That photo of the input rating confirms that the amp has automatic mains voltage selection – my theory of the voltage selection relay switching, as being the fault is looking good, but precisely what is causing it is another issue.

Nice that you didn’t need the printed input ratings to identify the voltage selection feature. I see that you were tipped off by the wires pouring out of the transformer. If this unit did NOT feature the auto-voltage selector, what would I be seeing instead? Just two wires and a ground?

 

The margin of safety purchased for the few cents more to get the higher voltage rating is worth the extra cost.

 

I am not sure increasing the voltage rating of those capacitors improves any margin of safety. Yes, they may survive a higher voltage spike for awhile, but is that spike normal? Consider the situation of a fuse that keeps blowing in a household circuit. Replacing that fuse with a higher current rated fuse may stop the fuse from blowing but doesn't address the cause and certainly doesn't do anything good about safety. Those capacitors may have just acted like fuses.

Now, you need to consider whether those voltage spikes may cause other problems, such as causing the insulation on the transformer or relay to fail? The fact that the manufacturer chose not to use capacitors with a higher voltage rating should raise at least a yellow flag. Do newer versions of the amplifier have a similar circuit? What voltage rating is used in them?

 

Since the primary of the linear transformer is auto switching for the correct input voltage rating – I’m reasonably certain that this high voltage generation/arcing is due to the incorrect switching of the input voltage selection relay.

It will be difficult to fault find this without a circuit diagram since the amp will not power on in a stable state, therefore I offer the following fix.

Disable the voltage selection relay in the connection orientation for your mains input voltage. If you are lucky, with the relay de-energised will be the required state.

To determine this, measure the resistance between Live and Neutral at the mains plug (this should be quite low of the order of 10s of ohms or less). Then disconnect one of the blue (Neutral) transformer wires at the PCB. If the resistance between the mains plug Live and Neutral increases significantly, then with the relay un-energised the unit is configured for 240V operation. If when one transformer Neutral wire is disconnected, the resistance reading doubles – then the relay is wired in the 120V state un-energised.

To fix the relay in the un-energised state, cut one of the relay coil PCB tracks – if you refer to the relay data sheet in jpanhalt’s post #5 you will see that the two coil contact are separated from the six switched relay contacts – which should allow you to identify the relay coil tracks.

Once the relay is fixed in the required de-energised state (with the coil track cut), I would recommend that you replace the amplifier mains fuse with one of 1.6A (just in case), then power the amp. If all is well then put back the original fuse.

If the amp needs the relay energised to be in the correct voltage selection for you mains voltage supply – please advise and I will advise how to effect a fix in this mode.

 

The spikes that the capacitors are intended to protect from arrive from OUTSIDE the amplifier. They travel on the mains. They are not generated inside the amp, and so not much way to prevent them. They do not hapen very often, either.