A few comments -
The NTE FET mentioned is probably a nice device, though I've never used it. I'd hate to have to shell out for 50 or so devices to do some matching, however. The prices I've seen for this FET are more than 3 times the price I paid for my 2SK170s from MCM.
For open loop applications, the high gm of the 2SK170 can be a bit of a liability, requiring use of source degeneration to restrain the gain somewhat. Gm is definitely not the only criterion I use in device selection. By this standard of measurement, all tubes should be relegated to the dumpster... Rather, I will use what I think is the appropriate tool for the job. For some applications, (anMC preamp may be one of them, but I would also look at the J110) the 2SK170 or one of its high gm, low noise bretheren is an obvious choice. For closed loop applications, it will also be easy to use. In many other applications, I will reach for a lower gain, perfectly servicable device.
The NTE FET mentioned is probably a nice device, though I've never used it. I'd hate to have to shell out for 50 or so devices to do some matching, however. The prices I've seen for this FET are more than 3 times the price I paid for my 2SK170s from MCM.
For open loop applications, the high gm of the 2SK170 can be a bit of a liability, requiring use of source degeneration to restrain the gain somewhat. Gm is definitely not the only criterion I use in device selection. By this standard of measurement, all tubes should be relegated to the dumpster... Rather, I will use what I think is the appropriate tool for the job. For some applications, (anMC preamp may be one of them, but I would also look at the J110) the 2SK170 or one of its high gm, low noise bretheren is an obvious choice. For closed loop applications, it will also be easy to use. In many other applications, I will reach for a lower gain, perfectly servicable device.
Re: Re: Vendetta phono stage
I never had the pleasure of looking into a Vendetta or audition it. The design is highly regarded in audio circles like the Absolute Sound.
My own MC stage is also using FETs (2SK389) at the input and it is very low noise.
See also:
http://www.diyaudio.com/forums/showthread.php?postid=378061#post378061
and below.
😎
Hi Petter, That's not the point. The point is that it possible to design a low-noise transformerless FET input MC stage.Petter said:
I never had the pleasure of looking into a Vendetta or audition it. The design is highly regarded in audio circles like the Absolute Sound.
My own MC stage is also using FETs (2SK389) at the input and it is very low noise.
See also:
http://www.diyaudio.com/forums/showthread.php?postid=378061#post378061
and below.
😎
I don't know the current price of the NTE458, but I paid about $1.20US for the ones I have. For MM use, I haven't had to parallel them to get the noise below audibility at any humanly reasonable volume setting. For MC, I follow my own advice and use a transformer.
Yes, one can brute force a FET input to be reasonably quiet without a transformer, but you guys know that the noise figure is not even close to being at a minimum with a source impedance of a few ohms. A bipolar will do better there, with all due respect to John's excellent work.
Yes, one can brute force a FET input to be reasonably quiet without a transformer, but you guys know that the noise figure is not even close to being at a minimum with a source impedance of a few ohms. A bipolar will do better there, with all due respect to John's excellent work.
Degeneration........in an open loop circuit? Horrors! Who could imagine using passive components to set the operating parameters of a circuit.
Noise figure is useful number for a front-end in the RF world, but is of no use in the AF world. Not even for the front end.
Jocko
Noise figure is useful number for a front-end in the RF world, but is of no use in the AF world. Not even for the front end.
Jocko
Jocko, yer a riot .... All irony aside, you do anything it takes to get the results you need, given the design constraints. For the project I mentioned, the constraints were threefold - no global feedback, use of JFETS throughout, and a decent harmonic distribution in simulation. If you go to the thread I cited and read it, you will understand where I'm going with the project - I won't burden everyone else with it here.
Once you abandon global feedback, it takes a bit of dancing around to get the approximate value of gain you need, plus proper voltage centering and decent harmonic distribution. It's one thing to do a high gain MC preamp or a follower - it's quite another problem to do a line amp with gain <10 without using global feedback, and meeting requiremts of voltage centering and pleasant harmonic spectrum. At least in simulation, a modest FET like the PN4303 worked a lot better in this application than the 2SK170.
I'm not knocking the 2SK170 at all, especially as I have used it in a lot of applications. I just don't think it's the be-all and end-all. Alternate choices can yield excellent results for a lot of applications. I like to keep a broadly stocked war chest of devices (especially if I can get them for cheap), and blanket statements aren't going to get me to change that practice any time soon.🙂
Once you abandon global feedback, it takes a bit of dancing around to get the approximate value of gain you need, plus proper voltage centering and decent harmonic distribution. It's one thing to do a high gain MC preamp or a follower - it's quite another problem to do a line amp with gain <10 without using global feedback, and meeting requiremts of voltage centering and pleasant harmonic spectrum. At least in simulation, a modest FET like the PN4303 worked a lot better in this application than the 2SK170.
I'm not knocking the 2SK170 at all, especially as I have used it in a lot of applications. I just don't think it's the be-all and end-all. Alternate choices can yield excellent results for a lot of applications. I like to keep a broadly stocked war chest of devices (especially if I can get them for cheap), and blanket statements aren't going to get me to change that practice any time soon.🙂
Hmmm. Three pages of comments, but nobody has stated the obvious.
Only the Japanese companies ever made JFETs as true complementary pairs. Both Hitachi and NEC used to make some very nice parts, but they have discontinued them many years ago. Only Toshiba is left.
If you like the colorations of capacitors, this is meaningless to you. If you prefer the neutrality of a DC-coupled circuit, complementary FETs are very nice. Which pretty much leaves you with Toshibas...
Only the Japanese companies ever made JFETs as true complementary pairs. Both Hitachi and NEC used to make some very nice parts, but they have discontinued them many years ago. Only Toshiba is left.
If you like the colorations of capacitors, this is meaningless to you. If you prefer the neutrality of a DC-coupled circuit, complementary FETs are very nice. Which pretty much leaves you with Toshibas...
It would be possible to servo some single ended circuits that use split supplies, but that just opens up another storm of controversy. Tube fanciers are experiencing both "tube glow" and capacitor coloration, but that's another story as well. I have done some closed loop designs that balance out just fine without complementary FETs, too - it's just a question of careful design. It all boils down to personal design philosophy and the sonic palette you prefer.
So far, I'm getting along just fine with the capacitor coloration in my new preamp, probably because my former preamp (an old Nikko) was brutalizing the signal in ways I don't like to think about... The two pieces of gear are as unlike as night and day. The next circuit I try in the preamp test bed sound totally different from the first two - that's the fun of it.
Just for safety's sake, I will always place at least an input coupling cap on anything I do, as I don't trust signal sources I haven't built and tested. I personally saw a Phase Linear amp literally set on fire by a combination of an output short and a Sony receiver that placed a huge, long DC pulse on its preamp outputs at power up. Being a friend of the poor party in question, I had to fix that Phase Linear, and needless to say, I put an input capacitor on that thing.
It's true that there aren't any real acknowledged complementary JFETs coming from the US, but the PN4392/PN4393 and the J174/175/176 series are close enough so you can fake it - I've done it, at least for second stages. I will admit, however, that on that particular project, I used 2SJ73 and 2SK146 matched pairs for the inputs, so I cheated a little.
So far, I'm getting along just fine with the capacitor coloration in my new preamp, probably because my former preamp (an old Nikko) was brutalizing the signal in ways I don't like to think about... The two pieces of gear are as unlike as night and day. The next circuit I try in the preamp test bed sound totally different from the first two - that's the fun of it.
Just for safety's sake, I will always place at least an input coupling cap on anything I do, as I don't trust signal sources I haven't built and tested. I personally saw a Phase Linear amp literally set on fire by a combination of an output short and a Sony receiver that placed a huge, long DC pulse on its preamp outputs at power up. Being a friend of the poor party in question, I had to fix that Phase Linear, and needless to say, I put an input capacitor on that thing.
It's true that there aren't any real acknowledged complementary JFETs coming from the US, but the PN4392/PN4393 and the J174/175/176 series are close enough so you can fake it - I've done it, at least for second stages. I will admit, however, that on that particular project, I used 2SJ73 and 2SK146 matched pairs for the inputs, so I cheated a little.
Charles,Charles Hansen said:
Right. For compelementary circuitry only 2SK389 and 2SJ109 are available. These devices are low noise but large geometry high capacitance devices so you have to cascode these avoiding the Mlller effect.
Wrenchone, you are correct that you can make some pretty decent complementary pairs by matching. This is best done with a curve tracer rather than relying on the manufacturers' published curves (if they even exist, which is another problem with US made JFETs).
One problem with servos is that they amplify the coloration of the capacitors used in the servo loop. I'd prefer just to cap-couple instead of use a servo. A really good cap has quite low levels of what are fairly benign colorations. However, really good caps are few and far between.
Elso, the dual monolithic parts (2SK389 and 2SJ109) have been discontinued as of late last year. You will probably be able to get them through distributors for a while, but not forever. The complements that still are being made are 2SK170/2SJ74 and 2SK246/2SJ103. That's it...
One problem with servos is that they amplify the coloration of the capacitors used in the servo loop. I'd prefer just to cap-couple instead of use a servo. A really good cap has quite low levels of what are fairly benign colorations. However, really good caps are few and far between.
Elso, the dual monolithic parts (2SK389 and 2SJ109) have been discontinued as of late last year. You will probably be able to get them through distributors for a while, but not forever. The complements that still are being made are 2SK170/2SJ74 and 2SK246/2SJ103. That's it...
Discontinued?
Charles,
Too bad. John Curl is using the 2SK389 and 2SJ109 by the thousends in the Halo amplifiers.
Charles,
Too bad. John Curl is using the 2SK389 and 2SJ109 by the thousends in the Halo amplifiers.
HOW ABOUT GROUP BUY ON 2SK389, 2SJ109
Does anyone have a good source for getting a few hundred 2SK389, 2SJ109 for a group buy? This may be our last chance to stock up on ~ 50 pairs or so.
Idss BINNING:
GL: 2.6-6.5ma
BL: 6-12ma
V: 10-20ma
Hand matching JFETs is very time consuming so I always use 2SK389, 2SJ109 BL grade matched pairs in diff amps.
GROUP BUY TIME??
Does anyone have a good source for getting a few hundred 2SK389, 2SJ109 for a group buy? This may be our last chance to stock up on ~ 50 pairs or so.
Idss BINNING:
GL: 2.6-6.5ma
BL: 6-12ma
V: 10-20ma
Hand matching JFETs is very time consuming so I always use 2SK389, 2SJ109 BL grade matched pairs in diff amps.
GROUP BUY TIME??
The big question would be, from whom could we order? I might also be up for a group buy if I'm able to get some specific ranked devices.
Right now, I have 2SJ109V and 2SK389GR. These would be ok for single-ended stuff, but I would like some matching rankings for complementary projects.
Right now, I have 2SJ109V and 2SK389GR. These would be ok for single-ended stuff, but I would like some matching rankings for complementary projects.
I've been to the Toshiba Semiconductor USA site, and they do not acknowledge the existance of Toshiba small signal devices, though the main site in Japan does. BTW, the 2SJ109 and 2SK389 are now officially gone from the home site, though the 2SJ74 and 2SK170 are still alive and well. Part Miner seems to indicate that there may be the devices we need at Toshiba Semiconductor USA. Whomever is interested in ramrodding a group purchase might want to drop a call or two to Toshiba headquarters in Irvine to see what is going on...
Just to add to the list of the audio fets I will mention couple other pairs:
2SK187: S~22mS@3mA, Cin~40pF, Cr~8pF(@10V), Rjunc-amb=0.42°/mW, 40V, (letters: C- 2.5-5mA, D- 4-8, E- 6-12, F- 10-20) ----- 2SJ69: Cin~82pF, Cr~19pF.
2SK186: S~13mS@3mA, Cin~20pF, Cr~4pF (@10V), same Rj-a, -----2SK68: Cin~32pF, Cr~7pF, also a 40V pair (B- 1.6-3.2mA, C- 2.5-5, D- 4-8, E- 6-12).
These are discontinued long ago, but still can be found here and there (except, probably, J68...).
2SK187: S~22mS@3mA, Cin~40pF, Cr~8pF(@10V), Rjunc-amb=0.42°/mW, 40V, (letters: C- 2.5-5mA, D- 4-8, E- 6-12, F- 10-20) ----- 2SJ69: Cin~82pF, Cr~19pF.
2SK186: S~13mS@3mA, Cin~20pF, Cr~4pF (@10V), same Rj-a, -----2SK68: Cin~32pF, Cr~7pF, also a 40V pair (B- 1.6-3.2mA, C- 2.5-5, D- 4-8, E- 6-12).
These are discontinued long ago, but still can be found here and there (except, probably, J68...).
As long as hole and electron mobilities are so different, there never will be truly complementary FETs.
SY said:
Uhh... yahbut, this is also true of bipolar transistors.
So are you saying that *any* circuit with "complementary" devices should be automatically discarded because the devices aren't *perfectly* complementary? Should we similarly reject stereo because the two channels can never be matched *perfectly*?
Maybe Bricolo's signature is more appropriate here. "Just remember: in theory there's no difference between theory and practice.
But in practice there usually is quite a bit of difference..."
By the way, what does your palindromic signature mean?
No, I'm arguing that we should choose devices based on performance in a given circuit, not nominal notions of "complimentarity." Remember, I'm the guy who mixes pFETs with tubes!
BTW, congratulations on noting the Latin palindrome. Ten more points if you notice the other features of it. The literal meaning is something like "the farmer Arepo holds the wheel."
BTW, congratulations on noting the Latin palindrome. Ten more points if you notice the other features of it. The literal meaning is something like "the farmer Arepo holds the wheel."
Nonsense! All you have to do is build P-channel devices out of antimatter. I think maybe if we scrape together all the antimatter made so far we might have enough to make a small one...SY said:
- Status
- Not open for further replies.