azira said:
I think thats exactly it. Did you post it here a few months ago asking the same question as this? If so, then yep, thats what I'm remembering.
HDTVman said:
Don't worry, I'm used to that.
OPA541 running at +/-38VDC driving a load under 4 ohms. the chip literally snapped in half and the lights dimmed as the transformer hummed. A little longer and the breaker probably would have popped.
azira said:
Ya Danny, it is a bad idea.
1. the two outputs are at half the dc voltage of the power supply for each chip. This means that the top chips output is at +30 VDC and the bottom chip is at -30 VDC. That's at no signal by the way. Remember the output is from chip output to chip output so now you have 60 VDC at the speaker terminals.
2. You could use output cap coupling but the turn on thump would be something else. Remember that you also need to drive the inputs 180 degrees out of phase and with +- 30 volt offsets.
3. Normal bridge circuit would give the same power output by the way.
The difference that the circuit you posted has is that there is only one output, not two and the load is referenced to ground.
Does that help?
BZ
HDTVman said:
1. This is not correct, you assume that an opamp by default centers its output on the rails but in truth it does not. The only reason it appears that way in standard circuits is that the - input or lets call it the reference input is set at the midpoint between V+ and V- and the + input varies around that value.
If instead, the reference input was set at the V+ rail and the + input was varied around that, the output would start at the V+ rail and attempt to vary around that as well, however, obviously the chip cannot deliver higher than V+ out.
That is also not correct. The class-B output stage is two seperate current drivers connected at the outputs and inputs just like the opamp circuit I drew. If you black boxed the current drivers in both circuits you would see that they are identical. The key to making them act the same is to configure it so that each current driver is active for only 1/2 the cycle.
--
Danny
I don't think I posted this idea. I tried simulating it a couple times and ran into some implementation difficulties that I never got around to solving. One of the interesting problems with this configuration is that many opamps don't behave nicely when their + input is drive above (or below) the rails.
Yeah, this circuit is pretty cool... I lost the link to that article about 6 months ago. NP posted a few GC circuits with similar bootstrapping ideas as well.
--
Danny
hardly any chip amps are able to survive any terminal swinging beyond the rails
most devices have a parasitic diodes to one power rail (body-substrate diode) and will conduct large currents when any terminal swings > 0.6 V beyond the substrate rail
on the other rail you may be able to pull some terminals a little beyond the rail without catastrophic short circuit current - but not by enough to be useful - the exception is "Open-Collector" outputs which make really poor linear amps
some devices even form parasitic SCR structures with the substrate, if you exceed the max current limit into some terminals then the chip will try to short the power rails together, and latch up in that condition
in any event exceeding the Vmax will risk avalanche bvreakdown
Spice macro models are really usless for modeling breakdown/reverse bias/crowbar power supply effects, they are simply not designed to model these phenomina
most devices have a parasitic diodes to one power rail (body-substrate diode) and will conduct large currents when any terminal swings > 0.6 V beyond the substrate rail
on the other rail you may be able to pull some terminals a little beyond the rail without catastrophic short circuit current - but not by enough to be useful - the exception is "Open-Collector" outputs which make really poor linear amps
some devices even form parasitic SCR structures with the substrate, if you exceed the max current limit into some terminals then the chip will try to short the power rails together, and latch up in that condition
in any event exceeding the Vmax will risk avalanche bvreakdown
Spice macro models are really usless for modeling breakdown/reverse bias/crowbar power supply effects, they are simply not designed to model these phenomina
PC modeling software is fun to play with and can be a useful tool, but it needs to be tempered with a dose of the real world.
If you want to build something as an experiment, stacking 2 chip amps like that go for it but don't expect it to sound like much. There will be a big crossover notch during the switch over from one chip conducting to the other. Thats if none of the problems JCX brought up never existed, and they do.
40 years of experence does make a difference. I use modeling software too, but I temper my designs with that experence. It's fun to model stuff as a starting point, but if it can't work in the real world it has no point other than playing at the computer.
BZ
If you want to build something as an experiment, stacking 2 chip amps like that go for it but don't expect it to sound like much. There will be a big crossover notch during the switch over from one chip conducting to the other. Thats if none of the problems JCX brought up never existed, and they do.
40 years of experence does make a difference. I use modeling software too, but I temper my designs with that experence. It's fun to model stuff as a starting point, but if it can't work in the real world it has no point other than playing at the computer.
BZ
How about the voltage difference from the ps voltage and what the output will do. Check the specs on some power opamps. somewhere from .7v to 2.5v x 2. The power supply is set up as +60 0 -60 in this example correct? One chip powered by the +60 to 0 volt supply and the other from 0 to -60 correct?
If the chip amp we use can swing to 1.5 volts of the rails then there will be a 3 volt step in the output as the input signal crosses 0.
No amout of feedback can help you here.
We haven't started talking about what the recovery of the chip amp looks like after being clipped, which is what is happening at the 0 crossing in this example.
It's still a bad idea.
BZ
If the chip amp we use can swing to 1.5 volts of the rails then there will be a 3 volt step in the output as the input signal crosses 0.
No amout of feedback can help you here.
We haven't started talking about what the recovery of the chip amp looks like after being clipped, which is what is happening at the 0 crossing in this example.
It's still a bad idea.
BZ
HDTVman, I am not saying you are wrong, as I suspect you are right. I recall however one webpage explaining opamp use fundamentals; there was a class-B output stage on the opamp the fellow was using... the output went from horrible crossover distortion to <0.1% with a wire (follower) for feedback.
If you could avoid clipping, or exploding chips, it could work eh?!
If you could avoid clipping, or exploding chips, it could work eh?!
Stocker said:
The problem is you are clipping as part of the design here. This is not at all like a discrete output stage. The design is being forced to be nonlinear at the 0 crossing because the + chip output can't go below some voltage greater than 0 volts, 1.5 v for example and the - chip can't go more + than -1.5 v. A 3 volt notch you can't avoid and the output can never be 0 volts.
With a class B amp stage at the 0 crossing no current flows and output voltage is 0 but as you start to turn on one of the transistors the voltage starts to increase. Yes you can help the nonlinearity of this type of output with feedback.
I would guess the page you refer to in the quote was using a single opamp in a more normal mode, and yes feedback could help a " Class B " output stage. The basic problem with the described circuit is that in reality it is not " Class B " but a push/pull "Class C " amp near "Class B".
BZ
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.