Paralelling multiple devices for the same Idss won't lead to a significantly higher Ciss compared to a single device. Otherwise said, Ciss depends significantly on Idss (higher with higher Idss).
However, Crs depends weakly on Idss. By paralleling devices, Crs will increase. As Crs strongly depends on Vds, it is to be expected that paralleled devices will have a poorer frequency response and a poorer linearity.
Thanks, this is what I thought. Now as to the second part, how significant is it? I.e., for the devices we're talking about here, what's the difference in frequency response and distortion (as an estimate, assuming a 25k source impedance from a worst-case 100k pot) from paralleling two lower Idss parts compared to a single higher Idss part?
But linearity of transfer characteristics is much better near to Idss. And even for FB designs there are no stability issues - depends on overall design. If you compare high and low idle, you get both lower distortion and better sound))at high idle.
Somewhere in this forum someone posted a chart of the ID vs. VGS of the 2SK170BL with 0V < VGS < 0.6V (ID> IDSS). In this case we would have even greater linearity but soon came to me one question: how are the capacitances of fet in this situation?!
eD
See attached. As expected (as long as Crss doubles), the 3dB cutoff is about half (from 128KHz down to 67KHz).
Also as expected, distortions are increasing with about the square root of the Crss ratio (here, 2), from 1.5% to 2%. This is for 100mV input and gain as in the attached picture. Idss is about 20mA.
Edit: Of course, green is single JFET, red is two paralleled JFETs, for about the same Idss. Bottom line, John is right but for all the wrong reason. It is the Crss that degrades the parallel JFET configuration, not the (input) Ciss.
Attachments
Last edited:
John brought up the input pot; the worry was apparently the doubling (which does not seem to be the case) of Cin. That's where I was having trouble understanding.
For a phono stage, it doesn't make sense to consider a 25K source impedance. For a realistic 50ohm @ 20KHz MC impedance, the distortion gap widens to about 3:1, but numbers are already low (0.01% to 0.03%). Bottom line, for MC preamps, paralleling JFETs has practically no impact on distortions and bandwidth (because of the low source impedance).
EDIT: for MM, the situation is significantly different, both because of larger input signal (2.5...5mV) and large equivalent source impedance @ 20KHz.
Last edited:
But where is the limit then?
When the drawer labeled 2sk369 is empty?
Magura
Not sure about what "limit" you are talking here, but to me the 2SK369 (and the 2SJ109 counterpart) rush after syndrome doesn't make any sense. They are nothing but matched pairs. They have nothing special and there's absolutely no reason why they couldn't be replaced with sorted 2SK170 and 2SJ74, at least for audio purposes (instrumentation applications could be different). Actually, you can easily sort a single JFET population to closer matching than the dual part spec.
Syn, I think you got me wrong, I wrote 2sk369, not 2sk389.
As for "limit", I meant the number of fets in parallel.
Magura
But where is the limit then?
When the drawer labeled 2sk369 is empty?
Magura
Recycle, and think of all the burn-in time.
Seriously, do you keep every preamp in a drawer when you go on to the next one? There is a new reality show "Hoarders", you know some but they have found some stunners.
Last edited:
Sorry, my tired eyes are playing tricks... I still have to get used with these progressive lenses.
Depends on application. For a MC preamp, there is no practical difference between one 2SK369 and e.g. 2 x 2SK170 in parallel. For a line preamp, I would choose one 2SK369 (if available).
Last edited:
Recycle, and think of all the burn-in time.
Seriously, do you keep every preamp in a drawer when you go on to the next one? There is a new reality show "Hoarders", you know some but they have found some stunners.
Oh no, I think this came across the wrong way. I usually give them away
I meant the drawer with the jfet's. The reason for the question is that I'm currently building a phono pre (MC), and have just used 4 in parallel, as that seems to be the standard, but if there is any benefit of more fet's in parallel, then I might as well do that, as I have loads of the 2sk369bl.
Magura
Oh no, I think this came across the wrong way. I usually give them away
I meant the drawer with the jfet's. The reason for the question is that I'm currently building a phono pre, and have just used 4 in parallel, as that seems to be the standard, but if there is any benefit of more fet's in parallel, then I might as well do that, as I have loads of the 2sk369bl.
Magura
There is another limit, and that's stability. Paralleling more than 4-6 x 2SK170BL with no degeneration is looking for trouble. Depending on the layout, you can get an 100MHz oscillator.
Oh, so I will have no doubt when the limit is reached.
That's kinda what I was trying to figure.
Thanks.
Magura
- Status
- Not open for further replies.