Rail-splitting a wall wart

I like the little CUI dual-output dc/dc converters. Or one of the LM266x inverting charge pumps. But they might be noisy.

The little 3-terminal switchers will invert too. Some are around one dollar.

I'd expect that a TIA wouldn't need much current, so a rail splitter is sensible.

Wouldn't a negative regulator, an LM337 or a 7905, work as a rail splitter?

fredag den 3. februar 2023 kl. 21.50.16 UTC+1 skrev John Larkin:

Lm317 + LM337

Wall warts are only isolated at DC! Some have horrible HF noise.

Only you know the kind of users but a lot of folk have trouble getting their mind around the idea of the non-positive supply rail not being ground. The near standard barrel DC jack with center pin positive encourages that notion so you may want to consider using a non-standard power connector?

piglet

----------------------------------------------

** Being made to meet Class 2 means having galvanic isolation from the AC supply and broad immunity against insulation failure without reliance on safety grounding. ** Indeed, many SMPS warts have a 1000 to 2000pF class Y cap from the hot side to the output.

.... Phil

A regulator typically can't sink current, so while it would hold the voltage down, it won't prevent the voltage from rising.

My audio mixer amp uses a BUF634P to split a 12VDC wall wart into ±6VDC:

Also, the audio mixer amp's notes contain a lot of useful links:

Danke,

You aren't in need of high current, are you? Two zeners and a bit of R for bias, and C for zener noise. Go ahead and waste a watt. 50 mA and 9.1V

This should work:

Maybe, but if ground-to-minus current is over 5 mA, the negative regulation vanishes (it needs bias current to stabilize the zener voltage).

Most situations more V+ than V-.

If this circuit won't work for you, don't use it. Or add a dump resistor somewhere.

Yeah, it's possible to bodge regulators to do that sort of thing after a fashion, but getting decent performance in all four quadrants takes a lot of parts AFAICT.

I'm leaning towards using a TCA0372 and a voltage divider to do the actual splitting, and then fixing it up for overcurrent with a dual comparator with mega hysteresis and back-to-back MOSFETs to cut the power.

Thanks

Phil Hobbs

I suggest reading Jim Williams' LTC AN-70 about push-pull power supplies. The circuits use transformers which some people balk at. If it doesn't meet your needs it is still a good read.

yeh, an opamp is easiest and it looks like that one is thermally protected so it should be pretty safe maybe add poly fuses to the input supply in case someone does something silly

We have that problem already, even in products with inverting switchers making the negative supply.

Simon has been hacking KiCad to make it easy to plunk in our standard IP blocks, such as cap multipliers for various current ranges, fast quiet PGAs, MCUs with Modbus over RS-485, and so on.

We have these nice machined brass boxes with feedthrough caps that we're using for lots of stuff these days. There's a 3D printed section attached that has space for connectors and line filters, so that crap basically never gets inside the box.

So filtering the wart's output before letting it into the box is fairly easy.

Cheers

Phil Hobbs

you machine the boxes? then machine a separate cavity for the inverter?

Yeah, it's pretty cheap to do machined lead brass, because it cuts so easily. (You lot can't use it on account of the regime, but oh well.)

We could do separate cavities and all that, but due to the getting-off-the-couch setup charge, it's better to use one box design and hang 3D printed stuff off to the side. (It all looks like DKNY black/gold bling anyway.)

Cheers

Phil Hobbs

SMD RF shield cans are pretty cheap too

Yup. We use them a fair amount to control local interference, but just letting a noisy wire into the box creates problems from unknown outside stuff.

Cheers

Phil Hobbs

would you have the same issue with a wall wart?