Originally Posted by Hyperion
I'd just like to make it clear that a Thyristor is not the same as a Triac, and only the Triac is suitable for this application.
Technically speaking, for the non-engineer observers following this thread, a thyristor is any semiconductor switch exhibiting bistable action dependent upon p-n-p-n regenerative feedback. Thyristors include a variety of devices: SCRS, GTOs, TRIACs, LASCRs, etc. Hence, a triac is a thyristor, but a thyristor is not a triac.
Now, while a triac is a very good choice, it is not the only thyristor suitable for this application. Paired, back-to-back, SCRs would work quite well, but require different gating circuitry. There are other thyristors that could be designed in as well. just know that whatever device you design in, you will require enough gating current to accommodate the the necessary di/dt (rate of change of current flow through the device over time) that occurs upon initial power-up of the amp (charging the big power supply caps), or you will eventually (possibly a significant period of time) punch through the thyristor through accumulated burned paths. If the gating current is too low, you will not get conduction paths, as the thyristor is turning on, of sufficient area to handle the inrush currents. Higher gate current results in more rapid expansion of conduction paths through the device as it is turning on each cycle.
Another competing parameter would be the on-state current necessary to ensure that the thyristor latches early enough in each 1/2 cycle for full cycle conduction, or you will get buzzing and noise. If you have insufficient current through the device, in its on state, for the device to latch, or stay latched, it will drop out, and the gate will drive it back on, and on and on, until good latching takes place, resulting in noisy hash at the beginning of each cycle of conduction (and possibly at the end of each cycle).
Devices that can tolerate higher di/dt, such as that experienced during amp power supply cap charging, generally require higher gating current and latching current for reliable full-cycle turn on and conduction. If you are taking percentages of the voltage you are switching, you must also consider what kind of gating current will be available at peak voltage, as well as at the desired start of conduction per half-cycle. Higher gate current at the start, may result in excessive available current at peaks, and early demise of the device.
When powering an audio amp via thyristor switches, you MUST consider the fact that there is a very wide range of currents drawn by the amp, from inrush currents of 10s to 100s of amps at initial power-up, to milli-amps at very low output levels. It is much easier to design thyristor switch circuits for constant/predictable loads like lighting systems or industrial controls. Depending on the amp model, it might require quite a delicate balancing act to design a suitable switch, and impossible to design one that will work for most or all amp models.
For reliable control of power thyristors, where low "control" currents or voltages are available, you might consider an opto-triac (an opto-coupler with a small triac as the output device).
Also know that if you want to power an audio amp via thyristor switch, you are feeding a transformer, which is significantly inductive. Because of that, you will have a shifting of phase between the current through the switch and the voltage across the switch. If there is enough phase shift, you will have trouble with commutating the switch (getting it to turn off), also resulting in hash noise. A properly designed snubber may be required.
In my opinion, a better choice would be an adequate switch, or use a relay, controlled by a less than adequate switch. The NRE for a solid-state switch exceeds the value of the resulting switch, unless the value is in the designing exercise, itself.
Carry on gentlemen. Good luck.
Enjoy,
Rich P
this mod on has almost no silver contact nub left and after more than a year it is still OK. The good thing about the Alps rotary switch on the LR 9090 is that the NC contact for shut down never carried much current and is always in new shape. Worst case is a totally fried power contact. Just add an ice cube relay N/C for the shut down trip circuit and use the old ALPS N/C contact in the N/O position for the TRIAC gate circuit. Contacts are interchangeable. I used the almost burned up one as a test.
Knowledge is based on amount of equipment destroyed 
