Putting in 96's into an old 700B
- Thread starter WOPL Sniffer
- Start date
To answer your question Lee, I was not there so I can only speculate. I think, yes, that they were aware of it but could only solve for it with more basic methods at that time. SPICE modeling was just in its infancy in the early to mid 70s, it is still being developed even today (I purchase a yearly update).
My speculation is that they only had empirical ways to solve the stability issues. Normally a Sziklai pair consists of one PNP and one NPN transistor to emulate a higher power PNP transistor (or NPN, the reverse works too). In the case of the PL application, they combined a single PNP transistor with a Darlington NPN structure to emulate the PNP power transistor. The Darlington NPN in this topology has huge voltage gain when compared to the normal single NPN transistor in a Sziklai pair. It is very hard to compensate for that huge voltage gain and keep the construct stable. When the second NPN kicks in in the SPICE model, the instability becomes much more apparent. The trace that I displayed was with only a 1V P-P square wave signal. If I increase it to a 5V P-P signal, the second stage of the Darlington turns on and things degrade.
The full comp does not have these drawbacks. They would have applied the full comp configuration if they had power PNP devices available at the time. Evidence of that is that they switched to that configuration in the last stages of Series 2 production.
The purple heatsink phenomenon is real and it seems that the only way they knew to solve it was put in higher wattage Zoebel networks in an attempt to contain it. That was not solving the problem.
My speculation is that they only had empirical ways to solve the stability issues. Normally a Sziklai pair consists of one PNP and one NPN transistor to emulate a higher power PNP transistor (or NPN, the reverse works too). In the case of the PL application, they combined a single PNP transistor with a Darlington NPN structure to emulate the PNP power transistor. The Darlington NPN in this topology has huge voltage gain when compared to the normal single NPN transistor in a Sziklai pair. It is very hard to compensate for that huge voltage gain and keep the construct stable. When the second NPN kicks in in the SPICE model, the instability becomes much more apparent. The trace that I displayed was with only a 1V P-P square wave signal. If I increase it to a 5V P-P signal, the second stage of the Darlington turns on and things degrade.
The full comp does not have these drawbacks. They would have applied the full comp configuration if they had power PNP devices available at the time. Evidence of that is that they switched to that configuration in the last stages of Series 2 production.
The purple heatsink phenomenon is real and it seems that the only way they knew to solve it was put in higher wattage Zoebel networks in an attempt to contain it. That was not solving the problem.
- Joined
- Jan 14, 2011
- Messages
- 74,264
- Location
- Gillette, Wyo.
- Tagline
- Halfbiass...Electron Herder and Backass Woof
Don, the RCA410 has pretty lousy GBW which is likely why it was picked. It limited the gain of the overall Darlington construct. You can stabilize it by dumping the loop feedback (changing the 22 ohm normal resistor to 100 will dump a lot of the loop gain). I am pretty sure that has other negative consequences to items like distortion however.
It's actually what wattsabundant writes. Bypass capacitor with 33 pF between base and collector of Q7 and Q10, also base stoppers on the outputs with 2,7 ohms - nothing works. Modern and old does not want to harmonize and why, is a mystery to me (probably the same phenomenon as with the RCA 410). I can recommend not to invest a lot of time in this.
As already said one channel was equipped with older ISC BU608 (whoever put them in back then, but they work). For fun, I bought a completely new set of BU608 from the same company - and that also oscillates. My 700B rejects anything newer.
The longer I have dealt with troubleshooting the more stupid I find this quasi complementary circuit with all its disadvantages today, so that this becomes a WOPL. This also increased the desire to rebuild everything. I have had restored and owned a couple of late 60s / early 70s US amps, but none of them was as stubborn as this 700B.
Nevertheless, I'm impressed about the forum, if someone is are interested I will show my the WOPL build!
As already said one channel was equipped with older ISC BU608 (whoever put them in back then, but they work). For fun, I bought a completely new set of BU608 from the same company - and that also oscillates. My 700B rejects anything newer.
The longer I have dealt with troubleshooting the more stupid I find this quasi complementary circuit with all its disadvantages today, so that this becomes a WOPL. This also increased the desire to rebuild everything. I have had restored and owned a couple of late 60s / early 70s US amps, but none of them was as stubborn as this 700B.
Nevertheless, I'm impressed about the forum, if someone is are interested I will show my the WOPL build!