LM3886 as a PSU?

[eccdbb]

Is it a good idea to use a power op-amp like the LM3886 to split a supply? For example 24v SMPS into a V-divider G=1/20, then LM3886 G= +20; output is (ideally) +/- 12v.

Would it be better to also include a pole at <20hz to filter out noise?

[eccdbb]

Oops I meant 1/40, not 1/20. How do you edit posts?

[AndrewT]

"edit" icon in bottom right of the post. But, it is time limited to 30minutes.
You should always proof read your posts during posting.

The 3886 is not low noise.
That may exclude it from PSU duty.

[eccdbb]

Interpolating from the THD+N vs. freq plots, does .1 THD+N at DC mean too much noise?

[Michael Bean]

It's probably not the best way to go, trying to use a device in a way for which it wasn't designed usually ends up in disappointment. If you need a voltage regulator for higher currents, you can add a pass transistor to any garden variety three terminal regulator or parallel multiple regulators. Take a look at some of the National or LT regulator data sheets for ideas.

Mike

[jackinnj]

I note that the folks from Belleson used an LM3886 to demonstrate line rejection in their regulators. This chip isn't that fast, (like the AD825, AD797) but why not give it a try and report back! Keep mindful that it does like to be well heat sinked. This would be a TLE2426 on steroids.

[eccdbb]

I don't know how to use transistors, don't have a LM3886, and don't have a scope

Experimenting with a ON 0372DP1 now...

[godfrey]

Originally Posted by eccdbb

Is it a good idea to use a power op-amp like the LM3886 to split a supply?...

Yes that works, but I would use a circuit something like that shown below. R1 and R2 act as a voltage divider to get the mid-point of the input supply voltage, then the LM3886 is used as a unity gain follower. R3 and R4 are just there to reduce the feedback around the LM3886 to keep it stable.

R5, C1 and C2 are optional, but make life easier for the LM3886. Assuming the split supply is for an audio amp, most of the speaker return current on the center rail will flow through the capacitors instead of the chip.

One serious warning though - For a split supply, you'll probably want to ground the center rail. In which case you need to make sure that neither of the other two are grounded inside the 24V SMPS. If it is (quite likely), you'll need to disconnect that ground.

[Johno]

ecdbb,
Don't let these unimaginative, conservative guys put you off. Your idea is a good one, while their reservations might be valid only in some situations.

Dedicated voltage regulator ICs are notoriously full of noise, high impedance and horridly low frequency response. High quality op amps, including high power audio amps, overcome all these deficiencies. Plus the 3886 et al are bullet proof with over current, thermal and voltage protections.

FYI my pre amp uses LME49720 plastic op amps after the regulators in unity gain buffer mode feed the rails of the amplifier chips which are also LME's.

[eccdbb]

Got my 0372 working, but it gets hot if I include output caps (C1 and C2 = 1uF, no R5). Why is that happening? If I take them out, it runs cool.

Is the LM3886 and the rest of the LM's unity gain stable?

Is it possible to use 2 stages of inverting op-amp as a very low pass filter, and then pass that to the V-divider? If the virtual ground and 0v ground is clean, but +24v is dirty, does that mean a half-clean PSU?

godfrey: how did you pick the ratio of R3 and R4, and how do they work? Is the power dissipation in R5 equal to R5*Iout^2 ? Why do you need to ground the center rail? What if I can't verify the SMPS?

Johno: conservative is good for now =) I don't exactly want to start a fire yet; no tools to find what is going wrong... Are you using a single or multiple LME49720's nested to increase output current? Is that for a headphone or speaker amp?