Can someone educate me on "....buffer Vb"

[jimmybjj]

bean on March 25, 2011, 07:13:59 AM
You've got half a TL072 free....why not use it to buffer Vb?

Thanks for sharing!
Quote from: bean on March 25, 2011, 09:57:18 AM
R4 and R5 are your voltage divider for Vb. Connect their output to pin5, then connect pins 6 and 7 together to buffer. Connect R3 from pin3 to pin7. The second half of the IC will act as a regulator for 4.5v. It's not necessary, but it's a nice feature that doesn't require anything other than moving some parts around.

I had this question in a different thread but it got lost in the mix. I really would like to learn about this, can someone tell me or direct me to a link that explains what bean is talking about. How/Why is buffering Vb(voltage bias?) beneficial?

[R.G.]

Reference voltages are one of those places where almost everything matters. They need to be quiet, because they are fed to amplifying stages that are supposed to reproduce the internal voltage exactly. Any noise comes right out the output. They need to be at least moderately accurate in DC terms because they also set DC levels, which was important enough to make a reference voltage in the first place. No reason to do it in a sloppy, inaccurate way.

If you read my article on choosing resistors for a Vref supply at geofex, you'll find some of the reasoning about keeping a low impedance for Vref. Doing Vref with resistors is a balancing act between low enough resistances to keep the DC voltage accurate and how much current you can pull from the Vref node. This doesn't matter much when all you're feeding is 1M bias resistors to the inputs of JFET opamps like TL072s, but it matters a lot when you feed current to other circuits that aren't so high impedance.

Enter the buffered reference voltage. Using an active Vref means that the Vref DC level will remain the same even if some of the loads pull a lot of current from it. That means less interaction between those loads connected to Vref, and a generally more accurate setup. That's why one might do it, especially if one has a free opamp for that.

There are down sides as well. A buffered Vref has noise from the amplifying device that a resistor/capacitor reference may not have, and has an impedance that gets higher as the buffer's feedback gets worse at high frequencies. And you can't necessarily just cram lots of capacitors on the output of an opamp without making it oscillate, so you can't just solve the noise/impedance thing with more capacitors.

Like most things, it's a balancing act - better/ more accurate DC performance and less load interaction versus high frequency considerations.

[jimmybjj]

Thanks RG, an explanation I can just about understand  I will read and try to absorb all that you have presented. Thanks for the reference to the other reading material on your website. There always seems to more material on there that I haven't discovered, it may be that I'm still at that steep learning curve point and I don't understand all that I read so when I have something to relate to it, it looks new. 

[jimmybjj]

Reference voltages are one of those places where almost everything matters. They need to be quiet, because they are fed to amplifying stages that are supposed to reproduce the internal voltage exactly. Any noise comes right out the output. They need to be at least moderately accurate in DC terms because they also set DC levels, which was important enough to make a reference voltage in the first place. No reason to do it in a sloppy, inaccurate way.

If you read my article on choosing resistors for a Vref supply at geofex, you'll find some of the reasoning about keeping a low impedance for Vref. Doing Vref with resistors is a balancing act between low enough resistances to keep the DC voltage accurate and how much current you can pull from the Vref node. This doesn't matter much when all you're feeding is 1M bias resistors to the inputs of JFET opamps like TL072s, but it matters a lot when you feed current to other circuits that aren't so high impedance.

Enter the buffered reference voltage. Using an active Vref means that the Vref DC level will remain the same even if some of the loads pull a lot of current from it. That means less interaction between those loads connected to Vref, and a generally more accurate setup. That's why one might do it, especially if one has a free opamp for that.

There are down sides as well. A buffered Vref has noise from the amplifying device that a resistor/capacitor reference may not have, and has an impedance that gets higher as the buffer's feedback gets worse at high frequencies. And you can't necessarily just cram lots of capacitors on the output of an opamp without making it oscillate, so you can't just solve the noise/impedance thing with more capacitors.

Like most things, it's a balancing act - better/ more accurate DC performance and less load interaction versus high frequency considerations.

ok, so I'm trying to understand this. here is my super summary of this, is it correct?
vref needs to be quiet so regulating it (keeping it at a stable voltage) makes it less noisy (as I understand it voltage fluctuations
equals noise). by using an buffered (active?) vref you can avoid some noise, to a point. buffered vref will be less affected by a load on the circuit.

so what I am getting out of this is buffered vref could potentially be a less noisy circuit, right?

thanks for any help.

[merlinb]

so what I am getting out of this is buffered vref could potentially be a less noisy circuit, right?

Potentially yes. The buffer can source/sink plenty of current while holding the voltage firm.
However, there aren't many pedals that ever need such current from Vbias, so a buffered voltage is usually unnecessary, or just a nicety when there's a spare opamp left over.

[Gurner]

so what I am getting out of this is buffered vref could potentially be a less noisy circuit, right?

Hmmm, did you not digest this bit....

There are down sides as well. A buffered Vref has noise from the amplifying device that a resistor/capacitor reference may not have,

I'll leave the eloquent explanations to others...I like a buffered Vref becuase I can depend on it (I can't be done with having to ponder whether a circuits current draw on a simple potential divider is going to make my required reference voltage stray)...I can't say I've noticed the extra noise, but as RG points out ...it'll be there.

[petemoore]

  RG stated the LM386 makes an excellent stiff, 1/2vref.
   + to supply, Gnd. pin4, the output is 1/2v. Very simple and effective, takes an 8pin socket space on the board.