Grounding resistors and CM feedback
Grey, I think it might be best to use both resistors from outputs to ground and the bridge feedback to the input diff pair Sources. Otherwise, since the the latter represents a fairly low impedance input, you will need pretty low value CM feedback resistors. Simulation indicates that approx 2K resistors are required to achieve much the same effect as the 30R resistors to ground. Perhaps using both, but with larger values might be the optimim solution?
Grey, I think it might be best to use both resistors from outputs to ground and the bridge feedback to the input diff pair Sources. Otherwise, since the the latter represents a fairly low impedance input, you will need pretty low value CM feedback resistors. Simulation indicates that approx 2K resistors are required to achieve much the same effect as the 30R resistors to ground. Perhaps using both, but with larger values might be the optimim solution?
Nelson,
my post about the prize was all in good fun, I hope you didn't take it the wrong way.
Ian, Grey or whoever understood the secret DC regulator,
there have been many hints about how to implement this control that to my simple mind are not very understandable/useful. Is the Ian's CM circuit useable as shown or what? When all the dust is settled, if it's not too much to ask, it would be nice for someone that knows what time it is to distill all these hints into some type of useable circuit.
my post about the prize was all in good fun, I hope you didn't take it the wrong way.
Ian, Grey or whoever understood the secret DC regulator,
there have been many hints about how to implement this control that to my simple mind are not very understandable/useful. Is the Ian's CM circuit useable as shown or what? When all the dust is settled, if it's not too much to ask, it would be nice for someone that knows what time it is to distill all these hints into some type of useable circuit.
Perhaps a comment here is appropriate. The operation
of the SuperSymmetric topology is not intuitive, which is why it
took me 10 years to work it out, and even after I got
the patent, I was still discovering things about its behaviour,
which is why the Aleph circuit made it to market earlier,
although it was invented later.
The common-mode feedback to the diff pair sources for
reducing absolute DC offset works better than I would have
expected, but there it is. In conjunction with resistive
loading of the output, DC drift is well under control.
of the SuperSymmetric topology is not intuitive, which is why it
took me 10 years to work it out, and even after I got
the patent, I was still discovering things about its behaviour,
which is why the Aleph circuit made it to market earlier,
although it was invented later.
The common-mode feedback to the diff pair sources for
reducing absolute DC offset works better than I would have
expected, but there it is. In conjunction with resistive
loading of the output, DC drift is well under control.
grataku,
I'm afraid I'm going to have to default here for a few days. In addition to a death in the family, followed by even more bad news, then yet more, then more still (egad, what a week)...I sat down to work on the resistor situation (specifically to nail down a good resistance value) and found that my prototype had mysteriously died. Dunno what's wrong yet, but I confess to being somewhat lacking in motivation at the moment. Perhaps when the smoke clears I'll be able to see better.
Rats.
I know that Ian was saying 2k, but I want to test that in the real world, as I believe that he's still working from simulations. My feeling is that unless I can find a value that allows me to completely delete the grounding resistors, there's no point in putting in the resistor bridge. My experience here was that the amp was really quite stable once it was adjusted, so unless the grounding resistors can be removed entirely, there's little to be gained by throwing in more parts.
I'll get there as soon as I can. If anyone else has any other suggestions for resistor values, go ahead and post, and I'll take a look at it...well, whenever...hopefully soon.
Grey
I'm afraid I'm going to have to default here for a few days. In addition to a death in the family, followed by even more bad news, then yet more, then more still (egad, what a week)...I sat down to work on the resistor situation (specifically to nail down a good resistance value) and found that my prototype had mysteriously died. Dunno what's wrong yet, but I confess to being somewhat lacking in motivation at the moment. Perhaps when the smoke clears I'll be able to see better.
Rats.
I know that Ian was saying 2k, but I want to test that in the real world, as I believe that he's still working from simulations. My feeling is that unless I can find a value that allows me to completely delete the grounding resistors, there's no point in putting in the resistor bridge. My experience here was that the amp was really quite stable once it was adjusted, so unless the grounding resistors can be removed entirely, there's little to be gained by throwing in more parts.
I'll get there as soon as I can. If anyone else has any other suggestions for resistor values, go ahead and post, and I'll take a look at it...well, whenever...hopefully soon.
Grey
Grey, Grataku, attached is a simplified schematic based on Grey's basic circuit, but showing the common mode feedback resistors. Grey is correct, so far I have been using a simulator, but it is one that has proved trustworthy, at least in determining whether something will work. Distorion analysis is something else however. The circuit shown has reduced currents (I am building a headphone version) and uses 3.9K CM feedback resistors and no output resistors to ground. This gives about the same performance as the 30R resistors from output to ground on their own. Sensitivity to diff pair input current is about 3.2V/mA. Personally I would use both techniques but with larger resistors but I wanted to prove that it would work without the output resistors. I've also included a simulated frequency response - make of it what you will.
Ian.
P.S. Grey, sorry to hear about your family problems.
Ian.
P.S. Grey, sorry to hear about your family problems.
Aleph X head
Unfortunately it does not seem to be so simple. The 68K resistors from input to ground appear to be essential when the common feedback scheme is employed. Lowering their value improves the stability of the circuit, whilst removing them or placing them as suggested causes the whole thing to become jumpy. Presumably this is because they form the DC reference against which the CM feedback is applied? Whatever the reason, these become much more important than the output resistors to ground with this variation on the topology.
Unfortunately it does not seem to be so simple. The 68K resistors from input to ground appear to be essential when the common feedback scheme is employed. Lowering their value improves the stability of the circuit, whilst removing them or placing them as suggested causes the whole thing to become jumpy. Presumably this is because they form the DC reference against which the CM feedback is applied? Whatever the reason, these become much more important than the output resistors to ground with this variation on the topology.
placing them as suggested causes the whole thing to become jumpy
Hmmm.... If the 10K input resistors are AC coupled, moving the 68Ks to input sides would look like 78K to ground instead of 68K to ground for the input feedback resistors at DC. That really drives the circuit nuts? Do you also have the 2K resistors to the outputs and the 30 Ohms to ground for your model. I actually like the idea of these 30 ohm resistors as they are also helping to buffer to open loop output gain against variations in impedance with frequency. Particularly at bass resonance frequency. I am tempted to drop them further down in value.
H.H.
Hmmm.... If the 10K input resistors are AC coupled, moving the 68Ks to input sides would look like 78K to ground instead of 68K to ground for the input feedback resistors at DC. That really drives the circuit nuts? Do you also have the 2K resistors to the outputs and the 30 Ohms to ground for your model. I actually like the idea of these 30 ohm resistors as they are also helping to buffer to open loop output gain against variations in impedance with frequency. Particularly at bass resonance frequency. I am tempted to drop them further down in value.
H.H.
Anyone tried to build the aleph x using the pcb provided by grey ?
I have just finished one channel, but it seems impossible to adjust the dc offset, the outputs have 6-8v, any attempt to adjust, will just change the polarity to the opposite, so the voltage is 6-8 v negative
I buildt the other channel on veroboard, this channel behaves normally. Is there any bugs on the pcb provided by grey?
I have just finished one channel, but it seems impossible to adjust the dc offset, the outputs have 6-8v, any attempt to adjust, will just change the polarity to the opposite, so the voltage is 6-8 v negative
I buildt the other channel on veroboard, this channel behaves normally. Is there any bugs on the pcb provided by grey?
Um...I know I'm a little preoccupied at the moment, but I don't recall posting a board layout. Someone else posted one, but I haven't done more than glance at it. Whether it works or not, I don't know. I'm still working on my own.
As for DC offset, what values are you using for R24, R26, and V2? The original ones or the newer values I posted later? The original values will give a wider range of adjustment. The newer ones will give you more control over a narrower range. I'm using a multi-turn pot here, which helps.
I partially dismantled and rebuilt the circuit (still on push boards) and managed to shoot myself in the foot. I thought I'd get rid of a few jumpers. All well and good. But in so doing, I ended up with the Source resistors closer to the active devices. The heat is driving them crazy, so I'm having trouble maintaining the DC balance I was able to get so easily with the other configuration. When I have time, I'll re-rebuild the circuit and try to get it back the way it was, because this is intolerable.
Or perhaps I'll just forge on ahead and try to put together a board, even if it ends up being double-sided. There are far too many connections and bare wires flopping about for my comfort.
Grey
As for DC offset, what values are you using for R24, R26, and V2? The original ones or the newer values I posted later? The original values will give a wider range of adjustment. The newer ones will give you more control over a narrower range. I'm using a multi-turn pot here, which helps.
I partially dismantled and rebuilt the circuit (still on push boards) and managed to shoot myself in the foot. I thought I'd get rid of a few jumpers. All well and good. But in so doing, I ended up with the Source resistors closer to the active devices. The heat is driving them crazy, so I'm having trouble maintaining the DC balance I was able to get so easily with the other configuration. When I have time, I'll re-rebuild the circuit and try to get it back the way it was, because this is intolerable.
Or perhaps I'll just forge on ahead and try to put together a board, even if it ends up being double-sided. There are far too many connections and bare wires flopping about for my comfort.
Grey