OK. Read.

First of all that is a good piece. Not the most advanced but so what.

The amp is a standard quasi comp design, not hard at all.

Now, first of all for the outputs 2N3773s would be great. that is a JEDEC (US) number though and you might not be able to get them there. the originals are TO-3, try to stick with that and not use the plastic equivalent, I don't like how they mate to the original holes. the original design calls for less than a certain thermal resistance there and upsetting that is not good. If you got a friend in the US those 2N3773s are not expensive and fit alot more than just that amp.

So take the outputs out. Short the base and emitter terminals of each as well as TR 624 and 604 (?) the bias transistors emitter to collector, if unsure of the basing just short it all the way around. Not with the outputs though, the basing is clear enough, the pins are E and B, those are what you short. (not to the other transistor...)

Now use a DBT for the fuse whether it is blown or not. When powering up the lamp should light then dim gradually until it is about out. The voltage at the speaker outputs should be zero or close. If not, one of the drivers may be shorted, if you care to that can be checked with no power using the DVM on the -|- scale of course.

With a scope you should even be able to see an audio waveform at the output, put on some rock music, the waveform should be pretty much symmetrical up and down. DO NOT connect a load like a speaker or dummy or you will blow the drivers if they aren't already.

Once you got that all straightened out and have it with not outputs in then put in the new ones. Leave the short across that bias transistor and connect speakers. There will be much distortion at low levels getting almost clean as you approach maximum power.

Once it passes that test, remove power and discharge the filters. Remove the short from one of the bias transistors, power on, if that light does not dim as it did before there is a fault in the bias, do not run it like that load or not.

All this assumes the resistors that join at the speaker outputs have been checked and are not open. Usually there is no smoke or burn so they need to be tested electrically. forget the exact value, to measure it accurately you need the right equipment, either a meter on which you can null the lead resistance or a Kelvin connection. If it is not open, 95 % of the time it is good.

Those are the live tests mainly, there is not reason not to check the transistors and low value resistors before even starting the powered tests. Main thing you want id no open transistors and no shorted transistors.

These methods have made me a bunch of money because I do not waste silicon. And BTW, did you say something about 3 of them ? They need to be replaced in pairs. If one shorts, the feedback in the circuit tries to pull the voltage back to zero and that can exceed the design maximums of the other output in the channel.

If the original AC fuse is not blown use that meter and look for open resistors all over those output circuits, if anything shorts something has to give. If what gives it in the bias circuit you could lose your new outputs in less than a second, that is why the first run is done with zero bias.

As far as making sure the problem is in the output stage, just take a pair of RCA cables and switch left and right. If the problem switches to the other channel then all this is moot and now we are in the preamp. The scope will come in VERY handy if that is the case.

If it does go to the other channel a mono switch would be nice but I do not wee one. In this case you can just short the volume controls together, either the top (not grounded) side or the wipers, doesn't matter, if you get both channels then the problem is before that. At that point there ain't much left. (ain't is a word used only by those with exceptional command of the English language lol)