Hi.
I've been puzzling over the problem described here, and looking at the circuit diagrams trying to figure out what the problem could be. A clue is the report that shorting pins 2 and 3 of IC7 makes the audio come through OK.
IC7 is an op-amp configured as a non-inverting unity-gain buffer, to isolate the FM multiplex signal from the tuner by way of the MAX4544 switch from the outside world. The RF level signal comes directly from the tuner module, so if there was a problem with the multiplex signal it is feasible that the software could report a full-strength reception while producing a faint audio output.
A unity gain op-amp is produced by having the gain-producing feedback component (normally a resistor) set to 0 ohms, thereby forcing the amp to a gain of 1, but with a very high input impedance. This means that pins 6 (output) and 2 (inverting input) are directly connected together, since a convenient 0 ohm resistor is simply a PCB track. If subsequently pin 3 is shorted to pin 2, this effectively means that 2,3,and 6 are all connected together, therefore bypassing the buffer completely. If at this point the signal level comes up to the correct value, it strongly implies that there is a fault with the op-amp.
This could be a power supply fault, so check that there is +9v present on pin 7. Also, for the sake of completeness, check that is actually IS a CA3140 at IC7! Mistakes can be made
Assuming all this is OK, also check that R13 is the correct 220 ohm resistor, and that you haven't accidentally used a different one.
If this all checks out, the only thing I can really think is that the chip itself is actually bad. If you can remove it and replace it with another one which then works, this would prove it. However, a workaround is as you suggested, which is to bridge pins 2 and 3. It's a little dirty, but I can't see any particular reason why it would cause problems, other than perhaps an increased susceptibility to induced noise from external sources, especially if the cable between the empeg and tuner is extended. The tuner module is still isolated by the MAX4544, so it should be safe enough.
If you want I can send you a replacement CA3140, but they are only a few cents apiece so if you can get one locally it would save a lot of effort. I have to admit I am curious to know if the chip really is bad. I can't see any other possibility, but bad chips are normally pretty rare.
pca