> I don't know that the diodes need to be high current
I don't know either. But a hi-current diode will give a lower bias and transistor current. A really-small diode on high-current transistors opens the chance of HIGH idle current and quick failure.
Then I noticed the BJTs were TO-66 (remember that?) 25W 1A 40V 3MHz parts. Bigger base than the 100mA TO92 jellybeans used in smaller gear. Use a big diode.
And 1N4007 should be dirt-cheap in any repair shop. So it should be a no-cost safe first try. If it actually biases too cold, we can work from there.
I have no clue how the diodes could go bad yet the transistors stay OK. I don't see a way for excess current to flow, or for reverse voltage to happen. So there's still a mystery.
On third thought: I'd be real tempted to lose the entire buffer and strap two 5532 chips as a parallel unity-gain buffer. Use individual 10 ohm output series resistors for current-sharing. This will drive the apparent 150 ohms. All the bias and protection is already in there. Yes, this "adds 5532 sound" to older vintage gear. But is there any complete (mike to speaker) reproduction chain which does NOT already have multiple 5532s in the path? You still have whatever goodness is in that A1000 chip: it will still do the input sensing and all the voltage gain.
I don't know either. But a hi-current diode will give a lower bias and transistor current. A really-small diode on high-current transistors opens the chance of HIGH idle current and quick failure.
Then I noticed the BJTs were TO-66 (remember that?) 25W 1A 40V 3MHz parts. Bigger base than the 100mA TO92 jellybeans used in smaller gear. Use a big diode.
And 1N4007 should be dirt-cheap in any repair shop. So it should be a no-cost safe first try. If it actually biases too cold, we can work from there.
I have no clue how the diodes could go bad yet the transistors stay OK. I don't see a way for excess current to flow, or for reverse voltage to happen. So there's still a mystery.
On third thought: I'd be real tempted to lose the entire buffer and strap two 5532 chips as a parallel unity-gain buffer. Use individual 10 ohm output series resistors for current-sharing. This will drive the apparent 150 ohms. All the bias and protection is already in there. Yes, this "adds 5532 sound" to older vintage gear. But is there any complete (mike to speaker) reproduction chain which does NOT already have multiple 5532s in the path? You still have whatever goodness is in that A1000 chip: it will still do the input sensing and all the voltage gain.