My PWM fan controller project, continued....

[plasticuser]

Hi all,

A few of you may remember I started a simple PWM fan controller project last year. I have an F150, and wanted to control two fans with a simple, affordable PWM controller that would work to minimize current draw while providing safe cooling.

After a while, I sidelined my prototype because of a few issues, and replaced it with a simpler relay system that isn't at all optimal.

I have decided to return to the PWM project, but I have decided to start over from scratch, and put all our heads together to make something great. It'll be an open design anyone can make, but when tested and known to be reliable, I will also make as many as needed and sell them at a low price, enough to cover my costs and time.

The market price for good quality controllers is $250+, with around $350 as the "sweet spot" where most good products are priced. I believe a decent PWM controller can be made and sold, as a complete kit, for under $100.

The savings of PWM fan control, especially on cars and trucks that have a mechanical fan/clutch, can be quite large (I saw a larger gain from using a PWM controller than from using an aerocap)...

So, fishing for interested people to work with, or help test units on their cars/trucks and give me good feedback. Also looking for people who know their PWM circuits, so we can improve the design or replace it altogether.

Who's interested?

[Ang84Indy]

I'm interested! I have a 1976 Ford F100 that I'm converting to an electric fan.

[plasticuser]

Here is a quick breakdown of the current functionality/design spec of the PWM controller:

Inputs:
1. 12VDC 40A max
2. Temp sender, sets fans to 40-100% power across range 195-205F
3. A/C clutch input, sets fans to 50% power when A/C is operational

Outputs:
1. PWM output to fan(s), 40-100% (16-40A)

General specifications:
PWM frequency: 3 KHz
PWM working range: 0-100%
PWM acts as current controller up to 100%, so fan speed remains constant with varying load up to 99% - an improvement here would be detecting 100% condition, then activating a relay to bypass the PWM circuit and reduce load even further.

Recommended power be fed through headlight relay, min 50A rating, relay controlled from ignition switched source. This stops the fans running after the engine is shut off...

[hawk2100n]

Have you specced out any components yet? I'd be happy to help with this project. How would you keep the fan speed constant without being at 99% load from an airflow perspective?

[plasticuser]

I think my current design is weak and falls short in a lot of areas - it's not reliable, efficient, or economical. It's two designs smudged/bodged together. So, this being a new, ground up design, no components are specced.

As airflow varies with movement, fan speed isn't important when moving. However, when stationary, other load variations in the car can cause the fan to speed up or slow down. This also makes no difference, but some are annoyed by the change in sound as, say, the A/C clicks in and out. Personally, I don't mind, but it's a requested feature.

[AndrzejM]

I was thinking to build PWM for fan control in Berta, but have other more important things to finish before I'll start working on that. But I have some experience with electronics so I can help you.

I was thinking ATMega or ATTiny with two DS18B20 and all you need is high current end of the unit.

[plasticuser]

I looked at a couple of microcontroller solutions, being a Raspberry Pi fan. Unfortunately, I couldn't see a cheap way of doing a small switch mode PSU suitable to run them, and I didn't want to go the inefficient 7805 route...

I guess that's my natural instinct - to play to my strengths. When I meet a few more people who have other strengths, like yourself, we'll have a much better (and more relevant) skill pool.

[hawk2100n]

I think a Raspberry Pi may be a little overkill. I have 2 myself however an arduino would be perfect. You could have the AC input signal on a digital input and read temperature via analog in and output a PWM signal to a motor controller which would simply be a high current transistor array that metered out the 12V with a 5V/3.3V low current signal. I'll take a look and see what I can find for pre-engineered transistor arrays. I'm sure there is something out there.

[plasticuser]

I have explored a lot of options here, and though some think a Raspberry Pi is overkill, I think it provides a great expandable platform. Advantages it has include that it can run a small LCD display in cab, and that it can run several other tasks too, like operating a grille block, monitoring OBDII, etc. The clincher for me is that the RPi has a hardware PWM output.

This makes the RPi 'monitorable' with a display, and also allows seasonal or vehicle customization options with just the plugging in of a USB keyboard. Also, with ethernet and USB, wifi can be trivially added, and the Pi could automatically upload and make available data logs of all activity each time it hits the driveway.

What is needed at this point is generic hardware designs for two areas:

1: an automotive +5V regulated supply that is low cost, heat tolerant and highly efficient, whilst rejecting the worst abuses of an automotive 12v system

2: a transistor array that can switch a large current, efficiently, from a 3.3v 10mA source (the PWM GPIO output on a Pi)

Separately, it would be required to develop an OPEN SOURCE, FREE code library for controlling various systems with simple code segments.

Any offers or suggestions?

[plasticuser]

Addendum:

It's my intent to design a daughter-board that fits onto the Pi's GPIO pins, and to make it generally available. Cigarette lighter USB power options don't work - they're usually cheap and inferior, and mechanically too fragile. I'd like to incorporate a high quality power supply onto the PCB. It could provide 5v and 3v3 directly, but I'd definitely like to bypass feeding the Pi power through the micro-USB port, for obvious reasons.

Also, I'd like it to be able to monitor, for example, the A/C compressor circuit, so a bullet-proof way to, again, drop the extremely hostile +12v to a clean 3v3 would be nice.