It may not even be necessary...typically only important if you expect high temperature operation. Most likely source of leakage current would be Icbo of Q1.
Data sheet shows Icbo with Vcb=120 Volts and IE=0 of 50 nA at 25C, and 50 uA at 100C. So, it goes up, lets say, 1000X for a 75C temp change.
That says it goes up 10X for every 25C change.
50 nA at 25 C
500 nA at 50 C
5 uA at 75 C
50 uA at 100 C
Assume Tambient max is 75 C, and only plus/minus 45 volt rails. Then, 5 uA might really only be 2.5 uA. Let's be conservative...say 3.5 uA, and we want to keep the voltage from base of Q10 to Vneg less than 0.7 Volts. Then we could put 0.7/3.5uA=200K resistor from base of Q10 to Vneg.
That should barely change operation, except in the off state, where it may keep you out of leakage current issues.
Data sheet shows Icbo with Vcb=120 Volts and IE=0 of 50 nA at 25C, and 50 uA at 100C. So, it goes up, lets say, 1000X for a 75C temp change.
That says it goes up 10X for every 25C change.
50 nA at 25 C
500 nA at 50 C
5 uA at 75 C
50 uA at 100 C
Assume Tambient max is 75 C, and only plus/minus 45 volt rails. Then, 5 uA might really only be 2.5 uA. Let's be conservative...say 3.5 uA, and we want to keep the voltage from base of Q10 to Vneg less than 0.7 Volts. Then we could put 0.7/3.5uA=200K resistor from base of Q10 to Vneg.
That should barely change operation, except in the off state, where it may keep you out of leakage current issues.
Let me add that many (most?) amps start "stupid" at turn-on, for a few milliseconds, usually until most capacitors (specially the one in the feedback loop, but also the one at the input and the bootstrap one, if used) charge to some reasonable voltage and during that time, the DC compensation does not work.
It may very well slam the output rail against either + or - V rail for that time, which is heard as a turn-on thump.
Your spice simulation showed that, it did not lie at all.
How long did it take the amp to be stabilized ... 500 msec?
Fine, why not trust spice?
Such result agrees with real world experiences.
It may very well slam the output rail against either + or - V rail for that time, which is heard as a turn-on thump.
Your spice simulation showed that, it did not lie at all.
How long did it take the amp to be stabilized ... 500 msec?
Fine, why not trust spice?
Such result agrees with real world experiences.
I ran some sims with the original circuit. Spice was getting fooled, I think.
The initial solution at power up showed -36 volts across C3, the dc block in the feedback loop. I don't see how this would happen in the real world. More interestingly, you would get different answers for 5 Ohms or 6 Ohm loads. If you changed the input stage transistor model, you got a different initial solution. If you changed the simulator engine, you got a different solution
Given the original post, "Amplifier can't drive 5, 7, and 13 Ohms" of course I'd pun...that's odd
The amp, even without the diode across C3 still does converge to a good solution, but it takes about 0.7 seconds...
The initial solution at power up showed -36 volts across C3, the dc block in the feedback loop. I don't see how this would happen in the real world. More interestingly, you would get different answers for 5 Ohms or 6 Ohm loads. If you changed the input stage transistor model, you got a different initial solution. If you changed the simulator engine, you got a different solution
Given the original post, "Amplifier can't drive 5, 7, and 13 Ohms" of course I'd pun...that's odd
The amp, even without the diode across C3 still does converge to a good solution, but it takes about 0.7 seconds...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.