Steven said:
why do you intend to make dummy circuit work in class A as output stage works in class AB
1.this wouldn't reduce crossover distortion of output
2.this would make splif amplifier less amplifier less efficient than any amp in any class
Charles Altman's splif topology
OFF TOPIC....OFF TOPIC....OFF TOPIC....OFF TOPIC
"Great spirits have always encountered violent opposition from mediocre minds" - Albert Einstein
OFF TOPIC....OFF TOPIC....OFF TOPIC....OFF TOPIC
OFF TOPIC....OFF TOPIC....OFF TOPIC....OFF TOPIC
"Great spirits have always encountered violent opposition from mediocre minds" - Albert Einstein
OFF TOPIC....OFF TOPIC....OFF TOPIC....OFF TOPIC
darkfenriz said:
Ad 1. Right, but operating the dummy stage in class AB neither. It makes no real difference. The current zero crossings of dummy and output stage are on different moments anyway.
Ad 2. Wrong, keeping the Splif amplifier dummy stage in class A by omitting the load gives it the highest efficiency. Adding a load to the dummy stage may move the dummy stage into class (A)B, but then that stage will dissipate more power than without load.
Steven
Steven said:
Steven,
Reviewing all this does bring up the question, why DO we need or want that dummy stage? It doesn't help for xover distortion, it doesn't help control the speaker, so WHAT does it, if anything?
Jan Didden
Current dumping not the same!
The Quad 405 current dumping amplifier does have a superficial resemblence to the composite V-I amplifier mentioned earlier, but it operates open loop. Its second amplifier only tries to current share with the voltage amp.
The V-I composite amplifier uses the current sensing resistor to develop the error feedback voltage for the current output amplifier. The current output amplifier can be thought of as a global feedback amplifier with a constant zero reference signal, and the sense resistor voltage as the error feedback signal. If the basic current amplifier is capable of say 0.1% distortion, then the voltage amplifier will only see something on the order of 0.1% of the load, including any EMF currents fed back from the speaker.
I think this puts the composite amplifier in a totally different ballpark from the Quad 405. Just my humble opinion.
Don B.
The Quad 405 current dumping amplifier does have a superficial resemblence to the composite V-I amplifier mentioned earlier, but it operates open loop. Its second amplifier only tries to current share with the voltage amp.
The V-I composite amplifier uses the current sensing resistor to develop the error feedback voltage for the current output amplifier. The current output amplifier can be thought of as a global feedback amplifier with a constant zero reference signal, and the sense resistor voltage as the error feedback signal. If the basic current amplifier is capable of say 0.1% distortion, then the voltage amplifier will only see something on the order of 0.1% of the load, including any EMF currents fed back from the speaker.
I think this puts the composite amplifier in a totally different ballpark from the Quad 405. Just my humble opinion.
Don B.
Re: Current dumping not the same!
Don,
IIRC (don't have the schematic handy) the 405 doesn't operate open loop, the L-C parts feed back dumper currrent action to the class A stage to lower the class A gain, so as to keep the total amp gain constant, that's really the crux of that design.
Jan Didden
smoking-amp said:
Don,
IIRC (don't have the schematic handy) the 405 doesn't operate open loop, the L-C parts feed back dumper currrent action to the class A stage to lower the class A gain, so as to keep the total amp gain constant, that's really the crux of that design.
Jan Didden
Hi Jan,
I'm looking at page 223 in J. L. Hood's book "Audio Electronics" ( 2nd ed. ) with a simplified schematic of the Quad amplifier. From what I can tell, the feedback is used to ensure constant transconductance in the overall amplifier, mainly because of the large crossover distortion in the unbiased Q1 and Q2 power output stage. I think any modern implimentation of the Quad design would use a biased class AB like stage instead. In that case, the feedback would be enforcing some current ratio between the amplifiers. But, correct me if I'm wrong, but I think this current split is set by a resistor ratio, R3 / R4 in the design, so is really an open loop like setup in effect. The V-I composite ampl. on the other hand sees any current from the V ampl. as an error to be zeroed. (I probably need to analyze the Quad a little more thoroughly with a simulator to be sure. )
Don
I'm looking at page 223 in J. L. Hood's book "Audio Electronics" ( 2nd ed. ) with a simplified schematic of the Quad amplifier. From what I can tell, the feedback is used to ensure constant transconductance in the overall amplifier, mainly because of the large crossover distortion in the unbiased Q1 and Q2 power output stage. I think any modern implimentation of the Quad design would use a biased class AB like stage instead. In that case, the feedback would be enforcing some current ratio between the amplifiers. But, correct me if I'm wrong, but I think this current split is set by a resistor ratio, R3 / R4 in the design, so is really an open loop like setup in effect. The V-I composite ampl. on the other hand sees any current from the V ampl. as an error to be zeroed. (I probably need to analyze the Quad a little more thoroughly with a simulator to be sure. )
Don
smoking-amp said:
Indeed, I said gain but you are right it is really the transconductance IIRC. I don't have the diagram here now, I will look it up.
Jan Didden
janneman said:
Jan,
Personally, I don't think we need that dummy stage. I think the feedback could have been derived from the VAS output if the purpose was to keep the output transistors out of the loop.
Anyway, what it does is linearizing the input stage and VAS of the amplifier without linearizing the output transistors. The reason for doing so could be to close the door for loudspeaker induced currents.
I certainly would not use the Splif for a low biased class AB amplifier, as I don't believe that the dummy stage would mimic the outputs stage to such extent that it improves the linearity of the output stage, including the crossover reagion. This problem is far less severe in a class A amplifer.
Steven
a little daydreaming while out hiking in the woods today
I was just thinking about some other way to remove the effects of speaker EMF feedback from a conventional amplifier (assuming thats useful).
By putting a small current sense resistor in series with the speaker, say on the ground connection side for convenience, one could extract the information of what the speaker is drawing current wise. This spkr fdbk signal would be a simple copy of the input if the speaker were a purely resistive load. The difference between the input signal and the current sense signal would then be the speaker EMF feedback and nonlinearities, and reactance etc. One could just subtract this computed spkr fdbk out from the conventional feedback signal if it were useful to do so.
Now, it seems, this scheme is already well known. Just the same current derived feedback component that some old tube ampls used to use to set a specific output impedance. Sometimes used for setting a critical damping impedance for the speaker. Only takes a few resistors to implement this on any amplifier.
Don
(no, I didn't crash into any trees today while thinking of this brilliant idea!) 😀
I was just thinking about some other way to remove the effects of speaker EMF feedback from a conventional amplifier (assuming thats useful).
By putting a small current sense resistor in series with the speaker, say on the ground connection side for convenience, one could extract the information of what the speaker is drawing current wise. This spkr fdbk signal would be a simple copy of the input if the speaker were a purely resistive load. The difference between the input signal and the current sense signal would then be the speaker EMF feedback and nonlinearities, and reactance etc. One could just subtract this computed spkr fdbk out from the conventional feedback signal if it were useful to do so.
Now, it seems, this scheme is already well known. Just the same current derived feedback component that some old tube ampls used to use to set a specific output impedance. Sometimes used for setting a critical damping impedance for the speaker. Only takes a few resistors to implement this on any amplifier.
Don
(no, I didn't crash into any trees today while thinking of this brilliant idea!) 😀
Re: Charles Altman's splif topology
That's funny, I always like to say the mediocde minds will always encounter violent opposition.
On a more serious note here... I don't see what the problem with back emf is. I am not an engineer here, so please correct me if I'm wrong. I'm here to learn as much as I am to give my two cents.
The amplifier is supposed to be a voltage source with a low output impedance. The way a low output impedance is achieved is with a very high open loop gain, and lots of feedback. Now I totally agree that this isn't the best way to do it because the feedback is delayed, and the more gain stages, the greater the delay. And this delay will cause intermodulation distortion.
Now the way I see the amplifier speaker relationship is this:
The amplifier maintains a voltage multiple of the input at the output. The speaker cone moves with a velocity that is proportional to voltage. As the velocity increases it creates a back emf, that as far as I know, is proportional to the velocity. If the cone moves too slow in relation to the velocity, the back emf is too low. This causes an increase in current, which creates a proportional increases in force on the cone to accelerate it. The opposite happens when the cone velocity is too high in relation to the voltage. As the cone begins to displace, the suspension causes a restoring force on it.
I'm drawing all of this from permanent magnet motor theory ofcouse.
I think the best amplifier for driving a cone speaker is a voltage source.
darkfenriz said:
That's funny, I always like to say the mediocde minds will always encounter violent opposition.
On a more serious note here... I don't see what the problem with back emf is. I am not an engineer here, so please correct me if I'm wrong. I'm here to learn as much as I am to give my two cents.
The amplifier is supposed to be a voltage source with a low output impedance. The way a low output impedance is achieved is with a very high open loop gain, and lots of feedback. Now I totally agree that this isn't the best way to do it because the feedback is delayed, and the more gain stages, the greater the delay. And this delay will cause intermodulation distortion.
Now the way I see the amplifier speaker relationship is this:
The amplifier maintains a voltage multiple of the input at the output. The speaker cone moves with a velocity that is proportional to voltage. As the velocity increases it creates a back emf, that as far as I know, is proportional to the velocity. If the cone moves too slow in relation to the velocity, the back emf is too low. This causes an increase in current, which creates a proportional increases in force on the cone to accelerate it. The opposite happens when the cone velocity is too high in relation to the voltage. As the cone begins to displace, the suspension causes a restoring force on it.
I'm drawing all of this from permanent magnet motor theory ofcouse.
I think the best amplifier for driving a cone speaker is a voltage source.
Re: Re: Charles Altman's splif topology
You wrote something about delay.
Just think of a small room with a speaker producing an acoustic wave. The wave 'bumps' of the wall and comes back to cone after (2*distance to wall)/velocity of sound and moves the cone which procuces voltage.
How do you feel about distortion issue???
I thinnk tha t velocity is proportional to diff(voltage)/diff(time).hummhoom said:
You wrote something about delay.
Just think of a small room with a speaker producing an acoustic wave. The wave 'bumps' of the wall and comes back to cone after (2*distance to wall)/velocity of sound and moves the cone which procuces voltage.
How do you feel about distortion issue???
Re: Re: Charles Altman's splif topology
It's even more hilarious. There are people that, if they meet somebody they don't understand, they think he is a "Great Spirit".
😀
Jan Didden
hummhoom said:
It's even more hilarious. There are people that, if they meet somebody they don't understand, they think he is a "Great Spirit".
😀
Jan Didden
But there are always redeeming features - often mediocre minds are good leaders, or astute businessmen, or imaginative marketers. No-one ever protested Nikola Tesla was an accomplished businessman, after all. Is there an irony here?
Good engineers and insightful intelligence are wonderful qualities, but there is nothing like being proved 'passionately wrong' to bring people back to earth......
I understand speakers can be designed for high or low impedance sources; it is a matter of designing for a particular mechanical Q and, of course, judiciously applying damping with the enclosure.
It bemuses me that we argue about the semantics of Charles' SPLIF concept. It's not global, it's not local, it defies ready description, why not build it and listen to it? If we agonize over what it is, the tail wags the dog......
Cheers,
Hugh
Re: Re: Re: Charles Altman's splif topology
I would classify the splif amplifier as global feedback. It might sneak into a unique subcategory of of its own, but it's still global feedback. You might not be drawing the feedback from the same amplifier that's driving the speaker, but it still comes from the output, and goes into a difference amp. That's my understanding of the circuit.
I'm pretty sure the velocity should be proportional to the voltage, and that diff(velocity)/diff(time) is proportional to diff(voltage)/diff(time). I could be totally out to lunch on this ofcourse.
I would classify the splif amplifier as global feedback. It might sneak into a unique subcategory of of its own, but it's still global feedback. You might not be drawing the feedback from the same amplifier that's driving the speaker, but it still comes from the output, and goes into a difference amp. That's my understanding of the circuit.
darkfenriz said:
I'm pretty sure the velocity should be proportional to the voltage, and that diff(velocity)/diff(time) is proportional to diff(voltage)/diff(time). I could be totally out to lunch on this ofcourse.
Re: Re: Re: Re: Charles Altman's splif topology
think about offset voltage, which is constant. this would cause constant velocity and bend the membrane to infinity with time....
certainly we care a lot to make this dc low, but our situation isn't this tragic😉
hummhoom said:
think about offset voltage, which is constant. this would cause constant velocity and bend the membrane to infinity with time....
certainly we care a lot to make this dc low, but our situation isn't this tragic😉
one more vote for velocity prop. to voltage
I think the voice coil velocity proportional to voltage is right (at least for a low ohmic resistance voice coil). The DC offset voltage just ends up being dissipated as I squared R loss in the voice coil, with the resultant current causing a force against the suspension which gets pushed off center until forces balance.
Was just thinking a little about using a dual voice coil speaker to get separate feedback info on V.C. velocity. (I know! You have to use a carefully balanced external air core xfmr. between the two voice coil leads to cancel out the internal speaker direct mutual inductance coupling between them, to get isolated signals.)
Trying to come up with some way to use in a SPLIF like idea, but I just keep ending up with the usual velocity servo idea.
How about connecting the current amplifier output in the the V-I composite amplifier idea to the second voice coil. Probably don't even need the isolation then. Looks like one gets velocity servo for free then, along with the lack of loading on the voltage amplifier.
Still uses the current sense resistor in the voltage amplifier output (in the ground side output connection most practical) to control the current amplifier for zero current load on the V amplifier. This has the nice effect of zeroing out the ohmic voice coil resistance for the voltage amplifier too. Just call it the superconducting speaker!
Don B.
I think the voice coil velocity proportional to voltage is right (at least for a low ohmic resistance voice coil). The DC offset voltage just ends up being dissipated as I squared R loss in the voice coil, with the resultant current causing a force against the suspension which gets pushed off center until forces balance.
Was just thinking a little about using a dual voice coil speaker to get separate feedback info on V.C. velocity. (I know! You have to use a carefully balanced external air core xfmr. between the two voice coil leads to cancel out the internal speaker direct mutual inductance coupling between them, to get isolated signals.)
Trying to come up with some way to use in a SPLIF like idea, but I just keep ending up with the usual velocity servo idea.
How about connecting the current amplifier output in the the V-I composite amplifier idea to the second voice coil. Probably don't even need the isolation then. Looks like one gets velocity servo for free then, along with the lack of loading on the voltage amplifier.
Still uses the current sense resistor in the voltage amplifier output (in the ground side output connection most practical) to control the current amplifier for zero current load on the V amplifier. This has the nice effect of zeroing out the ohmic voice coil resistance for the voltage amplifier too. Just call it the superconducting speaker!
Don B.
- Status
- Not open for further replies.