speaker jack shorting or not?

[Bluefinger]

I have seen both versions. It it better to make the jack shorting or not? Of course as long as the speaker cable is inserted it doesn't make a difference and it should be plugged in all the time. It's just in case someone forgets to plug in the speaker ... is the risk of damage smaller with the output wires shortend or open?

thanks!

[R.G.]

Tube amps and solid state amps have opposite reactions to load impedances.

In simple designs, solid state amps die almost instantly from shorted outputs. It's so fast and total that almost all solid state amps have designed-in protection from shorts that in turn, they're essentially immune to any load resistance from an open to a short.

Tube output stages can die from high voltage spikes generated when too high a resistance is connected to their output. In simple tube designs, they're almost immune to shorted outputs, but can die from open circuits, or any load resistance above some amount. The exact amount is hard to determine, but open circuit is generally bad for them.

I think the best thing might be to put only NON-shorting output jacks on solid state amps. Tube amps might use dead-short or resistance-connecting jacks. Both would be better than non-shorting jacks for a tube amp.

The problem with speaker cables is that external forces can either short the output cable by squashing it and cutting the insulation, or cut it open entirely, depending on Mother Nature's mood at the moment, and how Her assistant Murphy feels about it.

[Bluefinger]

Thanks, that makes sense!

[Bruce / Mission Amps]

I have seen both versions. It it better to make the jack shorting or not? Of course as long as the speaker cable is inserted it doesn't make a difference and it should be plugged in all the time. It's just in case someone forgets to plug in the speaker ... is the risk of damage smaller with the output wires shortend or open?

thanks!

Solder a 360 ohm (if an 8 ohm winding) to 470 ohm (if a 16 ohm winding) 5 watt power resistor across the highest impedance secondary to chassis ground. Then forget about it.

[Dave H]

On a tube amp I’d leave the jack switch open but have a 220-470ohm resistor wired permanently across the OT secondary (not through the switch). There’s always the possibility that the speaker cable could become disconnected at the speaker cab but still be plugged into the amp. A shorting jack on the amp won’t offer any protection in that case.

[Dave H]

Bruce must have posted as I was typing. All my amps have a 470ohm resistor across the 16ohm tap.

[trem]

Solder a 360 ohm (if an 8 ohm winding) to 470 ohm (if a 16 ohm winding) 5 watt power resistor across the highest impedance secondary to chassis ground. Then forget about it.

If I can just piggy-back onto this question.....
I rebuilt a BF/SF Bassman for a friend. His OT was bad. I installed a 4-8-16 Ohm OT from Mag Comp.
At present there is no "protection" from an open circuit.
So, with that OT, I should run a 360 Ohm, 5 watt resistor from the "tip" of the 16 Ohm speaker jack to ground.?
Thank You

[Bluefinger]

ok, thanks. that's easy, I'm going to do that.

[jpfamps]

Bruce must have posted as I was typing. All my amps have a 470ohm resistor across the 16ohm tap.

I do this too.

[R.G.]

I was just thinking about this general topic. That's always dangerous.

I did some reading about tube amp outputs, open circuit loads, OTs and such. I did not find a specific cause of death for tube amp outputs open circuited in a way that was definitive. I speculate that output death in tube amps with open circuit loads comes in one of two ways, sudden opening of the output, and oscillation of the output stage.

I *know* that simply opening the output side of an OT in an amp without feedback is not uniformly deadly or even something to write home about. I know this because I did this accidentally on one of the prototypes for the Workhorse amps for about an hour a couple of times. There was no change in the amp at all, no bad effects whatsoever.

I speculate that there are two causes of death for an unloaded tube amp: high voltage spikes and over-dissipation of the output tubes. The high voltage spikes would be a consequence of the leakage inductance and perhaps primary inductance of the OT primary having its current suddenly interrupted or reversed, and the overdissipation would be caused by RF oscillation in the power tubes, with perhaps similar high voltage effects in the leakage inductance. The actual death syndrome for spikes would be arcing of the power tube, tube socket, or internal insulation of the OT itself. Small, low energy spikes would arc and cause minimal damage, but repeated arcing would reinforce the damage until it was so bad that normal operation would be impossible. Symptoms would be short tube life (from internal arcing), arc trails on the socket, and OT performance between "ugly sound" and death.

Leakage inductance is by definition inductance that does not and cannot couple to the secondary, so the mechanism of it killing power devices is well known in power-conversion circles. Switching power supplies in particular have to use snubbers and other spike-catching mechanisms to prevent sudden-death to power devices attached to magnetics. But leakage inductance is relatively independent of the secondary, since by definition it's not coupled to the secondary, so secondary open or shorted should not matter.

But if the output tubes themselves try to turn off the current into the OT too fast, both the leakage and primary inductances go all V-equals-L-dee-I-dee-T on it and make big spikes, with punctures, sickness and death resulting.

The primary of a push-pull is wound with two different windings for the "primary". Actually, there are two primaries, one for each tube, and these have leakage inductance between them, flux that is not coupled between the two half-primaries, as well as flux that is not coupled between each half-primary and the secondaries. Each of these leakages may be and usually are different values unless the winder is either lucky or extremely skilled *and* lucky.

In the normal course of operation, a power tube pulling its half-primary voltage down towards ground makes the other half-primary go up above B+ by transformer action. The ends of the OT always go in opposite directions from B+.

This is the origin of the protection circuit which puts 3000V worth of diodes from ground to the ends of the OT. The diodes are there, not to break over at 3000V on positive voltage spikes, but to clamp the negative-going half to between ground and B+. This forces the opposite side to not go above two times B+ by transformer action. The high voltage breakdown on the diodes is to keep the high-side diodes from being killed. It's the low-side diodes that do the protecting.

That's all well and good, but transformer action cannot clamp leakage inductance, by definition. These diodes protect against sudden death from spikes from interrupting the primary inductance, but can't help against leakage inductance spikes. This is valuable, as the primary inductance may be 10,000 to 100,000 or more times the leakage inductance in OTs, but the smaller E = (1/2)L*I*I spikes from the leakages, which can add up over time.

A capacitor between the output tube plates, or a capacitor and resistor between the output tube plates can act like a snubber for both primary and leakage spikes. The cap converts the spikes into part-sine wave rings (which they are anyway from the internal capacitance of the primary; but slower with the external cap) that cut down the speed of the voltage change, and hence the peak voltage of spikes. A cap+resistor actually damps the ringing down faster as well as cutting the peak. This is good for saving transformers, but may lower frequency response and possibly upset Nyquist stability.

My favored solution is a string of one or more MOVs from primary to primary. These have some capacitance when open, but suddenly snap over to conduction at their protection voltage and not only keep spikes from going much higher, but eat the energy from the spikes. This seemed to work well in the amps I've tried it on.

RF oscillation is a special case. I think this primarily (... sorry... ) happens in power stages with feedback. Feedback around a transformer is always chancy from a stability standpoint. The usual feedback from secondary to PI means that the forward gain of the output stage must be limited to keep it stable, and so to get benefits from feedback, the output stage is always near the edge of Nyquist stability, at least for a design where this has been thought about.

The gain of a power tube can be thought of as its transconductance times the load on the plate. The plate load is normally the leakage inductance in series with the parallel combination of the primary inductance (which is as big as it can be made within the economics of the design) and the secondary reflected load. When the reflected secondary load gets huge, as it does when the secondary is open circuited, the plate load seen by the output tube becomes just the primary inductance. This is not only much larger than the normal load, but is essentially an undamped inductor/capacitor load.

The gain rises hugely at the resonance of the unloaded primary winding. The phase shift is very different on each side of the resonance, and the feedback connection to the secondary is not unloaded, so the secondary gets feedback unshunted by an external load. It is very likely that this sets up a runaway oscillation at or near one of the resonances of the primary winding. L-C resonances can build to very high voltages, limited mostly by any internal resistances (that is, the wire resistance of the winding). These high voltages can easily build up until the catch diodes or MOVs limit them. A primary-side capacitor doesn't help, it just changes the frequency of resonance, unless it moves the resonance down to a frequency where the loop gain is below unity. A primary side cap-resistor may well help, as it adds loading and a lower Q at high frequencies.

In all of these cases, the death mechanism is probably high voltages on the plates of the tubes. It can kill the tubes, arc over the sockets, or punch through the insulation of the OT.

Putting a resistor at the output jack may help or may not. It depends on the value of the resistor. Since this resistor is always connected to the feedback point as a load, it both damps the primary resonances through transformer action, and lowers the load seen by the output tube plates, which lowers the forward gain when there is no speaker attached, and may keep the amp out of RF oscillation.

I suspect that a parallel resistance may work GREAT in some situations, but not in some others. The "others" will be amps that are already near a cliff of stability, and any change is enough to let them run away. But it sure can't hurt. The only down side to this would be the loss of power to the resistor and the heat given off by the resistor in normal operation.

I wonder if a series capacitor+resistor on the output jack might be juggled to load the transformer at above-audio (or above-guitar!) frequencies and avert oscillation-death, much as zobels protect solid state amps.

[trem]

I was just thinking about this general topic. That's always dangerous.

I did some reading about tube amp outputs, open circuit loads, OTs and such. I did not find a specific cause of death for tube amp outputs open circuited in a way that was definitive.

OK, I am glad YOU wrote this.
Will be the first to admit that some of it is beyond my understanding, but I get the gist.
Anyway.....
back to your word DEFINITIVE.
I think we have all read claims of an OT burning down "because the bass player tripped over my speaker cord". And how not even the heroic efforts of the guitar player, to shut his amp down in 10 seconds, was enough courage to save the OT of the amp.
I have never personally seen that happen. I have played amps with either one side of the speaker cable disconnected or the other (which would create either an Open or a Short, correct me if I am wrong) and have never seen any catastrophic failures. I am not saying there was no, subtle internal damage to some part of the winding(s). But the amps all seemed to go on living.
Is there a Real Answer.?
What stars really have to align to kill an OT.?
On a side note, I recently repaired a BF Bassman and a Tweed Deluxe that had a bad OT.
There was a newish Heyboer in the Bassman and newish Merc Mag in the Tweed. Both amps were bought used, and the cause of OT abuse is not known. So I am aware that it does happen.
Thank You

[mooreamps]

Just do it the way Bruce does it....He knows what he is doing....

-g

[Dave H]

I think we have all read claims of an OT burning down "because the bass player tripped over my speaker cord". And how not even the heroic efforts of the guitar player, to shut his amp down in 10 seconds, was enough courage to save the OT of the amp.

I doubt it would burn out in 10 seconds if the guitar player stopped playing immediately. I once had an amp on for an hour without a guitar or speaker connected and it was fine. Here'a a simulation of an over driven amp (single ended for max effect) into 16ohm, 470ohm and open circuit loads. I expect something would break down before it got to 45kV!


VJ 16ohms.pdf
VJ 470ohms.pdf
VJ open.pdf

[jpfamps]

Another excellent and thoughtful post (as ever) from R.G.

A few of my own observations on the subject.

Although this thread is about OTs, it's worth noting that the screen grid current increases significantly with increased loading, so this is another good reason for not running your amp without a load.

A perusal of the load lines for any pentode clearly demonstrates the effect of increasing the loading dramatically, as a horizontal load line, ie infinite impedance, would result in infinite positive voltage swing with very little signal voltage. I assume this would encourage oscillation as only a very small signal would need to be required to swing a very large voltage out. Presumably this would not be Nyquist instability, but caused by some form of parasitic oscillation (assuming of course you are employing negative feedback).

Open loads was certainly a know issue in the 30s; indeed one of the advantages of the "ultra-linear" output configuration that Alan Blumlein mentions in his 1938 patent, is that greater signal voltage swing is required to generate such high voltages with very high loading (again easy to see from the load lines), so an ultra-linear output stage should immune to open loads.

My guess is that wire insulation in the 30s was not as good as it is now, so transformer damage due to no loading may have been a more common occurrence than today.

It's also occurred to me that the "conjunctive filter" (ie Zobel network) described in the RCA Receiving Tube Manual should provide some protection for the transformer from an open load.

I see quite a few amps with damaged OTs.

When I have the inclination to investigate the nature of the failure (by dissembling the transformer), there is usually some evidence of arcing, which implies an insulation breakdown.

The other cause of transformer failure is see is a broken wire, which I assume either due to poor construction or mechanical damage.

In amps known to run with open loads (customers can be quite coy regarding the events preceding amplifier failure), I see both damaged OTs, and valve with evidence of external arcing.

I put a 470 ohm resistor across the 16 ohm tap of the amp I build as a precaution again an open load (although of course the load is much higher than is "should" be). The VOX AC50 also does this too; this amp was designed in an era when valves were the prevalent technology.

Regardless, it's cheap and makes me feel good......

Of course I've accidentally run amps without a load, and no harm has resulted, so I would have to conclude that whilst running and amp without a load is a bad idea, it doesn't inevitably lead to destruction of the OT (or indeed valves), although have seen examples of both these failure modes in amps.

[Enzo]

It is important to note that getting away with something does not mean it is a good idea. These potential failure mechanisms for transformers are just that - potential. A transformer can punch through and arc in an instant. It is not a sort of failure like slowly warming up until it melts. Any particular unloaded peak MIGHT damage the transformer. If it does not, great, you dodged a bullet. 10 minutes? Not a factor, other than the amp has 10 minutes worth of opportunities to fail the transformer. There is no one-to-one do this and your transformer burns out every time.

SO if an amp is working fine over time, then we forget to plug in a speaker one time and POOF the tranny dies, it is not unreasonable to think them related. On the other hand, "I did it before and nothing bad happened" is no guarantee.

MY analogy is driving home from the bar while drunk. Many people do that every night of their lives, and do not kill themselves. That does not mean it is a good idea or that there is no risk because "I done it a million times." It will be the one time all the factors add up against that tragedy happens. It won;t be because the drive was 10 minutes instead of 12.