Fan controller +bells+whistles+diy(?)

[ecat]

I'm currently developing a fan/temperature controller that provides some interesting features and addresses some of the problems, to my eyes, in the current commercial offerings. In this thread I'd like to gather some opinions as to the features of the basic hardware and discover if there is any interest from the community in a diy electronics project such as this. In addition if there are any electronics types out there feel free to correct any mistakes or bad assumptions on my part.

1) I'm attempting to build this entirely from off the shelf parts making it as plug and play as possible, still assembly will require:

a) Some basic wiring, sleeved or un-sleeved .

b) Soldering, I hope to limit the soldering to wires and plugs and sockets but this will still involve an amount of fairly fine detail, around 0.1" for those in the know, eg http://www.dummies.com/how-to/conten...ering-101.html

c) Preparation and use of stripboard/veroboard eg http://uk.geocities.com/ronj_1217/link/lk7.html

d) Case work or other mounting mechanism. Yay, another excuse to wiggle your Dremel.

2) Inputs

a) 4 maybe 6 inputs for the standard foil/thermistor temperature sensors

b) Support for many digital sensors, specifically the Dallas/Maxim DS18B20. How many is many ? Would 30 be enough ? I doubt I'll be able to afford this many for testing but once again I'll see what I can do.

c) Measurement of 5v and 12v lines

d) A couple of inputs connecting to motherboard fan headers. If your motherboard has built in fan/temperature control it should be possible to use the voltage on these inputs to indicate CPU, chipset etc temperatures without the need to run Windows or additional applications.

e) Standard fan speed monitoring, pump speed where possible. I don't feel the need to monitor every fan individually, you may disagree. Would 6 be enough ?

f) Flow monitoring. I've not used a flow monitor but I'll see what I can do. Would 2 inputs be enough ?

3) Fan/Pump(?) control. I've abandoned PWM in favour of 90%+ efficient switch mode analogue/voltage control.

a) 4 channels. Would 6 be better ?

b) 25 Watts per channel.

c) Each channel is fully protected against short circuit and over temperature.

d) 4 Amp short term peak per channel.

c) Channels can be connected in parallel (! further testing required).

d) Option to power the fans from an external PSU. This allows for much greater maximum load, say 120Watts, it also allows for 13v or 14v to be used at the input for those worried about not getting the full 12v at the fan.

4) Main power ie not fan power

a) USB

b) External PSU, the unit can run while the PC is off

5) Form factor

a) For mounting in a PC, 2 x 5 1/4

b) Single 95mm x 40mm cut out for the display

6) Display. I want to view temperature traces at all times, including during boot up, so

a) 4.3" 480x270 colour LCD, the same screen used in the Sony PSP

b) Touch screen so no need to drill holes for switches

c) For the strange people among us who do not like to view temperature traces all day the will be able to show static pictures. I am trying to come up with a way of displaying video and WinAmp visualisations etc but at the moment it is not looking good.

d) Wiggly lines without windows, Linux users rejoice.

7) Alarms, alerts etc

a) With the exception of any data linked to third party applications, SpeedFan etc, the controller will run independently of windows. It may flash, it may beep, it will certainly be able to switch the PC off

8) Data logging

a) I aim to log all values from all connected inputs every second.

b) Logged data may be viewed on the display.

c) Logged data may be uploaded to the PC.

d) I've not yet decided on where the data will be stored. There is not much ram available but flash is cheap and it should be possible to support some types of memory cards.

Some approximate costs / possible suppliers. Better parts? Better prices? Let me know :nod:

a) Main controller £30 / $40 eg http://www.sparkfun.com/commerce/pro...oducts_id=8971

b) DS18B20 about £4 / $4 each eg http://robosavvy.com/store/product_i...roducts_id/413

c) Switch mode voltage control module £19 / $25 per channel eg http://www.dimensionengineering.com/DE-SWADJ3.htm

e) Display £30 / $50 eg http://www.depot888.com/brand-origin...ght-p-701.html

f) Touch panel for screen £15 / $20 http://www.sparkfun.com/commerce/pro...oducts_id=8448

g) Display controller £45 / $70 eg http://forum.lcdinfo.com/viewtopic.php?t=2602

h) Connectors, wires, misc. bits £25 / $35 at a guess

I'll update this post with anything I've forgotten.

Thanks for reading. Over kill? Over priced? Over reaching? Barking mad?

What are your thoughts ?

Whoops. I forgot to mention that fan control will be manual and/or automatic - temperature curves work best for me

[matsjuw]

That is quite extreme!! But I'm already drooling if you can make all of this come true.
The channels look very nice, possible to control 2 pumps on 2 channels and about 6 fans on 1 channel for a push/pull configuration.

I was just thinking about the size of the thing, if it would be possible I'd try to make it the size of a HDD, that way it can be mounted in a HDD rack where most of us have a fan blowing in.
Pricing looks not too bad either, 244$ (170€) vs 140€ for an aquaero over here. Not considering shipping ofcourse.

[Waterlogged]

If I were you, I would start a little smaller and work my way up first. 30 probes is vary overkill, IIRC, Martinm210 had 32 probe hookups on his testing rig and only used like 18. . .the standard user isn't going to need even that many, maybe 6. . .tops. 2A per channel would be a vast improvement but 1.5 would probably be a better starting point. I've got a few more things to add to this list but unfortunately, work beckons me to show up. I'll add them in a new post later.

[eponymous]

I reckon you could do most of this with a PIC and a PIC programmer.

[Rise]

I reckon you could do most of this with a PIC and a PIC programmer.

i was thinking you could do it in a PLC (programmable logic controller), but i don't know if that's what you were referring to.

still, anything that we don't already have is worth developing, IMO.

[noobieocer]

i was thinking you could do it in a PLC (programmable logic controller), but i don't know if that's what you were referring to.

still, anything that we don't already have is worth developing, IMO.

no thier differnt a plc is a programmable interface controller and i forgot what PLC is used for but i belive theier for factories? PIC is easyier to program i used some for my college course

edit: i was corect:
http://en.wikipedia.org/wiki/Program...gic_controller

i will love to see how this gets on but one thing i dont understand is how are you going to change the voltage on those regulators, they have a pot to control voltage. unless you going to have a dac from you pic to the where the pot is and there for controlling the voltage set. or on second thought maybe a transistor and the base connected to the dac.

anywayz love to watch this

[denali]

I've been looking for something similar to this. Basically, I want something that will keep the pumps and fans running after the system is shut down. Once the fluid gets down to a certain temperature, then the pumps and fans would power down.

[Kibbler]

I've been looking for something similar to this. Basically, I want something that will keep the pumps and fans running after the system is shut down. Once the fluid gets down to a certain temperature, then the pumps and fans would power down.

If you don't mind, why?

[Hondacity]

ecat

awesome list

addin some huge heatsinks and fan mounts for the heatsink

i was thinking you could do it in a PLC (programmable logic controller), but i don't know if that's what you were referring to.

still, anything that we don't already have is worth developing, IMO.

PIC microcontroller...they're good stuff i used them in my robotics class

If you don't mind, why?

we use the same principle with our systems...when our system(120kw) powers down...the chiller...has a power off delay of 5minutes...

[ecat]

If I were you, I would start a little smaller and work my way up first. 30 probes is vary overkill, IIRC, Martinm210 had 32 probe hookups on his testing rig and only used like 18. . .the standard user isn't going to need even that many, maybe 6. . .tops. 2A per channel would be a vast improvement but 1.5 would probably be a better starting point. I've got a few more things to add to this list but unfortunately, work beckons me to show up. I'll add them in a new post later.

Some good points...

I picked 30 as an example figure. The devices communicate over a single wire and this single wire can connect to a virtually unlimited number of devices, if you write the code to access 2 devices on a single wire the same code will access 100 devices. In reality I think splitting the devices over, say 4, separate channels will simplify both the timing requirements of the code and the cable runs.

I was thinking of people like Martin when putting the design of this together. This unit will sit on a desk, log n channels of data to better than 0.1C resolution, display the data that has been logged and at the end of the day upload the data to a pc.

This thread is directed at the 'standard user', at least the standard XtremeSystems user. I'll do things one way if I build this exclusively for myself, I'll do things a different way if I build this so others can follow the instructions. One of the reasons for this thread is to find out how many, if any, are willing to drop $200 on a box of bits and soldering iron.

25Watt, 2 Amps per channel just happens to be the rating of the ready built modules I'll be using. The same company makes a 10w unit with the same form factor that costs about £5 less, I think the 25w unit is the correct choice. For reference, my original do it all yourself design was rated for around 5 Amps per channel but this would have required a custom circuit board and some very tricky soldering. The off the shelf module comes with a guarantee and can tested in isolation before making a small modification, again I think this the right choice.

As an aside, if you're happy setting the speed with a small screwdriver or you get inventive with a plastic rod and some glue the 25w module makes a lovely manual speed controller. I have one set up here just now, stripped naked of all plastic and aluminium, running six Nexus fans at 9v the unit is not even warm.

I reckon you could do most of this with a PIC and a PIC programmer.

<chuckle> The controller is a PIC, a PIC32MX460F512L in fact, built into a ready assembled module. It is quite a powerful beast to be honest, the main reasons for choosing it are:

1) The pre-assembled module saves messing around trying to solder the tiny surface mount PIC.

2) This particular module has a number of nice features including: inbuilt voltage regulation, inbuilt USB port, some switches and LEDs and comes pre-programmed with some useful software. It can be tested straight out of the box with nothing for then a USB cable.

3) The pre-programmed software allows the module to be programmed with nothing more than a small windows/linux application and a USB cable. No special programmer is required.

4) I needed more than 16 control lines and this was the cheapest option I could find .

N.B. You'll find the phrase 'tested out of the box' or 'tested before modification' appears several times in my replies. I'm trying very hard to ensure that each individual module used in this project can be tested with the absolute minimum amount of work. You buy the module from wherever you like, you test the module with the minimum amount of fuss, you send it back if it doesn't work. Once a module has been modified you are on your own, well, I'll help diagnose where I can.

Edit:

ecat

awesome list

addin some huge heatsinks and fan mounts for the heatsink

Thank you Huge heatsinks and fans eh? At 90% efficiency and 25w the power modules may need a small heatsink to help with the 2.5w dissipation, to be honest they are better than 90% and designed so the whole pcb acts as a heatsink. As for fans, the last thing I need is another one

ecat
PIC microcontroller...they're good stuff i used them in my robotics class

I don't like PICs all that much, at least I don't like the MicroChip complier and some of the library functions always feel somewhat broken. Still, the software is my problem, once you get these PICs working they work well.

[DarkCow]

This looks like an awesome project. Can't wait to see it.

[kamieldehond]

maybe you can interface i2c from the mobo to your diy fancontroller. Dont know how easy/hard this would be and if this would be usefull.

[ecat]

maybe you can interface i2c from the mobo to your diy fancontroller. Dont know how easy/hard this would be and if this would be usefull.

I don't think there are many motherboards that have an i2c connector, the only place I remember seeing it mentioned is on my baby VIA file server, I imagine there are also problems supporting all the different data formats. I'd be delighted if you can prove me wrong on this, reading the i2c directly would be so much better than trying to divine temperatures from the output voltages of the mb fan headers.

[Charles1961]

Good project! With I2C inetrface you can work with MAX6651 chip for control and monitoring the fan in linear voltage.See data sheet http://datasheets.maxim-ic.com/en/ds/MAX6650-MAX6651.pdf[/URL]
And for temp. you can use one like MAX6697 see datasheethttp://www.maxim-ic.com/quick_view2.cfm/qv_pk/4597

[astrodanco]

Just make sure it works better than the Koolance TMS-200 + TMS-EB200. If you don't have one of those, get one, it's your competition. The main failings of Koolance's product are not enough fan headers (6), buggy non-open source software/firmware and non-user upgradable firmware. Other than those failings, it's one heck of a good product and works a treat! I'd recommend you don't go too hog wild on the cost.

[ecat]

Good project! With I2C inetrface you can work with MAX6651 chip for control and monitoring the fan in linear voltage.See data sheet http://datasheets.maxim-ic.com/en/ds/MAX6650-MAX6651.pdf[/URL]
And for temp. you can use one like MAX6697 see datasheethttp://www.maxim-ic.com/quick_view2.cfm/qv_pk/4597

Thanks for the links. The temperature chip looks very interesting The control chip has one big problem: the output is linear this means the drive transistor could end up dissipating a lot of heat. Here's a nice link I found explaining heat dissipation in transistors, it's a description designed for people into water cooling

http://answers.yahoo.com/question/in...0120914AA8qraX

Just make sure it works better than the Koolance TMS-200 + TMS-EB200. If you don't have one of those, get one, it's your competition. The main failings of Koolance's product are not enough fan headers (6), buggy non-open source software/firmware and non-user upgradable firmware. I'd recommend you don't go too hog wild on the cost.

How many fan headers would be enough? How many fan speeds need to be monitored?

[astrodanco]

How many fan headers would be enough? How many fan speeds need to be monitored?

My maximum requirements: Software power level control and monitoring of sixteen three-pin fans, four three-pin pumps, four flow meters, and sixteen temperature probes.

My minimum requirements: Software power level control and monitoring of six three-pin fans, one three-pin pump, one flow meter and three temperature probes.

I don't actually care whether the interface for the pump is three-pin (ala PMP-400) or molex power plug with separate RPM lead (ala MCP-350). I can use a CBL-CTR103P to convert in either direction.

I'm easy to please.

[Charles1961]

Your welcome for the link
If you want build this fan & temp. controller (FTC) for all the need (overkill) don't forget the guy with quad radiotor,this mean each channel shoud be able to drive 8 fan in push/pull configuration WOW! Maybe .25 amp/fan already 2 amp. per channel.With 3 channel we are able to drive 24 fan i think nobody wants more!
I know when we use transistor we have to put heatsink on it, but for me your FTC can easyle fit in 5 1/4 drive bay for the heat dissipation you can use one like that.
http://www.mountainmods.com/120mm-ha...alu-p-326.html

For the pump i don't see the needs for the voltage control. It's the heart of your water system and nobody want to reduce the waterflow. but it's a good idea to monitoring it. Let' see 3 channel for 3 loop 1 pump per loop, again nobody want more then that.
Now we have 6 channel.

For temperature monitoring what's the needs? Assume we have 3 loop i want see the temp. at the start and at the end of the loop. What else? CPU,NB,SB,Voltage regulator X2,GPU X2 maybe 3,1 for the inside temp. of casing I think i go all around. Like WATERLOGGED 18 probe is more then sufficient.

Don't tell me it's a overkill description it's ONE
Sorry for my english i'm Canadian frenchguy

[Waterlogged]

25Watt, 2 Amps per channel just happens to be the rating of the ready built modules I'll be using. The same company makes a 10w unit with the same form factor that costs about £5 less, I think the 25w unit is the correct choice. For reference, my original do it all yourself design was rated for around 5 Amps per channel but this would have required a custom circuit board and some very tricky soldering. The off the shelf module comes with a guarantee and can tested in isolation before making a small modification, again I think this the right choice.

Well, when you put it that way, I can't really argue with your reasoning. Kind of a no brainer.

[eponymous]

Good project! With I2C inetrface you can work with MAX6651 chip for control and monitoring the fan in linear voltage.See data sheet http://datasheets.maxim-ic.com/en/ds/MAX6650-MAX6651.pdf[/URL]
And for temp. you can use one like MAX6697 see datasheethttp://www.maxim-ic.com/quick_view2.cfm/qv_pk/4597

That features SMBus support which is a plus too.