Buffer with Dual N-Ch Jfets and Dual Power Supply

[vdi_nenna]

Like to use a buffer on output of a R2R DAC.
BF861 SMD is an inexpensive, available, low-noise N-CH Jfet.
Are there any other variation beside what is in the attached pics?
Pics from Erno Borbely's JFET Frontier's articles.

I tried the B1 v1 in between the DAC and BA3-Pre and the sound was much better. Like to build something good for inside the DAC case and avoid the extra caps, pot, switch and interconnects.

Thanks,

Vince

Notes:

[Hikari1]

I don't believe there is a complimentary P channel version of bf861/862. So pretty much all designs will use a coupling cap.

Are you going to use the DACs power supply? Is it single or dual? That will make a lot of the decisions for you. Also, most DACs use a lower voltage supply than most of these buffers.

B1 rev2 would be the smallest/easiest.

[Zen Mod]

Iron

later in thread , final version

[vdi_nenna]

I'm not look for complimentary jfets.
This is the point of my post.
Just asking if there are any alternatives to what is posted in the pic, e.i., what Erno propsed.

Thx

V.

[vdi_nenna]

Thx ZM. Will check it out.

[Nelson Pass]

With the 2SK170 / LSK170 you will find that 15 a,b,c will all work very well, and put
you in the .00x distortion category or better. The current source version is preferred,
and you don't need to use expensive Jfets for that. J113 will do the job nicely.

[Zen Mod]

Originally Posted by vdi_nenna

Thx ZM. Will check it out.

also some of it (plain buffer or Borbely's WCF) explained in opening post of tips 'n' tricks for B. M25 and SissySIT

with some SOT23 adapter pcbs and veroboard , you can do your own , including shunt reg

buy 100pcs of JFet I used and you're good for a decade , for peanuts , without any loss of performance (comparing to more expensive options)

[vdi_nenna]

...you will find that 15 a,b,c will all work very well, and put
you in the .00x distortion category or better. The current source version is preferred...

Do 15 b and c both use current sources, one just a discrete current source and the other a JFET?

So in other words, use a 2sk170 for the buffer and use a J113 for the current source. No matching needed? Just use same Idss class?

[Zen Mod]

J113 is JFet , internally connected as 2 pole

though , irrelevant - -use whatever CCS option you want ,taking care of TempCo (think DC offset) , according to your needs

[PRR]

You should note that all of these plans will have some DC at the output. Unless you are comfortable feeding DC out the line to the next box (which may change!), an output DC blocking cap or other technique is wise.

15C has "potential" to be DC-free. If the two FETs are exactly matched the DC nulls. If they are a bit off-match, trimming the bottom resistor will null DC. At least for today, or this-hour. I had a Heathkit 'scope which used this plan as input buffer, and I found the DC drift annoying enough that I put a then-sexy chip in place.

15B would null if the two different parts happen to match, unlikely. 15A can be nulled by changing Rs, but it may need large change and will drift with temperature and rail voltage.

16A "could" be nulled if Rd were put down to its proper value (few dozen Ohms) and another couple resistors added, which modestly reduces its virtue in low-Z loads. 16B "would" null if you could match P- and N-types- unlikely without a barrel-full of FETs and an automated sorter, and tempco may not be totally compensated.

Stick a cap on the input, run the Gate resistor to a +/-1V trim, they all null today, with some residual drift.

Stick a cap on the output and DC is avoided. Cave-man, but effective. For typical hi-fi loads, I'd go for 15A, because it is simple, real-real clean despite not being "best", and the small curvature is not likely offensive.