Little PC has a DC powered computer that can work in a car.

Four inputs probably means it uses a four bit binary to control the outputs.

Really one USB can easily handle one to infinite outputs. Just need to come up a way to do it. I already gave you my idea. Adding optoelectronic chips is just an added feature. Another added feature, is creating ten boards that contains four relays and sensing circuitry on each module board. Use any socket type for the module boards. This will make it easy to service when a part fails.

You said your room mate will be helping you.

These people

http://www.controlanything.com/Merch..._Code=32-RELAY

have 32 relay boards, serial, USB and ethernet. Assuming you don't want lots of CPU power you could run it off one of these:

http://www.soekris.com/net4501.htm

and save some $$$.

http://www.robotics.com/highload.html

Here's some more basic information about controlling electrical loads via relay, and microcontroller-based, and could be adapted to controlling relays and the like with the output(s) of serial or parallel ports.. The instructions are really basic, using a few resistors and some MOSfet transistors. Obviously you would need to multiply the basics here by 20 or 40 times, based on your previous posts' indicated requirements/outcome, but---it's the principle of it anyways.

According to G'fruit's link, if a $3 MOSfet can switch 10A why use a clicky-sticky mechanical relay at all?

But you still need the logic to decode the serial/USB output, so perhaps it is easier to look for a commercial solid-state relay board. Google tells me such things exist, but I'm getting out of my depth.

Thanks for the links, I'll go over them tonight.

I'm familiar with the Soekris boards, but I was planning on putting a touchscreen display and GUI on all of this, so I was looking at mini-ITX solutions. I believe it's mini-box.com that has some good looking ones for around $300, that are designed for in-vehicle applications, fanless, run off of a 12v DC power supply, and even have ignition inputs to time the power cycle.

What is the required response-time of this 'contraption' when done? I mean, how fast, and with how much precision, do the controlled events need to occur??

I wonder.. It's been some time since I messed with this sort of stuff, but.. I'll try to explain.
Let's assume that the output of the serial port is either ON or OFF ('digital').
Now, let's take a DAC chip, such as the type used to drive an LED bargraph display for showing voltage, for example. So, for simplicity, we'll say it can display on the LED-bar a level of voltage from 0 to 5 VDC, in 0.5 volt increments (giving us 10 possible readings).
Instead of connecting each of the 10 outputs of the LED-driver IC to an individual LED element, you connect each one to the gate of a MOSFET, thereby giving you 10 on-off switches.
By varying the frequency at which the DAC input is oscillated by the serial port, you will vary the analogue output level indicated by the 'LED-driver', controlling one of the 10 switches.

So, now you have control over 10 'devices', albeit individually.

There are atleast 3 pins from a serial port which are 'outgoing', pins 3,4, and 7. I'm no serial port expert, but *IF* one is able to oscillate all 3 of these pins individually at a given frequency, then by either (A) replicating the above outline of the LED-driver stuff for each of these 3 pins, OR (B) by using wider-range DAC's, you can now have either control over (A) 30 individual devices, or (B) 30+ devices, controllable in groups of 3-at-a-time.
OR, by transistorizing the outputs from the three 10-stage-DAC's and multiplexing the whole nine yards, by my calculations you could control 1000 individual devices..

All in all, I agree with the above post--- there must be a commercially available way to to this. I think I've begun reinventing the microprocessor here

I drew a rough diagram that your room mate should be able to create a schematic. Go to http://www.mediamax.com/electrojmu/H...yframework.ppt. I created it in OpenOffice and saved in MS PowerPoint format.

You need the following parts.
(4) 74145 Binary to Decimal converter
(2) 74164 Serial-in, 8-bit Parallel-out (SIPO)
Microcontroller (ex. Parallax Basic Stamp II or Parallax Basic Stamp II PE)
Dallas Semiconductor (Maxim) RS-232 transceiver for Microcontroller
5 volt Voltage Regulator

The microcontroller is used to clock the SIPO chips and buffer serial transmission, so the SIPO chips does not hang or mess a bit. The microcontroller can also help in diagnosing if current is flowing through the LED strings. One thing I forgot to include is the enable wiring, so the microcontroller can tell the two SIPO chips to accept data.

maroonbaboon and GrapefruiTgirl, jantman did not not want to use an MOSFET because he or she wants all the brighness for the LED strings. The consequences of using a MOSFET is it does act like a small resistor. What I forgot to say, it produces heat when it is on. Dissipating 60+ watts of heat for each LED strings can make this project harder and very bulky.

A DAC will not work because they are not predictable and both digital and analog ground have to be separate. Analog components are unpredictable when heat is added to the equation.

BTW, my (other) hobby is electronics. I could create a rough schematic in about a month. I do not have any simulation programs, but if I do, I could create a schematic in a few weeks.

Thanks!

I'll have my roommate look over all of this when he gets a chance, if he's still game for the project.

One thing that may have thrown some of you off - the LEDs need to be switched by external relays. I guess I shouldn't have mentioned LEDs at all. I'm not talking about one-light units, I'm talking about something more like:
http://www.lshlights.net/catalog/whe...ds_5428699.htm
To abstract it, we'll assume each relay on the controller should be capable of switching a 2A@12V DC load, or controlling an automotive relay.

I don't know a whole lot about this stuff - so most of the suggestions here were over my head.

In terms of time, it needs to be relatively low delay - to put a realistic benchmark on it, it should be able to turn on all 40 relays in under 2 seconds or so, and should be able to run through a diagnostic test (turn relay on, check for current on associated input, turn relay off) in under 10 seconds.

Some of the outputs will have to be controller by automotive relays - one will be controlling the power to a large solenoid, and a few others will ultimately be controlling 10-40A loads. I don't know how current sensing works, but I'll probably need to add another later between the inputs (current sensing) and the actual loads to be sensed, being that many of them will also be in series with 12v DC 5-30A devices.

I'm not sure what you really mean by precision - time isn't too important. What's important is that when the controlling computer sends a signal to turn relays X-Y on, they must go on, within a reasonable amount of time (a second or two maximum).

As to heat - I guess I should have mentioned this. We can assume that some relays will be powered all of the time that the vehicle is running, and that it won't be too uncommon to have EVERY relay powered for a few hours at a time.