SexyMF DIY controller - so far

[SexyMF]

As some of you know I decided to attempt to make a controller for a single system.

The goal was to make it open source, easy to build and cheap.

I decided on these specs:

0.2C steps -55 to +50C
Hardware settable condensor overtemperature set-point (activates an output) 0 -> 50
Hardware settable computer on set-point (activates an output) 0 -> -55
Hardware settable condensor temperature optimal point (for the fan control) 0 -> 50
Automatic fan throttling based on condensor temperature

Thermistor sensors
LCD display

This is the first time I have used thermistors and so it was a learning curve. A thermocouple implementation would have been a lot easier but price of

thermocouples and the required CJC (cold junction compensation) ICs is to high - compared to the thermistor.

The system is simple. It works using 3 potentiometers to alter the set-points. You hold down the only button while turning it on and it goes to a visual

calibration mode. Turn the pots and when done release the button and the system moves on to a startup routien.
The startup loop checks your setpoint against the current evaporator temperature. A pulse is sent out a port to trigger something (ie a relay). During the

startup loop the fan is throttled and the condensor temperature is checked against the overtemperature setpoint.
Once the on set-point is reached the condensor and evaporator temperatures are shown. The fan and overtemperature checks are also polled.


Technical wise:

Implementation of the temperature calculation was done using a massive (1k) lookup table. This was calculated in excel using manufacture datasheets for the

thermistors I choose. Subsiquently I found there to be many thermistor manufacturers which supply software or precalculated tables (R/T curves).
It became apparent that with such a large temperature range the thermistor resistance values change dramtically on the low scale but very little on the

higher end. It was not possible to pickup the subtle change in voltage with the 10bit ADC on-chip. So I had to do some oversampling. The program uses an

effective 5bit oversample for an effective 15bit ADC.

The safety cutout is designed in the way that the controller has dedicated power. In an overtemperature situation the computer is first shutdown then the

other port triggers. I had intended for this trigger to shutdown the compressor. The fan control coming from the uC remains on. During the pull-down before

computer on, an overtemperature situation will not trigger the computer-on output. Only the cutout output.


This is a work in progress. I am entirely self taught for AVR assembler programming. I built my programmer as many of you will.
There are probably still some bugs I have not picked up on. This is were I would like your help.

I offer you a barebones setup. Customisable because there is fully commented code.

Enjoy this first installment.

The prototype



Temperature testing against a thermocouple (borrowed)



Basic schematic - for now at least



Code so far:





Some references for some light reading:

www.atmel.com
www.avrfreaks.net
http://avr-asm.tripod.com/
http://www.geocities.com/dinceraydin/lcd/intro.htm
http://www.sixca.com/resource/micro/index.html
ADC oversampling http://www.atmel.com/dyn/resources/p...ts/doc8003.pdf

AVRStudio http://www.atmel.com/dyn/products/to...p?tool_id=2725
Pony Prog http://www.lancos.com/prog.html
YAAP http://www.myplace.nu/avr/yaap/

STK200 dongle http://sbprojects.com/projects/stk200/stk200.htm

http://www.alldatasheet.com/datashee...I/74LS164.html
http://www.atmel.com/dyn/resources/p...ts/doc1477.pdf

[SexyMF]

Here are some very poor quality AVIs of the controller...


Calibration... press the button and turn the pot

Download Calibration.AVI

Pull-down.. on at your setpoint.

Download Pulldown.AVI

[Trouffman]

This can be very interesting

I search about using thermocouple for an extended range like -100*C to +100*C.

[Blaster]

sounds interesting although i dont understand half the stuff

.... i hate electronics heheh

[Big SturL]

Nice work there MF. Could you possibly give us some wiring schematics? I know this is a work in progress, but it'd be nice to know how to wire it up, and of course what components to use.

[{.bLanK} GoD]

Nice work there dude. Do you plan on making any yorself, for sale on any scale? or do you simply just want to design it? I have 3 SS DD units about to start, and probably more after that, I "could" be interested in a few.

[star882]

I planned on building something like that. Can you please detail which IC pins go to which lines in the ATX connector (i.e. what wires to intercept and which to tap)?

[Big SturL]

My guess Star, is that it is connected to the PWR-pin on the motherboard, not the ATX-cable.

[Xeon th MG Pony]

Op amps rock espesialy the 310 series from SG

[SexyMF]

SoddemFX:

I had considered swapping ranges. The uC can be used as differential ADC inputs with gain. So no extra parts required. But the added bonus of oversampling is that it takes a good deal of time. Slowing the whole update rate down. Thermistors have a long settling time anyway. The fan automatical speed change doesn't go hunting because of too fast an update.
I had intended the condensor temperature to be measured at the output. So if you hit 50C there then things are not going well.

Big SturL:

That is the full wiring schematic shown except for what you put on the PWM/Cutout/Computer ON outputs. Join the red labels to red labels and that is it.
Vcc = 5V
3 x 10k resistor (1/4 or 1/2W)
1 x 1nF cermanic capacitor
1 x ATMEL ATTiny26
1 x 74LS164 8 bit shift register (alternatives will work, like HEF4094)

This is technical but easily implemented.
These parts are different depending on your needs:
You must generate the 15 bit ideal ADC voltage tables for your chosen thermistor. This is done with the datasheet or software from the manufacturer.
I used
2 x 470K 1% tolerance resistor (1/4 or 1/2W)
2 x Thermistor EPCOS B57861S0502J40 (5k 5%)

The 470K [R1] is chosen as follows:
the thermistor is in a voltage divider with 5V source.
the thermistor has a resistance of ~481k @ -55C
we want the voltage across the thermistor for input into the ADC to be just under 2.54V (that is the top voltage reference for the ADC)
So you solve this 2.54 = 5 x R1/(R1+thermistor resistance) to get R1.

I used an excel table to work out the nominal resistance vs temperature for my chosen thermistor.

The PWM can be directly tied to something like an IRF7805 MOSFET. This is capable of handling both fans (condesonr and vent) in parallel.

The computer on signal could be tied into a solid state relay like OMRON G3VM-351E N. Or drive a small signal transistor which is switching a normal 12V/5V relay. This would be connected in parallel with the POWER switch on the motherboard. That way you can still override the power on.

The cutout signal probably should go to a small signal transistor switching a 12V/5V relay which is in series with your compressor. So as to shut it off on overtemperature fault.

I simply used a 12V plug pack and a 7805 voltage regulator to power my prototype. If you used a 1.2A or so plug pack that is enough to run the fans at 12V and the controls at 5V. Something like an old scanner power supply.

The LCD and controller use little current. It is the LCD backlight and fans which gobble all the juice. But they are only say 1/4 amp per fan and 1/4 amp for the backlight. My LCD didn't have a backlight. If you use a backlight then I would run it off another 5V regulator so as the keep the heat down.


So to sum up. To build this, at minimum you require to build a programmer so you can insert the code and eeprom data into the uC. These are dead simply to build. The STK200 dongle is the standard but it can be done using just a few wires from the parallel port. Google is your friend.


Once I'm happy with this I will fully surface mount it. The footprint should not exceed the size the the LCD.

[metalazzo]

tip : there are nice and cheap temp sensors like LM35 for exemple (or alternative like LM335...)
they give you 10mV/K (2.73V@0°C) and are rather linear
well, just and idea
nice work !

[SexyMF]

tip : there are nice and cheap temp sensors like LM35 for exemple (or alternative like LM335...)
they give you 10mV/K (2.73V@0°C) and are rather linear
well, just and idea
nice work !

Yes but you require more wires and have to insert an entire IC into the evaporator/condensor. They are designed to measure ambient in a dry environment. They will fail. They are also non linear, and the wire leads running out of them will affect the temperature reading.
Such ICs are used to measure temperature on a PCB, not as a remote sensor.

[quintus]

promis to open some alcohol and drink thinking at your kindness(i rarely drink alcohol 1time a beer in a month or less)

hope my electronist friend will dig every litle thing and make this thing hapend

thank you
q

[Big SturL]

SoddemFX:
Big SturL:

That is the full wiring schematic shown except for what you put on the PWM/Cutout/Computer ON outputs. Join the red labels to red labels and that is it.

Hehe, I guess I'm still a nub in electronics I've been in an electronics class for about three months now, but I guess I've still got a lot to learn. I'm used to seeing the "wires" and such on wiring diagrams. But one of my teachers used to program chips such as that before, so he can hopefully help me.

[SexyMF]

Still going with this project. Just waiting on parts. They will be here in a fortnight or so.
In the meantime I have updated the code to shift the outputs together to make routing the PCB easier.
I'm working on the artwork, but this can not be finalised until I have the parts in my hand so I can check their component footprints.

Not to far off now... (if anybody cares!)

[girth_maul]

Will be a interesting read when all is done, though last time i did any electronics was way back at school in the form of a crappy parking sensor lol!

[4Qman]

Not to far off now... (if anybody cares!)

I Care, Keep this up fella

[aussie_guy00000]

Great work so far shaun , if you need any help figuring out stuff, gis a yell and i'll pick the brains of the electrical post grads at the uni here, I know quite a few (i don't have enough knowledge to help you myself though ) . Keep up the good work.

P.S how's the NL11F goin?

[jinu117]

Good stuff
Here are some parts that might interest people wanting to expand with this...
AD595,
http://www.micromegacorp.com/downloa...rmocouples.pdf
Basically thermocouples, you amplify and filter (which is not easy to do accurately without dedicated chip) than read the value through A/D.
Problem is... the part for amplifier is not cheap at all... was like $9-10 a piece I think last tie I checked unless purchased in volume.
I would love to see thermocouple implementation of this eventually though... thermistor of high accuracy and big range does get mighty expensive.

[star882]

What about using an AD590 like I'm going to do?

[jinu117]

I just don't like -55c limitation that's all :P But it would make it hell lot easier to implement for sure...
one thing I noticed is... thought -55c limitation might be moot point for ss as we keep tuning it for higher load...

[Big SturL]

Now, what would be nice is if you set up a step by step guide for those who have NO experience with this at all, like myself Like this:

-Buy: this, and that, and this, and so on
-Connect:this here and that there
-Program: Using this prog, and these values
-Be: Happy!

[SexyMF]

There are many transducers but they are all very expensive compared to the thermistor. Yes, I have to make tables and more complicated code but I don't mind. It is what I can contribute to the forum.

The AD range you are talking about is 4-10x the price of a best thermistor.

This is just the first go at a controller. I will make other which use different sensors. But for now things like thermocouples are just to expensive. Currently the LCD is the biggest cost of the whole project.


@jinu117

I would have liked to go lower but if you look at most all the sensors -55 seems to be the decided number. Most are ranged -55 -> 150. I can't get thermistors which go lower than that. I would if I could. For a SS I expect under load people won't reach -50C (for the most of us!)

[jinu117]

Heh I know exactly what you mean about -55c as I was thinking about making one myself while back That LTC2413 looks very good! For same price as the thermocouple amp, you also get D/A converter which definitely will reduce cost on microcontroller side (well not too much but every single bit helps I guess). I guess it could be used multiplexed as well easily?
PM me on few things SexyMF... could give you some pointers for some things I've found after searching for nearly a year on many components

[SexyMF]

Seeing if believing (post #2 updated)