Self biased JFET source follower design: what JFET do I use?

[Fancy Lime]

Hi there!

Based on another discussion I started wondering if it was feasible to build a phantom powered JFET buffer into a 1/4" plug to get a buffered guitar cable. The concept is far from new. Don Tillman has implemented it as a JFET booster instead of a buffer (http://www.till.com/articles/PreampCable/) and there is even a patent (http://www.muzique.com/lab/patent3.htm) for the buffer version. The issue with the buffer is, that it uses a dual power supply, which is not the biggest of deals but it is a tiny bit cumbersome. We could of course put a biasing network in the plug as well but we would like as few components here as possible because it is going to be a tight fit anyway.

So I thought, how about just using a self biased single supply JFET for that? That of course means that we need a rather specific kind of JFET if we want to get a clean signal from common (bass) guitar pickup output voltage swings. What we need is a large(ish) (negative; for n-channel) Vgs(off) and good noise characteristics. And if it'd not be made of unobtainium, that'd be great, too. Unfortunately, all JFETs that have usable info on noise characteristics in the datasheets are really low Vgs(off) types (~-0.3 to -1.5V), for some reason. Does anybody know why that is? For our purposes here, something on the order of -6V would probably be good. That would mean we can bias the gate to ground (which we don't even need a resistor for because that is exactly what the guitars volume control does) and then bias the source to 3V (assuming 9V supply) by adjusting the source resistor. This should give us almost +/- 3V of swing before clipping, correct? That should suffice for the purpose, no? I am unaware of any JFET that has reasonably tight specs at such Vgs(off) values, so manual testing will be unavoidable. R.G.'s improved JFET matcher (http://www.geofex.com/Article_Folders/fetmatch/fetmatch.htm) can be adapted to find the right JFET and the right source resistor, I think. Using a very large value (say 1M) for Rset would give us a read of Vgs(off). Once we find a suitable one, we adjust Rset until the meter reads half of the Vgs(off) value and then use that Rset value as source resistor. Is my thinking sound here? I'm making things more complicated than they need to be again, aren't I?

The 2N5457 has a max Vgs(off) of -6V but since I only have a finite supply of those, I would hazard the guess that none of them is particularly close to the theoretical maximum. What low noise JFET has a "typical" value of -6V, if such a thing even exists?

Cheers,
Andy

[swamphorn]

Tayda sells the J105 with a V_GS(off) range of -4.5 to -10 V. I can't speak for its performance as a linear amplifier, however. I wouldn't be surprised if most JFETs with a large V_GS(off) are specified for switching applications. You could always buy surface-mount '5457s. They're much more available so you don't have to worry as much about cherry-picking and they'll be an even easier fit inside a jack.

[1] J105/J106/J107 https://www.taydaelectronics.com/datasheets/files/A-3211.pdf

[jonny.reckless]

I pretty much always use a J112 as a self biased source follower. With a 4k7 resistor I find the source typically biases in the 2 to 4 volt DC range, allowing plenty of headroom for passive pickups, and a nice low noise, low impedance output for the next stage. You can buy them in many places and they are not expensive. If you want even more voltage or a lower impedance, try the J111. Although they are sourced from a process designed for analog switches they are both surprisingly quiet.
$0.14 for 100 off at Mouser. I usually buy 100 at a time.
https://www.mouser.com/ProductDetail/ON-Semiconductor-Fairchild/J112?qs=sGAEpiMZZMvAvBNgSS9LqnPOL556k3vz
BTW the J113 is great for common source gain stages in a guitar preamp. I prefer it to the J201 tonally. See the Boba Fet or Rosie guitar amp designs for details and a demo.

[merlinb]

All JFETs have noise lower than a TL071, which is already low enough enough for guitar work, so don't worry about it. Add J310 to the list.

I'm making things more complicated than they need to be again, aren't I?

Yes. The buffer circuit you're describing is already the perfect test jig to find the right JFET for the buffer circuit! In other words, just build the preamp and plug in a few JFETs until you find the best one.

[Fancy Lime]

Thanks guys!

I have a few J111 and J112 (and a bunch of J113) but no J310 and J105. I'll just test them all and see if any of them are suitable.

All JFETs have noise lower than a TL071, which is already low enough enough for guitar work, so don't worry about it. Add J310 to the list.

I'm making things more complicated than they need to be again, aren't I?

Yes. The buffer circuit you're describing is already the perfect test jig to find the right JFET for the buffer circuit! In other words, just build the preamp and plug in a few JFETs until you find the best one.

Now that you mention it, that sounds very reasonable. For some reason my instinct is never to just try it and see what works even if that is often quicker than figuring out the theory, then measuring, adjusting and finally finding that if I had just stuck with my first impulse, I would have gotten the same result but would have been done by last Thursday.

Cheers,
Andy

[amptramp]

There is so much variation in JFET characteristics that trying out a circuit with the JFET's you have on hand is usually the only way to get sensible results.  Not many pieces of equipment use JFET's anymore for just this reason - mass produced JFET circuits have to take into account the entire variation in the range of pinchoff voltage and Idss.

[Fancy Lime]

There is so much variation in JFET characteristics that trying out a circuit with the JFET's you have on hand is usually the only way to get sensible results.  Not many pieces of equipment use JFET's anymore for just this reason - mass produced JFET circuits have to take into account the entire variation in the range of pinchoff voltage and Idss.

Yet there are applications where JFETs are just superior to anything else. Ultra low noise, high impedance, MHz frequency inputs are one of the biggest challenges in a lot of scientific sensor equipment, to just name one example. These days I would think that it should be fairly easy for manufacturers of transistors to automatically measure all parameters and allow ordering in very tight groups with Vp, Idss, noise specs and whatever else all closely matched so that customers can order, e.g. 10,000 of *exactly* the JFET they need for a premium. Not the thing for customers like us but I'm guessing if large customers came aknocking, that should be a viable business model. Then again, the area where JFETs reign surpreme may be too small for that to fly. If I had to pitch an idea in a business meeting at onsemi or the like, I'd try that one.

Andy