Power amplifier thermal protection

[R.G.]

In thinking about how to set up the power switching PCB for my Thomas Vox Beatle resurrection, it occurred to me that having a thermal switch on the heat sinks shut down the whole mess was a Good Idea, and I thought of stringing the wires for the thermal switch from the power switch (on the preamp assembly in the Thomas Vox amps) to the power amp chassis, putting in two additional wire positions in the connectors, etc, and it struck me - if I put a triac and a zero-crossing switch out on the power amp chassis all by itself, it could be made to refuse to "eat"any AC provided for it if it was too hot.

This kicked up a dim memory of turning triacs on by connecting the gate to the MT2 through a resistor. Sure enough, it works.

So you can do an AC power switch by connecting/disconnecting about 10ma on a triac, controlling 10s of amperes. And if you make the switch that controls the gate current be a thermal switch, you can make the triac turn on/off with temperature without running the full current of your power amp through the thermal switch.

Three parts: thermal switch, resistor, triac; full thermal protection.

NOTE NOTE NOTE: I'm talking about AC power line wiring. If you don't already know how to do AC power wiring safely, don't try this.

[PRR]

Thermal switches were somewhat common around 1973.

The standard type was 5/8" diameter with mounting ears and good for 15 Amps. We just put it in the power line.

There is a smaller type found in Mr Coffee (where did Joe DiMaggio go?); these also have high power rating (they break a 1,500W heater).

I have not autopsied a lot of high-stress amps lately, but I think thermal switches are gone out of style. First: transistors are VERY tough now. Second: SOA specs usually suggest many more devices than dissipation specs, prices are low. It has perhaps become difficult to over-cook transistors now.

> Thomas Vox Beatle resurrection

If you are cloning a bad-old antique design, then yeah you face all the problems which made early tranny amps less reliable than good tube amps. I bet many thermal cutouts were blind-guess "fixes" for SOA failures which were not well understood.

[R.G.]

Thermal switches were somewhat common around 1973.

The standard type was 5/8" diameter with mounting ears and good for 15 Amps. We just put it in the power line.

The venerable Klixon and its clones. What I disliked about that arrangement was that while electrically the thermal switch was in the power line, physically the power line was in the middle of the amplifier, with all the safety and hum issues that implies.

Not that this will be a whole lot better. At least the only thing that goes to the heat sink will be the two wires for the gate current, not the full current for the power supply.

I have not autopsied a lot of high-stress amps lately, but I think thermal switches are gone out of style. First: transistors are VERY tough now. Second: SOA specs usually suggest many more devices than dissipation specs, prices are low. It has perhaps become difficult to over-cook transistors now.

Yep. It could also be that the people are better. Second breakdown was either unknown or the subject of patents and doctoral theses in the mid 60s. And there is now another fifty years of accumulated knowledge on heat transfer and why it does all those horrible things to you.

If you are cloning a bad-old antique design, then yeah you face all the problems which made early tranny amps less reliable than good tube amps. I bet many thermal cutouts were blind-guess "fixes" for SOA failures which were not well understood.

Perhaps especially the Thomas Vox Beatle. They did not have transistors that would take the abuse. They used 2N3055s in parallel, but the 3055 just doesn't have enough SOA to survive reliably in the conditions, primarily in terms of the high BVceo and the associated structures that increase SOA. The answer to that one is to get a bigger horse.

The Thomas Beatle had other issues, too. They put the heat sinks on the most interior, airflow-blocked part of the enclosure possible, then tried to cope by putting small slit-style air vents. In effect, the compounded their SOA problems with real thermal problems. I suppose that it was like sky-high stilleto heels, a sacrifice in function for the sake of good looks, but it does cause thermal issues.

I originally was using a 12F508 reading a thermal sensor and controlling the Triac with an opto. Then I found a batch of surplus thermal switches. 

[PRR]

> the 3055 just doesn't have enough SOA

You know that there are ORIGINAL 2N3055, _thick_ ("through") diffusions baked 22 hours; and modern "2N3055" with typical thin surface diffusions?

The deep-fried devices were pretty tough. Of course designers pushed until field-failure rate rose. By the time it got to be an epidemic, RCA was ramping-down the deep-fry process, most "2N3055" were less-stern stuff and failed much quicker than the original.

"...the wafers often became stuck despite the alumina coating, and had to be soaked in hydrofluoric acid for possibly days to separate them.  Manual intervention— using a razor blade pushed between the wafers—was sometimes required."
"...at 4.6 4.6 mm the chip area is large. There may not even have been a dozen chips on a 3/4-in wafer"
"While one-inch wafers were about 170 m thick, an 8-in wafer is 725 m thick. To use the same manufacturing method would have required diffusions of weeks, which would have warped the larger wafers. The technology had little or no prospect of reducing costs by using bigger wafers. Competition from epitaxial base technology became an unwinnable battle."

They are fairly different devices in ways that could affect "sound":


IEEE Transactions, Nov 2001, 'The 2N3055: A Case History'

[R.G.]

You know that there are ORIGINAL 2N3055, _thick_ ("through") diffusions baked 22 hours; and modern "2N3055" with typical thin surface diffusions?

Yep.

Unfortunately, neither version had enough SOA for how Thomas used them. At least, not enough for high reliability.

[merlinb]

The standard type was 5/8" diameter with mounting ears and good for 15 Amps. We just put it in the power line.

but I think thermal switches are gone out of style.

The hifi company I worked for used (and still use) thermal switches just like that, attached to the heat sinks. They operated a mains power relay rather that interrupting the mains directly.

[R.G.]

The hifi company I worked for used (and still use) thermal switches just like that, attached to the heat sinks. They operated a mains power relay rather that interrupting the mains directly.

If they were high power amps - like, over about 300VA at the rating plate - I'm not surprised. The Klixon and its ilk were rated for several amps, but if you wanted it to be reliable, you'd downrate it. A big amp, possibly with a toroid power transformer like many hifi setups use, can easily exceed the surge rating for a bimetallic switch. And if there's a power relay already, like many hifi amps use, why not open it?

The thing that struck me about this approach was the economy; one thermal switch, a cheap-ish triac (or preferably alternistor because of the inductive nature of the power transformer) and one 1/4W resistor, and it can be put on the power chassis with the power transformer and heat sinks, no control leads and so on.

What it doesn't give you is annunciation and other "be polite" aspects. The uC approach lets you light up an indicator showing what happened. You pay for that privilege with having to power the uC, put in a circuit board for it, all the other overhead.

[amptramp]

Modern microwave ovens have a number of Klixon-clone bimetallic switches in them - I count three in one I just repaired last week.  The problem?  The Klixon had corroded and the contacts were loose on the spade terminals.  Sandpaper and a file and crushing the spade terminals with pliers solved the problem.  They have to interrupt power to a 1000 watt magnetron, so the current is about 13 amps on the 120 VAC inputs.  A bit better than those rated at several amps you are familiar with.

[wavley]

I actually have a friend that called me this weekend because his amp is having problems and I believe that it's going into thermal shut down.  He is supposed to send me his copy of the schematic, if this is indeed the source of his problem then I would be happy to send you a copy and maybe the circuit will be of some help to you.