2
\$\begingroup\$

I am trying to use an LM358 opamp to amplify an electret mic to something that I can read with my attiny85's ADC. (What can I say? I like DIP-8's :)

I feel that I need a gain of about 100 to get a decent signal.

I'd like to use a single-supply configuration, with a voltage divider to provide a virtual ground.

My first design used one half of the LM358 as a basic inverting amplifier with gain 100, and the second half to provide a virtual ground as in:

http://www.swarthmore.edu/NatSci/echeeve1/Ref/SingleSupply/SingleSupply.html

Now, the LM358 has a gain-bandwidth product of 1 MHz or so, which means that my 100 gain will start to be affected at around 10 kHz.

What I'd like to do is to use both halves of the LM358, each with a gain of say 10 and a coupling capacitor between them.

The page I list above indicates that the virtual ground current of an inverting amplifier is zero, which suggests that I should be able to use a resistor-divider for both amps.

Does that sound right? Or am I missing something?

\$\endgroup\$
  • \$\begingroup\$ TI application note 31, figures 13 and 14. \$\endgroup\$ – Ignacio Vazquez-Abrams Sep 12 '13 at 15:00
2
\$\begingroup\$

If the resistors making up the voltage divider are of low enough value, and the supply rails are sufficiently low impedance, yes, a single resistor divider can be used to provide virtual ground to multiple op amps.

In order to stiffen up the ground a bit, a capacitor may be added between the supply and ground rails at the resistive divider, and a second capacitor between the virtual ground point and either of the rails.

schematic

simulate this circuit – Schematic created using CircuitLab

Alternatively, a dedicated virtual ground IC such as the Texas Instruments precision rail splitter TLE2426 comes in handy for the purpose.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ Thanks for the response. The resistors you show there are 220 ohms. With Vcc at 5 V, we're talking about more than 10 mA. That seems high to me. Would 100k or 10k resistors work? \$\endgroup\$ – user2557313 Sep 12 '13 at 15:30
  • 1
    \$\begingroup\$ @user2557313 The actual resistor values are dependent on what impedance is expected to be feeding off the virtual ground. For experimenting I generally bung in 220 or 470 Ohm resistors because that takes away any concern about the impedance of the next stage. For a prototype or release design where 10 mA is a concern, I wouldn't cut corners by using a resistive divider anyway. However, if this virtual ground is needed just for a couple of LM358 inputs, then sure, go with higher resistances. Keep in mind that the higher the resistors, the more there will be sagging and cross-talk. \$\endgroup\$ – Anindo Ghosh Sep 12 '13 at 15:35
  • 1
    \$\begingroup\$ @user2557313: There's no free lunch. If you use higher resistor values and there's a tiny amount of current imbalance flowing into the vground, it will shift it proportional to the difference you made in the resistor values. I wrote an article on this topic. Right up at the top, it covers a case where using 4.7kΩ resistors can cause a substantial virtual ground shift due to the load the circuit is driving. The rest of the article talks about ways to fix this other than lower resistor values, including the TLE2426. \$\endgroup\$ – Warren Young Sep 12 '13 at 19:24
  • 1
    \$\begingroup\$ I went with a couple of 10k resistors and a 1k mutiturn trimmer pot between them to get the divider as close to 1/2Vcc as I could. Then both halves of the LM358 in basic inverting amplifier configuration. with 0.1uF capacitors and 10k resistors on the inverting input and 100k feedback resistors. I get a nice clear signal at my ADC. Thanks so much for everyone's help. \$\endgroup\$ – user2557313 Sep 13 '13 at 11:12
2
\$\begingroup\$

If you cascade two inverting op-amp sections where both circuits have self-contained feedback, then both LM358 op-amps can share the same resistor divider mid-point - it will connect to their non-inverting inputs and values can be reasonably high in value providing some ground decoupling is used such as 1uF or thereabouts.

I wouldn't go higher than 100k for each because bias currents may offset the midpoint by over 0.1V. Also beware with the LM358 that it won't swing it's output up to anywhere near close to the positive supply rail so if you are considering ruiing from a low voltage you need to read the data sheet and possibly bias your "mid-rail" closer to 0V to get a decent symmetrical voltage swing on the output (or use a rail-to-rail o/p device).

| improve this answer | |
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.