effect of main 'reservoir' capacitors before or after standby sw in tube rect. amp

[Leo_Gnardo]

Leo, if its a slow warmup tube rect or not, and you get the tube rectifier all warmed up and happy, then flick the standby off, won't that cause the waterfall out of the tube into that empty 32uf cap?

Yes, actually an electron burst, and in truth the electrons are headed in the other direction but hey, so what. It is a stressful moment for the rectifier. If it's going to put on a lightning display, that's the instant it will happen. For those amps that have a reservoir on the rectifier side of the switch, it's less of a crisis as the charged reservoir contributes more or less half of the electrons set into motion at the instant standby is switched into operate mode. So yes, that method is less taxing to the rectifier tube.

If the tube rectifier does pop, do they usually go short or open? Or no way to tell?

Short, the line fuse pops. About a week ago I was working on a 1975 Deluxe Reverb. It was popping fuses after 15-30 minutes warmup. A quick exam showed 1) I had been in this one before, left a tag dated 1998. 20 years without maintenance, I guess that's a pretty good run.

2) Bias current running a bit high @ 30-40 mA on the pair of tough old RCA 6V6GA in the power department. Bias voltage around 32V. Discarded the white Mallory bias filter, installed a new 100 uF cap, bias rose to 37V and bias current dropped to a much more comfortable 22 mA on one, 30 on the other 6V6. I found a good match to the cooler 6V6 in my stash of pulls, also a nice old RCA, dialed bias down a tad to 19 mA, plenty enough to be out of crossover distortion territory. So far, so good.

3) One - original - filter cap was leaking, had a popped blister in its seal. Can we trust 43 year old filter caps. Hm, not so much. So I installed a fresh batch, warmed up the amp, worked fine. Then I let it sit there, idling but with no output tubes, for a while whilst I hit the kitchen, got a bite of lunch, checked email. Half an hour later, back to the amp. It was off, fuse blown.

now 4) What's left? The rectifier tube was a Ruby 5AR4C, I probably installed that 20 years ago. Swapped in a fresh Ruby 5AR4C, no more fuse poppin'. Let's see if this lasts another 20 years. The owner of the amp and I will be 85 years old then... what a couple of geezers. Let's see if he complains about it then! "Y'know I only got 20 years out of that last rectifier you installed. I wish you would stop selling me that made in China garbage! Now pass me that bottle of Geritol..."

This Deluxe as they all do, is a "hot switch" circuit. 20 years life out of any rectifier in constant use is pretty respectable. Sometimes I see venerable and highly praised Amperex Bungle Boys and Mullard, also Telefunken, RCA, Sylvania, GE and even Matsushita 5AR4/GZ34 rectifiers still plugging away, working just fine thanks very much, in service 40 or 50 years in spite of being in "hot switch" amps. Could we get more life out of them? How long ya gonna live - let's find out! The old tough ones still can't be beat. (None of my customers want to pay the going rate for them though.) The current champ, Ruby, not at all bad. JJ, well they did ship bad batches in the early 20teens and earned themselves a black eye for that but the currently made ones seem OK - only time will tell. And Sovtek/EH/other New Sensor brands although lauded by some techs, seem to poop out quickly so I avoid 'em.

[pdf64]

...About a week ago I was working on a 1975 Deluxe Reverb...This Deluxe as they all do, is a "hot switch" circuit...

Are you sure?
BF and SF Fenders generally have the reservoir cap on the 'always hot', rectifier side of standby, so the reservoir will get (and keep being) charged up whilst the amp is in standby mode.
https://el34world.com/charts/Schemat..._schematic.pdf
Unlike the nasty current Korg era AC30 https://el34world.com/charts/Schemat...ice_manual.pdf
or the earlier Marshall designed and build AC30 https://el34world.com/charts/Schemat...x_ac3093pa.pdf

[Leo_Gnardo]

Are you sure?
BF and SF Fenders generally have the reservoir cap on the 'always hot', rectifier side of standby

Can't say I've seen a Deluxe Reverb that has that feature. Maybe the ones destined for export to 240V countries? I've definitely seen some of the early 60's brown/tan/cream/white amps with reservoir caps on the rectifier side of the switch. Any way the factory made it, it's easy enough to alter the circuit to make life easier for the rectifier tube. In a couple of cases I've substituted a film reservoir cap say 30 uF at 800 or 900V so there's no chance of wrecking an electrolytic with excess voltage nor do we have to worry about the typical electrolytic breakdown with age. Might take a bit of ingenuity as far as cap placement but those I've fixed sound fine & owners are happy. Those film caps also come in handy with old Ampegs. I wedge them into the back corners of the chassis, epoxy them in place, use a couple blocks of styro insulation to make sure they won't rattle around in case the epoxy ever lets loose, bob's my uncle & maybe yours too...

[Enzo]

I have replaced rectifier tubes, but I have replaced far more power tubes, and far more preamp tubes.

No, I have never rewired a standby switch, I see no point. I know not to assume that one failure indicates a pattern. If I get a bad rectifier tube, I am not going to assume it dies because of faulty circuit design.

All these theoretical tendencies of circuits, well, they are just that: theoretical tendencies. So we flip the standby and a momentary current surge happens. So? It has happened every day for decades, and the amps soldier on. If the design was faulty, then that model would be eating rectifier tubes all the time. And yet, NO amp out there has a reputation for eating those tubes.

When I was in college, my frat brother was in med school, and like so many med students, as they studied each unit, they tended to see the symptoms in themselves. "OMG, I must be catching cancer TOO!!!" Then next week it is polio, and the following week, the Yaws. I see it here all the time too. Someone becomes aware of some phenomenon and that assumes it is a real important factor in his amp. We haven't heard anyone wailing about skin effect for a while, we are about due.

[R.G.]

Hi R.G., have you ever built one of these? Is it safe to put inside a Fender style chassis, gets pretty hot in there. Maybe wire the mosfet outside the chassis someplace, with suitable safety shield around the HV wiring, e.g. cap can?

Yes, I have. As with any power circuit, there are some caveats that go with it.
First - it's a power circuit, and it can get hot. You have to think about getting the heat out of the MOSFET and how much heat there will be.
In most cases, there will be minimal heating. In normal operation, the MOSFET is turned full on. The voltage across it is just the voltage needed to enhance the gate to pass that much current. This can be substantial for some MOSFETs, so you need to look at the device you buy and get a low threshold voltage MOSFET.

The first one I listed, the IRFBE20 has a threshold voltage of 2.0 to 4.0V. So the MOSFET will turn on as soon as the gate is 2 to 4 v positive compared to the source. This voltage comes through R1. In normal operation there is no current flowing through R1, so the gate is at the same voltage as the drain. When the drain (and gate) gets 2-4 V higher than the source, the gate turns on the channel and current runs through from drain to source. The spec on the IRFBE20 says it has a transconductance of 0.8S, which is 0.8 amperes per volt of gate-source voltage more than turn on voltage, so the voltage source will sag a volt or two below the drain as more current runs in. This will eat 4-6V from your B+, all the time.

The power spent in the MOSFET will be the time integral of the current and voltage across the MOSFET, which depends critically on the transformer, filter caps, rectifiers, yada, yada. But the RMS is on the order of 1.2-1.4 times the DC average current out of the filter cap to the amp. So the MOSFET power will be at most maybe 5V times 1.2 to 1.4Amps, and that kind of threshold-guessing is how I got to the 5-10W of dissipation.

There is a disaster-mode issue that occurred to me as I thought about this. If there is a shorted tube or cap on the power distribution line in the amp, the circuit will still clamp the current to the desired level, but it will now be holding off all the power supply current, as the limiting transistor will be regulating the gate voltage to keep the correct current limit. That means that the MOSFET will be handling a lot of watts. A thermal cutout here would keep this from happening and also as a side effect protect the amp against shorts. But that needs some thought.

In looking at the schematic, one of those button-style thermal cutouts could be arranged to open the gate drive to the MOSFET and have it shut current off entirely. Or the button could just open the main positive from the rectifiers.

[mikepukmel]

Ah this is what I was looking for, thanks to Rob Rob: https://robrobinette.com/Fender_Deluxe_Models.htm
Of course, I don't know about production, but the schematic for a1172 has the standby on the rectifier side, but A1270 its back to the spot The Good Lord intended that the standby switch be placed.

[mikepukmel]

Yes, I have. As with any power circuit, there are some caveats that go with it.
First - it's a power circuit, and it can get hot. You have to think about getting the heat out of the MOSFET and how much heat there will be.
In most cases, there will be minimal heating. In normal operation, the MOSFET is turned full on. The voltage across it is just the voltage needed to enhance the gate to pass that much current. This can be substantial for some MOSFETs, so you need to look at the device you buy and get a low threshold voltage MOSFET.

The first one I listed, the IRFBE20 has a threshold voltage of 2.0 to 4.0V. So the MOSFET will turn on as soon as the gate is 2 to 4 v positive compared to the source. This voltage comes through R1. In normal operation there is no current flowing through R1, so the gate is at the same voltage as the drain. When the drain (and gate) gets 2-4 V higher than the source, the gate turns on the channel and current runs through from drain to source. The spec on the IRFBE20 says it has a transconductance of 0.8S, which is 0.8 amperes per volt of gate-source voltage more than turn on voltage, so the voltage source will sag a volt or two below the drain as more current runs in. This will eat 4-6V from your B+, all the time.

The power spent in the MOSFET will be the time integral of the current and voltage across the MOSFET, which depends critically on the transformer, filter caps, rectifiers, yada, yada. But the RMS is on the order of 1.2-1.4 times the DC average current out of the filter cap to the amp. So the MOSFET power will be at most maybe 5V times 1.2 to 1.4Amps, and that kind of threshold-guessing is how I got to the 5-10W of dissipation.

There is a disaster-mode issue that occurred to me as I thought about this. If there is a shorted tube or cap on the power distribution line in the amp, the circuit will still clamp the current to the desired level, but it will now be holding off all the power supply current, as the limiting transistor will be regulating the gate voltage to keep the correct current limit. That means that the MOSFET will be handling a lot of watts. A thermal cutout here would keep this from happening and also as a side effect protect the amp against shorts. But that needs some thought.

In looking at the schematic, one of those button-style thermal cutouts could be arranged to open the gate drive to the MOSFET and have it shut current off entirely. Or the button could just open the main positive from the rectifiers.

Thanks RG. Is the thermal cutout something you could mount on the heat sink, like bolted on to the metal tab of the TO220 case?

[Leo_Gnardo]

Ah this is what I was looking for, thanks to Rob Rob: https://robrobinette.com/Fender_Deluxe_Models.htm
Of course, I don't know about production, but the schematic for a1172 has the standby on the rectifier side, but A1270 its back to the spot The Good Lord intended that the standby switch be placed.

Well there you go: Fender was not consistent in the way they wired up their amps. Small wonder...

Remember the old hindoo tale of six blind men and the elephant? Or was it seven, I forget. No matter, in this case it's two: I've seen it done this way. No no no, I've seen it done that way. Turns out both are correct.

[R.G.]

Thanks RG. Is the thermal cutout something you could mount on the heat sink, like bolted on to the metal tab of the TO220 case?

Yes. It's a roughly 0.5" diameter button that has bolt-down tabs.

Your questions made me go back and do some sim work.
- using a 4N25 opto isolator works for switching the MOSFET on and off. Happens cleanly, no funny stuff.
- I had to use two 100R resistors and a 100pF cap on the gate of the MOSFETs to make it stable for all versions of MOSFET. These are pretty cheap and the actual values don't matter much, as all they're doing is damping the gate-source's tendency to UHF oscillation. But this may have been a quirk of the simulator.
- Turn on power in the MOSFET is about 12W for 300mS. This is likely not to damage the MOSFET on any kind of heat sink.
- Steady state power on the MOSFET is around 2W with the clamp level set to 2A. A TO220 and especially a TO3P package would do this without a heat sink, but the sink should be there just to be sure.
- Shorting the output of the thing ramps power up to about 200W in the MOSFET. Bad ju-ju indeed.You can't heat sink any normal device to get that much heat out of it. So a thermal cutoff of some kind is needed.

The button cutout for sensing temperature would work. But there are now some 3/4/5/6/8 pin ICs in either SMD or through hole that act as thermostats. They sense their chip temperature and turn an output pin off or on when a threshold is hit, and stay that way til things cool down. The ones I looked at are CMOS and suitable for low voltages, so I might be able to rig one up to live on the board with the other parts and operate parasitically on the leakage from the high voltage. Still thinking about this.

[R.G.]

Mouser has tiny, flat bimetal cutouts for about $1 each that could be epoxied to the MOSFET and have wires to the cutoff circuit TBD.

Looks promising and comes in a variety of trip temperatures.

[R.G.]

Since there's so much interest, I'll update the article at geofex with the later thoughts, including a designed, but not tested, thermal cutout. The parts won't be here in time.

[mikepukmel]

Since there's so much interest, I'll update the article at geofex with the later thoughts, including a designed, but not tested, thermal cutout. The parts won't be here in time.

Thanks, very interested!

[mikepukmel]

So sounds like, at least, for those building a JTM45 'clone' from the original schematic, would be a good idea to move the first big filter cap to the rectifier side of the standby switch, and getting a little fancier, use Merlin's suggestion of putting a leak resistor across the standby switch to keep the tubes from getting damaged if long periods at standby are expected. I had not heard of "interface resistance" before:

Hoping Mr Merlin doesn't mind a quote from his web site:
"First of all, don't make the same mistake Vox did. If you have a valve rectifier then the standby switch must be placed after the reservoir capacitor so the cap can charge up slowly as the tube warms up. ... Leaving the valves totally cut-off, while heated, encourages interface resistance. A simple way to discourage this is to add a resistor in parallel with the switch to allow a trickle current to flow at all times, while still keeping the amp more-or-less muted. A 47k 2W device is a reasonable compromise. You can also add a 100nF (or so) capacitor across the switch to reduce arcing inside it."

[R.G.]

If I remember correctly, interface resistance is the more correct term for "sleeping sickness", one of the processes which can degrade the efficiency of cathodes. The tube literature is divided on how bad this is. It's a cumulative thing, where having the tube heated but not conducting slowly accumulates.

To evaluate what the issues are with .whole mess, you need to go back and dig through the concepts involved in standby. Why do we have standby on guitar amps? The best explanation in my mind is that it's a convenience to silence the amps during breaks without having a 15-30 second delay for warmup upon resuming. I think standby may have been justified as "saving tube lifetime" perhaps at some time. But guitar amps generally are not designed to be gentle to tubes in the first place. It would be interesting to know whether any guitar amp tube ever had to be replaced because of interface resistance. Fact is, interface resistance was diagnosed as a reason tubes died in vacuum tube computers, where every tube was guaranteed to be, on average, turned off but heated 50% of the time. The "sleeping" hours accumulated very quickly indeed. Hours of "standby" accumulate very slowly. Even an hour or two of standby every time an amp is used, which would be a lot, or forgetting to take it off standby overnight would not be all that many hours.

But still, if you're building an amp, you get to decide what you want the amp to do. It's one thing to use "exactly the same as the oriiginal 1959 Belchfire amp" as an advertising point; it's another to just have the amp do what you personally need. My personal use of amps has not seen a lot of need for standby. The very few times I needed it, I used it as a mute.

There are much simpler ways to mute an amp than turning off the high voltage. One is simply to ground the signal at some place in the signal chain. The phase inverter input is a good spot to do this. You may have to toss in a cap or two to keep DC off the switch to prevent pops on turning back on, but using this scheme means that your standby switch won't ever arc over, as it carries no high voltage or current.

If you insist on having a "power down" standby, put a 10k in series with the cathodes of the power tubes, and shunt this 10K with a power MOSFET (remember those? ) when you want the amp to make noise. This leaves the power supply with its soft power up characteristics if it has a tube rectifier, leaves the power tubes with a trickle of current during "sleep", and again makes the standby switch not a power handling component. Yes, you can use the standby switch instead of the MOSFET to do sleep (I think I'm going to call the condition of no output but powered hot sleep instead of standby), but using the MOSFET lets you do other things with the MOSFET.

With the MOSFET at ground, you can do things like sensing tube currents and turning off the MOSFET to save the power rectifier, power transformer, and/or output transfromer if a power tube shorts. Just a thought.

[pdf64]

So sounds like, at least, for those building a JTM45 'clone' from the original schematic, would be a good idea to move the first big filter cap to the rectifier side of the standby switch, and getting a little fancier, use Merlin's suggestion of putting a leak resistor across the standby switch to keep the tubes from getting damaged if long periods at standby are expected...

Yes, an exact clone may not be the best way forward.
I also suggest that grid stoppers (eg 5k6) are added for the power tube control grids.

...If you insist on having a "power down" standby, put a 10k in series with the cathodes of the power tubes, and shunt this 10K with a power MOSFET (remember those? ) when you want the amp to make noise. This leaves the power supply with its soft power up characteristics if it has a tube rectifier, leaves the power tubes with a trickle of current during "sleep", and again makes the standby switch not a power handling component. Yes, you can use the standby switch instead of the MOSFET to do sleep (I think I'm going to call the condition of no output but powered hot sleep instead of standby), but using the MOSFET lets you do other things with the MOSFET...

Yes, the 'xxk in series with the power tube cathodes' is now my preferred preferred standby arrangement, if standby there must be; though just with a regular switch to short it out in the 'operate' mode.

...With the MOSFET at ground, you can do things like sensing tube currents and turning off the MOSFET to save the power rectifier, power transformer, and/or output transfromer if a power tube shorts. Just a thought.

Unfortunately I don't think that will help with a common power tube failure mode, ie in which an internal tube short seems to send HT fault current to 0V common via the heater circuit. It seems to be quite common to encounter 100ohm heater balancing resistors fused, probably due to this.