The paradise is a hard to beat design.. 4 brains and a lot of experience... came together in the best creative process.
Noise is always a problem in balanced.
My FPS has a 3dB noise disadvantage only though so the input stage i use there could help.
My FPS has a 3dB noise disadvantage only though so the input stage i use there could help.
ok this doesn't help 🙂 !
We are in 3 but really not....anyway.....if people want to join in...they are more than welcome!!!
Let's try to carry this on with whoever is interested!
EDIT: this is probably going to be a more complex desing than paradise as you surely have realized already! With this thought in mind...it is obviously going to require more efforts too...just as an aside note.
Is the front END a balance simpler version of the second version I have put on this attempt?
With this regard, would you consider an input stage like that one followed by transimpedance just swapping the time constants?
I am also wondering, I get the issue of Audio Linear, but if you don't post the topology here, I mean why could I?
I mean, I don't know copyrights and stuff, but I think you could be posting a general without values structure just for the purpose of this thread.
Not sure how it works...anyway...want to make sure that if you don't post it I can post it.
I still don't understand why you don't go into a LTP design with the interfet parts. Has it been possible for you to get hold of models for them....??
are you referring to my conversion of the transimpedance to the LTP configuration?
Yes I would love too. I got a hold of them, pricing and everything.
Models are not avaialable and the parts it's not avaialable for at least other 12weeks...that is why.
Plus this is my though and please feel free to comment on it:
it is true that IF part has a 20time the gain of k170, however it has also more than 20times the capacitance of it.
In few words, wouldn't IF part be exactly equal to parallel 20 K170?
From a sonic standpoint and from capacitance thus speed stanpoint as well.
Am I missing something?
It is more or less this way. When you drive up Gm in a J-Fet you drive up capacitance.
The high Gm devices are so called wide gate and that drives up the capacitance up of physical reasons. That can not be eroded totally but some devices have a more favorable combination of high Gm and low input capacitance. One of this more modern ones is the BF862. So i would simply parallel 8 of them per side of the MiiB One Shot so you end up with around 0.5nV/qHz, the best you can hope for in a balanced input.
Do not forget to put 1uH coils in each gate ore you end up with an oscillator.
The Interfet part may be nothing less then a parallel arrangement of may Fets on one die. I am not afraid to parallel many Fets any more. Many years ago i thought also like you, that this effects the transient response especially in the bass. The outstanding bass in the Paradise proves otherwise.
The high Gm devices are so called wide gate and that drives up the capacitance up of physical reasons. That can not be eroded totally but some devices have a more favorable combination of high Gm and low input capacitance. One of this more modern ones is the BF862. So i would simply parallel 8 of them per side of the MiiB One Shot so you end up with around 0.5nV/qHz, the best you can hope for in a balanced input.
Do not forget to put 1uH coils in each gate ore you end up with an oscillator.
The Interfet part may be nothing less then a parallel arrangement of may Fets on one die. I am not afraid to parallel many Fets any more. Many years ago i thought also like you, that this effects the transient response especially in the bass. The outstanding bass in the Paradise proves otherwise.
Interesting!
why then instead of using it on the one shot design I don't try to parallel more on the only N-channel version of masterpiece which I think it R7 (attached here for reference) this is a fully balance topology already and I have custom values for the RIAA so that I could make a direct comparison against Paradise.
Thoughts on this?
To comment on parallel, when paralleling more fet ini my experience I don't find the bass to get worse, rather the opposite.
What I feel I loose by that is the air around instrument the openess the directness of the sound. But bass in all cases increases and sounds more muscolar and powerful.
Paradise is a good compromise of bass and highs all around.
There is no way around paralleling when you want to use J-Fets balanced unless you get yourself some Interfets. I can not second your opinion. I made several Fet stages, all parallel symmetric, some commercial, that have 4 x 2SJ74 plus 4 x 2SK170 that sound wounderful in the midrange and treble.
This is what I don't understand. Why wouldn't paralleling more BFs would equal having interfet from a gain standpoint?
Because Gm doubles with each doubling of Fets. The Interfet inside may just be several JFets in parallel on the same chip.
Prof. Faulkner in England was the first that had the paralleling idea in the 60th, it was with BJT at the time but equally valid with Fets. Noise impedance falls by 3dB with each doubling and Gm goes up by a factor of two.
Maybe you should study a book like this :Browse Results - SpringerLink
Maybe you should study a book like this :Browse Results - SpringerLink
Right...so 10 pairs of bf would have 450mS gm so what I was saying what is the difference...beside matching and thermal tracking, between paralleling 10 pairs of BFs and having a pair of IF? I don't get it..
The difference is that one is a BJt design and the other a Fet design so you can direct couple the Fet without problem plus self bias so no elcaps like in the Paradise .
The BJT version will likely be better in terms of 1/F noise so the bass is cleaner of noise.
Fans of Fet argue that this difference is swapped because of psycho acoustics ( A filtering ).i can hear the difference, ear on speaker, but in the listening space my Fet phonos are dead quiet. The BF862 is very good on 1/F noise by the way. Better then the 2SK170 plus lower input capacitance. It is certainly my first choice for low noise phono with only N-channel. how good the 1/F noise is on the Interfet i can not tell.
The BJT version will likely be better in terms of 1/F noise so the bass is cleaner of noise.
Fans of Fet argue that this difference is swapped because of psycho acoustics ( A filtering ).i can hear the difference, ear on speaker, but in the listening space my Fet phonos are dead quiet. The BF862 is very good on 1/F noise by the way. Better then the 2SK170 plus lower input capacitance. It is certainly my first choice for low noise phono with only N-channel. how good the 1/F noise is on the Interfet i can not tell.
I posted in the same moment you posted yours.
Maybe that book answers to my question...dunno! 🙂
Maybe that book answers to my question...dunno! 🙂
Maybe I wasn't clear. I wasn't asking what is the difference between the bjt paradise style and the jfets approach.
I was saying lets say I try the r7 schematic only n ch and use bf862. Would I be able to parallel 10 and get the gain needed? You said I can only do it with the IF part and I was asking why not with BFs?
Thanks you answered. So I can move forward and try out the only n ch design and paralleling 10 BFs and see if I can get up to the needed gain.
Do you think that this circuit has potential?
I can not tell if it has potential.
To make it clear YOU do not want to parallel, so the Interfet is the only choice.
If you accept paralleling forget the Interfet. 10x BF862 is not the same but to a first order approximation extremely similar.
Yes, that paralleling solves the gain and the noise problem-
To make it clear YOU do not want to parallel, so the Interfet is the only choice.
If you accept paralleling forget the Interfet. 10x BF862 is not the same but to a first order approximation extremely similar.
Yes, that paralleling solves the gain and the noise problem-
- Status
- Not open for further replies.