Sophos9 - Diy "poor Mans" Reef Cooling

[sophos9]

During the summer (did we have summer?), I added a fan into the hood of my tank, nothing special just a 3.5inch computer fan. I measured the difference.

This fan was able to take over 3dC off the tank by evaporative cooling which was great (as the tank never exceeded 29dC) - now this got me thinking, the thermostat can heat the water - the fan can cool the tank, water evaporation is not that much of a problem as we have got the auto top off coming but all the variables are well, variable

So what I want to do is link the fan up to a thermostatic device so that as a predetermined temperature is hit, the fan will activate and as the temp rises, the RPM of the fan rises also

In my simplistic brain, all I would need is a thermistor in the 12v circuit that runs the fan but I really need help on the calculations of resistors and locations

OR

if (Andy ) has any ideas to design something like a LED or LCD temp display with probe with a POT for variable control in the circuit that would also be great (and possibly more easy than the auto top off)??

 

[sophos9]

The specs...

Reef temperature - 26dC
Min fan RPM on temp - 26.2dC
Med fan RPM on temp - 26.6dC
Max fan RPM on temp - 27dC

The above may not have enough hysteresis built in, can someone advise?

A thermistor (or temperature) probe to be submerged in tank to feedback current temp with a detachable connection to the 'black box'
A switch for 'water change/cleaning' mode
The fan must be off when temp is < 26.2dC

 

[xxBarneyxx]

Simple option would be to get a cheap Pc fan controller. Make sure you get one with a Temp probe (which you will need to water proof).

This will give you a nice looking panel which shows you the fan speed and temperature. You can also then manually set the fan speed. Runs on a 12v current and is nice and easy.

http/www.overclock3d.net/reviews.php?/ca...an_controller/2
Thats probably a more expensive one but you should be able to pick a cheapo unit up for about £20. This can also be used to control multiple fans and normally 2 fans running at half speed makes less noise then 1 fan at full speed.

A simple adjustable Thermistor would do the trick as well.

 

[sophos9]

Ok, so what do you think about this?

This is not my work, credit goes to Peter Newman

So this circuit allows a temperature controlled on/off and gradual increased RPM of the fan. I believe the temperature range we are after is smaller than this circuit so ratings may have to change. This circuit has LED's indicating the state of RPM speed.

How easy would it be to drop a LED/LCD temp display in here?

xxBarneyxx, thanks for your reply - those things look great. I would however like to at least price this up to get an idea of cost and ease!!

 

[sophos9]

Right, so what about this?

Use a temperature module to create the high/low outputs with LCD display, connect the additional temperature probe to go into the tank.

Put this in an enclosure with some additional buttons to connect to the solder edge connection then manually set the high switching setting

Groovy, temp control!!

So, I would not have the slow spinning of the fan, it would either be on or off which is OK for me
The module has adjustable 10 second sample - would this be enough hysteresis? If not could the outputs go into a 555 timer?
Would also add a 2 color LED so that when its off its low - its red, when its high - its green...

What do you think?

 

[Matthew5664]

Try this

 

[sophos9]

Try this

Thats real nice, problem for me is I'm not running a chiller - only a 12v extract fan for evaporative cooling so once I have to buy the additional parts (enclosure, glands, wiring, plugs etc) it will be a bit more than I want to spend.

If I had a chiller and the like I would get one straight away!!

Thanks for the post!

 

[Underwurlde]

yo!

To the untrained eye, these kind of systems appear at first glance to be simple: You have a set temperature you want something to target.

Dig a bit deeper & thing start to get complicated. This is a basic feedback and control system and can be applied to anything with a time delay involved (Think speed control, directional control, central heating systems etc). The time delay I am talking about here is called 'thermal impact'.

It goes something like this (e.g. central heating):

Room is cold (thermostat goes on) -> Heating demand ->boiler goes on: Takes 5 mins to heat up -> Radiators heat up: Takes 15mins to heat up -> Air heats up & circulates: Takes 30 mins to heat up -> Room is warm (thermostat goes off) -> boiler goes off -> does the room cool down? NO! Because the air will take time to COOL DOWN (30 mins), oh and the rads..... So the system hunts!

..../\/\/\/\/\/\/\/\/\/ <- ON/OFF/ON/OFF/ON
../ <ON
/ <- ON

Such systems are controlled using INTEGRATORS. The amount of energy (heat) applied to the system is an integral of the measured heat: So when its COLD = loads of heat, when its COOL = Not so much energy, when were getting to the desired heat (some time later) the amount of energy applied ALMOST exactly matches that required to maintain the set temperature:


..............................ooooooooooooooooooo <-set heat matched, use amost the amount of heat we need to stabilize the system
.....................ooooo
.............oooo
.........oo
......oo <- getting there = not so much heat
....o
..o
.o
o <- steep curve = lots of heat (full on)


(Note I say 'almost' - it is a function of such systems that the desired setting is NEVER reached).

The speed at which the system appears to get to the set heat is called the 'damping' of the system. (Too little and the heat overshoots and bounces about, too much and it takes too long / never gets to the desired setting).


....oo
...o..o......ooo
..o....o...o......ooooooooooooooooooooo
..o.....ooo
.o
.o
.o <-curve too steep = overshoot = at some point the tank will get too hot!
o
o


Hence for such a system I think an integrating controller is needed for which we need some means of adjusting the damping (the 'speed' if you like). For this to work, an analogue representation of the tank temperature is required (e.g a thermistor) and an analogue means of controlling the temperature (as you have, the speed of a cooling fan).

Does that make any sense?

Andy

PS
Of course for a COOLING system such as this, invert all of my crappy graphs!

 

[sophos9]

Does that make any sense?

Hang on... does that mean a thermistor and fan circuit is easy?

No seriously, I do get you... the fan would be spinning up every 10 seconds right (10 second sample time)?

 

[Underwurlde]

the fan would be spinning up every 10 seconds right (10 second sample time)?

No.

‘Sample time’ is a term not relevant here unless we are talking micro processor / controller based system. Yes such systems can be modelled in software. These are PID (Proportional Integral Differential) controllers (or in this case a simple PI controller). The sample time for such a system I’d set at a nice easy say 1s, control loop (for the PI) say every 10ms.

We (I think ) are talking about an analogue control system (same as a linear amplifier for example – same principles). In such a system (and yes, it would I think be a very simple system) the fan would be varying, not simply going fully on / fully off. The speed of the fan would vary, to an extent where it would be:
Fully off (no cooling required).
Fully on (can’t cool enough).
Speed changing from fully on to a set speed (system integrating towards the heat set).
Speed changing from fully off to a set speed (system integrating towards the heat set).
Speed steady (system cooling reached a ‘steady state’ heat set reached).

Andy

BTW Matthew5664 was spot on controller wise. This is essentially what you are trying to design here!

 

[sophos9]

Ahhh.... I was referring to the electronic temperature module that takes a temperature sample every 10 seconds...

You examples of on/off/steady speeds sounds great!! Is it easy to link a LCD/LED readout of current temp and throw some LED's in to show its fan state?

Thanks again Andy - see you are too good at this stuff!!

 

[Underwurlde]

With you.

That temperature module will give you the same hysteresis problem as the float switch, but on a much longer time scale, this time because the output (demand) is either on or off. This is OK, if you can live with the noise of the fans being unnecessarily on and the fact that the cooling system will be fighting (if you like) the heater and would be quite a horrible waste of energy.

Is it easy to link a LCD/LED readout of current temp and throw some LED's in to show its fan state?

The digital meter could be used to show temperature - I'm thinking of this as a separate system and will have nothing to do with the working of the cooling control system you want to design here.

As for the LEDs, not easy really. You'd need an analogue display to represent an analogue fan speed: It is the amount of variance of the fans you'd want to be displaying - think it through in your head as to what is happening to the fans when the cooling point is met on a sunny day... You don't really want to show that the system is 'cooling' per se (what does that mean anyway) - you'd want to know that there is:
demand for cooling / fans are therefore on (at whatever speed) & cooling
fans on but in a steady state (constant speed) = cooling temperature is met / steady state (but note the fans are still cooling).

The display can be driven by looking at the 'error' signal in the system (error = [set temp - actual temp] If encroaching 0, temp met)

Again, when you get down to the nitty gritty of electronics engineering, things start to get a LOT more complicated. Nothing is ever easy, is it?

Andy

 

[sophos9]

Do you think it would be best to buy one of those ATC800's for £40 all in, I'm guessing the cost will go over £20 just for the parts on a DIY project?

 

[Underwurlde]

Dare I say, I think you're right there.

Andy

PS
Ha! Now you are thinking like an engineer. (Don't do it, you're a young man!!!)

 

[sophos9]

- Guess thats that then!!

Matthew5664 - thanks for your post, will probably end up getting a ATC800

Thanks to everyone!