Somewhere in the Monolithic SuperSymmetry Nelson was teasing us with a Current Feedback version of an X circuit. More to the point an Aleph or AlephX circuit.
The idea was tempting enough to give it a try, at least conceptual. I couldn’t find a similar X circuit with CF on the whole forum. Correct me if someone else tried this before. I’ll be happy to learn.
So I tried to design something. Don’t build it; it will probably not work flawless.
That’s why I putted up here. If there’s enough interest, we might make it a real X Current Feedback amp.
I took bits and pieces from what I could find on the net, took me two weeks to try many different combinations of transistor pairs and finally, this circuit came out as best.
I didn’t add current sources for the moment, wanting to keep the concept simple. I tried with mosfets but didn’t succeed at all.
I can tell for now that the hardest part is taming the differential DC at the outputs. For the moment I can’t get it lower then 6V, which is far too high.
Please correct me in every possible way, I don’t even know if the term CF applies, or even if it’s an X.
I know some will say CF is a dangerous approach but I’m fearless for now.
/Hugo
The idea was tempting enough to give it a try, at least conceptual. I couldn’t find a similar X circuit with CF on the whole forum. Correct me if someone else tried this before. I’ll be happy to learn.
So I tried to design something. Don’t build it; it will probably not work flawless.
That’s why I putted up here. If there’s enough interest, we might make it a real X Current Feedback amp.
I took bits and pieces from what I could find on the net, took me two weeks to try many different combinations of transistor pairs and finally, this circuit came out as best.
I didn’t add current sources for the moment, wanting to keep the concept simple. I tried with mosfets but didn’t succeed at all.
I can tell for now that the hardest part is taming the differential DC at the outputs. For the moment I can’t get it lower then 6V, which is far too high.
Please correct me in every possible way, I don’t even know if the term CF applies, or even if it’s an X.
I know some will say CF is a dangerous approach but I’m fearless for now.
/Hugo
Attachments
Netlist,
Why not remove the Aleph part and make it a full comp. circuit, nice idea though.
Regards,
Jam
Why not remove the Aleph part and make it a full comp. circuit, nice idea though.
Regards,
Jam
jam said:
I think the challenge is to try it with the SE amp.
There are many examples of push-pull designs everywhere and I had the feeling we would end with Nelson's first drawing in the Monolithic Susy thread.
Common guys...shoot!!!
/Hugo - not sure he will be able to hold the bullets 🙂
Two comments.
This circuit needs communication between the Sources of the
input diff pair system in order to be an X
Nothing wrong with current feedback - the caveats about
current feedback are mostly b.s. - I find it very easy to
use, fast and stable.
This circuit needs communication between the Sources of the
input diff pair system in order to be an X
Nothing wrong with current feedback - the caveats about
current feedback are mostly b.s. - I find it very easy to
use, fast and stable.
Not me, yet...Nelson Pass said:
Thanks for declaring this attempt B.S.-proof 😀
Two things puzzle me:
1) You didn’t comment on the approach of the diff pair. May I assume the CF part of the drawing is correct?
2) How can we get rid of the differential DC and get the whole thing stable. Maybe something’s missing.
I added a resistor between the inverting inputs. That would probably make it an X. The scope figures are much worse though.
/Hugo
Well, you simply have placed followers in front of the diff
pair devices, no?
Without some emitter degeneration on the diff pairs, you'll
have an exciting time controlling the bias, since you cannot
assume accruate enough matching.
pair devices, no?
Without some emitter degeneration on the diff pairs, you'll
have an exciting time controlling the bias, since you cannot
assume accruate enough matching.
Followers, diffpair, emitter degeneration, controlling the bias.
You are definitely able to capture all problems in one phrase.
Thanks a lot, I'll have to contemplate a bit now.
/Hugo 🙂
You are definitely able to capture all problems in one phrase.
Thanks a lot, I'll have to contemplate a bit now.
/Hugo 🙂
Hmm...perhaps not a bad idea to start with and see how things behave. But in the end I would prefer discrete.Bernhard said:
/Hugo
I’ve been puzzling around a bit more on this CF thing and among the many things I don’t get, here the most significant ones.
Looking at the principle of CF and VF and trying to integrate it in the AlephX, Grey succeeded in building the AlephX with only two input devices. When trying the same with CF up to eight input devices come to mind as seen in the schematic on page 24 and in the drawing in my first post. Not to mention all the current sources. Found some very complex Accuphase schematics on page two here. This is all far away from the Zen-ish approach of AlephX. Do I see things too complicated?
Also, I cannot find an example of SE application of CF. All are push-pull, something I would like to avoid. Maybe it’s just not possible to apply CF in a SE application?
/Hugo
Looking at the principle of CF and VF and trying to integrate it in the AlephX, Grey succeeded in building the AlephX with only two input devices. When trying the same with CF up to eight input devices come to mind as seen in the schematic on page 24 and in the drawing in my first post. Not to mention all the current sources. Found some very complex Accuphase schematics on page two here. This is all far away from the Zen-ish approach of AlephX. Do I see things too complicated?
Also, I cannot find an example of SE application of CF. All are push-pull, something I would like to avoid. Maybe it’s just not possible to apply CF in a SE application?
/Hugo
Could be, don't remember that, I thought the mono-susy amp was push pull??? Anyway, I'd like to build it discrete for now.
/Hugo - 😕
/Hugo - 😕
I’ll promise not to hammer anyone, be nice and ask things very polite, so while you guys are at it could you have a look at my CF attempts? 😉
I don’t pretend to know much but willing to learn and understand things.
I could copy an existing schematic but as stated earlier, the challenge is trying to make a CF AlephX.
Could someone explain that emitter degeneration issue, is it possible to use only two input transistors per side as shown in the picture or should the schematic of the first post be continued?
Any comments greatly appreciated.
/Hugo
I don’t pretend to know much but willing to learn and understand things.
I could copy an existing schematic but as stated earlier, the challenge is trying to make a CF AlephX.
Could someone explain that emitter degeneration issue, is it possible to use only two input transistors per side as shown in the picture or should the schematic of the first post be continued?
Any comments greatly appreciated.
/Hugo
Attachments
Hugo,
will you stay with 15V rails ?
Then why not simply use a f§$%&#g cfb op amp ?
Or have a look at the internal schematics of the CFBs in the data sheets and try to simplyfy...
Bernhard
will you stay with 15V rails ?
Then why not simply use a f§$%&#g cfb op amp ?
Or have a look at the internal schematics of the CFBs in the data sheets and try to simplyfy...
Bernhard
A possible simplification would be the diamond transistor.Bernhard said:
But I think it has some tempco of ofset issues
Got myself a ‘working’ version.
Not perfect, there will be too many errors I guess but:
The starting point was the CF amplifiers note from National.
This was the most simplified one I found.
I went to Peranders’s site as well and tried to mix what I found in his schematic.
1) Now the DC is under control. It can be adjusted by R33/R34.
2) I think it’s an X: the inputs for the gain devices are cross connected and according to Nelson’s remark I made the communication between the emitters from the input buffers.
I can't find that connection in the normal AlephX. Is it not there, or is it me who doesn't see it?
3) I didn’t bother to make it more complicated than necessary. No cascodes, no current sources for the input devices, keeping the Zen idea in mind.
4) One of the many things I have to figure out is the bandwidth. –3db occurs at about 9.5Khz at the collector of Q5.
Bernhard, thanks, I tried with LM6181 but I really would like to achieve it with simple transistors. Again, the simplicity of Zen has bitten me. 😉
Bricolo,
Please elaborate a bit. Link?, schematic?
/Hugo
Not perfect, there will be too many errors I guess but:
The starting point was the CF amplifiers note from National.
This was the most simplified one I found.
I went to Peranders’s site as well and tried to mix what I found in his schematic.
1) Now the DC is under control. It can be adjusted by R33/R34.
2) I think it’s an X: the inputs for the gain devices are cross connected and according to Nelson’s remark I made the communication between the emitters from the input buffers.
I can't find that connection in the normal AlephX. Is it not there, or is it me who doesn't see it?
3) I didn’t bother to make it more complicated than necessary. No cascodes, no current sources for the input devices, keeping the Zen idea in mind.
4) One of the many things I have to figure out is the bandwidth. –3db occurs at about 9.5Khz at the collector of Q5.
Bernhard, thanks, I tried with LM6181 but I really would like to achieve it with simple transistors. Again, the simplicity of Zen has bitten me. 😉
Bricolo,
Please elaborate a bit. Link?, schematic?
/Hugo
Attachments
- Status
- Not open for further replies.