Nelson Pass said:
CS with gain? Are we talking about a current mirror here or something more exotic?
2.5 stages? I suppose I could go for that given that one is only concerned with current gain 🙂
Ian Macmillan said:
I understood that Papa talks about cascode (CS) , not CCS .........forgive me if I'm wrong
CS = cascode?
I suspect you are right but I can't say I have come across the notion of a cascode with current gain. This sounds interesting and I will have to think about it a bit more.
Ian.
I suspect you are right but I can't say I have come across the notion of a cascode with current gain. This sounds interesting and I will have to think about it a bit more.
Ian.
Ian Macmillan said:
you know - I'm to dumb to answer this without looking at precious papers ; but - these papers are not near me in this moment , so I presume that someone will answer before me
ZM Dumb(o)
I believe that this refers to a "common source" connection for a MOSFET or JFET. In this application I would guess that the CS would be the output side of a current mirror. The operational word here is "guess".
Graeme
Graeme
Sorry, I meant Common Source, where the input is the Gate,
the Source goes to the supply rail through a resistor, and the
signal comes off the Drain.
A cascode would be a Common Gate, and a follower is a
Common Drain.
😎
the Source goes to the supply rail through a resistor, and the
signal comes off the Drain.
A cascode would be a Common Gate, and a follower is a
Common Drain.
😎
Nelson Pass said:
Thank you for the clarification Mr Pass.
I think we have already been given more hints that we could really expect and certainlly enough to get started, so I think we need to move to the next stage and start developing some schematics. I believe Grey is already planning to do this and I will try and find some time this weekend to have a go too. It will be interesting to see how these and of course the real XA.5 differ. Assuming we come anywhere close...
Ian.
It's been a helluva week. I'm running behind on everything and it doesn't look as though it's going to lighten up for at least a week or two...or three...
I'll try to get some time tonight if things are quiet at work.
Grey
I'll try to get some time tonight if things are quiet at work.
Grey
GRollins said:
I am sure nobody doubted you would Grey - you have a good reputation on this forum and have contributed more than most. Unfortunately, most of us have full time jobs and other committments that prevent us spending as much time as we would like on this very interesting hobby of ours.
Ian.
Seems I get to take first shot at posting a schematic. Well here goes for my first attempt which I will admit right now that I haven’t built (yet). Nonetheless I am fairly confident that it will work although it is harder to guess at how well. I’m also not claiming any great credit for this design being as it is essentially a cobble up from the X5, Aleph0 plus some UGS stuff here on the forum. Don’t pay too much attention to the choice of devices either, for example the ZVP2106 would be running rather close to its maximum voltage and hence ZVP2110 might well be better. Same goes for the N-channel device.
Note that the output stage (the components running on the lower voltage supply) needs to be replicated as appropriate for the power output you require. I was anticipating have 6 sets of each, i.e. a total of 36 devices including the SE bias devices.
I’ve arranged the JFET diff-pair front end to run at around 10mA in total (5mA per side) and the drivers (the ZVP and ZVN devices) are running at about 30mA as the Master says he arranged for a current gain of 5. I believe this should be enough to drive up to six sets of output devices.
I’ve no idea whether this circuit is even vaguely like the latest XA.5 amplifiers but I hope it is reasonably close given the various hints that have already appeared. Either way I am sure if could use a good few tweaks and I trust these will be forthcoming shortly as there are others more expert in this hobby than I who usually contribute to this forum. I look forward to some interesting discussion…
Ian.
Note that the output stage (the components running on the lower voltage supply) needs to be replicated as appropriate for the power output you require. I was anticipating have 6 sets of each, i.e. a total of 36 devices including the SE bias devices.
I’ve arranged the JFET diff-pair front end to run at around 10mA in total (5mA per side) and the drivers (the ZVP and ZVN devices) are running at about 30mA as the Master says he arranged for a current gain of 5. I believe this should be enough to drive up to six sets of output devices.
I’ve no idea whether this circuit is even vaguely like the latest XA.5 amplifiers but I hope it is reasonably close given the various hints that have already appeared. Either way I am sure if could use a good few tweaks and I trust these will be forthcoming shortly as there are others more expert in this hobby than I who usually contribute to this forum. I look forward to some interesting discussion…
Ian.
Hi Ian,
Perhaps you wouldn't mind offering a bit of a tutorial for those of us who aren't in your league?
🙂
Perhaps you wouldn't mind offering a bit of a tutorial for those of us who aren't in your league?
🙂
carpenter said:
I don't my league is that special - I'm just a hobbyist 🙂 I'll see if I can find a few minutes to put together a simple explanation though. I did consider it yesterday but ran out of time and just posted the schematic whilst it was fresh. I am sure that there must be others more expert than me who can offer some improvements though...
Ian.
Hi Ian,
Thank you for "having a bash" at it. Very nice indeed.
A couple of thoughts:
1) I suspect that a follower design does not require the heavy loading on the amp output to enhance absolute offset stability. I would think that more modestly sized load resistors to ground off the second stage mosfet outputs would be more appropriate.
2) I suspect that the second stage is a current mirror with a current gain of 5 - not a simple CS stage as shown.
3) I would scale up a number of the resistor values. 10K and 390K for the gain setting resistors and 22K for the Macmillan resistors for starters.
So are there any plans for a proto?
Cheers,
Graeme
Thank you for "having a bash" at it. Very nice indeed.
A couple of thoughts:
1) I suspect that a follower design does not require the heavy loading on the amp output to enhance absolute offset stability. I would think that more modestly sized load resistors to ground off the second stage mosfet outputs would be more appropriate.
2) I suspect that the second stage is a current mirror with a current gain of 5 - not a simple CS stage as shown.
3) I would scale up a number of the resistor values. 10K and 390K for the gain setting resistors and 22K for the Macmillan resistors for starters.
So are there any plans for a proto?
Cheers,
Graeme
Thank you Graeme for the kind words.
I haven't yet explored the common mode DC stability of this design so can't really comment on the need for the load resistors at the output. I did wonder about the need and since your have raised the topic I will investigate further.
Current mirror as the second stage. Yes, I wondered about this as well since the Master referred to a current gain of 5. However, this would result in a lower value of resistor in the front end (since it no longer carries the Vgs voltage) and would require more gain from the 2nd stage. I couldn't convince myself that this would be an improvement but I haven't experimented with this option.
Scaling up the resistor values? Perhaps as the values shown are very low but we would need to see what this would do the high frequency response which is currently fairly respectable.
Plans for a proto? Of course 🙂 Just need to get hold of some 2sk389 and 2skJ109 devices which is a bit tricky at the moment. Ian Mackenzie (our friend in Oz) has also volunteered to bread board (he has the devices).
Ian.
I haven't yet explored the common mode DC stability of this design so can't really comment on the need for the load resistors at the output. I did wonder about the need and since your have raised the topic I will investigate further.
Current mirror as the second stage. Yes, I wondered about this as well since the Master referred to a current gain of 5. However, this would result in a lower value of resistor in the front end (since it no longer carries the Vgs voltage) and would require more gain from the 2nd stage. I couldn't convince myself that this would be an improvement but I haven't experimented with this option.
Scaling up the resistor values? Perhaps as the values shown are very low but we would need to see what this would do the high frequency response which is currently fairly respectable.
Plans for a proto? Of course 🙂 Just need to get hold of some 2sk389 and 2skJ109 devices which is a bit tricky at the moment. Ian Mackenzie (our friend in Oz) has also volunteered to bread board (he has the devices).
Ian.
Hi Ian,
I forgot to add R11 and R22 to the list. I think they should be brought up to 10K to 22K or so.
Let me check my transistor supply. I might be able to help you with those parts. There's someone about 30km south of me that sits on a mountain of the things. Maybe "he" could spare a couple?
Graeme
I forgot to add R11 and R22 to the list. I think they should be brought up to 10K to 22K or so.
Let me check my transistor supply. I might be able to help you with those parts. There's someone about 30km south of me that sits on a mountain of the things. Maybe "he" could spare a couple?
Graeme
Hi Graeme,
I've experimented with your suggestions and was pleasantly suprised by the results.
Removing the output common mode load altogether (not replacing at output of Vas) and upping the Macmillan resistors to 22K shows the circuit to remain reasonably DC stable.
I've increased the gain resistors as suggested and R11/R22 and although there is a decrease in the high frequency response, it still looks good to about 150kHz.
I also tried the current mirror idea and it seems that the resistors can remain at roughly the same value (must have been a flaw in my earlier reasoning). This makes no obvious difference to the performance of the circuit in terms of simulation but I expect a real version might be different. Time to move to a prototype perhaps? Can you shed any light on what the anticipated benefit of the current mirror might be? BTW, I expermented with using the same MOSFETs as before not wishing to introduce bipolars into this part of the circuit. Probably just my particular prejudice.
Ian.
I've experimented with your suggestions and was pleasantly suprised by the results.
Removing the output common mode load altogether (not replacing at output of Vas) and upping the Macmillan resistors to 22K shows the circuit to remain reasonably DC stable.
I've increased the gain resistors as suggested and R11/R22 and although there is a decrease in the high frequency response, it still looks good to about 150kHz.
I also tried the current mirror idea and it seems that the resistors can remain at roughly the same value (must have been a flaw in my earlier reasoning). This makes no obvious difference to the performance of the circuit in terms of simulation but I expect a real version might be different. Time to move to a prototype perhaps? Can you shed any light on what the anticipated benefit of the current mirror might be? BTW, I expermented with using the same MOSFETs as before not wishing to introduce bipolars into this part of the circuit. Probably just my particular prejudice.
Ian.
