uF, the time constant is t = 390*47E-6 = 18mS. Hmmmm. That's no much bigger than the 16mS time foa cycle of AC. If we get the time constant up to maybe half to one second, the cap will ride it through without much sag. But the cap needed gets to being something like 47000uF at 500V. Big dollars involved there.

That is, it's impractical in terms of space and money to put in a second "first" cap that will prevent sag on the time scales of long, loud passages. The ready source of current does help the circuit not "see" high B+ impedance, OK, and that's good for stability and removing oscillations, that kind of thing, but it's not much good for the power supply sagging from low to high power out.

Putting the series resistor in series with the high side or the low side of the second "first" cap doesn't change anything at all. As a practical matter, the rectified high voltage in the first set of filter caps will be cut loose from "ground" and the chassis and signals will be referenced to "ground" at the second "first" filter cap. Changing the end of the second "first" filter cap that the resistor goes in makes no difference at all.

[half_smith]

Thank you for your replies RG, Martin, and others.

this is the closest I could find (there are some small differences between this schem and my unit)
https://el34world.com/charts/Schematics ... e_U320.pdf

I've second guessed myself a bit on this because it's been modded by someone and the filter caps were all a bunch of parallel and series cobbled together...
I've replaced them with single 22 or 47uf 500v F&Ts, but I noticed that when I switch on the HT the voltage in B+ ramps up to over 600v (I turned down on the variac to not damage all the components rated for lesser voltages)
B+ eventually settles into a workable but still too high 550v range - still too high for my caps or 7591's.

[sluckey]

Japanese Univox (Lafayette) U-320 - designed with a voltage doubler and meant to run 7591's
Someone earlier changed the transformer and rewired it for 6L6GC's
Owner wants it returned to 7591's
The new transformer is putting out 268vac across the secondaries, so running through the voltage doubler gets pretty high.
New OT is costly and may require fitting...

You're wasting a lot of time and brain power trying to come up with a suitable way to drop 100V. I don't think you'll find one. The real issue is the new transformer is not suitable. The proper thing to do is use the proper transformer and rectifier. Looking at both of the U320 schematics in Hoffman's library, I see one uses a 180V PT with a voltage doubler, and the other uses a 390-0-390 PT with a conventional full wave rectifier. Either PT produces about the same B+. So just get a proper PT (I prefer the 390-0-390) and be done. Sometimes you just can't make a silk purse from a pig's ear. Let the owner know it's gonna cost to return the amp to it's original glory.

[R.G.]

[...] The real issue is the new transformer is not suitable.

Oh, sure; drag us back from theoretical considerations to the real world...

You're right, of course.

[johnnyreece]

• Japanese Univox (Lafayette) U-320 - designed with a voltage doubler and meant to run 7591's
  Someone earlier changed the transformer and rewired it for 6L6GC's
  Owner wants it returned to 7591's
  The new transformer is putting out 268vac across the secondaries, so running through the voltage doubler gets pretty high.
  New PT is costly and may require fitting...

There are two versions of the U320: the U320A (Lafayette), and the U320APB (Univox). The Univox uses a 780VCT PT.

https://el34world.com/charts/Schematics ... 320APB.pdf

I have a PT for the Univox version which I would be willing to sell or trade for the 250V PT currently installed.

This seems like the easiest solution. Someone has the correct version that they're willing to trade for the one you have. Did this maybe get overlooked?

[half_smith]

With the current transformer, which most likely came out of one of the 1200 series Univoxs, you can expect to see a voltage in the high 600s. I have seen a few 1200s just over 700V. I attached a U1221 for reference (which eccentrically is what this modified U320 is now).

To hit a target of 450V, you need to shed over 200V. If you target the max of 550V for 7591s, then over 100V. U320s with stock transformers typically run about 500V.

https://el34world.com/charts/Schematics ... ox1221.pdf

Thanks Astronomicum! I think you're correct, it looks like the amp has been modded to the 1221 specs. It has the 6AN8 PI (not 12AX7)

Maybe I'll just conform it to that design and move on with life.
*I've DM'd you re: original PT trade, thanks again