Symmetrical power supply from a single rail

[TrickyNekro]

Hallo again with another question,

This of course is related to some extent to my previous one but that´s not the point.

I need some pointers here only, I need to build a symmetrical power supply, really low currents 100 - 200mA
worst case scenario. The outputs can be whatever from +/- 7.5V to +/- 12V, I will use linear regulators after
the switching one. The extra rails will be powering precision op amps so the need for linear regulators is mandatory.
The problem is that I don´t know what to look for, as in the name of the topology, or if
someone did it before he/she can suggest a part number.
I am not really sure about the input voltage that I want. But it can range from +5V to 12V, unfortunately, I
can not be sure yet, there are many more things to consider.

So, the question? What am I looking for? Sure enough it´s a DC/DC converter but I get lost sometimes with
topology names! And a IC suggestion will be much much appreciated!!!

BTW there is not really a space restriction, I am building this for a robotic platform, meaning that I don´t really
need a one chip solution. It can be two also. And I don´t have a problem with using inductors, I´d rather stay
away from charge pumps, I don´t really like them, they don´t really have a high enough current output and
the voltage swing can be terrible, so no thank you.

[Alana]

https://www.eevblog.com/forum/projects/layout-review-inverting-buck-converter/ <-- something like this?

[Skimask]

Just got done building a few of those.
LM2588.
4v-6v input, +12v/-12v output @ ~300mA.
It's all in the datasheets.  Works like a champ if you follow the directions.

[dannyf]

Most dc/dc converters can be configured to produce a negative voltage.

[TrickyNekro]

https://www.eevblog.com/forum/projects/layout-review-inverting-buck-converter/ <-- something like this?

Yes, why not, but! You got a small but potentially big problem with your design. I don´t know if it´s a coupled inductor you used,
but if not, you have to remove the ground plane below the inductor and a little bit larger area below the
inductor, plus maybe more the inductor a little bit further away from the other components.
You have to provide the inductor with a magnetic field return path, it´s almost like a must. Right now it´s like inductively heating
the copper below, losing a lot of efficiency and potentially creating noise sources!
This not me telling you this, these are Maxwell´s equations. The magnetic field is formed around a closed path. Eventually it will
close sure enough, but you are effectively killing the "nominal" value of your inductor, usually by a lot.

But anyhow that´s a friendly advice from someone who studied physics 

Just got done building a few of those.
LM2588.
4v-6v input, +12v/-12v output @ ~300mA.
It's all in the datasheets.  Works like a champ if you follow the directions.

Heh, good one! thanks! But the circuit suggested in the datasheet can be tremendously improved with a higher though
part number. But really good find thanks!

Most dc/dc converters can be configured to produce a negative voltage.

Hey, come on, give me some credit 
I´ve worked with MC34063 before, but I don´t have a vast knowledge on DC/DC converters. I am more asking about
successful romantic stories 

Anyhow, thanks for your time and replies!
Best Regards,
Lefteris, Greece

[tom66]

I've used a DC-DC buck with a negative charge pump running off the switching waveform. Works acceptably up to 30~40mA but the output is equal to the input minus a few diode drops, it's not regulated, and you need to ensure a minimum load on the positive rail OR choose a converter which never enters pulse skip mode like one designed for audio systems. (Constant switching frequency = easily filtered.) There's also a very nice LT part with dual boost/inverting converters I've used before, constant frequency if you choose the right version but it's only available in a tiny DFN package which is a pain to work with. LT3471 from memory.

[TrickyNekro]

Thanks for the suggestions, especially the LT part seems promising although more advance, I would prefer to
use something more easy at the beginning. Still, the double feedback is really a good feature! These packages
can really be a big pain especially when you are building your own PCBs at home, but as long as they are not
BGA everything is possible! :-)

And yeah, I kinda really hate to use charge pumps. They are a good solution when you only need a reference,
or just power something small but not for my case, plus I consider them messy. You are depending a lot on these
capacitors. I need really stable and reliable rails, I have to use linear regulators after the switching ones.