I'm finally getting around to finish the two Onos I started a couple of years ago. Got severly delayed , for several reasons..
I was measuring my way around 100+ of 2SK170GRs for the MC section, and the main questions is simple-- two what tolerances are you guys mathing these JFETs???
I got some very good quads out of this, in the range of 50-65 mV across the 22 ohm resistor.
I remember NP said something about these tolerances, but when I started searching I couldn't find it- what I did find, though, was a statement from NP that they used the BL version...???
In all the older treads "everybody" is looking for GR - which I have, and stating that the BL will cause the MC cascode to malfunction..??
What did I miss here, if anything???
I was measuring my way around 100+ of 2SK170GRs for the MC section, and the main questions is simple-- two what tolerances are you guys mathing these JFETs???
I got some very good quads out of this, in the range of 50-65 mV across the 22 ohm resistor.
I remember NP said something about these tolerances, but when I started searching I couldn't find it- what I did find, though, was a statement from NP that they used the BL version...???
In all the older treads "everybody" is looking for GR - which I have, and stating that the BL will cause the MC cascode to malfunction..??
What did I miss here, if anything???
Hi,
I have build the ono with BL types. and matched the Id current against a Ugs of 0V within 0,1mA.
In the running versions, the drain current is within 0,5mA and that is sufficient for correct operation of the frontend.
I wonder if you would let me buy some of your GR types, as I modified the drain resistors and the source resistor to controll the current with the BL types in the frontend, as well as used a 400E drain resistor instead of 1K (to have sufficiend voltage reserve).
Also the output fet of the frontend is not working well with the BL type.
So I would be very interested to buy some of your GR types.
I have build the ono with BL types. and matched the Id current against a Ugs of 0V within 0,1mA.
In the running versions, the drain current is within 0,5mA and that is sufficient for correct operation of the frontend.
I wonder if you would let me buy some of your GR types, as I modified the drain resistors and the source resistor to controll the current with the BL types in the frontend, as well as used a 400E drain resistor instead of 1K (to have sufficiend voltage reserve).
Also the output fet of the frontend is not working well with the BL type.
So I would be very interested to buy some of your GR types.
AuroraB:
I started out with 80 2SK389BL and managed two quads that have a total current draw of less than 10mA. It is my understanding that the BL are preferable when you can achieve this. If not, then use the GR grade. Here are my final, in-circuit measurements:
Idss for 2sk170gr
=============
Q_13_R = 2.35 mA
Q_12_R = 2.32 mA
Q_11_R = 2.27 mA
Q_10_R = 2.22 mA
Q_13_L = 2.54 mA
Q_12_L = 2.46 mA
Q_11_L = 2.47 mA
Q_10_L = 2.47 mA
I would reference Algar_emi's schematics as he gives node voltages for both types and has excerpts of posts in the download that discuss this somewhat.
Take care!
I started out with 80 2SK389BL and managed two quads that have a total current draw of less than 10mA. It is my understanding that the BL are preferable when you can achieve this. If not, then use the GR grade. Here are my final, in-circuit measurements:
Idss for 2sk170gr
=============
Q_13_R = 2.35 mA
Q_12_R = 2.32 mA
Q_11_R = 2.27 mA
Q_10_R = 2.22 mA
Q_13_L = 2.54 mA
Q_12_L = 2.46 mA
Q_11_L = 2.47 mA
Q_10_L = 2.47 mA
I would reference Algar_emi's schematics as he gives node voltages for both types and has excerpts of posts in the download that discuss this somewhat.
Take care!
Well, - I have all the Borbely articles....
Point is, that the GR version is the one with the lowest Idss - not the BL. If the total current of all four is supposed to stay at or below 10mA, the BL cannot be used easily. That's why I wonder about the conflicting messages.
I have managed to get 6-7 good quads out of my 100+, but since I'm building two simultaneously (one for my brother), I think I need the extra sets as spares, but I'll do another test setup.
Point is, that the GR version is the one with the lowest Idss - not the BL. If the total current of all four is supposed to stay at or below 10mA, the BL cannot be used easily. That's why I wonder about the conflicting messages.
I have managed to get 6-7 good quads out of my 100+, but since I'm building two simultaneously (one for my brother), I think I need the extra sets as spares, but I'll do another test setup.
Thanks alot, Mr. Pass, for clearing this up.
I figured this was the case, but if 2.5mA is the designed current level for the JFETs, the yield out af a BL batch must be quite low.
Some have reported voltages across the source resistor in the order of 70-100 mV, indicating currents in the order of 4 mA +/-, not 2.5.
It would be clarifying to know the designed level ( and also for the Pearl...)
Moderators-
For some time I've seen the necessity of some sticky fact sheets, also for others of the more popular projects. I know there are Wikis for some pfrojects, but what about some sort of "Design Facts" section ??
I figured this was the case, but if 2.5mA is the designed current level for the JFETs, the yield out af a BL batch must be quite low.
Some have reported voltages across the source resistor in the order of 70-100 mV, indicating currents in the order of 4 mA +/-, not 2.5.
It would be clarifying to know the designed level ( and also for the Pearl...)
Moderators-
For some time I've seen the necessity of some sticky fact sheets, also for others of the more popular projects. I know there are Wikis for some pfrojects, but what about some sort of "Design Facts" section ??
I don't think the JFets need to be that closely matched to each other in this position. Instead I would generally match components channel to channel so that the values are as close as possible. For your JFets this would mean swapping two of them from L to R so that your total current draw is about the same (right now it is 10% off, maybe not a big deal but if you want to be accurate be accurate where it makes sense).
Thank-you for clearing that up, Nelson
Dear Nelson,
Thank-you for the final word on the type of Jfet you used for the Ono. I have watched the Ono/Xono threads for a long time and have seen grown men have apoplexy because they could NOT get the GR versions!
- all they could ever find was those "dreadful" BL's 
It is nice to be put at ease.
Regards,
George.
Dear Nelson,
Thank-you for the final word on the type of Jfet you used for the Ono. I have watched the Ono/Xono threads for a long time and have seen grown men have apoplexy because they could NOT get the GR versions!
- all they could ever find was those "dreadful" BL's It is nice to be put at ease.
Regards,
George.
Regarding "Sticky" fact sheets
Back again,
Regarding "Sticky" fact sheets, I have seen them NOT work so many times, that I think the idea is not so good.
I think that the Wiki's is the way to go. It requires thought and a degree of enthusiasm to add to a Wiki, whereas the "sticky" posting gets messed up pretty easily, partly because it is the first thing anyone sees - unless it is heavily limited as to who can post to it ... and that then places the onus on "the few" to keep it up to date.
Just my tuppence ha'p's'orth ...
Regard,
George.
Back again,
Regarding "Sticky" fact sheets, I have seen them NOT work so many times, that I think the idea is not so good.
I think that the Wiki's is the way to go. It requires thought and a degree of enthusiasm to add to a Wiki, whereas the "sticky" posting gets messed up pretty easily, partly because it is the first thing anyone sees - unless it is heavily limited as to who can post to it ... and that then places the onus on "the few" to keep it up to date.
Just my tuppence ha'p's'orth ...
Regard,
George.
MRupp said:
well, I will have to confess my ignorance, but I've been matching transistors for quite some time and have never matched them channel-to-channel... why would this matter? In this case, gain is a function of three resistors and the transconductance of the JFets. For the 2SK170, a typical value of g_m is 22mS, regardless of the grade. So by matching them as close as possible in terms of Idss, we're looking to ensure low-noise...
at least this is what I always thought, feel free to correct me
I think the confusion comes from the term “matching”. I am aware that you match parallel transistors in output stages (for proper current sharing), in differential pairs, etc.. For the Ono MC stage I would normally “select” Jfets in the current range that fits into the circuit, but not match them to each other.
The transconductance of the 2SK170 is a function of the current that runs through it (see graph in datasheet) and there is very little variation between devices. So if you have a difference of, say, 10 % of current thru a (single!) Jfet you also have a difference in the transconductance and hence in amplification. Again, this may not be a huge variation but in the lower current range the transconductance curve is fairly steep. Btw., your “typical” g_m value is given at Vgs = 0V, in other words it varies with the Idss rating of the individual device.
Now regarding channel matching: For stereo cicuits it is a good idea to have the tolerances between the channels as low as possible. It all depends on the effort you want to make, and the results you are striving for, but this being DIY you can off course go to a great length. For phono stages you could obviously match the filter caps as close a you can, if possible well below 1% (see for instance the recommendations from Alan Wright). But then you should also match those other components where you would otherwise have high tolerances. So if you have exact values for your Jfets you can obviously distribute them such that you have similar total current on both sides.
Again, this is my personal approach.
The transconductance of the 2SK170 is a function of the current that runs through it (see graph in datasheet) and there is very little variation between devices. So if you have a difference of, say, 10 % of current thru a (single!) Jfet you also have a difference in the transconductance and hence in amplification. Again, this may not be a huge variation but in the lower current range the transconductance curve is fairly steep. Btw., your “typical” g_m value is given at Vgs = 0V, in other words it varies with the Idss rating of the individual device.
Now regarding channel matching: For stereo cicuits it is a good idea to have the tolerances between the channels as low as possible. It all depends on the effort you want to make, and the results you are striving for, but this being DIY you can off course go to a great length. For phono stages you could obviously match the filter caps as close a you can, if possible well below 1% (see for instance the recommendations from Alan Wright). But then you should also match those other components where you would otherwise have high tolerances. So if you have exact values for your Jfets you can obviously distribute them such that you have similar total current on both sides.
Again, this is my personal approach.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.