DIY Mini-chiller
- Thread starter Keyoke
- Start date
Just a quick update - what with the weekend over, i'll have to slow down on this some.
The arduino got here today, and i've been messing around with the thermistor circuit this evening. I'm doing everything P2P at this point, the clips in my breadboard are a wee bit corroded and i think that's interfeing with my circuit some. I'll have to head over to Frys or radio shack and pickup a new one tomorrow.
I ordered some 6x6 inch .020 Ti sheet from McMaster today for about $15.00, and a 1/8" 12x12" sheet of acetal (derlin) as well.
So the new plan is to use the derlin as a 'housing' for the tec. I'll essentially cut this to size, and then cut a 40x40mm hole in the middle clear through. i'll face the underside with the Ti, and hold it all together with some thermal expoxy.
I'm still deciding as to wether or not i need to fashion a new waterblock, but i don't think so. At least this time around.
Heatsink wise i think i'm OK. The one I have now is a 3x3x3 cube, with a 80mm 12/1a fan that blows through the fins. I salvaged this a few years ago from some lab equipment at previous employment and i think this should work out fairly well. I did test the TEC with this and the PC power supply and I was able to get the unit to "freeze", so i think it should be plenty big. if it's not well.. i'll just have to go get a bigger one.
The arduino got here today, and i've been messing around with the thermistor circuit this evening. I'm doing everything P2P at this point, the clips in my breadboard are a wee bit corroded and i think that's interfeing with my circuit some. I'll have to head over to Frys or radio shack and pickup a new one tomorrow.
I ordered some 6x6 inch .020 Ti sheet from McMaster today for about $15.00, and a 1/8" 12x12" sheet of acetal (derlin) as well.
So the new plan is to use the derlin as a 'housing' for the tec. I'll essentially cut this to size, and then cut a 40x40mm hole in the middle clear through. i'll face the underside with the Ti, and hold it all together with some thermal expoxy.
I'm still deciding as to wether or not i need to fashion a new waterblock, but i don't think so. At least this time around.
Heatsink wise i think i'm OK. The one I have now is a 3x3x3 cube, with a 80mm 12/1a fan that blows through the fins. I salvaged this a few years ago from some lab equipment at previous employment and i think this should work out fairly well. I did test the TEC with this and the PC power supply and I was able to get the unit to "freeze", so i think it should be plenty big. if it's not well.. i'll just have to go get a bigger one.
you can make the TEC freeze with no heat sink on it at all, the power supply is plenty big enough, but you have ot remember they work off a temp differance. When you start warming up the cold side with your tank water the other side will get hotter, and thats where your heat sink can and probably will fall short.
ReefEnabler
Premium Member
A ranco would certainly work, just wouldn't maximize the potential of a peltier based system.
The ranco will either be on or off, whereas using pwm, the peltier can be controlled to only apply a certain amount of energy.
Of course, it will probably work just as well as a cycle-on/off chiller.
The ranco will either be on or off, whereas using pwm, the peltier can be controlled to only apply a certain amount of energy.
Of course, it will probably work just as well as a cycle-on/off chiller.
MarkS
Premium Member
He seems fully able to complete that project. If he needs help, I'm here as are others with electronics knowledge. Worst case scenario is that I give him all of my files to my temperature controller and let him go from there. I doubt it will come to that though.
Yes, as I stated earlier. A relay based system is not the way to go. PWM using a simple PID loop and a MOSFET is the way to go.
BeanAnimal
Premium Member
I do not see any benefit in using a feedback loop to control the power delelivered to the TEC.
Our operating environment is a very narrow band. Each specific TEC has a balance point at which a certain applied power produces the most cooling per watt consumed. Because our "cold" side and "hot" side ambient temperature will fall within a very narrow band, the best operating point for the TEC will change very little. An "ON" OFF" controller and power supply tailored to the average efficiency point will be more than suitable.
Feedback loop controllers makes sense where the capacity of the controller has the potential to greatly change the process temperature in a very short period of time and the temperature need to be held at a constant. Think of a HUGE heater in a tiny furnace. On "high" the heater can raise the temperature in the furnace in a matter of seconds or minutes. We can use a feedback loop to keep that furnace at a MUCH LOWER constant temperature without the large swings that an "ON" "OFF" thermostat would cause.
Our operating environment is a very narrow band. Each specific TEC has a balance point at which a certain applied power produces the most cooling per watt consumed. Because our "cold" side and "hot" side ambient temperature will fall within a very narrow band, the best operating point for the TEC will change very little. An "ON" OFF" controller and power supply tailored to the average efficiency point will be more than suitable.
Feedback loop controllers makes sense where the capacity of the controller has the potential to greatly change the process temperature in a very short period of time and the temperature need to be held at a constant. Think of a HUGE heater in a tiny furnace. On "high" the heater can raise the temperature in the furnace in a matter of seconds or minutes. We can use a feedback loop to keep that furnace at a MUCH LOWER constant temperature without the large swings that an "ON" "OFF" thermostat would cause.
Actually, I can't wait to get more into the microcontroller. I don't have any intention of shelving this project, that's for sure. The evaporation rate that my little BC29 is going through (over 1gal a day) is a little on the high side IMO.
I looked at the ranco, but finding the arduino.. well.. I just couldn't pass that up.
The Ti and Derlin sheet showed up yesterday. I'm in Nebraska this weekend though for the 4th. I'll pick everything else back up when I get back into town.
I looked at the ranco, but finding the arduino.. well.. I just couldn't pass that up.
The Ti and Derlin sheet showed up yesterday. I'm in Nebraska this weekend though for the 4th. I'll pick everything else back up when I get back into town.
BeanAnimal
Premium Member
I prefer to allow the Ranco to control my tank temps... I am confident in my coding skills, but am more confident in a commercially tried and proven temperature controller.
You get well over 8,000 BTU's of cooling per gallon of evaporation.
That is about 333 BTU/h on average or more like 550 BTU/h during the hot part of the day and 150 BTU/h during the night.
Before I changed some of my pumps out for a larger single sequence pump, I was evaporating 5-7 gallons a day on a ~150 gallon system. I used small fans to keep the air moving over the surface of the tank.
You get well over 8,000 BTU's of cooling per gallon of evaporation.
That is about 333 BTU/h on average or more like 550 BTU/h during the hot part of the day and 150 BTU/h during the night.
Before I changed some of my pumps out for a larger single sequence pump, I was evaporating 5-7 gallons a day on a ~150 gallon system. I used small fans to keep the air moving over the surface of the tank.
OK, so I've been planning this thing out a bit more whilst I wait for the DS18S20 to get here (should be tomorrow or Thurs).
In the mean time, I started working out how I want to cut the Ti sheet and Derlin.
Here's what I came up with (no dimmensions yet, can't find my steel ruler. I think the kids did something to it)
"TEC" Side of the unit:
Click for full size
"Waterblock" Side:
Click for full size
I'm not a SketchUp guru, franky I find the software awfully hard to use (I want 3D Studio R4 Back dagnabit!) so try not to laugh too terribly hard.
As I mentioned, I plan on pasting the Ti sheet to the derlin with the help of some thermal epoxy.
I also came upon a rather nifty little heatsink. Open box special on it - spent a whopping $5.00, so I couldn't pass it up. It's probably STILL not beefy enough, but it does do a much better job in my tests than the other one (and it's much much quieter!).
I've also modified my temp controller design slightly. I will, i believe go the PWM route for the TEC/Fan assy, but I'm still going to go ahead and use a 5VDC/120VAC relay so I can power on/off the associated pump. I dunno if that's a bit overkill or not, but what the heck.
In the mean time, I started working out how I want to cut the Ti sheet and Derlin.
Here's what I came up with (no dimmensions yet, can't find my steel ruler. I think the kids did something to it)
"TEC" Side of the unit:
Click for full size
"Waterblock" Side:
Click for full size
I'm not a SketchUp guru, franky I find the software awfully hard to use (I want 3D Studio R4 Back dagnabit!) so try not to laugh too terribly hard.
As I mentioned, I plan on pasting the Ti sheet to the derlin with the help of some thermal epoxy.
I also came upon a rather nifty little heatsink. Open box special on it - spent a whopping $5.00, so I couldn't pass it up. It's probably STILL not beefy enough, but it does do a much better job in my tests than the other one (and it's much much quieter!).
I've also modified my temp controller design slightly. I will, i believe go the PWM route for the TEC/Fan assy, but I'm still going to go ahead and use a 5VDC/120VAC relay so I can power on/off the associated pump. I dunno if that's a bit overkill or not, but what the heck.
One more pic - an exploded view I forgot to put in my last post.
Click for full size
A little off-topic blurb too.. in looking at how to accurately cut the Derlin, I came across these little desktop CNC machines.. There are a ton of these things from different vendors, DIY kits, etc out there.. Wow. I never knew. Granted it's a bit overkill to use just to cut 1 piece of plastic.. but still.. it's pretty neat.
Anyway, cutting wise I'll be using my trusty Dremel and a end-mill bit I got from Drill Bit City. Nifty little site for those teeny drill bits and end-mills for doing circuit boards, etc.
Click for full size
A little off-topic blurb too.. in looking at how to accurately cut the Derlin, I came across these little desktop CNC machines.. There are a ton of these things from different vendors, DIY kits, etc out there.. Wow. I never knew. Granted it's a bit overkill to use just to cut 1 piece of plastic.. but still.. it's pretty neat.
Anyway, cutting wise I'll be using my trusty Dremel and a end-mill bit I got from Drill Bit City. Nifty little site for those teeny drill bits and end-mills for doing circuit boards, etc.
The 1-wire temp sensor is finally here, and so are some other misc. electronic parts I ordered.
I'm quite impressed with the little thing, as it's awfully easy to read the temp from the unit compared to all the darn math that the Thermistor needed, and the circuit is *extremely* simple when hooked to the Ardunio.
Click for full size
I'm not using the transistor shown, since I don't need the sensor to do internal temp conversion, so I don't really need to keep the 1-wire buss "High" for that 750ms that the datasheet suggests.
The cool thing, is that I can just keep sticking DS18B20's onto the bus, and not have to worry about any additional parts. Each 18B20 has it's own serial number/address combo. It's very slick.
I've also nearly completed the fabrication of the unit that will house the TEC. I discovered that JB Weld and Derlin don't "stick" to each other very well. After about 24 hours of drying time, I was able to just "peel" the TI right off of the derlin, with all the epoxy stuck to the Ti.
Took forever to sand that stuff off, and it smelled like unwashed feet when I did it too!
I roughed up the derlin a bit and am trying again. It's been drying for about 20 hours so far. I've got about 60 lbs of brick stacked on the unit right now.
I've had to modify the heatsink too a bit. After I test-fit everything, I realized that the mounting feet that the heatsink uses raise it up a little too high for the TEC to come into contact with the copper heat pad on the bottom. I broke out the trusty dremel and sawed these off and now it fits perfectly. Without the mounting feet however, I've had to make some changes as to how I'd planned on getting the heatsink combo attached. Luckilly, the screws that hold the fan unit to the top of the heatsink are removable, so I'm going to be using those instead. I picked up some 1/8" aluminum tube, and a 4-40 tap, and I'll use tube to act as a sort of.. retainer. Screws from the bottom of the derlin/TI into the AL tube, and screws "down" from the heatsink fan assembly into the tube from the top side and i should have a fairly decent mounting system.
I'll post pics as soon as my camera recharges.
I'm quite impressed with the little thing, as it's awfully easy to read the temp from the unit compared to all the darn math that the Thermistor needed, and the circuit is *extremely* simple when hooked to the Ardunio.
Click for full size
I'm not using the transistor shown, since I don't need the sensor to do internal temp conversion, so I don't really need to keep the 1-wire buss "High" for that 750ms that the datasheet suggests.
The cool thing, is that I can just keep sticking DS18B20's onto the bus, and not have to worry about any additional parts. Each 18B20 has it's own serial number/address combo. It's very slick.
I've also nearly completed the fabrication of the unit that will house the TEC. I discovered that JB Weld and Derlin don't "stick" to each other very well. After about 24 hours of drying time, I was able to just "peel" the TI right off of the derlin, with all the epoxy stuck to the Ti.
Took forever to sand that stuff off, and it smelled like unwashed feet when I did it too!
I roughed up the derlin a bit and am trying again. It's been drying for about 20 hours so far. I've got about 60 lbs of brick stacked on the unit right now.
I've had to modify the heatsink too a bit. After I test-fit everything, I realized that the mounting feet that the heatsink uses raise it up a little too high for the TEC to come into contact with the copper heat pad on the bottom. I broke out the trusty dremel and sawed these off and now it fits perfectly. Without the mounting feet however, I've had to make some changes as to how I'd planned on getting the heatsink combo attached. Luckilly, the screws that hold the fan unit to the top of the heatsink are removable, so I'm going to be using those instead. I picked up some 1/8" aluminum tube, and a 4-40 tap, and I'll use tube to act as a sort of.. retainer. Screws from the bottom of the derlin/TI into the AL tube, and screws "down" from the heatsink fan assembly into the tube from the top side and i should have a fairly decent mounting system.
I'll post pics as soon as my camera recharges.
Ok, so here's the nearly completed cooling element thing. Sorry about the lighting, I'm using the cam on my laoptop.
Top:
Click for full size
Bottom:
Click for full size
Without the heatsink, you can see the "well" that the TEC fits into:
Click for full size
And one last shot of just the Ti back/waterblock assembly:
Click for full size
It's awfully hard to tell, but I "lapped" the Ti on the inside of the TEC 'well' area to a smoother surface (not mirror finnish, but it's way better'n the dull finish it was before) and the back side is polished up a wee bit.
And now I get to wait till this evening so I can claim the bathtub for a few hours to do a leak test.
Top:
Click for full size
Bottom:
Click for full size
Without the heatsink, you can see the "well" that the TEC fits into:
Click for full size
And one last shot of just the Ti back/waterblock assembly:
Click for full size
It's awfully hard to tell, but I "lapped" the Ti on the inside of the TEC 'well' area to a smoother surface (not mirror finnish, but it's way better'n the dull finish it was before) and the back side is polished up a wee bit.
And now I get to wait till this evening so I can claim the bathtub for a few hours to do a leak test.
MarkS
Premium Member
<a href=showthread.php?s=&postid=12938286#post12938286 target=_blank>Originally posted</a> by Keyoke
Click for full size
I'm not using the transistor shown, since I don't need the sensor to do internal temp conversion, so I don't really need to keep the 1-wire buss "High" for that 750ms that the datasheet suggests.
That transistor, actually a MOSFET, is not for internal temperature conversion. It is necessary if the DS18B20 is running off of parasitic power. In other words, the Vdd pin on the sensor is not connected to a 5V supply. The MOSFET pulls the data line high to supply power to the sensor to begin sampling. If you are using parasitic power, which I cannot imagine why you'd feel the need to, you'll definitely need this. Either way, do not forget the 4.7K pull up resistor on the data line.
BeanAnimal
Premium Member
To add to what Mark said, the DS18S20 is a digital sensor that needs to be triggered to take a sample... you can't just hook it up and read continuous data from it like the analog sensors. The "internal conversion" is this sampling process. The sensor is triggered, it does a conversion and write the data to the scratchpad so that the user can read it. If you don't do a conversion, then you are reading whatever was on the scratchpad last.
Bean
Bean
