Switching speaker cabs with a relay

[trixdropd]

If I wanna take a 100 watt tube amp and use a relay to switch between 2 speaker cabs, how big of a relay do I need? Any relay you can recommend?

[Unlikekurt]

the more important question is "should i do this?"
You will no doubt find conflicting responses.  but with a tube head, it might not be the best idea to have the OT no-loaded, even if for only that moment while switching occurs.

[PRR]

100 Watts at 8 Ohms is 28.28 Volts and 3.535 Amps.

Headlight relay is probably ample.

Agree about never un-loading a tube amp passing signal. There is a trick.

[Transmogrifox]

the more important question is "should i do this?"
You will no doubt find conflicting responses.  but with a tube head, it might not be the best idea to have the OT no-loaded, even if for only that moment while switching occurs.

I am one of the conflicting responses

The technical reason for not having an output transformer unloaded is because it's a freakin' huge inductor.  A discontinuous current results in a very high voltage which can arc across relay contacts, putting stress on your relay, potentially welding the relay contacts to one side or the other.

Some may argue potential damage to the output tubes -- I am doubtful this is likely.  This high voltage and dis continuous change will be reflected through the OT, but it won't be much worse than when no speakers are connected.  The OT is designed to be more of an ideal transformer and not an energy storage tank, so the inductive kick won't be out-of-bounds for the output tubes.

The output transformer won't likely be permanently damaged by such a short-duration low energy spike (low energy in comparison to the energy dissipated in the process of delivering power to speakers).

In the end, I think this is actually pretty benign, especially since it is likely to occur when no AC power is being delivered to the amp (usually you switch during silence).

That said, I fall among those who disagree there is much risk in doing this --even when amp is passing signal.  It might weld relay contacts at worst, but that is not catastrophically destructive to your amp.

28 volts at 3.5 amps as PRR states -- peanuts.  I work in an industry where we test thousands of cycles on relays passing 30 amps at 250 volts.  We use a hybrid switch (semiconductor parallel with relay contacts) but that is only because it's 30 Amps...5 amps, just put a 50 Volt bipolar clamp in parallel and you're all good.

I hope this adds another angle to let you know this isn't magic.  It's just an argument about something that might happen, but probably none of us have engineered this to the level of diligence where we can say there's any guarantee.   

The careful side says, " This is a bad idea".

The careless side (me) says "This is no big deal, the electric utility industry depends on switching 40 mH inductive loads on a 50A rated relay at 30 Amps...what is 100 Watts at 3.5A on a similar relay?

Just get a relay rated at something more than 250V, 5A, use a protective TVS on each pole of the relay (say >50V), and hail Mary.  It won't blow up your amp.  

[PRR]

> won't be much worse than when no speakers are connected.

That's the problem.

If current flows in an inductor, and is then cut-off (tube clips), ideally the voltage will go to *infinity*.

In iron-core coils the Q is rarely better than 10, and the peak voltage is no more than 10 times the DC supply.

A 100W tube amp may have a 500V supply. 10 times 500V is 5,000 Volts.

The paper classically used to insulate OTs is good for 1,000V for short time. This is why many chokes carry a 500V rating (coil to core). They may use 2 layers in hi-power tube OTs, but that's just 2,000V. They can't use many layers because we need room for copper. They now use plastic, with higher V/mil, but in thinner layers, so the voltage rating is not a lot higher.

I believe it has been proven many times that HEAVY (American Woman) overdrive with no load will kill OTs in years months or even a day.

Insulation breakdown depends on time. Whether real harm can happen in a few hundred 10-mSec switchings is an open question.

However a breakdown in a 100W tube amp OT is something to cry about, or not risk in the first place.

[greaser_au]

I've had to replace a few output  transformers over time (for friends) because of them being run without load for a short while.

At one point it was cheaper to unstack, take them to me local transformer place to get rewound,  and restack, but not any more...

davidt

[Transmogrifox]

Insulation breakdown depends on time. Whether real harm can happen in a few hundred 10-mSec switchings is an open question.

This is the real point of my response.  I don't think it's a big deal, but as PRR stated, it is an open question.  We're just stating conflicting opinions, while I concede there are real consequences to letting it fly unloaded.

In the end, I still don't discourage you from making a speaker cabinet switcher.  There is a way to mitigate the concerns expressed in this thread:

Put together a couple of relays and form a guaranteed make-before-break scheme so the result is 2 speakers connected in parallel for that <10ms while the contacts are flying.  The general idea is to ensure both relays are engaged during the transition.  If you need help probably any of us geeks who like a diversion from work can spin up a circuit to implement a dual-relay make-before-break scheme.

[R.G.]

I'm fully in agreement. It is by no means a certainty, but some tube amps go flat-out nutzo when unloaded. This can - but sometimes doesn't - fry the output transformer.  It's likely enough that it ought to be avoided unless you like taking the amp to your amp tech for new transformers every so often.

The technical reason for not having an output transformer unloaded is because it's a freakin' huge inductor.  A discontinuous current results in a very high voltage which can arc across relay contacts, putting stress on your relay, potentially welding the relay contacts to one side or the other.

Some may argue potential damage to the output tubes -- I am doubtful this is likely.  This high voltage and dis continuous change will be reflected through the OT, but it won't be much worse than when no speakers are connected.  The OT is designed to be more of an ideal transformer and not an energy storage tank, so the inductive kick won't be out-of-bounds for the output tubes.

The output transformer won't likely be permanently damaged by such a short-duration low energy spike (low energy in comparison to the energy dissipated in the process of delivering power to speakers).

In the end, I think this is actually pretty benign, especially since it is likely to occur when no AC power is being delivered to the amp (usually you switch during silence).

That said, I fall among those who disagree there is much risk in doing this --even when amp is passing signal.  It might weld relay contacts at worst, but that is not catastrophically destructive to your amp.

So far, so good, but the real issues have to do with the OT insulation, not any relays. Transformers in general die when their insulation fails. The iron and copper is pretty sturdy, but the insulating layers are not. The danger in generating inductive kickback voltages with OTs is that it can damage the insulation, either by punching through it or by establishing carbonized tracks between places.

One instance of this is not fatal, but the insulation is permanently damaged at the punch-through spots. This generates a slew of (generally) carbon-containing decomposition products in the small area of the punchthrough and an open spot in the insulation. And it's almost never only one instance of high voltages. In the intended use, the player will switch back and forth at will, so in a design for this, we have to assume that it will happen many times, perhaps many times a night.

But a single punch through isn't the worst. Some tube amps go into frenzied oscillation when their output transformer is opened. This can and sometimes does result in a blast of high frequency oscillation at any unintended resonances of the OT and stray parasitics. Amps that do not use feedback from the speaker side of the OT are almost immune to it. Amps that use feedback are more prone to it, but it's not an every-amp thing. So it's easy to find counter-examples of "well, I opened up my speaker output and my amp lived through it fine" testimony.  But it's also easy to get the word of techs who replace OTs after their customers left the speakers unplugged during long sound checks, too.

The effects are worst if the OT is hot. Heating lowers insulation resistance to punch through, and any punch through is made worse by heating and moving around those carbon-based artifacts of earlier punch-throughs. Once you get carbonized trails between defects, you can start getting localized heating from currents through the trails.

So IMHO, one open-output incident isn't fatal - but it's not a good thing to do, and the next time *might* be the one that sets off an OT death spiral. Again, ask an amp repair tech.

I hope this adds another angle to let you know this isn't magic.  It's just an argument about something that might happen, but probably none of us have engineered this to the level of diligence where we can say there's any guarantee.

The careful side says, " This is a bad idea".  

I agree there. There is an exposure.

The careless side (me) says "This is no big deal, the electric utility industry depends on switching 40 mH inductive loads on a 50A rated relay at 30 Amps...what is 100 Watts at 3.5A on a similar relay?

Just get a relay rated at something more than 250V, 5A, use a protective TVS on each pole of the relay (say >50V), and hail Mary.  It won't blow up your amp.  

That's absolutely true if the question is about damage on the secondary side. The problem is that transformers are a big tank of magnetic field driven by active devices. The secondary side of an OT is low voltage/high current by design, and is on the outgoing side of the tank. The primary is high voltage and low current, and is driven by devices that can go into oscillation, and is isolated from the transformer's M-field by leakage inductances which can both store M-field, and which prevent the active devices from having high frequency control of the M-field. The danger from open secondaries is from high voltage punch through by a couple of different syndromes, and the energy involved in a punch through doesn't have to be very big to start the chain of failures. It just has to make one hole at a weak spot. The next few thousand stresses enlarge the hole.

It is FAR safer in switching speakers to tube amps to SHORT the speaker outputs, or even better to parallel up a dummy resistor load before the switch happens. This completely avoids the issue of starting a punch through or making it worse. Tube amps will be fine when shorted for short periods. Solid state amps may go nuts, so using a dummy resistor is the best design choice because someone, someday will hook up the speaker switcher to solid state amps.

So a speaker switcher ought to (1) add on dummy resistors to the amp output, (2) make the switch of the hot amp output to the new speaker load and finally (3) release the dummy load on the amp. It takes some timing and some careful reading about relay switching times, but as you noted, you can speed things up a bit by using an SSR to connect up a dummy load resistor in the interval until the mechanical stuff has time to move.

[wavley]

Something that can help with the unloaded transformer problem is a 180 ohm 20 watt resistor like Traynor used to put on it's heads (Pete starting out as a rental repair guy was tired of fixing amps so he made bulletproof as possible amps to rent)  Of course this is only for the occasional "Shoot I forgot to plug in the speaker" moments, not repeated switching so something like an 100 watt load resistor might be a better choice for this situation, of course like Pete I'm a really big fan of overbuilding things because I have enough stuff in my life to repair.

What you want can totally be done if you just take care to do it right.

[amptramp]

This may be a good place to use a TrensZorb element which is a zener diode (back to back internal connection for AC) which limits the overhoot but does nothing until the voltage gets high enough.  That way, you don't have a continuous load from a resistor.  The capacitance may be an issue but antiparallel diodes in series would keep that to a negligible level.

There are also some make-before-break relays that could solve the problem with a momentary state of half the regular impedance.  Not good for solid state but not a problem for a tube amp.  There used to be a huge number of relay variations but these have disappeared over the years.  You might be able to bend a relay contact into a make-before-break on a standard relay.

[R.G.]

In the Workhorse amplifiers, I used both a "safety load" and a MOV on the primaries for transient suppression.

I once accidentally left a prototype on and unloaded overnight. No problems. Of course, they did not use feedback, so it may not have oscillated.

The stack of reverse biased diodes from ground to each end of the OT primary is there to catch an over-voltage transient. It's not all that effective, because it has to protect the opposite side by transformer actions, so the leakage inductances still can do punchthrough. A MOV or transzorb is better.

[PRR]

> a couple of relays and form a guaranteed make-before-break scheme so the result is 2 speakers connected in parallel for that <10ms while the contacts are flying.

Lot of parts to do that with relay-logic and caps.

Some will say "computers are cheap". But my expensive hard-wire smoke-alarms just got recalled because of a computer bug. (If 110V power is lost the instant that self-test starts, they go dead forever!) Who of us is sure there is NO bug in our code?

SPDT relay. While contacts are flying there is still a 2Z load on the amp.



Yes, there is a "bug" that if user does not connect both speakers, it goes open while-flying and in the no-load position.

Also the "upper" output can have full signal voltage on the jack shell. Which is a reason to stop mis-using cheap approximations of telco plugs and use proper no-live-parts plugs (but that battle goes slowly).

[karbomusic]

Fried a vintage Marshall head OT in less than 60 seconds about 15 years ago not realizing the cab wasn't hooked up FYI. 

[Transmogrifox]

> a couple of relays and form a guaranteed make-before-break scheme so the result is 2 speakers connected in parallel for that <10ms while the contacts are flying.

Lot of parts to do that with relay-logic and caps.

^^Very true

SPDT relay. While contacts are flying there is still a 2Z load on the amp.

Brilliant .  I love to see things that are so dead simple it's a no-brainer.

All the same, here is my attempt which leans more to the complicated side of things, but it does provide the Z/2 during switching instead of the 2Z.  It relies on the Carling 2-pole momentary pushbutton switch.


Here is a Spice model of the 1-shot that pops the latching relay over.  The timing is critical to avoid making it chatter (this is an oscillator if it had power to coils for more time than needed to trip it only once).


And here is the simulated response of the 1-shot.  The Blue trace is the battery voltage, emulating the turn on of a switch.  The green trace is the current through the latching relay coil.


As can be seen it doesn't use battery very quickly because the battery is only turned on to trip the latching relay.

Thanks a bunch to RG for a long response.  I will admit I didn't even consider transformer insulation as being the primary concern here, but makes much sense when pointed out.