LM1875/TDA2050 adding output transistors.

[wg_ski]

My guess is the reason you get massive distortion when you connect the other pair is lack of global feedback. If you take the global feedback from the first output, what is correcting distortion on the other half of the bridge? If the answer is nothing, there’s your culprit. If the NPNs and PNPs were well matched to each other, the distortions would cancel - corrections on one side of the bridge would by default be corrected on the other. TIPs are pretty bad about match to each other, let alone NPN to PNP. If you are running the whole output stage open loop, with feedback around the 1875 only, it would surprise me if it’s any good even with only half the bridge.

 

[johnnyhou]

I built 2 transformers that were identical for the bridge mode. The output is not a perfect sine but it satisfies the requirements for my particular project. I did find a short circuit on the base of the transistors and corrected the issue so I am not certain if there was an issue feeding back from the opposing pair into the transformer. I will conduct further tests at some time and report my findings.
I doubled up on the output transistors using a total of 8 (4NPN 4PNP) transistors putting out over 500 watts RMS from a single LM1875T on 30-0-30 PSU. I agree that transformer coupled amplifiers are not the cleanest form of audio power but I simply had tons of these chips and power transistors on hand so I found a way to get what I wanted out of what I had. I will continue building variants of this particular topology and post specs with pics when I am able.