well good i guess i feel a bit better about that. maybe??? I need to re run my test with the VBE transistor more centrally located in the heatsink. then let it sit for a while after a 1/3rd power run and see if the the bias comes back down to the normal operating point.
The other interesting thing is that ARC set the bias with the TT outputs by trimming R37 and that Q11 is not mounted to the heatsink. it is on top of the board with its own little heatsink and even air currents can change the bias. so if i add a VBEx in place of the TT diodes. it sort of has 2 places to set bias? or....should a resistor just be put in place of the TT diodes and Q11 get mounted to the heatsink?
OK, But. at ARC we trimmed R37 to set the bias at 20mv across the emitter resistors. now that we are not using the diodes in the TT devices. what is R37 doing? should it be trimmed to some specific current through the VBEx?
Ok, actually it`s a bit nifty solution they engineered, the voltage left by the string of eight diodes can`t be totally predicted, so they made Q13 a adjustable current source by making R37 adjustable. Q13 sets the bias for current dumpers Q11 and Q12. Those again pumps more - or less - current down the diodes to finally end up with the bias they wanted. So, again, leave R37, add Vbe as previous explained and adjust bias there.
Which in turn makes me wonder why they couldn`t solve the problem that the defective TT introduced! I can only surmise that at that time the clever engineers had left the sinking ship.
R
Which in turn makes me wonder why they couldn`t solve the problem that the defective TT introduced! I can only surmise that at that time the clever engineers had left the sinking ship.
R
More testing tonight. I measured 24ma of current through the VBEx. I added an 18 ohm resistor in series with the collector of the VBE transistor and found it to help quite a bit to stabilize the bias have a hot thermal event. even with the amp running at 1/3rd power for an hour straight, bias was steady and then letting it idle it gradually settled back to a normal bias. i feel pretty confident that this solution is now stable. I am going to make up some boards to permanently fix this.
This is a quick photoshop markup of the ARC schematic with the UN-Thermaltrack changes made. the design is straight from the Onsemi Thermal track white paper and uses a MPSA06 transistor mounted to the heatsink. I added the 18 ohm resistor in the collector leg HOWEVER i realized later that the connection should have been made at the X location instead. However it seems to work well as is. I need to retest with the correct connection.
