Opamp-based VB Reference Voltage

[lars-musik]

Again I am revealing my sketchy knowledge of opmaps.

For a King of Tone I am looking for a way to genetrate a stable ½ VB reference voltage (here). The self biasing LM386 introduced a lot of noise into the circuit (even though the inputs were grounded – here). So I jerry rigged a TL071 based one that works fine. But I'd like to know a little more about this (mainly because I'd like to create myself a nice little standard if something goes wrong with a VB again in the future).

Now I found the artice by RG about voltage divider networks and the dimensioning of the corresponding resistors (here). The datasheet of the TL071 gives me an input bias current of 7 nA. Is this the value I calculate the voltage divider with? That would result In approx. 1.2MOhm resistors for R1 and R2 (possible of course there's an error because there are quite a lot of zeros involved). The LM321 has an bias input current 45 nA, resulting in 200 KOhm for R1 and R2.

Apart from that I am not sure if I am ob the right track at all, what would be the best choice op amp for that task? Low input bias current? Low power consumption? Low price? JFET input? LM321? TL071? LM741?

Thanks for your input!

Lars

[jubal81]

Check out the TLE2426. It's an IC dedicated to that very thing, comes in a TO-92 package and I've had great success with them.

http://www.mouser.com/ProductDetail/Texas-Instruments/TLE2426CLPR/?qs=sGAEpiMZZMuBck1X%252b7j9fK0Q2jxjfQGgP40LcngJCCM%3d

[lars-musik]

Thanks Jubal,

while looking for a datasheet to your suggested virtual ground chip I found that article:

https://tangentsoft.net/elec/vgrounds.html

that covers everything I need. At least I hope so.

[PRR]

I would fearlessly use 1Meg in any TL07x input. They are stable with 100Meg.

Larger R suggests smaller C for the same low-pass filtering; but at 1meg you are already into low-cost caps.

I can't see any "better" plan than the real-cheap hi-Z TL07x.

[antonis]

As Paul said, use any TL07X and any value resistor pair..

7nA input bias current should be considered negligible for a divider current greater than 70nA - that stands for a pair of 64ΜΩ resistors..
(or 700nA divider current, if you prefer 1% accuracy, which is the best you can get with commercialy availiable resistors - that give us a pair of 6MΩ divider resistors for 1%, a pair of 12MΩ for 2%, a pair of 33MΩ for 5.5% e.t.c. )