I am designing a new board, based on a Atmel AVR for my college project. And I want to have the 65SIB port on it. However I got to fit it all in a footprint of 2.5x3.8" and I was wondering if anyone had a idea to make a tiny +/-12V supply that I can put on that small board and still not take over the whole board space?

The power will be off car batteries, (3 in series, so 12/24/36V inputs are possible).

Any ideas would be great.

Dimitri

Most of the things that would use the ±12V do not need it accurate at all, and that makes the job much easier. The reasons are given in the voluminous thread where we deveoped 65SIB. I usually only run mine at around ±9V, and they aren't really matched either.

So it sounds like you already have the +12V, although if charging could run it a lot higher (I had a car that ran at 16V, and the only problem was that lights kept blowing out-- the Interstate battery even lasted 10 years!) then a simple linear regulator to hold it down would be in order. An LM7812T, or better yet, an LM340T-12, in a 3-pin TO-220 package would bring it down to 12V but wouldn't really regulate until the input is around 13.8V or higher. If you bring it down to something like 10V with an LM317T which allows you to set the output voltage, you should do fine.

A "cheap 'n' dirty" way to derive the -12V from the +12V line could be with an LTC1044A switched-capacitor voltage inverter in an 8-pin DIP, which I've used. A lot of related switched-capacitor units cannot handle that much voltage, but this one is specified for up to 12V. The negative voltage needs are for pretty low currents like LCD bias or some D/A converters and op amps. National's related LMC7660 is spec'ed for up to 10V. They can be paralleled for more current capacity. If you had somethng that needed a lot more (like say a quarter of an amp), the solution would be a little more complex; but any 65SIB device that needs much more than that should generally have its own power supply, because of the current-carrying limitations of the ribbon cable.

You have a string of batteries giving you 0, +12, +24, +36 volts
Make battery +12V the board ground, battery 0V the board -12V
and battery +24V the board +12V

Since SPI uses all uni-directional inputs, and if you choose to go with Bogax's recommendation, then you will need to use opto-isolators between the board and the peripherals you attach to it. Otherwise, one device's 0V GND, when connected to the 65SIB bus GND (which would sit at +12V relative to earth), would likely result in hilarity, anger, and lots of escaped smoke.

BTW, this problem is endemic with all transportation-based controller devices. There's a reason why most automotive, aeronautical, and astronautical devices use balanced signals instead of GND-referenced, unbalanced signals.

He could also use something like a Fairchild KA278R12C LDO regulator, which works well when the battery voltage is only slightly higher than 12 volts.

_________________
x86? We ain't got no x86. We don't need no stinking x86!

Good point, although in this case the voltage is seldom critical, and I don't recommend LDOs for the casual user who may not want to put the effort into understanding and preventing the potential stability problems. Oscillation can be quite a problem if you don't keep your nose clean, so to speak.

This is a electric vehicle, so I was planning on using the 24V of 2 batteries to get my power. Since there is no "charging" system onboard, that means that my 12V batteries, no matter how high they are will end up dropping to about 10V or less when the vehicle stops operating.

I've never done too much with power supplies other then to get 5V to run electronics using things like the 78XX series. So I am a little in the dark, which is why I asked.

The off the board connections will only be RS-232 (Terminal interface to a PC), the CAN Bus, and the 65SIB. I am putting in the 65SIB more for interest then for use right now cause I have little need to have more I/O then the Atmel chip offers.

Dimitri

If I were you Dimitri, I would plan on worst case. Worst case being battery failure, and charging failure from the alternator.

At this point what will charge your CPU? Maybe a CPU that is in charge of many things..

I would count on a light source like a small solar panel maybe for power at this point...

A solar cell would not need a regulator. Although, you might need a power supply watchdog though if you pursue this idea.

Its totally a electric vehicle, so once the batteries go, the car ain't going anywhere. Its a college project, so, 100% reliability when the car is driving on the track is all I need. No safety margin for "what ifs".

No Alternators since there is no gas engine, and no other energy sources are allowed but the batteries, so no solar panels, and other things.

Dimitri

So the batteries are expected to die? unlike hybrids?

Yes, being fully electric, once they reach their max range, they can't produce a charge high enough to run electronics and they are no good after that.

Its a competition, based on X amount of pounds of batteries per team as a limiting factor, and you try to push your design to the limit to see how far you can drive in one hour.

Dimitri

Sounds like an excellent competition!

May want to investigate a power supply watchdog for your control center.

Or is this function built into the AVR you're using?