I need help with a basic circuit to allow variable gain (basically, an analog volume control) with a single-supply op-amp. I've done dozens of dual-rail audio op-amp circuits, but single supply makes things considerably more difficult with AC signals, and I've never really learned how to cope with that.
Here's what I want to do: There's a buffer stage, then I need a volume control (unity to neg. infinity), then various filters. Oddly enough, I can find good examples of HP/LP/BP filters from TI app notes, but a simple potentiometer gain stage is nowhere to be found. Anyone have a good example, or at least better Google-Fu than I?
I've attached the little I have so far, with the "?" being the gain stage. VCC/2 (the signal ground) is accomplished via a voltage divider feeding an op-amp to provide a stable 2.5v reference (given a 5v rail).
Cin is required so the Input Source can reference it's own Ground... same with Cout.
If you'll be connecting a 2nd stage after this, which also uses the same virtual ground as 1st stage, then you can remove Cout. Otherwise, if interfacing with another system with its own ground, you need Cout.
Virtual Ground generation using opamp... though you can use 2 resistors if you want to further simplify.
** not shown here are bypass caps for Vcc, use 0.1uf to ground.
** not shown are bypass caps across R2... 47pf to 100pf will do for most opamps. (check with scope for any oscillation at your max. gain setting and pick appropriate value)
** you need a REVERSE audio log pot, also sometimes called CCW audio log pot. (not just an ordinary audio log pot)... Otherwise, your volume control will be "reversed" if you just use a normal audio log pot.
Actually the first, inverting amp ckt will work ok, just use the virtual gnd opAmp
ckt of the 2nd diagram for Vcc/2 biasing.
Secondly, in order to work, the second noninverting ckt needs DC biasing on the + terminal
side. Eg, a 22K to Vcc and another 22K to gnd will work there.
Thirdly, vasquo is correct about putting a cap across R2 to produce a 1st-order low-pass
filter. The 3db frequency is simply given by F3db = 1/(2piR2*C). You may wonder why
R1 is not in the equation, but the virtual ground point at the - terminal effectively removes
it from the issue.
Guys, thanks so much. I'll have to wait a little bit before I can get into simulations, but I see where this is headed and I get it.
I'm with you on the input and output Cs, that's standard practice on any of my audio circuits.. despite, so far, being ground-referenced. You never know what you're hooking up to. I understand it's sometimes required between stages of VG circuits, since any gain affects the DC bias as well.
Regarding the anti-log pot, I usually use a linear pot + parallel resistor to fake-log anyway, so I'll just end up swapping the placement of that R to achieve the same effect. No problem.
Your virtual ground buffer example is almost exactly what I used (I left it out because hand-drawn schematics in Paint is a chore), so that's crystal clear.
For audio circuits, I use unity-gain stable op-amps, so if I'm not mistaken, the C across R2 is not required to prevent oscillation, right? I've never used one before, and so far (knock on fiberglass substrate) everything works fine. If I'm being naive, say so. (I have started adding RF-blocking C filters on the input, though.)
For audio circuits, I use unity-gain stable op-amps, so if I'm not mistaken, the C across R2 is not required to prevent oscillation, right? I've never used one before, and so far (knock on fiberglass substrate) everything works fine. If I'm being naive, say so. (I have started adding RF-blocking C filters on the input, though.)
Yes to oscillations. I think if the gain is more than 1, you should always have a cap
across R2. Also, as mentioned, it makes a good low-pass filter too.
I just realized, the example above will never have a gain less than 1. I was hoping for 0 to 1.0 gain - a volume control. In a normal dual-rail design, the first stage output would feed a pot's right terminal, left terminal to ground, and the wiper connected to the non-inv. input of the next stage. I don't know what I was doing wrong before with my simulations, but with the left pin connected to the virtual ground, my circuit would glitch and oscillate. Now it's behaving exactly like it should. I must've just overlooked something obvious. (I didn't think it should be so hard.)
Well, hey... at least I got some good ideas out of the discussion. Thanks for that!
Next step is to take this to the breadboard and make sure it still behaves the way I expect.
I must've just overlooked something obvious. (I didn't think it should be so hard.)