amplifier and rail-splitter forming a filter?

[Dominique]

Hi,

this is half-way a power supply, half way an amplifier problem...

as I wanted to design an easy microphone amp for a single battery but don't like those big electrolytic caps necessary in an op-amp circuit with single supply, I thought of using a dual op-amp:
One channel to split the voltage and to provide the virtual ground, the other to be the mic-amp.

So I simulated the circuit and it seems to work fine except one thing:

If the output resistors R3 and R4 are both 100 Ohm for example, the whole thing seems to be a filter, the frequency response shows a bump at ... well 10Hz or so and is far from being flat. Also this happens at R3=100 Ohm and R5=10 Ohm ( I played a bit with the values to see how the frequency response would look like), but not for the shown configuration, R3=1k, R4=100Ohm.

Unfortunately I don't see why that is, so my question is:

* Do you see a reason for the "not flat" behaviour?

* How to choose the values / modify something in my circuit in order to prevent those problems.

Please help!

Cheers,
Dominique

PS. Ok, my micamp circuit isn't well made yet, and the "dual" op-amp is named Op27 here , but that's not important for the problem...

[Dominique]

Here's the frequency sweep when "it works".
e.g. R4=100R, R3=1k

[Dominique]

Here's the frequency sweep with the bump!
e.g. R4=100R, R3=100R

[Dominique]

PS. If anyone doesn't know what I mean by "rail-splitter", please look here:

[Fixup]

Believe me, you don't need that X1 rail-splitter. Change R2 and R8 to 4.7K, parallel them with 4.7uF tantalum caps, remove X1, the middle point of R2 and R8 is your virtual ground.

The tantalum caps work also as power bypassing for low frequency for X2. Do simulation again, everything will be fine. If low frequency is still not flat, increase the values of the tantalum caps.

Active rail splitters never work in hi-end audio. In circuits that do not drive heavy loads (like this one), two resistors filtered by caps do the job without introducing other troubles such as noise and phase shift. In circuits that drive heavy loads, then use X1 for the load return, and only for the load return.

[Fixup]

The problem with active splitter is: the power bypassing caps are loads to the splitter. And we all know: most opamps cannot drive capacitive loads.

[darkfenriz]

You have two ples in transfet function- one is internal pole of an op-amp and the second is created by those bypass capacitors to ground C1 and C4. My advice is skip those caps if you want splitter.

[Dominique]

@darkfenriz
leave the the ps bypass caps away?
Ok, if that solves the problem...
I'll try that on breadboard tomorrow.

@Fixup
Ok, I think I'll try that easy solution too!

Unfortunately I forgot that the circuit needs a resistor, about 4.7k from the positive supply to the signal of the mic (right before the input capacitor).
So it would introduce an imbalance. Is it ok to choose the resistor from the positive rail to the mic 5.6k, the pos splitting res. 4.7k and the negative splitting res. then 2.7k? Then ground should be more or less in the center again?
Or is it better now to make both of those splitting resistors smaller?

Thank you darkfenriz and Fixup!

Cheers,
Dominique

[Fixup]

Originally posted by darkfenriz
You have two ples in transfet function- one is internal pole of an op-amp and the second is created by those bypass capacitors to ground C1 and C4. My advice is skip those caps if you want splitter.

Like I said, these bypass caps cause trouble to the active splitter, but the right solution is to get rid of the splitter, not the caps. These caps are absolutely necessary for the opamp.

[Dominique]

Hi again,

getting back to my previous concerns about that resistor which I forgot in the first schematic:

Here's the updated schematic, but I'm not sure if is it ok with the adapted split resistors...?

Cheers,
Dominique

PS. In my first mic preamp, I left those caps at the supply pins away and didn't have any problems at all. I used low impedance batteries, an opa2228 and everything was together in a grounded metal case, so that possibly helped.
Since everybody says to use those caps, I included them on all my later circuits.