Society of Robots - Robot Forum
Mechanics and Construction => Mechanics and Construction => Topic started by: gamefreak on October 10, 2009, 07:35:11 PM
-
For a school project I am working on a mini-fridge using peltier units. I intend to have a heat sink on the hot side either submerged in some fluid or with a fan blowing across it and on the cold side either just leave it as it is or use a fan to help circulate the air. I'm having a bit of trouble figuring out materials to use and how to mount the thing due to its size and my lack of knowledge on anything thermal. I got a couple of heat sinks donated by friends and got some peltiers online.
The peltiers are 40mm x 40mm x 4mm
The heatsinks basically have the same 2 in by 2.5 in base but the fins are different heights and widths. One of the heatsinks has bent fins and the bottom is scratched and pitted due to getting shot off the top of a car due to forgetfulness.
The main problem is that I cant quite figure out a material that will go in that 4 mm gap between heatsinks that wont conduct too much heat but is rigid enough to let me build the box out of it. Also i've been struggling to figure out what I am going to do with all of the heat generated, the original plan involved mineral oil but the research i've done seems to point to it having a low thermal conductance, so does anyone have ideas as to a good way to be rid of the heat?
-
I've used Peltiers for cooling CCD imaging chips. The colder the CCD is the less noise is generated during a long exposure from imaging astronomic objects.
There is some information on modifying CCD cameras to add a Peltier (also called a TE), google will find them. Also google "cookbook camera".
Here is one web site with picture for ideas:
http://www.heavensgloryobservatory.com/my_cookbook_camera_project.htm (http://www.heavensgloryobservatory.com/my_cookbook_camera_project.htm)
The second picture shows the TE between the cold finger (where the CCD is mounted) and the hot side spreader. The two plates are connected by two screws. The metal screws do transfer a little heat but not enough to worry about. Also Nylon screws could be used that would transfer almost no heat.
In the cookbook camera the hot side has a hollow water jacket in which cool water is pumped through to carry away the heat. Standard heat sinks with a fan also work but don't remove quite as much heat.
The damaged heat sink can be fixed. Straighten the bend fins with pliers, they don't need to be perfect. The surface that mates to the TE does need to be very good however. Rub it across a sheet of sand paper that is placed on the flat surface. Use finer and finer sand paper until the heatsink's surface is flat and smooth.
Here are some more links:
http://tocobs.org/ccd.htm (http://tocobs.org/ccd.htm)
http://ghonis2.ho8.com/rebelmod450d16c.html (http://ghonis2.ho8.com/rebelmod450d16c.html)
http://hackaday.com/?s=peltier (http://hackaday.com/?s=peltier)
http://adsabs.harvard.edu/full/1999ASPC..189...24T (http://adsabs.harvard.edu/full/1999ASPC..189...24T)
Hope this helps.
-
Connecting the two sides with screws could work, but then I'm still stuck without any surface to build a frame off of. It seems that this should be simple but I just cant wrap my mind around this.
Also, would having say a tub of water on the hot side work to absorb most of the heat? My original logic says it would, but the then the heat would build up faster then the water releases it into the container and it would slowly fail....
-
Yep, a tub of water will just heat up until the TE can't transfer any more heat.
With my cookbook camera I use a 5 gal pail of water and after running for 4-5 hours the water is noticeably warmer. I now run the water from the TE through a coil of copper tubing that has a fan blowing on the tubing. This does get rid of most of the heat before returning the water to the pail.
Prototype it first and just use a fan. Then measure the temperature difference to determine cooling.
For prototyping maybe build a box out of foamcore board. This stuff is cheap, easy to cut and glue and can be bought at most arts and craft stores.
Did you look at the links? On is a TE mounted through a container to cool a DSLR inside the container.
-
I did glance through the links, if I'm thinking of the right one then the peltier was ripped from a cooler and had a nice plastic piece surrounding it that allowed for easy attachment to the metal box. Also is there a slight gap between the peltier and the metal coffee can in the ghonis link? wouldn't that slow down the cooling process?
I think I might be a bit too worried about wasted heat since every material that has come to mind has quickly been dismissed. I do have some foam board lying about but no fun ways of measuring temperature(not even sure if I have a thermometer around my house I could use).
Also it seems that air already a ridiculous thermal insulator, so would it be best to just ignore that 4 mm gap and buy a block of some metal, attach it to the cold side, bolt it to the hot side, bolt/screw wood into the cold block, surround wood with styrofoam, and be happy?
-
thin aluminum angle and channels are available at any Ace hardware. just RTV the edges.
thermal grease used on heatsinks is good for 1 mm or so.
NASA stacked Peltier coolers together to cool the preamps on the antennas that tracked eath orbiting spacecraft. a noise reduction thing.
-
Alright, so I made a prototype out of duct tape, hot glue, and foam board, drilled some holes through the heat sink, forced some washers onto some nuts, and attached the heat sink to the foam board and the peltier.
It was constructed by cutting the sides out with a knife and then duct taping them so they hold the shape, I then hot glued the inside edges so that it would be "air tight".
it basically resembles this:
Green = foam board
dark grey = heat sinks
purple = peltier wires
yellow = bolts
red/blue = hot/cold
white = peltier
(http://img88.imageshack.us/img88/894/peltier.th.jpg) (http://img88.imageshack.us/i/peltier.jpg/)
the only difference is that the current design does not have a heat sink on the cold side.
Overall the foam board seems strong enough and if I tighten the nuts enough then I think it should squish down to the size of the peltier. But would there be a better way then having that area of foam board squished between one side that hot and one side thats cold? Or is foam board resistant enough to heat that I can basically ignore it?
-
The hot side should never get hot enough to melt the foam in the foam board. The heat sink should pull enough heat out and into the air.
Your mock-up sketch looks good. Have you applied power to the Peltier yet?
Try starting with a lower voltage (lower current) to check heat transfer.
Keep us posted.
-
Prior to building I set the peltier on top of the heat sink in a tub of water, within 30 seconds I could dip my finger in the water, touch the top of the peltier, and freeze my finger to it XD
Im reluctant to apply power since:
1. the current heat sink is fairly scratched and has a few pits in
a. that heat sink will probably become the cold side sink since I dont really care that much about that side
2. I cant touch the peltier so I am uncertain as to how much heat the heat sink is conducting
I need to drill holes in the good heat sink and procure some thermal paster, at that point i'll power it up and see how well the box works.
The problem with the power supplies I have access to is they only go up to 3 amps, so I can not do a full power test but that is a job for the electronics forum...
-
Here are a few tips . . . I've taken classes on heat transfer, thermodynamics, thermofluids, and fluid mechanics, so I know a thing or two :P
First, air is a horrible conductor of heat. That gap you have, you want to fill it with a polished and flush metal plate. Roughness means air gaps. Thermal paste should only be a last resort to get rid of air gaps.
Copper is the best. Aluminum is second best. Copper is more expensive, but it can conduct heat away almost twice as fast as aluminum.
There is only one good way to mount your heat sink, and that is *not* flat and parallel to the ground. You want it so that the fins are parallel with gravity. This is so that the heat sink can naturally cool by rising hot air (convection). Adding a fan in addition to that is even better (forced convection).
Oh and lastly, peltier chips have a flaw. They can get so cold that the moisture in the air will condense on them, forming water droplets that can short electronics. Keep that in mind. This is why they typically aren't used to cool processors.
-
Which gap? the ones on the damaged heat sink, or the one between heat sinks? If its the one between heatsinks then I dont want to conduct heat through there, I need a good solid insulator.
I intend to have the microcontroller, LCD, and power supply hiding in some side compartment that wont be cold in order to avoid the condensation problem altogether. Also, by the pins being parallel to gravity you mean that i need to rotate the doodle ~90 degrees around the inward vector and ~90 degrees along its length, correct? ie. if i were looking straight at it it would look like:
| || | | | || |
| || | | | || |
| || | | | || |
| || | | | || |
| || | | | || |
^look ascii doodle ;D^
If I wasnt busy wiring up 35 AOI logic gates I would experiment tomorrow, but sadly i have a circuit to build. My robotics team is working with a mentor who has access to a 5 DOF milling machine, if I remember correctly the guy said it has precision to like 1/20 of a human hair.... the competition provides us with 0.5" thick aircraft grade aluminum and we normally have extra. Think its worth a shot to get a chunk and mill it to have some small fins and be really smooth?
-
Think its worth a shot to get a chunk and mill it to have some small fins and be really smooth?
Nope. I tried (see attached image way below). Took like 5+ hours of manually using the mill, and we couldn't get close fin spacing (important!) because the mill bit would break. The correct way is to use a big metal cutting spinning blade saw (I don't remember the technical name) to slice the block.
As for heat sink positioning, do it like this (assuming the table wasn't blocking airflow at the bottom):
(http://www.pctechguide.com/images/tutorials/MBoard/Heatsink1.jpg)
and not like this:
(http://www.3cvillage.com/cayman/images/heatsink-cm-sock-a-a.jpg)