toxic,
I don't think you have to worry. Post 64 from http://www.diyaudio.com/forums/pass-labs/236111-transistors-needed-7.html
reports genuine curve traces. I got a batch from B&D on Thursday and I can tell you that I have measured the yfs at greater than 22ms on both n and p channel jfets with an Atlas DCA 75. Given the report from 6L6, and, since I don't know of any other high yfs P channel jfets so I haven't bothered to run any curve traces.
I don't think you have to worry. Post 64 from http://www.diyaudio.com/forums/pass-labs/236111-transistors-needed-7.html
reports genuine curve traces. I got a batch from B&D on Thursday and I can tell you that I have measured the yfs at greater than 22ms on both n and p channel jfets with an Atlas DCA 75. Given the report from 6L6, and, since I don't know of any other high yfs P channel jfets so I haven't bothered to run any curve traces.
I had purchased 50 2SJ74s and 50 2SK170s from B+K. They arrived 3 days after ordering, with no shipping fee.
I matched using an 11 VDC power supply and measuring voltage drop across a 100 ohm resistor to calculate Idss at a Vgs of 0 volts.
Idss values for all 100 transistors ranged from 6.7 to 11.1 mA, well within the range for the BL rating of 6 to 12 mA.
When sorting transistors to tenths of a mA (for example, 9.1 mA or 7.7 mA), I ended up with 22 pairs. Thus 44% of the purchased transistors matched to the same tenth of a mA. If I would extend the matching to ±0.1 mA (for example, match 8.1 P channel with an 8.2 N channel), then more matches would be allowed.
When just considering the 22 matches, the cost was about $9.10 per match. If I go the ±0.1 mA route, then the price falls further.
I recognize that I have not measured drain current at different Vgs values, but only at a Vgs = 0, and some builders desire to characterize matched pairs even more, especially if attempting to build balanced amplifiers. My impression from reading NP's writing is that this is the extent of matching he recommends for the F5, and I seriously doubt I would be able to tell the difference between my matched JFETs and those whose transductance curves might be a little bit more similar. I am very pleased with the results and want to thank everyone for the advice and the lead from 6L6 on the B+D products.
Steve
I matched using an 11 VDC power supply and measuring voltage drop across a 100 ohm resistor to calculate Idss at a Vgs of 0 volts.
Idss values for all 100 transistors ranged from 6.7 to 11.1 mA, well within the range for the BL rating of 6 to 12 mA.
When sorting transistors to tenths of a mA (for example, 9.1 mA or 7.7 mA), I ended up with 22 pairs. Thus 44% of the purchased transistors matched to the same tenth of a mA. If I would extend the matching to ±0.1 mA (for example, match 8.1 P channel with an 8.2 N channel), then more matches would be allowed.
When just considering the 22 matches, the cost was about $9.10 per match. If I go the ±0.1 mA route, then the price falls further.
I recognize that I have not measured drain current at different Vgs values, but only at a Vgs = 0, and some builders desire to characterize matched pairs even more, especially if attempting to build balanced amplifiers. My impression from reading NP's writing is that this is the extent of matching he recommends for the F5, and I seriously doubt I would be able to tell the difference between my matched JFETs and those whose transductance curves might be a little bit more similar. I am very pleased with the results and want to thank everyone for the advice and the lead from 6L6 on the B+D products.
Steve
I managed about 11 out of 20 for my matches . Some trading of unmatched singles could be arranged if we all agree on a test method so the number are close to the same .
Another concern, of course, is that exchanging of transistors doesn't become unintentionally contaminated with fake Toshibas.
Here is the first measurement of the LSJ74, as promised.
The sample came from Jan Didden of Linear Audio, and has an Idss of 7.34mA.
A 2SJ74BL was pulled from our data base and plotted side by side for direct comparison.
Test Vds was 9V, at room temperature.
Blue line is Toshiba, red is Linear Systems.
As you can see, Toshiba's have significantly higher Yfs (38mS compared to 27mS at Idss).
Also curve fitting reviews quite a bit higher (normalised) 3rd harmonics.
A pair of LSJ74s are on their way to more accurate THD analysis using a R&S UPV.
So, close but not quite the same.
Patrick
.
The sample came from Jan Didden of Linear Audio, and has an Idss of 7.34mA.
A 2SJ74BL was pulled from our data base and plotted side by side for direct comparison.
Test Vds was 9V, at room temperature.
Blue line is Toshiba, red is Linear Systems.
As you can see, Toshiba's have significantly higher Yfs (38mS compared to 27mS at Idss).
Also curve fitting reviews quite a bit higher (normalised) 3rd harmonics.
A pair of LSJ74s are on their way to more accurate THD analysis using a R&S UPV.
So, close but not quite the same.
Patrick
.
Attachments
But the characteristics displayed by the LS jfet are still very, very good, and will work quite nicely in the Nelson Pass and similar circuits. Essentially no other devices will do that.
My understanding is that noise performance of the LS parts is very good.
You see my point we now have 4 variables 2 74s 2 170 and 4 curves . Patrick has shown that the 2sj74 and 170 are not totally matching curves I now wonder how the LS curves compare to they . they are different but are they closer to each other than are the orginals ? Or does a mix and match produce better matched pair curve wise ? Then again it mater more in some applications than others as 6L6 has noted .
not to worry it all good . So matching the 74 and 170 is far lees of a problem then the some what neurotic over achieving diyer need to be fussing over.I apologize... the 'who cares' thing didn't come across great.Sorry.
My point is simply that with the selective degeneration (the pot) the issue is largely moot.
Hello Patrick,
While I don't want to doubt your results, I note that this is one Toshiba compared to one LIS. Maybe another Toshiba with another LIS would show a quite differnt picture.
We should compare an 'avarage' Tosh against an 'average 'LIS in my view.
jan
My record of Toshiba's does not show a Yfs variation of 25% between devices of similar Idss. 10% maybe.
So while your argument is correct, I do not believe that the Yfs are comparable even if I have a bunch of LSJ74s to measure.
And you were (past tense) the few person who had a large bunch of those.
Perhaps we should just wait till Linear Systems put up a proper datasheet.
And then we shall see.
Patrick
So while your argument is correct, I do not believe that the Yfs are comparable even if I have a bunch of LSJ74s to measure.
And you were (past tense) the few person who had a large bunch of those.
Perhaps we should just wait till Linear Systems put up a proper datasheet.
And then we shall see.
Patrick
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.