Yamaha Stagepas 500W

Have you checked for oscillation?

As distinct from bench testing without the clock signal input I suppose. Will do that but as bench current drain is only .7mA looks as though I'll have to defeat the ERR line from this bad amp and add some droppers in its power lines. At the moment the little bad PA is unplugged and powered from the bench ps. This amp , like the other ones of 150W rating , when on +&-30V settle with the speaker output at about -1.5 to -1.7V and both sets of gates , and D settle to near enough always the same readings. Then properly in circuit with +&-48V and clock , but still no speaker load they settle to 10mV DC or so. With bad amp in hte unit and powered up and before it cuts out I will monitor for oscillation. Unfortunately with triggering of this protect you have to wait 5 minutes for a cap to discharge, I must locate that so I can manually discharge it. Changing the preset value on this one makes no difference to the -1.6V. I may try a bench oscillator fed to the clock of this amp, unusual exercise if nothing else

I hate the chicken and egg nature of long-tail pair pre-driver stages.

At least I know its safe , like the 300W ones, to swap L an R PAs, confirming that a problem on the PA and not farther back. I've previously tried on 300W ones visually comparing L and R amps to find the extra transistor or trace . Everyone knows of the antiphase output of the L one but where is the balancing inversion. Its on the CONTROL3 board just prior to the PA header connectors, swapping of signals to INV and NON-INV lines. So as far as I can find, despite different identifying overlay labels on these PAs, there is no electronic difference. Not sorted out what the CLK/CLOCK means as tied to output_ground , with no connectors in the output 1/4 speaker sockets anyway. As cold DVM-D test comparisons of long-tail sections of each PA has not shown up anything amiss, at the moment I'm assuming a transistor is leaking a bit, especially as this amp is used outdoors. With +\-30V bench ps powering, one section is off, will try 30V powering of the good PA and see if the same with that one, no zeners around . Then if the same I'll have to fudge my usual bench ps to supply +\-45V by setting as isolated 30V & -30V with another single isolated 30V supply bridging it along with a couple of 15V zeners, with fingers crossed. I've decided its too awkward to try probing live, via the units ps ,on all this SMD stuff, in an awkward position even with 5w power line extender , having decided not to put droppers in there . It really needs a test frame making up to hold these very light boards in place.

good PA shows the same off pair of transistors in the long-tail , with

+/-30V so bench ps bodged to +/-44V coming soon perhaps. No oscillation on output at powerup of full rails until the moment the signal is fed through then the output steps DC a lot, then protect engages, whether signal is present at input or not and whether this bad PA is in the L or R position. So whether signal is fed to + or - input or just low level noise . Will investigate the power-down line in the PA as presumably power-up firstly. This enable area also involves the "off" pair so must try pulling that line low with +/-30V current limited supply on both good and bad amp first

Had to fudge +/-44V supply before longtail section would bias properly with mute off. Not in that area , between that and power section proper a 120V .1A SMD Q with 10V forward across B-E junction. Not an iota of difference DVM-D cold checking that Q and its good PA brother. Hopefully enough space between the two tiny PAs for a 120V E-line to go in there

Was still triggering protect. Took a closer look at the ERR area. I'd thought , confusing schematic lines, that it monitored the speaker output line. No it seems to monitor imbalance in the power lines. But if , as in this case , one or more of the 0.047 ohm fusible Rs fail then

44V is put across B-E junction with only 1K dropper in one case and 2.2K in the other case. So the stressed one , momentarily, is leaky