Well, my original plan was a small board for each amp, containing the psu and the circuit. I built one (finished it yesterday) and proceeded to do a test run on the bench.
Plugging in with no bias, everything's ok; the DC voltages are ok (exactly +/- 22vdc, wohoo!). I adjust DC offset first to 0v and then slowly increase bias up to 1,7A... the sinks get hot All fine 'till here... each transistor has it's own sink and every one measured 67-70ºC (mind this is one of the hottest days of the year here). I was happy!
So i smell heat... damn. I touch the sinks on the cap. multiplier and they're SCORCHING hot! I used small sinks, but i thought i was in te vecinity of 5ºC/W (30ºC worst case). Not only that, one of the drivers transistors has no sink (a BF258 replacing the original MJE371; this trans. has a really old case, like a "water tank"... think of a germanium one)... 100ºC there!!!
Now, i'll have to rebuild the whole thing... a separate board for the two psu's, with the cap-x transistors on a bigger sink and another one for the pair of amps (replacing the BF258 with a BD140, sinked). I HATE this, the amp was running fine and i was dissipating more heat on the driver/psu transistors than the output ones!!!!!!!
Lesson: don't ever, NEVER, underestimate the power dissipation on small transistors, and use sinks. Even a rather small piece of Al will do the trick. 2w made one crazy, while the other, with a minimal sink, just stood warm. So did the 7815 regulator (i added a resistor to ground just to make sure i always had the minimal drive current).
There's a bright side through. I adjusted the DC offset to cero before starting... and it increased *only* to 2mV after the whole mess! Which is rather amazing, knowing how hot the drivers got. I once readed someone here asking for a DC servo for this amp... from my experience, it's VERY stable in that sense. I dunno about the 1969 JLH, but the quiescent current control used in the '96 can be used there too. The turn-on and off thumps were small and fast too, through my digital multimeter's not the fastest.
Back to the workshop
Plugging in with no bias, everything's ok; the DC voltages are ok (exactly +/- 22vdc, wohoo!). I adjust DC offset first to 0v and then slowly increase bias up to 1,7A... the sinks get hot All fine 'till here... each transistor has it's own sink and every one measured 67-70ºC (mind this is one of the hottest days of the year here). I was happy!
So i smell heat... damn. I touch the sinks on the cap. multiplier and they're SCORCHING hot! I used small sinks, but i thought i was in te vecinity of 5ºC/W (30ºC worst case). Not only that, one of the drivers transistors has no sink (a BF258 replacing the original MJE371; this trans. has a really old case, like a "water tank"... think of a germanium one)... 100ºC there!!!
Now, i'll have to rebuild the whole thing... a separate board for the two psu's, with the cap-x transistors on a bigger sink and another one for the pair of amps (replacing the BF258 with a BD140, sinked). I HATE this, the amp was running fine and i was dissipating more heat on the driver/psu transistors than the output ones!!!!!!!
Lesson: don't ever, NEVER, underestimate the power dissipation on small transistors, and use sinks. Even a rather small piece of Al will do the trick. 2w made one crazy, while the other, with a minimal sink, just stood warm. So did the 7815 regulator (i added a resistor to ground just to make sure i always had the minimal drive current).
There's a bright side through. I adjusted the DC offset to cero before starting... and it increased *only* to 2mV after the whole mess! Which is rather amazing, knowing how hot the drivers got. I once readed someone here asking for a DC servo for this amp... from my experience, it's VERY stable in that sense. I dunno about the 1969 JLH, but the quiescent current control used in the '96 can be used there too. The turn-on and off thumps were small and fast too, through my digital multimeter's not the fastest.
Back to the workshop
Lisandro
Sorry to hear that you have had problems with your JLH. As you have found out the hard way, the BF258 was not a good choice for Tr5. According to my databook, it is rated at 1W whereas Tr5 dissipates around 850mW at idle and has a peak dissipation of 1.9W (1.1Wrms) at maximum output. If you cannot get the MJE371, the normally listed equivalent is the BD436 though a BD140 should be OK.
If you are rebuilding your amps, please consider using a ccs in place of the 7815, as shown on my 'JLH for ESL' page. I have been concerned about the 7815 injecting noise into the emitter of Tr4 (not to mention the 7815 stability problems) which is why I introduced the ccs in the ESL version. Another constructor has just replaced the 7815 with a ccs in his 1996 JLH (built with premium components) and has reported that the ccs is "cleaner, smoother and weightier".
I think you will find that when your amps are working correctly, the output dc offset variation between switch-on (cold) and normal operating temperature will still be significant (>100mV) even with the 1996 version.
Geoff
Sorry to hear that you have had problems with your JLH. As you have found out the hard way, the BF258 was not a good choice for Tr5. According to my databook, it is rated at 1W whereas Tr5 dissipates around 850mW at idle and has a peak dissipation of 1.9W (1.1Wrms) at maximum output. If you cannot get the MJE371, the normally listed equivalent is the BD436 though a BD140 should be OK.
If you are rebuilding your amps, please consider using a ccs in place of the 7815, as shown on my 'JLH for ESL' page. I have been concerned about the 7815 injecting noise into the emitter of Tr4 (not to mention the 7815 stability problems) which is why I introduced the ccs in the ESL version. Another constructor has just replaced the 7815 with a ccs in his 1996 JLH (built with premium components) and has reported that the ccs is "cleaner, smoother and weightier".
I think you will find that when your amps are working correctly, the output dc offset variation between switch-on (cold) and normal operating temperature will still be significant (>100mV) even with the 1996 version.
Geoff
Geoff said:Sorry to hear that you have had problems with your JLH. As you have found out the hard way, the BF258 was not a good choice for Tr5. According to my databook, it is rated at 1W whereas Tr5 dissipates around 850mW at idle and has a peak dissipation of 1.9W (1.1Wrms) at maximum output. If you cannot get the MJE371, the normally listed equivalent is the BD436 though a BD140 should be OK.
Yep... the BD436 is kinda tricky to get down here. I'll use a BD140 and place both drivers on a small sink, that'd ok.
If you are rebuilding your amps, please consider using a ccs in place of the 7815, as shown on my 'JLH for ESL' page. I have been concerned about the 7815 injecting noise into the emitter of Tr4 (not to mention the 7815 stability problems) which is why I introduced the ccs in the ESL version. Another constructor has just replaced the 7815 with a ccs in his 1996 JLH (built with premium components) and has reported that the ccs is "cleaner, smoother and weightier".
I think you will find that when your amps are working correctly, the output dc offset variation between switch-on (cold) and normal operating temperature will still be significant (>100mV) even with the 1996 version.
Yep, i considered that. My approach to the 78xx's stability issues was to put "large" caps to ground (10uf) in both the input and the output, and a 4.7k resistor to ground in the output, again, to ensure the minimal drive current. I will check on the current source and get back to you through, the idea is surely appealing.
About the dc offset... yes, i was expecting that too But i think the amp WAS running ok (despite the enormous heat), and i rechecked the offset quite a few times. I didn't want to plug a loudspeaker as you can imagine... the thing could destroy itself at any moment.
I already built a good sink out of Al sheet for the cap-x psu (this one will be internal), and will be building that part in a short while. Wish me luck!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.