I had looked at both of the ideas that Harry posted above. I think some variation of the first one should work, but the second, I believe, would not. Given that the node between R11 and R12 is a virtual ground of some sort, you are effectively shorting the Source of the current source MOSFET to ground. My solution to that was to use a Zener to offset the voltage at the Source, but then the bias current required to light up the Zener becomes a problem in and of itself, so I abandoned the concept. (Maybe Ian could breathe life back into the idea by using a MOSFET instead of a Zener, sorta like a Vgs multiplier circuit.)
My current thinking (pun intended) is to turn the whole thing upside down and use an N-ch MOSFET for the current source. Two 10k resistors for the bridge to the output. Two resistors (optional pot to fine-tune) form a voltage divider to the positive rail from the virtual ground. Set the values of those resistors such that it biases the current source MOSFET Gate to proper conductance. I've got some scratch values around here somewhere, but that scrap of paper seems to have wandered off.
Once you start thinking in terms of N-ch devices for the current source, it opens up all kinds of avenues, some of which I've discarded, some of which I still think might be worth pursuing. Yes, I know taking the current from the Source is an imperfect solution, impedance-wise, but with proper selection of resistor values, it should suffice.
Grey
My current thinking (pun intended) is to turn the whole thing upside down and use an N-ch MOSFET for the current source. Two 10k resistors for the bridge to the output. Two resistors (optional pot to fine-tune) form a voltage divider to the positive rail from the virtual ground. Set the values of those resistors such that it biases the current source MOSFET Gate to proper conductance. I've got some scratch values around here somewhere, but that scrap of paper seems to have wandered off.
Once you start thinking in terms of N-ch devices for the current source, it opens up all kinds of avenues, some of which I've discarded, some of which I still think might be worth pursuing. Yes, I know taking the current from the Source is an imperfect solution, impedance-wise, but with proper selection of resistor values, it should suffice.
Grey
You are doing it again, Harry
herm
I don't think so. I just prefer to disscuss actual circuits rather than speculative comparisons to what someone "might" be doing, sight unseen.
Grey
"Given that the node between R11 and R12 is a virtual ground of some sort, you are effectively shorting the Source of the current source MOSFET to ground."
I don't think so..... It is two resistors. The junction has zero AC volts and an impedance to ground of one half of R11. This impendance appears in parallel with R7. How hard would this be to try before deciding how it doesn't work?
H.H.
herm
I don't think so. I just prefer to disscuss actual circuits rather than speculative comparisons to what someone "might" be doing, sight unseen.
Grey
"Given that the node between R11 and R12 is a virtual ground of some sort, you are effectively shorting the Source of the current source MOSFET to ground."
I don't think so..... It is two resistors. The junction has zero AC volts and an impedance to ground of one half of R11. This impendance appears in parallel with R7. How hard would this be to try before deciding how it doesn't work?
H.H.
I guess the really clever bit is knowing one's good ideas from the bad. I couldn't decide whether my idea (which apparently isn't original according to Joe Berry) had any deleterious effects - that's why I am going to build some different versions and have a listen.
In any case, so we're clear, resistors from both outputs
can be fed to the node which attaches the two Sources
for the diff pair and the Drain of their current source.
Works quite well, as any absolute DC offset results
in opposing bias to the diff pair. This technique also
presumes a DC impedance to ground associated with
the input so that the feedback is not unity gain at DC.
can be fed to the node which attaches the two Sources
for the diff pair and the Drain of their current source.
Works quite well, as any absolute DC offset results
in opposing bias to the diff pair. This technique also
presumes a DC impedance to ground associated with
the input so that the feedback is not unity gain at DC.
originality cont'd
Ian, sorry if I wasn't clear earilier, but your idea of returning a common mode signal to the DRAIN of the CCS is in fact not identical to Hadley or other prior circuits I've seen. My reference was to returning a common mode signal to the GATE or control terminal of the CCS. Returning the common mode signal to the drain bypasses the CCS as a control element and turns the job over to the diff pair itself.
Or, as Emily Litella would say, "Oh, that's different. Never mind."
Ian, sorry if I wasn't clear earilier, but your idea of returning a common mode signal to the DRAIN of the CCS is in fact not identical to Hadley or other prior circuits I've seen. My reference was to returning a common mode signal to the GATE or control terminal of the CCS. Returning the common mode signal to the drain bypasses the CCS as a control element and turns the job over to the diff pair itself.
Or, as Emily Litella would say, "Oh, that's different. Never mind."
you think I might actually send you?
You know..... An autographed picture of the XA200 wouldn't cost too much to ship to England, but he would probably want a shot of the inside. I think everyone got a prize with the advice on the offset circuit. If the winner (or runner up lets say...) lived in the United States you could send him some cosmetically blemished heatsinks and/or chassis. He might even pay the shipping. He would probably settle for a XA200 picture and your continued advice on the forum though......
H.H.
You know..... An autographed picture of the XA200 wouldn't cost too much to ship to England, but he would probably want a shot of the inside. I think everyone got a prize with the advice on the offset circuit. If the winner (or runner up lets say...) lived in the United States you could send him some cosmetically blemished heatsinks and/or chassis. He might even pay the shipping. He would probably settle for a XA200 picture and your continued advice on the forum though......
H.H.
Okay, let me try to catch up, here.
Regarding the discussion of bias earlier, the rails are half of the voltage you'd see on a 'normal' Aleph, but the current is double. Purely by the laws of Thermodynamics, it's a wash. As to whether you'll get lower distortion from changing the rail voltage (higher is better, I believe) or the bias current (higher is also better)...I dunno. Perhaps Nelson could comment as to whether the tradeoffs favor one path over the other when reality (say, device dissipation) won't let you have both at the same time.
Ian,
Per your inquiry about the input impedance, imagine that you're an electron looking into the input. You see 10k, right off the bat, in series. Okay, tally 10k on the sheet. Once you travel through the 10k resistor (R10, or R28, your choice), you come to a node with the MOSFET Gate (plus the nominal 221 ohm resistance, hardly significant for present purposes), 68.1k to ground, and 100k out the back door (the output). We'll ignore the Zener diodes for the moment. The 68.1k to ground is fairly easy to see. The 100k to the output a little less so, perhaps...until we factor in the 30 ohm resistor to ground. (I'm still working from the original schematic, which doesn't reflect all this recent stuff.) The 30 ohm resistor to ground, like the 221 ohms into the Gate, kinda gets lost in the shuffle, so at this point, we're left with 68.1k and 100k in parallel. This is something like 40k. Add this to the 10k from the input resistor, and we end up with about 50k, plus or minus a smackeral.
But what about the MOSFET Gate?
Remember, you're still an electron. Are you actually going <i>in</i> there? No. Gate currents for a MOSFET are insignificant. True, the Gate is at ground potential, but it's not really an option for an electron looking for a way to get home. What's the Zin on a MOSFET? 10M or something? I don't remember (it's late, I'm tired, and I've had a rough two weeks, so no yak from the peanut gallery), but it's very, very high indeed. Saying that it's a virtual ground just means that it's at ground potential, not that you can actually push an electron in there. The situation would be different, of course, if we we talking about bipolars, as the emitter would be a valid option for an electron looking for a way out--but that's reflected in the much lower Zin for a bipolar emitter.
Basically, what it boils down to is that the electrons have to exit via the 68.1k or the 100k, or else pile up in front of the MOSFET Gate until overcrowding stops all incoming electrons from joining the party.
Harry,
My choice of the word 'short' was unfortunate in that it brings to mind images of sparks and such. I plead distraction; like I said, trying times hereabouts. Wouldn't putting the bridge to the Source of the current source be the wrong phase?
The ironic thing about this is that I had something like what I think Ian and Nelson are describing in my notes (as well as bridge to current source Gate--no, Harry, I don't recall that I had a bridge to Source, that one's all yours), but convinced myself that it wouldn't work. Hmmm. Makes me wonder how many other ideas I passed by along the way. Oh well, all's well that ends well, as the Bard said.
I prodded at the idea for a few minutes this afternoon before coming to work with inconclusive results. I'll take another whack at it tomorrow if nothing else comes unstuck in my life, as I like the idea of getting rid of those pesky grounding resistors.
Thanks, Ian and Nelson. Ian, you said something about being able to tell the good ideas from the bad ones--I reckon I failed on that one. <i>Sigh</i>
Grey
Regarding the discussion of bias earlier, the rails are half of the voltage you'd see on a 'normal' Aleph, but the current is double. Purely by the laws of Thermodynamics, it's a wash. As to whether you'll get lower distortion from changing the rail voltage (higher is better, I believe) or the bias current (higher is also better)...I dunno. Perhaps Nelson could comment as to whether the tradeoffs favor one path over the other when reality (say, device dissipation) won't let you have both at the same time.
Ian,
Per your inquiry about the input impedance, imagine that you're an electron looking into the input. You see 10k, right off the bat, in series. Okay, tally 10k on the sheet. Once you travel through the 10k resistor (R10, or R28, your choice), you come to a node with the MOSFET Gate (plus the nominal 221 ohm resistance, hardly significant for present purposes), 68.1k to ground, and 100k out the back door (the output). We'll ignore the Zener diodes for the moment. The 68.1k to ground is fairly easy to see. The 100k to the output a little less so, perhaps...until we factor in the 30 ohm resistor to ground. (I'm still working from the original schematic, which doesn't reflect all this recent stuff.) The 30 ohm resistor to ground, like the 221 ohms into the Gate, kinda gets lost in the shuffle, so at this point, we're left with 68.1k and 100k in parallel. This is something like 40k. Add this to the 10k from the input resistor, and we end up with about 50k, plus or minus a smackeral.
But what about the MOSFET Gate?
Remember, you're still an electron. Are you actually going <i>in</i> there? No. Gate currents for a MOSFET are insignificant. True, the Gate is at ground potential, but it's not really an option for an electron looking for a way to get home. What's the Zin on a MOSFET? 10M or something? I don't remember (it's late, I'm tired, and I've had a rough two weeks, so no yak from the peanut gallery), but it's very, very high indeed. Saying that it's a virtual ground just means that it's at ground potential, not that you can actually push an electron in there. The situation would be different, of course, if we we talking about bipolars, as the emitter would be a valid option for an electron looking for a way out--but that's reflected in the much lower Zin for a bipolar emitter.
Basically, what it boils down to is that the electrons have to exit via the 68.1k or the 100k, or else pile up in front of the MOSFET Gate until overcrowding stops all incoming electrons from joining the party.
Harry,
My choice of the word 'short' was unfortunate in that it brings to mind images of sparks and such. I plead distraction; like I said, trying times hereabouts. Wouldn't putting the bridge to the Source of the current source be the wrong phase?
The ironic thing about this is that I had something like what I think Ian and Nelson are describing in my notes (as well as bridge to current source Gate--no, Harry, I don't recall that I had a bridge to Source, that one's all yours), but convinced myself that it wouldn't work. Hmmm. Makes me wonder how many other ideas I passed by along the way. Oh well, all's well that ends well, as the Bard said.
I prodded at the idea for a few minutes this afternoon before coming to work with inconclusive results. I'll take another whack at it tomorrow if nothing else comes unstuck in my life, as I like the idea of getting rid of those pesky grounding resistors.
Thanks, Ian and Nelson. Ian, you said something about being able to tell the good ideas from the bad ones--I reckon I failed on that one. <i>Sigh</i>
Grey
Ah...I forgot...had some e-mail asking about setting the pots, and figured I would answer here so that anyone else who might have the same questions would be able to find the answers. My fault for not being clearer earlier.
V1 and V3 fine tune the bias current. In keeping with the traditional Aleph values, we're looking for a nominal .5V across the Source resistors of the output MOSFETs. Regardless of what you set it to, make sure both sides match. You'll find that the pots give you a fair amount of latitude in setting the bias. As a general rule, more bias current is a good thing, but remember to keep an eye on the power being dissipated in your output devices; don't want to cook them.
V2 sets the absolute DC offset. Attach a test lead to one output, the other to ground. Adjust V2 to get the output-to-ground reading as close to 0V as possible. Let the amp warm up for at least 30 minutes. An hour is better. Readjust. Yes, it will drift a little afterwards, don't worry about it.
I think I'm caught up now.
Maybe.
Grey
V1 and V3 fine tune the bias current. In keeping with the traditional Aleph values, we're looking for a nominal .5V across the Source resistors of the output MOSFETs. Regardless of what you set it to, make sure both sides match. You'll find that the pots give you a fair amount of latitude in setting the bias. As a general rule, more bias current is a good thing, but remember to keep an eye on the power being dissipated in your output devices; don't want to cook them.
V2 sets the absolute DC offset. Attach a test lead to one output, the other to ground. Adjust V2 to get the output-to-ground reading as close to 0V as possible. Let the amp warm up for at least 30 minutes. An hour is better. Readjust. Yes, it will drift a little afterwards, don't worry about it.
I think I'm caught up now.
Maybe.
Grey
input impedance
The gates to the input diff pair are virtual ground by the nature of negative feedback so the input impedance is about 10K for each input with respect to ground.
"Wouldn't putting the bridge to the Source of the current source be the wrong phase?"
No.
I hope things get better with your trying times and wish you the best. I went through two open heart surgeries on my mother about this time last year and know that there is nothing more draining than the illness of a loved one,She is doing quite well now though.
H.H.
The gates to the input diff pair are virtual ground by the nature of negative feedback so the input impedance is about 10K for each input with respect to ground.
"Wouldn't putting the bridge to the Source of the current source be the wrong phase?"
No.
I hope things get better with your trying times and wish you the best. I went through two open heart surgeries on my mother about this time last year and know that there is nothing more draining than the illness of a loved one,She is doing quite well now though.
H.H.
Prize
Just to put this one to bed. First off I wasn't suggesting that Nelson make any hints public domain, else why would it be a prize since everyone would have it? Also it was more of wish than a serious request. Actually, I am more than happy to settle for Nelson's continued support for our DIY efforts. And by the way, I made my contribution before there was any mention of a prize.
Just to put this one to bed. First off I wasn't suggesting that Nelson make any hints public domain, else why would it be a prize since everyone would have it? Also it was more of wish than a serious request. Actually, I am more than happy to settle for Nelson's continued support for our DIY efforts. And by the way, I made my contribution before there was any mention of a prize.