after a voltage regulator with push pull op.

Somebody, TI I think, used to make a rail-splitter thing; don't know if it's still around.

John

Hi, I was looking to simplify a voltage supply to generate +5/-5 v (+-10%)

Hello Colin

This is commonly done via a 50/50 resistor divider and an opamp as a follower. That can source and sink as much as the data sheet says and is cheap.

Just be careful with capacitive loading at that node. Some (most) opamps don't like that but there are work arounds in National's app notes and others. And in AoE I guess.

Regards, Joerg

Hello John,

Nice chip even if a bit on the pricey side. Colin should just keep in mind that it's a SOT23-5 and after so many hundred mW blisters and smoke will show.

Another option if he needs lots of power is to use an audio amplifier IC. These are really cheap.

Regards, Joerg

ooh thanks thats quite a nice little beastie :)

might just about handle the current diference, at 5v the thing shldnt vapourize too quickly even at short circuit. interesting that it handles such capacitance.

I forgot to mention I considered an op amp, I usualy use a spare op amp but they dont like it when you just try and hang a big capacitor on them, I need a low impedance al the way down to dc.

Im just surprised/disapointed there isnt a regulator that is designed for it, as im sure its a comon thing to want to do, would be nice to have the added protection of over temp, over current, over voltage with wildly unbalanced loads, over anything else etc. especialy when prototyping.

my idea of 7.5 zeners acros the +/- regulator inputs still seems quite atractive from this point of view. (especialy as the regulators are already there atm) or I cld use the op amp to balance the input voltage to the 2 regulators.

Colin =^.^=

Hello Colin,

That can be handled. Look through National's appnotes. Capacitive load + opamps.

Can be risky. Zeners have a lot of tolerance. If Vz is too high the central node meanders around. If too low they'll short the supply. With zeners there ain't much wiggle room in between these two situations.

Regards, Joerg

already

I would need vz to be in the range 7.5 to 9v with ldo regs thats assuming the 15v is steady wich it isnt realy, the supply has some decoupling resistance so it wouldnt hurt if they both conducted slightly anyway.

I also considered 2 transistors, normaly one or the other would conduct keeping the input centered within a volt or so.

-+--+-----reg---- | | | 5kr | | | c | +-b | | e |

300r +------+----- | e | +-b | | c | 5kr | | | | | -+--+-----reg----

gona use it in several places tho so would be realy neat to have a little soic 8 chip that did it all realy, maybe some semi conmpany is listening :D

Colin =^.^=

Hello Colin,

Nah, why should they if it can be done with amps? Ok, so now I looked it up for you. The magic trick is shown in figure 15 and the text around it at the end of this application note:

Regards, Joerg

little

:D

thanks thats a usefull appnote, although r4 doesnt afect the output resistance so much it would afect the output curent although 50ohm would just about do, it does make that other op amp look so cool.

Colin =^.^=

(+-10%)

way

but

5v 10ma

and

bias

oo thanks, thats got my hopes up, I spent some time looking today at voltage regs/references but didnt see any, maybe il look again .. for some reason not all devices seem to be easily found on some manafacturers sites unless u know the url or partnumber, maybe they want to keep them for themselves ? :g

Colin =^.^=

Hello Colin,

Remember that this is a very old app note. If you use a modern amp with a more stiff output R4 can be much lower. Plus, the mid rail wouldn't see that resistor because it is regulated out. The fast stuff is taken care of by, well, the cap.

Regards, Joerg

TI I'm sure and no doubt others make dedicated ICs to provide that midway reference.

You don't need 'single supply' type op-amps if you go that route.

Graham

some

inputs,

in

src

for

midway

voltage

to

oo after much searching i came across reg1118 and lt1118 it seems both originaly intended for scsi termination voltage source supplies, and utc560, however they al seem to be somewhat unavailable exept for the 2.85 v version, but they can source and sink plenty of current.

Colin =^.^=

This was the first thing I was going to suggest - it has all of the benefits you've mentioned, and you're going to need caps anyway.

Have you bothered to figure out how much current your negative rail really has to sink?

Good Luck! Rich

Search on SDRAM terminators, e.g. the LP2996. These are readily available and cheap.

Regards Ian

In article , colin wrote: [...]

This is a bad way of doing it. If the -5V supply gets overloaded, the +5V supply ends up being 10V. It is better to use a pair of 5V regulators like this:

-------- ---------+-------------+-----! LM7805 !-------- +5V ! ! -------- ! ! ! ! \\! ! ------- ! \\ ! ! 5VREF !-------! >---------+------------ 0V ------- ! / Push pull amp. ! /! ! ! ---------+-------------+------------------------- -5V

--
--
kensmith@rahul.net   forging knowledge

way

Hi, thnks yes good point, although for the op amps it wldnt be a problem, id thought of doing it that way but instead if the 5v rail gets overloaded then the -5 v rail becomes unregulated, but im also using it for a max038 wich I wouldnt like to risk overloading either rail so im thinking il use the pos and neg regulator aproach.

However idealy a single device would shut down totaly under any fualt condition, another aproach i might use a charge pump for the -5v this would mean lower suply voltage and also one side of the supply is gnd wich seems neater, maybe save me a wire too, they seem quite neat small package and only a smal capacitor just lose about .6v on the -5 rail, might be able to compensate, also fail safe.

still, some intersting little beasties have eventualy come out of the woodwork :)

Colin =^.^=

cool seems like these are actualy stocked by RS anyhow, thanks :) adjustable too. be real neat if they added a normal output

Colin =^.^=

problem,

overloaded

max038

use

gnd

might

Well yes ballpark figures anyway, for neg supply its virtualy just the quiescent supply for the op amps, few ma depending on what op amps I find perform best (ultra low noise), not sure I like the idea of the effects of switching near