[ricard]

Thanks for the tip, might well order a set, they don't seem too common, and the length and cross sectoin is probably more critical than in general, as it's the back of the belt that drives the left reel table in rewind mode, and the belt tension has to compete with the tension of the spring pulling it towards the left table.

That said, I found a makeshift solution, by rethreading the belt; I took a square section belt and threaded it around the knurled part of the left reel table that it normally just makes contact with when threaded round the idler. It then runs over the idler to get it in the right position for the trail towards the right (nylon) idler. I had to file a couple of mm away from the left brake arm so the belt wouldn't rub against it. Perhaps not an ideal solution, as it seems to add a bit of friction to the left reel table (but on the other hand I don't get loops in play mode when starting), but at least a way to get the mechanics working while I go through other parts of the machine to see what else looks looking to.

I'll see if I can post a picture, it's easer than describing it.

The quiescent current in the output stage (slightly oddly enough a class A output stage with an OC26) was off, causing the output volume to be too high and slightly distorted. A simple adjustment gave me a good sounding output. The output stage is typically Philips, with a three-coil output transformer; one is the primary and internal speaker secondary, as an autotransformer, another is for the external speaker output and also used for negative feedback. The third coil is just connected via a resistor and capacitor to +15.5V, so in normal use there's no current flowing through it. The only reason I can think of is some form of anti-thumping turn-on circuit (counteracting some other current flow in another winding), or possibly some form of bootstrapping.