Safety Resistor on output jacks

[angelodp]

I noticed on my Redplate that Henry has a cement R on the output jacks, just in case. Do you all do this, is it common?

[thetragichero]

traynor did this for the bass masters. results in a little less power output. seems folks are better at not running a head without a load attached these days

[Colossal]

That was a common safety mod within the AX84 crowd, some years back. 220R 25W wirewound resistor across the jack.

[TUBEDUDE]

220R seems awful high for even a 16R load. And at a high resistance 25W seems like major overkill.

[pdf64]

Consider that around its bass resonance freq, a 16ohm nominal speaker’s impedance will typically rise up above 160ohms.
Hence to avoid a safety resistor affecting tone / draw significant power away from the speaker, a value of around 25x the nominal impedance is typically used.
eg the 470ohm safety resistor used across the 16ohm output of 60s Vox AC50.
The key point is that the safety resistor is not intended to function as some kind of alternative load.
Its purpose is solely to maintain some reasonable degree of resistive path across the OT secondary, to eliminate the possibility of an open circuit load appearing on an operating amp. Even at idle, without one, just flipping an impedance tap switch can cause sufficient back emf to result in a spark in the primary circuit, eg pin3 anode to pin2 heater of a typical octal power valve.

[Colossal]

220R seems awful high for even a 16R load. And at a high resistance 25W seems like major overkill.

Yes, but as Pete explained, it was intended as a bleeder and meant to stay out of the way when not performing its protective duty. 220R were cheap*, a commonly available value, and the 25W wirewounds don't take up much real estate and can be bolted to the chassis (not to the jack). I said "some years back"....more like ten years now

* they used to be something like $1.35 and now can be had for $3-4 on Mouser for good quality wirewound/chassis mount, still not too bad for a little extra insurance against a no-load condition.

[R.G.]

[...]
Its purpose is solely to maintain some reasonable degree of resistive path across the OT secondary, to eliminate the possibility of an open circuit load appearing on an operating amp. Even at idle, without one, just flipping an impedance tap switch can cause sufficient back emf to result in a spark in the primary circuit, eg pin3 anode to pin2 heater of a typical octal power valve.

Good case for diode clamps on plates and MOV or TVS transient eaters across the primary. If normal operation is causing arcs ...

[pdf64]

[...]
Its purpose is solely to maintain some reasonable degree of resistive path across the OT secondary, to eliminate the possibility of an open circuit load appearing on an operating amp. Even at idle, without one, just flipping an impedance tap switch can cause sufficient back emf to result in a spark in the primary circuit, eg pin3 anode to pin2 heater of a typical octal power valve.

Good case for diode clamps on plates and MOV or TVS transient eaters across the primary. If normal operation is causing arcs ...

What advantage would those have over a safety resistor?

It’s just that even with my ruined hearing, I can switch a (25-30x nominal) safety resistor in / out and not detect any difference. Whereas that’s not the case for the typical ‘string of 3’ 1N4007 between power valve anodes and 0V, ie the overdriven tone seems a bit smoother with the diodes.

Bear in mind that many amps which have a NFB loop already have a certain degree of safety resistance already in place, provided it’s hardwired to the OT secondary nodes, ie not wired via any switches etc.

My guess is that a resistive load of up to around 1k may have some degree of benefit as safety resistor; whereas over 10k, perhaps not.

[TUBEDUDE]

220R seems awful high for even a 16R load. And at a high resistance 25W seems like major overkill.

Yes, but as Pete explained, it was intended as a bleeder and meant to stay out of the way when not performing its protective duty. 220R were cheap*, a commonly available value, and the 25W wirewounds don't take up much real estate and can be bolted to the chassis (not to the jack). I said "some years back"....more like ten years now

* they used to be something like $1.35 and now can be had for $3-4 on Mouser for good quality wirewound/chassis mount, still not too bad for a little extra insurance against a no-load condition.

I agree they are good prevention, and I use them, but between 50-82R and 20W depending. For the short time they will be in circuit I guess it doesn't really matter as long as some resistance is there.

[R.G.]

Good case for diode clamps on plates and MOV or TVS transient eaters across the primary. If normal operation is causing arcs ...

What advantage would those have over a safety resistor?

Whoa, down boy. I wasn't trying to say that clamps are better than safety resistors, and should be used instead of an emergency load resistor. There's room for both in the amp world, and using both would probably be a good idea.

I have my own ideas about the string of three 1N4007s. That approach relies on the OT's transformer action to catch spikes on the "on" output tube to ground. Leakage inductance cannot be caught by transformer action, so the spikes will still be there on the "off" tube of a push-pull, although with a lower total energy. That's why I prefer high voltage transient suppressors across the OT itself.

Disconnecting the load resistance can cause damage by two paths; one is spikes from sudden opening the other from runaway ultrasonic oscillation through the feedback loop. Both cause different failure syndromes. I only commented because switching the impedance selector was one that clearly exists, too, but that I hadn't thought about.

No slight intended.

It’s just that even with my ruined hearing, I can switch a (25-30x nominal) safety resistor in / out and not detect any difference. Whereas that’s not the case for the typical ‘string of 3’ 1N4007 between power valve anodes and 0V, ie the overdriven tone seems a bit smoother with the diodes.

Bear in mind that many amps which have a NFB loop already have a certain degree of safety resistance already in place, provided it’s hardwired to the OT secondary nodes, ie not wired via any switches etc.

My guess is that a resistive load of up to around 1k may have some degree of benefit as safety resistor; whereas over 10k, perhaps not.

I would go along with that in most ways. Amps with NFB are most prone to going into runaway oscillation when unloaded, for sure. Whether the resistance permanently attached to the secondary is enough to snub sudden unloading transients in all cases will of course depend on the cases - some circuits and OTs will be more sensitive than others. Some will have enough advantage from the NFB feedback resistors, others will need additional load.

[brewdude]

If one suspected that a safety resistor on the speaker jack might effect the sound or performance of the the amp, couldn't it be wired to the switched portion of a speaker jack such that it is only in the circuit if there is no speaker plugged into the output?

Hmmm, I suppose if the other end of the the speaker cable was not connected to a speaker it would be useless.

[R.G.]

You could, and I've seen amps wired that way. It would rein in most cases of the ultrasonic oscillation thing.

But every amp and design is different, just like every switch is. as well. If the amp is prone to making instant arcs when it's unloaded for a fraction of a second, like was described for what happens when an impedance selector is changed, even the fraction of a second between when the plug leaves the jack and before the switched section closes could cause an arc. And then there is always the issue of what happens when that switch contact gets coated in bar funk.

There is one taxonomy of switches that divides them in to break-before-make (bbm) and make-before-break (mbb). For switches without a lot of attention to make sure of one or the other, you usually get bbm, meaning there is a tiny fraction of a second while things are changing where there is no connection at all. Neither the position switched from nor the position switched to is connected. This is practically a prescription for transients. I've even seen this open-in-the-middle used to make very short pulses. It's funny to see a switch where two poles are wired together and the common grounded, while the common poles are expected to make a pulse. Tricky, that.

Make before break (MBB) are usually designed specifically so that before the current position opens, there is a shorting to the next position. Done properly, there is never an instant when the whole thing is open. If the impedance selector switched that pdf64 mentioned had been MBB, no transients would have happened. But they do cost more, and I've never seen a MBB jack.

The whole vein of thought about how to protect against failures is an exercise in imagining the microsecond mishaps that could happen sometime, and trying to outwit Mother Nature.

[brewdude]

Thanks.

[pdf64]

Good case for diode clamps on plates and MOV or TVS transient eaters across the primary. If normal operation is causing arcs ...

What advantage would those have over a safety resistor?

Whoa, down boy. I wasn't trying to say that clamps are better than safety resistors, and should be used instead of an emergency load resistor. There's room for both in the amp world, and using both would probably be a good idea...

Apologies , I perhaps worded that somewhat directly, but I didn't mean to convey irritation etc. I really was 'just askin'', and am grateful for your explanation and further thoughts on the matter.

How common is 'oscillation when unloaded' for amps with a NFB loop?
I'll get around to checking a few, but seem to remember Steve Conner over on MEF mentioning that the NFB typically acted to maintain stability in that scenario.
My JTM 45 builds oscillate unloaded, but they're probably something of a special case, as their NFB ratio is so high.

[R.G.]

No worries, mate. I'm just aware I may get a little hard-edged about my technical opinions. So I try to watch it.

The sensitivity to high frequency oscillation... varies.
Great... (I can hear you saying. )
Everything in the universe with a voltage difference to any other thing has a capacitance to the other things. Exactly how big the capacitance is, and how much signal can be transferred depends on the exposed area, the voltage difference, and the distance between them Hugely, the distance, as the capacitive energy transfer goes down by the square of the distance between the things. This is a long winded way of saying that component placement, wire position, etc, etc matter.

In general, there is always some amount of gain that will give enough phase-shifted feedback through some strange capacitive/inductive/conductive/power/ground path enough signal to make the amp oscillate. The trouble is knowing how much gain and what the attenuation of the feedback paths are. It's a truism that any amplifier has internal phase shifts. Unfortunately, in a tube amp with an OT, the OT itself has two out of the three time constants needed to allow oscillation, so one tube driving the transformer adds the third. More tubes in series add at least one time constant per stage. So any tube amp is primed and ready to do Nyquist/gain-phase oscillation if the gain it big enough and the feedback paths are thus-and-such.

So actually, I don't really know how common it is, as I don't do amp repairs and don't have the breadth of coverage that a working professional in the field does. I'm limited to knowing it can happen, and how.

One other thing that weighs in is that a mostly-current-source output like pentodes and power beam tubes have a voltage gain that includes load resistance as a term. So the higher the load resistance, the bigger the voltage gain. So feedback paths from the output tube plate wires to other things get more and more important as the reflected load impedance on the output tubes gets higher by unloading on the secondary. In the limit of an open circuit, the load on the output tubes gets to being just the primary inductance plus leakage inductance of the OT, and that goes up with frequency. It's a situation rife with possibilities.