i did mention a while back the gain would be between 10dB and 20dB. now, that is a pretty wide range, but i left it flexble so we coudl optimize the design. we could get away with very low gain stages if we used two of them (input and output). otherwise a single stage with 12dB sounds good to me.
dorkus said:
Perhaps consider an offset servo in the power amplifier. This would be more beneficial, because the DC offset of the pre-amp and the power amp can be nulled at the same time. This means actual speaker output offsets in the microvolt range are possible.
Tube Sound FX
Any comments/thoughts regarding transformer output coupling ?
I know that good line level transfformers are relatively expensive,
but could they be an icing on the cake ?.
The primary can also be polarity switched to give output polarity switching.
Eric.
Any comments/thoughts regarding transformer output coupling ?
I know that good line level transfformers are relatively expensive,
but could they be an icing on the cake ?.
The primary can also be polarity switched to give output polarity switching.
Eric.
Assuming that CD's are the input, 12dB sounds like a fairly realistic number. I was just curious.
Grey
Grey
Suggestions
Use a balanced topology. There are sonic improvments to be had even if it is in an unbalanced mode.
6 dB ought to be more than enough. Most modern electronics put out a healthy signal. You will probably end up putting some volume trim adjustments in each input to have everything the same level; and you will probably still have more gain than needed.
Cascodes are useful when you have high rail voltages. Should not be an issue in a preamp, regarless of gain.
Jocko
Use a balanced topology. There are sonic improvments to be had even if it is in an unbalanced mode.
6 dB ought to be more than enough. Most modern electronics put out a healthy signal. You will probably end up putting some volume trim adjustments in each input to have everything the same level; and you will probably still have more gain than needed.
Cascodes are useful when you have high rail voltages. Should not be an issue in a preamp, regarless of gain.
Jocko
Convergance
There is no reason that 6 to 26dB of gain cannot be achieved with the diiferential Jfet topology based on the Borbely design. I would suggest +/-24 volt rails and no global negative feedback which can be a bit tricky to finess for different gains and stability. Adjustment for offset would not be necessary for an AC coupled output. With reasonable rail voltages cascode operation would be an option also. A current source connection to the negatve supply is really needed for proper single ended input balanced output. I don't like the use of the 2SK246 for cascode as it does not allow enough voltage from drain to source for the 2SK389. Data sheet analysis show an improvement in output conductance and significant reduction in capacitance as one approaches 10 volts drain to source voltage. I sure would not use less than 5 volts drain to source because of these considerations. I highly recommend the jfet articles on the Borbely site as well as reading the data sheet for the 2SK389.
http://www.borbelyaudio.com/
http://perso.wanadoo.fr/jm.plantefeve/2SK389.pdf
There is no reason that 6 to 26dB of gain cannot be achieved with the diiferential Jfet topology based on the Borbely design. I would suggest +/-24 volt rails and no global negative feedback which can be a bit tricky to finess for different gains and stability. Adjustment for offset would not be necessary for an AC coupled output. With reasonable rail voltages cascode operation would be an option also. A current source connection to the negatve supply is really needed for proper single ended input balanced output. I don't like the use of the 2SK246 for cascode as it does not allow enough voltage from drain to source for the 2SK389. Data sheet analysis show an improvement in output conductance and significant reduction in capacitance as one approaches 10 volts drain to source voltage. I sure would not use less than 5 volts drain to source because of these considerations. I highly recommend the jfet articles on the Borbely site as well as reading the data sheet for the 2SK389.
http://www.borbelyaudio.com/
http://perso.wanadoo.fr/jm.plantefeve/2SK389.pdf
The TDR twins are back!
Finally we are getting somewhere.
How about mosfets for the cascodes and followers? 😉
Jam
P.S. Heck! Why not mosfets for the differential since noise and offset will not be an issue here.
Finally we are getting somewhere.
How about mosfets for the cascodes and followers? 😉
Jam
P.S. Heck! Why not mosfets for the differential since noise and offset will not be an issue here.
Re: Let's go round in circles!
it's ok. as long as we keep a S/N ratio of around -12dB (i.e. 1 constructive post for every 4 not-so-constructive ones) i'll be more than happy. 😛
jam said:
it's ok. as long as we keep a S/N ratio of around -12dB (i.e. 1 constructive post for every 4 not-so-constructive ones) i'll be more than happy. 😛
True, but there is no reason they need 26 dB of gain.
Right. 6 db with no feedback should not be that hard for a point of agreement.
Option, but not necesary. Some folks believe there is undesirable sonic signature to cascodes. Suppose you could put the pads on the boards for the adventurous. Doubt there will ever be convergence on this point.
Jocko
i think harry was only pointing out the flexibility of the implementation. i think 10dB of gain is fine.
as for cascodes or not, etc.... we do not need to converge on every design decision. we only need to agree on what is worth trying. so in the case of cascodes, i can just build two boards, one with and one without it. then i'll listen and see how they sound. others can try this as well. as i've said numerous times, i'm more than willing to entertain different implementations. there's more than one way to skin a MOSFET.
as for cascodes or not, etc.... we do not need to converge on every design decision. we only need to agree on what is worth trying. so in the case of cascodes, i can just build two boards, one with and one without it. then i'll listen and see how they sound. others can try this as well. as i've said numerous times, i'm more than willing to entertain different implementations. there's more than one way to skin a MOSFET.
Jocko and Harry,
Thanks for clear, good posts.
I happen to agree with Jocko that agreement on cascodes, etc. is unlikely, but a PCB layout that would allow for both might be a Good Thing. The only fly in the ointment I can see is that if the circuit is optimized for cascodes--let's say you're counting on a certain amount of feedback to meet objective specs--then not using them will throw the gain/performance out of kilter. Makes for a tricky design problem; surely good for another twenty or thirty pages of discussion.
However, as I believe at least one person has posted previously, the value in this thread may be more in the journey than the destination.
If it's any consolation, I'm trying to include variations in both the X-preamp and Aleph-X circuits and the permutations are killing me.
By the way, I vote for balanced. You can always use a balanced circuit as single-ended (with the option of a switch for switching absolute phase, for instance), but there's no way to use a single-ended circuit as a balanced one. Not to mention noise reduction, etc. To the extent that balanced operation seems to be making slow but steady inroads into sound reproduction, it would also be a more flexible design for the future.
Grey
Thanks for clear, good posts.
I happen to agree with Jocko that agreement on cascodes, etc. is unlikely, but a PCB layout that would allow for both might be a Good Thing. The only fly in the ointment I can see is that if the circuit is optimized for cascodes--let's say you're counting on a certain amount of feedback to meet objective specs--then not using them will throw the gain/performance out of kilter. Makes for a tricky design problem; surely good for another twenty or thirty pages of discussion.
However, as I believe at least one person has posted previously, the value in this thread may be more in the journey than the destination.
If it's any consolation, I'm trying to include variations in both the X-preamp and Aleph-X circuits and the permutations are killing me.
By the way, I vote for balanced. You can always use a balanced circuit as single-ended (with the option of a switch for switching absolute phase, for instance), but there's no way to use a single-ended circuit as a balanced one. Not to mention noise reduction, etc. To the extent that balanced operation seems to be making slow but steady inroads into sound reproduction, it would also be a more flexible design for the future.
Grey
Marc,
I think it's time to fix a few things in stone before this thread meanders off into nothingness.
My guess is that we need to fix in order
a) Basic topology of the design
b) Power Supply
c) Volume control
d) Switching
e) Physical Layout and Interface
f) Circuit Boards
I disagree with the journey and and not the destination comment because the journey so far for the most part has been pointless. Let us get down to brass tacks and create something that could possibly hold it's own with the best out there. We certainly have a pool of tatent that can do it. Here is also a chance to learn from each other.
I feel some of the long standing members of this forum might be intimidated by this and rather maintain the status quo and rather go into long diatribes that mean absolutely nothing, maybe some sort of ego thing. That is why I sometimes find those so called gratuitous images better than someone trying to promote his self importance. It is time to put up or shut up.
So my question to you is where do we go from here? You were brave ( or foolish in your words ) enough to start this and both Harry and myself pointed out the pitfalls you would encounter - time to manage the project.
Regards,
Jam
I think it's time to fix a few things in stone before this thread meanders off into nothingness.
My guess is that we need to fix in order
a) Basic topology of the design
b) Power Supply
c) Volume control
d) Switching
e) Physical Layout and Interface
f) Circuit Boards
I disagree with the journey and and not the destination comment because the journey so far for the most part has been pointless. Let us get down to brass tacks and create something that could possibly hold it's own with the best out there. We certainly have a pool of tatent that can do it. Here is also a chance to learn from each other.
I feel some of the long standing members of this forum might be intimidated by this and rather maintain the status quo and rather go into long diatribes that mean absolutely nothing, maybe some sort of ego thing. That is why I sometimes find those so called gratuitous images better than someone trying to promote his self importance. It is time to put up or shut up.
So my question to you is where do we go from here? You were brave ( or foolish in your words ) enough to start this and both Harry and myself pointed out the pitfalls you would encounter - time to manage the project.
Regards,
Jam
Who said anything about feedback?
If you do not use global feedback, you will not have to worry about optimizing thing as is common in that topology.
You should be able to build this thing on +/- 18 V rails, so you won't need cascodes. You should not need more than 6 dB of gain, so you can use degeneration instead of global feedback to get good distortion numbers. Unless you insist on driving 600 ohm loads, you won't need low output impedance with high current drive capabilities, so you still don't need global feedback. If you do, open-loop buffers are easy and numerous.
So........what is wrong with the Borbely design someone posted? Everyone knows his work, and it probably sounds very good. Bet if you put a current source and shunt regulator ahead of it, it would sound damn good. Try it before you decry it. But I'm not designing it.
Oooooops, said a bad word: "current source".
Jocko
If you do not use global feedback, you will not have to worry about optimizing thing as is common in that topology.
You should be able to build this thing on +/- 18 V rails, so you won't need cascodes. You should not need more than 6 dB of gain, so you can use degeneration instead of global feedback to get good distortion numbers. Unless you insist on driving 600 ohm loads, you won't need low output impedance with high current drive capabilities, so you still don't need global feedback. If you do, open-loop buffers are easy and numerous.
So........what is wrong with the Borbely design someone posted? Everyone knows his work, and it probably sounds very good. Bet if you put a current source and shunt regulator ahead of it, it would sound damn good. Try it before you decry it. But I'm not designing it.
Oooooops, said a bad word: "current source".
Jocko
Digital attenuator
Why do you guys want to try making a 0-70 dB attenuator with 0.5 dB steps? An enourmous pain awaits you.
What is wrong with a 0-70 dB master attenuator with 2.0 dB steps. And a 0-6 dB trim attenuator in 0.5 dB steps for each channel?
Jocko
Why do you guys want to try making a 0-70 dB attenuator with 0.5 dB steps? An enourmous pain awaits you.
What is wrong with a 0-70 dB master attenuator with 2.0 dB steps. And a 0-6 dB trim attenuator in 0.5 dB steps for each channel?
Jocko
I'm certainly not advocating feedback, particularly in large doses. I was just using that as an example.
Actually, I thought the Borbely already had a current source--perhaps I need to go back and look at the circuit again. In the event that someone would want to use the preamp as an unbalanced-to-balanced converter (phase splitter), a current source is pretty much a requirement...something I've always acknowledged as a perfect use for a current source.
Grey
EDIT:
Oh, and a friend of mine uses a CAT Signature preamp that he's modified to do the coarse and fine volume adjustment that Jocko describes (used the balance pot for the fine adj.). He's quite happy with it. I've been toying with the idea for my crossover, but haven't decided yet.
Actually, I thought the Borbely already had a current source--perhaps I need to go back and look at the circuit again. In the event that someone would want to use the preamp as an unbalanced-to-balanced converter (phase splitter), a current source is pretty much a requirement...something I've always acknowledged as a perfect use for a current source.
Grey
EDIT:
Oh, and a friend of mine uses a CAT Signature preamp that he's modified to do the coarse and fine volume adjustment that Jocko describes (used the balance pot for the fine adj.). He's quite happy with it. I've been toying with the idea for my crossover, but haven't decided yet.
Here's my preference.......
Folded cascode LTP JFET input with SE Class A mosfet follower outputs. Digital volume control located between stages. No negative feedback and use degeneration to gain decent specs. Rails would be at least +/- 50V.
This would allow balanced or unbalanced operation.
Just my 2 cents.
Grey, Harry, Jocko, Sonny and others what do you think about this?
Folded cascode LTP JFET input with SE Class A mosfet follower outputs. Digital volume control located between stages. No negative feedback and use degeneration to gain decent specs. Rails would be at least +/- 50V.
This would allow balanced or unbalanced operation.
Just my 2 cents.
Grey, Harry, Jocko, Sonny and others what do you think about this?
- Status
- Not open for further replies.