Adding pairwise fuse protection while still having external single tube bias check...

[Steve Conner]

No coincidence, it's the Real McKeen! :-)

In 2000 (so RG beat me to it I guess) I built a tube amp with a MOSFET regulator for the B+. The regulator was hooked up to the standby switch to provide soft start, and it also had current limiting, and a screen overcurrent detector that would latch the B+ off, requiring a cycle of the standby switch to bring it back.

I ended up ripping this feature out. Why? I didn't realize that guitar amps abuse their screen grids mercilessly, and the limiter would turn the amp off just as it was starting to sound good. The moral of the story is... well I don't know what it is to be honest! :-)

In hindsight I'd have only regulated the screen and preamp supply, or maybe I wouldn't have bothered. The amp is still working, however, regulator and all.

The original question seems a hard one to me. The only solution I can think of is to put the bias test resistors between cathodes and fuse, and accept the following:

1: They're going to get live if the fuse blows.
2: You're going to need two negative test jacks and four positive ones, because the negative ends of the test resistors aren't grounded. The voltage drop across the fuses would be significant.

[J M Fahey]

WOW !!!!!

I'm finding a lot of my (modern) design and service heros here.
Good forum to hang out and learn.

[R.G.]

No coincidence, it's the Real McKeen! :-)

Yep, guilty as charged.

In 2000 (so RG beat me to it I guess) I built a tube amp with a MOSFET regulator for the B+. The regulator was hooked up to the standby switch to provide soft start, and it also had current limiting, and a screen overcurrent detector that would latch the B+ off, requiring a cycle of the standby switch to bring it back.

Not to that one. I didn't get around to an integrated B+ regulator/startup/etc. for several years. I did equip a friend with a B+ regulator made from power MOSFETs and run by an LM317 floating up at B+ back in about that time.

I ended up ripping this feature out. Why? I didn't realize that guitar amps abuse their screen grids mercilessly, and the limiter would turn the amp off just as it was starting to sound good. The moral of the story is... well I don't know what it is to be honest! :-)

Probably the moral is - it's impossible to overestimate what a guitarist will do to an amp in search of "that tone".

My tinkering with current limits on the output tubes taught me that there are a lot of transients running around in there that are not brought to light by the coupling to the speaker. The first model reliably tripped exactly where I set it, at 10% more than the peak currents for the max output power.

And did it on remarkably UNremarkable output levels, too. Hmm. Spikes. Gotta toss in a little transient damping capacitance on the front end to slow things down a bit. Better, but still trips. OK, up the average trip level to near the max current for the tube. Better, but still trips. More capacitance. This went on until I worked out that it had to be slow enough to let the tubes overload for a few seconds, and high enough current level to let them go ahead and overcurrent, but count on the time delay to shut them off before they could really melt down. With microsecond shutdown available, I had to not use it because guitar players would not stand for it. sigh.

In hindsight I'd have only regulated the screen and preamp supply, or maybe I wouldn't have bothered. The amp is still working, however, regulator and all.

Yeah, preamps are much more predictable. They profit from low impedance supplies too.

The original question seems a hard one to me. The only solution I can think of is to put the bias test resistors between cathodes and fuse, and accept the following:

1: They're going to get live if the fuse blows.
2: You're going to need two negative test jacks and four positive ones, because the negative ends of the test resistors aren't grounded. The voltage drop across the fuses would be significant.

The devil that sits on my shoulder suggests another way: fuse each tube in its plate lead, then sense when a fuse is blown in one tube. Use that detection to fire an SCR to blow the matching fuse on the selected other tube... There is nothing which can't be added to with another bell and/or whistle.

[Enzo]

Offhand, It would seem to me to be a lot of effort to... well, not sure what. If the tubes are fused individually, so what if only one blows instead of two? SO the amp will have two tubes on push and one on pull. Still works. When you have that pair of tubes go out of service, will there also be an automatic impedance switchover to another output tap?

[DIY_Guy]

When you have that pair of tubes go out of service, will there also be an automatic impedance switchover to another output tap?

What a great idea

[DIY_Guy]

In case any one was curious, this is what H&K uses....

[R.G.]

In case any one was curious, this is what H&K uses....

Ah. That's a fuse-blown-indicator, not a limiter. The fuse blowing lets the tube provide drive to the transistor through the 100K resistor. The transistor then turns on the LED.

[DIY_Guy]

R.G. or anyone else following this....

On another subject, I am trying to design my bias adjustment for each individual tube. I know how to do it in a 2 tube PP design, but I'm a little stuck on a 4 tube design. Is it as simple as splitting the signal coming off of each side of the PI? So if my 2 tube 50w design looks like this-


I should just change it to this-


Please correct me if I'm wrong.

[black_labb]

if you bias them seperately you will have separate coupling caps coming off the PI and the separate bias supply into different gridleak resistors.

when i say separate bias supply i dont mean completely different, i just use 2 parrallel pots for the separate supply. you have to wire the bias to the wiper though if you want to keep them from being interactive. in a case of the wiper dieing you will lose your bias supply and the tubes with it if you arent quick to notice.

[R.G.]

R.G. or anyone else following this....

On another subject, I am trying to design my bias adjustment for each individual tube. I know how to do it in a 2 tube PP design, but I'm a little stuck on a 4 tube design. Is it as simple as splitting the signal coming off of each side of the PI?

It's as simple as splitting the signal to be DC isolated on each tube - and as complicated, as Black Labb points out.

Pots providing DC levels are notorious for having their wipers go open and developing dead spots. It's a beginner's mistake to trust the wiper of a pot to always be in contact. Where an open wiper would let the tube detonate, you you have to design it to fail safe. This is usually in the form of a resistor from the wiper/grid connection to the most-negative side of the pot for tube bias pots.

The dead-simple version is a single pot across the full bias supply with a resistor from the wiper to the full bias supply. When the wiper goes open, the tube is biased as off as the resistor can make it, so it may have crossover distortion but does not overheat and take your B+ fuse (if you have one) and possibly your OT, rectifier, and/or PT with it.

But sizing that resistor is a ticklish design problem. You want a resistor value high enough not to interfere with the action of the pot and low enough to actually drag the grid down.

Tubes have max values for the grid leak resistor. The resistor has to be low enough to actually dispose of the leaking electrons - "grid leak resistor" is an accurate name! And the bias pot/network has to be low enough in resistance to make sure that it can drag that grid leak resistor to the bias voltage. On top of that, it's a little counterproductive to have the pot itself strung across the full bias supply, as you almost never want any bias voltage smaller than some minimum for the tube. For example, a 6L6 may need bias voltages of -30 to -45, but probably never -10 to 0. The low end travel of the pot is completely unnecessary, and opens the possibility of setting the bias to a destructive value accidentally.

This leads to identifying another beginner's mistake - not limiting the effective range of a pot. You do this with fixed resistors on either or both ends of the pot. This limits the active range of the pot's wiper travel to the desired range of electrical change. In the bias networks I designed from scratch, I used a pot and three resistors. One was from ground to the clockwise side of the pot, another was from the CCW side of the pot to the full bias supply, and another was from wiper to the CCW end of the pot. The resistors were sized so an open wiper pulled the bias voltage to the CCW end of the pot, and the total resistance from tube grid pin through the grid leak resistor, through the "fail-safe" resistor and then to the bias supply was less than the maximum resistance spec'ed for the tube. As you can see, it took some dinking with resistor values to ensure that the resistances were small enough to make everything work out.

The final values were smaller than most people use for bias pots. Which brings up the last consideration, which is also a primary design issue: the bias supply itself must be able to provide the necessary current to the lower-resistance network on the bias adjuster to make the voltages work out as planned, and it must do this for as many bias networks as you put in parallel across it.

That may be tricky in a situation where you're not designing an amp and power transformer from scratch, but instead adapting an existing PT which may have a high-impedance (i.e. only low currents available) bias supply tap, or no bias tap at all, only a capacitor-limited supply derived from the high voltage winding. In that situation, you're faced with the bias supply simply being incapable of doing what it needs to for the tubes and the bias adjusters simultaneously. If you have to live with the PT and bias supply that you have, no luxury of being able to beef up the bias supply, you have to do the homework to design some stuff, figure out what happens under failure conditions, and make the failures as non-fatal as possible. This would probably come in the form of a bigger "fail-safe" resistor to the full bias supply per tube/wiper, and making this resistor and the grid leak resistor to the tube as small as possible, figuring it will cover many tubes if not all.

[defaced]

I should just change it to this-


Please correct me if I'm wrong.

Yes, this is correct. The individual caps isolate the DC bias voltage from each tube. The individual grid leak resistors go to each bias supply, and well you already need individual grid stoppers, so that stays the same.

Here's a quad bias supply I've been working on. I can't guarantee the values because I accidentally fried my prototype before I had a chance to fully test it and incoming voltage is very important when designing the resistor network. It'll allow independent adjustment of each tube, and the ability to swing all four tubes up and down together.

I'm definitely interested in critiques on this circuit.