I have a couple questions about this schematic posted awhile ago by DJK. I realize this is the basic circuit QSC has been using since the beginning, but I'm still a little shaky on the details.
This is the schematic/parts list i have:
http://i120.photobucket.com/albums/o200/ppiamp333/DJKopampdriven.jpg
A) R6 sets the bias current right? How much current needs to be available from the -15V supply used there? Is that just from the +/-15V supply used for the opamp?
B) The transformer CT for the main supply is not connected to ground or anything, as the rails need to float with the signal, right? This also means the +/-15V for the opamp supply needs to come from a separate winding or another transformer?
C) Because of the floating rails each channel should have its own winding, or it's own transformer; dual mono is mandatory?
D) Even though the output transistor collectors are at ground they still need to be isolated from the heatsink to avoid multiple ground loops through the heatsink to chassis and the ground trace on the boards?
Lastly, is the schematic shown here a fully working/reliable design? Do you have any pictures of your layout that i can work off of? Thanks for any help.
This is the schematic/parts list i have:
http://i120.photobucket.com/albums/o200/ppiamp333/DJKopampdriven.jpg
A) R6 sets the bias current right? How much current needs to be available from the -15V supply used there? Is that just from the +/-15V supply used for the opamp?
B) The transformer CT for the main supply is not connected to ground or anything, as the rails need to float with the signal, right? This also means the +/-15V for the opamp supply needs to come from a separate winding or another transformer?
C) Because of the floating rails each channel should have its own winding, or it's own transformer; dual mono is mandatory?
D) Even though the output transistor collectors are at ground they still need to be isolated from the heatsink to avoid multiple ground loops through the heatsink to chassis and the ground trace on the boards?
Lastly, is the schematic shown here a fully working/reliable design? Do you have any pictures of your layout that i can work off of? Thanks for any help.
Stormrider said:
Yup - this bias arrangement pulls current from the opamap output continuously - forcing the opamap's output stage to run class A. This should improve the sound compared to the usual bias scheme. Downside is not just any op amp can do this - you need one that can drive relatively low Z loads directly.
Stormrider said:
The usual practice is to mount the trannies directly to the heatsinks. You just have to be careful about ground loops. That may mean isolating the HS from the chassis (but the voltage differential will be low so there is no shock hazard if the HS is exposed).
No current limit if that's what you mean. If you do without that, use enough outputs so that the power supply collapsing under shorted conditions is enough to protect it before a fuse blows. That will take 8-10 MJ15024's per bank per 500VA of trafo. More if you oversize the reservoir caps.
Thanks for the help wg ski.
That clears everything up except for the transformer CT part. I'm pretty sure there shouldn't be, as no QSC schematic I've seen shows one anywhere for this topology. Just looking for confirmation.
Isolating the heat sink should be easy enough for any layout i use. I'm betting the heat transfer will be much better with only grease.
Would current limiting be simple to add, or would it be a redesign? I'd be using a 500-600VA transformer per channel with +/-60 to 70Vdc rails. I was planning four pairs of MJ21193/4 or 5/6 originally. It will only be me operating the amp and shorts with signal would be unlikely, but not having to replace entire output stages because of a simple mistake would be nice.
That clears everything up except for the transformer CT part. I'm pretty sure there shouldn't be, as no QSC schematic I've seen shows one anywhere for this topology. Just looking for confirmation.
Isolating the heat sink should be easy enough for any layout i use. I'm betting the heat transfer will be much better with only grease.
Would current limiting be simple to add, or would it be a redesign? I'd be using a 500-600VA transformer per channel with +/-60 to 70Vdc rails. I was planning four pairs of MJ21193/4 or 5/6 originally. It will only be me operating the amp and shorts with signal would be unlikely, but not having to replace entire output stages because of a simple mistake would be nice.
The trafo center tap isn't required, but advisable. If you don't use one, the put the usual equalizing resistors across the filter caps. If you do, it goes to the speaker output terminal (center point between the caps).
Current limiting is tricky with this topology. In the QSC amps, it uses two levels of current starvation to implement. The first is to limit the drive voltage from the opamp, with the trip point set by a pot. This limits instantly and is normally set to a very high value. The is second time-dependent. The QSC opamp supply is drived from the idle rail voltage - and thus relies on a couple 100uf or so bypass/bootstrap caps to supply current to the opamp during output swings above +/-15V. Where it limits depends on how long the conditions last. More base current = quicker limiting action.
If you wanted to use a standard load line limiter, it would take two transistors per rail. One to sense the voltage and current, which would be tied to the rail where the emitter resistors are. Then you'd have to use a second to translate the switching action back to where the output stage is driven from the opamp. Either that or put in "emitter resistors" in the collectors of the outputs, but that defeats the purpose of having the heatsink at ground.
Current limiting is tricky with this topology. In the QSC amps, it uses two levels of current starvation to implement. The first is to limit the drive voltage from the opamp, with the trip point set by a pot. This limits instantly and is normally set to a very high value. The is second time-dependent. The QSC opamp supply is drived from the idle rail voltage - and thus relies on a couple 100uf or so bypass/bootstrap caps to supply current to the opamp during output swings above +/-15V. Where it limits depends on how long the conditions last. More base current = quicker limiting action.
If you wanted to use a standard load line limiter, it would take two transistors per rail. One to sense the voltage and current, which would be tied to the rail where the emitter resistors are. Then you'd have to use a second to translate the switching action back to where the output stage is driven from the opamp. Either that or put in "emitter resistors" in the collectors of the outputs, but that defeats the purpose of having the heatsink at ground.
that is a very inetersting
circuit and even though was arround here since i dont remember any more but no one clone anything like that ....
of course this is no hi end amp but sure worth looking at it powerwise
my circuit ....the modified Hill chameleon is still working like hell serving small PA application ..... an AB +b design that doesnt say much quality wise but its solid rock and very ready to accept various forms of abuse ....
circuit and even though was arround here since i dont remember any more but no one clone anything like that ....
of course this is no hi end amp but sure worth looking at it powerwise
my circuit ....the modified Hill chameleon is still working like hell serving small PA application ..... an AB +b design that doesnt say much quality wise but its solid rock and very ready to accept various forms of abuse ....
wg_ski said:
So, if i understand this right, keeping the capacitance low on the output side of the +/-15V regulators I'm going to use would limit the output swing from the opamp well enough to help protect the output stage if the output is shorted, or does that just help protect the amp during long periods of high output - at clipping?
What base current are you talking about? Q5?
If you're using a separate regulated supply for the opamp, you can't use QSC's current limit scheme. The scheme relies on the fact that the opamp's supply physically goes away on every other half cycle beyond a certain output voltage, and all that can be supplied is a residual charge in the cap bypassing the opamp rail.
The base currents are that of Q1 and Q2. Ib = Ic/(hFE1*hFE3). Even if you don't use QSC's time dependent limiting, your hard limit will be when the opamp limits (around 30mA for the 5532).
I suppose it would be possible to do something like QSC does by setting a current limit on the opamap supply at a very low value, not much more than the quiescent current of the opamp. Then use 100uf bypass caps which could retain a charge and supply more for short periods, but get drained when the going gets tough. You'd just have to build it to see if it's effective or even stable. But even this woudln't work with the opamp biased into class A like djk did in his amp so that would have to change. If I were building this myself, I'd just incorporate a load line limiter referenced to the rail rather than the usual at the speaker output. The QSC PLX amps do it this way and the scheme is even in some of the really old Motorola app notes (the one for the MJ802/4502 which has example amp schematics).
The base currents are that of Q1 and Q2. Ib = Ic/(hFE1*hFE3). Even if you don't use QSC's time dependent limiting, your hard limit will be when the opamp limits (around 30mA for the 5532).
I suppose it would be possible to do something like QSC does by setting a current limit on the opamap supply at a very low value, not much more than the quiescent current of the opamp. Then use 100uf bypass caps which could retain a charge and supply more for short periods, but get drained when the going gets tough. You'd just have to build it to see if it's effective or even stable. But even this woudln't work with the opamp biased into class A like djk did in his amp so that would have to change. If I were building this myself, I'd just incorporate a load line limiter referenced to the rail rather than the usual at the speaker output. The QSC PLX amps do it this way and the scheme is even in some of the really old Motorola app notes (the one for the MJ802/4502 which has example amp schematics).
I have worked extensively with grounded collectors/drain/source/emitter amps
If you want to use one transformer than you need separate windings, you can also go for 2 transformers if you budget allows.
The collectors in this design act as reference earthing point as whole of the circuit is referenced to this very point. you can use this collectors ground as main power ground i.e you can connect the collectors of N number of channels on same heatsink.
QSC RMX4050 does this, shares same heatsink, both the channels share common ground with chasis also.
Stormrider said:
If you want to use one transformer than you need separate windings, you can also go for 2 transformers if you budget allows.
The collectors in this design act as reference earthing point as whole of the circuit is referenced to this very point. you can use this collectors ground as main power ground i.e you can connect the collectors of N number of channels on same heatsink.
QSC RMX4050 does this, shares same heatsink, both the channels share common ground with chasis also.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.