LT1010 rail splitter not splitting

The input bias current is specified as max 250 uA at no load, 500 uA at

Dropping the resistors to 1k instead of 100k will get you closer.

Cheers

Phil Hobbs

Thank you sir. So this part is no good for a rail splitter? There's a circuit in the datasheet that shows it beingused as a rail splitter, but it doesn't work.

Is this right: 100k is to high a value because it doesn't allow enough input bias current to flow, so I reduce it in order for the thing to work properly (250uA for no load).

So even 10k is too big because 250uA through the 10k will throw me off by 250uA*10k = 2.5V?!

Wrong part for the job. This is because the part has a bipolar transistor input isn't it? I'm going to try an OPA656 FET input Amp.

much thanks!

Right. The 1010 also has a bunch of offset voltage. It's sort of a beast.

If the current is low, a c-load opamp would be better. Or a voltage regulator, maybe, if the load polarity stays to one side.

Aren't there specific rail splitter chips?

Yup.

The OPA656 is a bit hot for that job--rail splitters usually have to drive a lot of capacitance, which will make the OPA656 oscillate. The LT1010 is an open-loop part, and is designed to drive a lot of capacitance, so it isn't a bad choice at all. It does have an output resistance of 12 ohms or so, so the regulation won't be perfect if you start drawing a lot of output current.

Since you're using a plug-in power supply, a few milliamps more or less won't worry you. I'd probably just use 500 ohm or 1k resistors instead of 100k, and stay with the LT1010.

TI also makes a dedicated rail splitter, the TLE2426, which has a much lower output resistance (about 8 milliohms) but only goes to 20 mA and needs at least 1 uF of capacitance on its output.

Cheers

Phil Hobbs

The LM8261 is c-load stable, and the LED looks nice.

This gates an analog signal into an acoustic-optic modulator, exact reason unknown to me.

There are some words that go with that circuit, like it is for flea-power loads such as split rail OA and comparators which draw relatively small current.

Not at small load currents where the bias is 100uA worst case. Then the offset induced by this current into the 10Ks is 0.1 mA x 10k||10k=0.5V which is 5% or less for supplies of 10V or more.

Yep- but that's your fault.

The LT1010 is intended for composite amp configurations so of course it looks ridiculous standalone. It's a Widlar creation.

TLE2426, (Sphero turned me on to those... that man's an IC pusher.)

George h.

^^^^^^ oops Spehro (sorry)

I don't think it's ridiculous, it's just old. I've used parts like that in the past, usually to move power dissipation away from a low drift stage. My fave used to be the LM6321 for slower stuff and the LH4009 or LH0063 for faster.

A lot of circuits aren't very sensitive to the exact supply rail voltages, so it could do a pretty good job there.

Cheers

Phil Hobbs

Has he gained weight recently? :)

Cheers

Phil Hobbs

You will be roundly criticized for starting another pun thread.

I've used it as an opamp power booster, inside the loop.

Hmmm, like like 18K pieces so far.

I could have used the TLE2426 in a design I did some time back. One of the upgrades I did was to improve the virtual ground. But the board has no room... literally *no* room for such huge parts as an SOIC-8. I wonder why they don't put this in a more space efficient package? I thought it might be the pin count, but the full die only has 4 connections so the 5 pin version of a SOT-23 would do the job. There seems to be a 4 pin version called that SOT143 too. Same size as far as I can tell, 3 x 2.5 mm.

I guess not so many who are cramped for space would be using a part like this. I notice it is a pretty old part too, so not likely to see any new packages.

Well, you're ahead of me there, but I still have a couple of dozen LH0063s and LH4009s. ;)

Cheers

Phil Hobbs

That's a whole 'nother ball of wax.

Cheers

Phil Hobbs

Yup. You can get more current out of John's fave LM8261, from a SOT23-5. With a big package like an SO16, you can get a TCA0372, which is a dual with a 1.4 A peak output per channel. It can't do that for long, of course, because the package thermal resistance is horrible.

Cheers

Phil Hobbs

Well, Mr Smarty Pants, I have a CK722.

In your hearing aid, right? ;)

Cheers

Phil Hobbs