Peltier Thermoelectric cooling modules - Has anyone used a Prop to rule these b
ElectricAye
Posts: 4,561
Ladies and Gentlemen,
I'm interested in using Peltier thermoelectric devices for cooling photomultiplier modules (and much more!). And it seems to me that I read somebody on this forum was designing a Peltier to cool their beer or milk or something, so I thought I might ask if anyone has had any useful experiences to share on this oh so important matter. I found the following via the wonderful search engine:
http://forums.parallax.com/forums/default.aspx?f=15&m=340337&g=340778#m340778
Helpful though it was, I crave even more information, especially as it might pertain to using a Propeller in my scheme. By myself, I've learned a few things:
Said Peltier devices consume about 12 volts at roughly 10 amps. Word on the street is that old computer power supplies might be good for kludging Peltiers. I found this interesting link that describes the pin outs of ATX type power supplies:
http://en.wikipedia.org/wiki/ATX
(EDIT: I should note that I heard Dell computers circa 1996 to 2000 had power supplies with wiring that LOOKED LIKE ATX standard but were NOT, so it would be prudent to test all the pin connections before using any of them - duh.)
I have no idea what ATX stands for, but it does appear that these kinds of power supplies have a MINIMUM requirement of current draw. That makes the control strategy kinda interesting, especially if you want to get the most power out of it as possible. Somehow a minimum dummy load must be used maybe? until you want to kick in the Peltier? Is there some way of gracefully switching from dummy to non-dummy loads? Being a Mech E, I'm thinking about some kinda duty cycling using bang-bang control, since the hammer metaphor suits me well. But maybe there are other approaches most wondrous and elegant?
These ATX type power supplies also can be turned on and off via a control pin, maybe just the thing for Propeller control, but I don't know if it would be wise to have an entire power supply clicking on and off day and night, especially if my temperature range is kept too tight and/or the temperature differential between outside and inside is so great that the heat loss is just oh so fast.
Naturally, of course, I'd like to place the controller on a PCB, but I don't know if your standard everyday PCB could gingerly handle guiding those ape-like level of amps through those tiny foil-like pathways? Anyone have experience with voltages and amp apes of this nature?
Too, I thought about using DS18B20 digital thermometer sensors to keep track of everything. But... I'm open to suggestions.
Any comments or insights will be absorbed with enthusiasm,
many thanks,
Mark
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Watching the world pass me by, one photon at a time.
Post Edited (ElectricAye) : 8/8/2009 3:38:43 AM GMT
I'm interested in using Peltier thermoelectric devices for cooling photomultiplier modules (and much more!). And it seems to me that I read somebody on this forum was designing a Peltier to cool their beer or milk or something, so I thought I might ask if anyone has had any useful experiences to share on this oh so important matter. I found the following via the wonderful search engine:
http://forums.parallax.com/forums/default.aspx?f=15&m=340337&g=340778#m340778
Helpful though it was, I crave even more information, especially as it might pertain to using a Propeller in my scheme. By myself, I've learned a few things:
Said Peltier devices consume about 12 volts at roughly 10 amps. Word on the street is that old computer power supplies might be good for kludging Peltiers. I found this interesting link that describes the pin outs of ATX type power supplies:
http://en.wikipedia.org/wiki/ATX
(EDIT: I should note that I heard Dell computers circa 1996 to 2000 had power supplies with wiring that LOOKED LIKE ATX standard but were NOT, so it would be prudent to test all the pin connections before using any of them - duh.)
I have no idea what ATX stands for, but it does appear that these kinds of power supplies have a MINIMUM requirement of current draw. That makes the control strategy kinda interesting, especially if you want to get the most power out of it as possible. Somehow a minimum dummy load must be used maybe? until you want to kick in the Peltier? Is there some way of gracefully switching from dummy to non-dummy loads? Being a Mech E, I'm thinking about some kinda duty cycling using bang-bang control, since the hammer metaphor suits me well. But maybe there are other approaches most wondrous and elegant?
These ATX type power supplies also can be turned on and off via a control pin, maybe just the thing for Propeller control, but I don't know if it would be wise to have an entire power supply clicking on and off day and night, especially if my temperature range is kept too tight and/or the temperature differential between outside and inside is so great that the heat loss is just oh so fast.
Naturally, of course, I'd like to place the controller on a PCB, but I don't know if your standard everyday PCB could gingerly handle guiding those ape-like level of amps through those tiny foil-like pathways? Anyone have experience with voltages and amp apes of this nature?
Too, I thought about using DS18B20 digital thermometer sensors to keep track of everything. But... I'm open to suggestions.
Any comments or insights will be absorbed with enthusiasm,
many thanks,
Mark
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Watching the world pass me by, one photon at a time.
Post Edited (ElectricAye) : 8/8/2009 3:38:43 AM GMT
Comments
I have just picked up a powerful peltier and a 12.288MHz Crystal. I am going to attempt to run the prop at 196.7MHz after my honeymoon. Crazy I am sure but will be fun experiment.
I have always used a car battery in my playing with peltier they have the right voltage and can supply very high currents for long periods of time. Battery chargers are not expensive and many are smart enough to be left on all the time.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5
Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.
Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
That's completely insane: you are supposed to overclock DURING your honeymoon!
Hmmmm.... not a bad idea! Bartender, get that man a positron mass spectrometer!
A solid state relay may be better than using the control pin to power the power supply on and off. They come in heat sink friendly packages and the leds inside need some 10 mA to conduct.
I'm not an expert in mass spectrometry yet but I use a tandem ms several days a week. (for the last 5 years)
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Visit some of my articles at Propeller Wiki:
MATH on the propeller propeller.wikispaces.com/MATH
pPropQL: propeller.wikispaces.com/pPropQL
pPropQL020: propeller.wikispaces.com/pPropQL020
OMU for the pPropQL/020 propeller.wikispaces.com/OMU
I would keep the PSU on 24/7, switch the 12v rail(s) you need with the appropriate duty cycle to keep your temperatures in check. I'd watch the hot side as well as the cold side to make sure your heat syncing is working. You could run the fan off separate PWM if you care to balance your cooling on the hot side... could be nice to reduce noise. Just be careful driving inductive loads (relays, fans, motors) because they can blow transistors if you do not protect the output with a diode.
Many source driver IC's have this built in for controlling relays and such... I have had good luck with this one:
Toshiba 8ch source driver "td62783a". It is a little weak to handle your 10A draw of the peltier's, but you can parallel as many of the 8 channels as needed to provide enough current to run you fan off 12v. You can also use 1 or 2 channels to drive your IGBT from the 3.3v the propeller puts out.
Hey, that's really cool. I didn't know that boosting the voltage will speed them up. That will help a lot! Thanks!
And to Tired2 and pmrobert, thanks very much for the tip on using an IGBT. I had never heard of these before but they definitely look like the way to go. Thanks a million!
bless you guys,
Mark
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Watching the world pass me by, one photon at a time.
Regarding your use of the ATX power supply to run a Peltier cooler, you may be able to use the power supply without any external high current switches by using the control signal of the power supply to turn it on and off as required. I believe there is also a low power supply that is always on so that a PC can "wake up" on an interrupt or other signal. This could be used to power the prop or other control circuit.
With the thermal inertia of the cooled volume, peltier cooler, and heat sink I doubt you will be turning the power supply on and off more than once every few minutes so I would try that first. What have you got to loose? ATX power supplies range from free to ridiculously cheap so if it dies after a while it is not a big loss, and the rest of the circuit can be reused if replacing it with something else becomes necessary.
1. They draw LOTS of current, but if you don't reach their max voltage, they wont draw the power they are supposed to. My device is rated for 230 watts with a max voltage of 15.4. At 12 volts, it only draws 7.5 amps for a dismal 90 watts. It still freezes over in about 10 seconds, but it is still not what I was hoping for.
2. You have to move the heat away from the hot side! This means a giant heat sink attached with thermal paste properly applied. You may also need a fan to keep it cool. If heat is not moved away from the hot side quickly enough, heat will leak through to the cold side and raise the temperature significantly. Too much heat and you will have a fried Peltier device.
3. If using a computer power supply, you need a dummy load on the 5 volt rail. On my power supply without the dummy load, I get about 11 volts out of the 12 volt rail. This jumped to 12.5 when the dummy load was installed (10 watt, 1 ohm power resistor), but dropped back down to 12 when the Peltier junction is attached. On my Dell power supply, you connect the green wire to any black wire to turn the thing on.
Other thoughts: Peltier devices move lots of heat, which means you need a very large heat sink. I bought a fairly large heat pipe heat sink designed for a cpu off of Ebay to cool my Peltier, but it was not enough to keep the surface frozen for more than a couple of minutes. Even with crazy server fans that sound like a mini shop vac, the heat sink cannot remove heat fast enough. It starts to melt around the edges and eventually makes its way into the center.
Peltier junctions are great for condensing water. With a PWM scheme, you could hold the temperature just above the freezing point, which is optimal for condensing water. I just plug in the peltier to the 5 volt rail to condense water, then switch it over to the 12 volt rail to freeze it.
Do not lick the cold side while the Peltier device is on! It gets very cold! In my picture, there is a LM34 temperature sensor frozen to the surface of the Peltier junction. It produces 10mv for every degree above zero (in Fahrenheit). My meter is measuring 95mv, so the temperature is 9.5 degrees Fahrenheit! The way I have the LM34 hooked up only allows temperatures down to 5 degrees, so the actual temperature may be even lower. If you're crazy, you can get the temperature down pretty low, as most Peltier junctions are rated down to -76 F.
Have fun!
Kevin
Thanks, Kevin. This is important wisdom you are sharing here! I have a question about your above quote: when you say "a dismal 90 watts", I presume that is the amount of power that the Peltier consumes at those given voltages and currents, but do you have any idea how to calculate how much heat/power the Peltier is moving from the cold side to the hot under those conditions? It would not be 90 watts worth, since the device is not 100 percent efficient, but any guess as to how to determine the Peltier's handling capacity under lowered voltage and current conditions? I've seen a spec sheet for these devices but it's not clear to me how to derive that information.
As for moving air at shop vac speeds for cooling the hot side, I guess it would make sense to use a liquid such as water since liquids are better than gases at carrying away heat. For my application that's okay but I bet a lot of people wanting to cool their CPUs aren't going to want anything liquid snaking around their electronics.
Yep, you're right on both counts.
thanks again,
Mark
I'll probably try both this way and the PWM method. Searching the internet, I have found numerous forums on which people have hotly debated whether or not PWM is the way to go with Peltiers vs. bang-bang control. People have pointed out that the rapid heating and cooling of the Peltier flexes its internal components, so bang-bang cycling can, under some circumstances, lead to accelerated thermal fatigue, causing the unit to fail prematurely. Apparently the same thing can happen with PWM if the frequency is too low. The vibe I'm picking up is that PWM is fine so long as the frequency is somewhere above about 2kHz. Also I'm hearing that it is important to "softly" start up the Peltier so there isn't a super-duper current rush at the start (when its internal resistance is low), which causes rapid heating and adds to the thermal fatigue problem. One person recommended that I use a "constant current source controlled by a PWM signal", though I'm not sure how that is different from just plain ole PWM mentioned above using an IGBT (maybe becuause Peltiers have a strong current-to-temperature relationship???).
One challenge to bang-bang control is what happens when the Peltier is unpowered: suddenly that thin little square of Peltier material can conduct heat. The Peltier is not an insulator, so if it's not actively pumping, it becomes a wide open door, thermally speaking. So that might cause bang-bang to really bang away, especially for a small device and/or thermal sensor attached directly to the Peltier's cool side.
So, long story short: getting a Peltier to do something cool is one thing, getting a Peltier to work optimally is one of those tasks I'll have to wake up and work at - darn it, yet another attack on my lazy nature...
many thanks for your help,
Mark
Mark (ElectricAye), may I ask what you're using your photomultipliers for ?·
am I correct that you'd like to cool them to reduce quantum noise effects?
These are interesting devices. I've seen several setups that were really sensitive. In one, they had tweeked it·(somehow*) so that the SNR·was so low that they could detect just a few quanta! In another at the NHMFL (National High Magnetic Field Lab) in Tallahasee, FL, some years back, they used liquid helium to supercool the coils. But that device did weird things they didn't understand yet [noparse]:)[/noparse])· (Electrons can disobey the Pauli-exclusion principle at those temps and can 'pair' up - imagine what that does to the multipliers!) They were just trying to get an effecient, ultra-low noise·system, but ended up with a whole new line of research. (This was years ago, and I'm not in contact with that group at the present... now I'm curious what came out of it.)
On the first set up - it still amazes me that the EE guy was able to reduce the noise that much using just off the shelf resistors and caps. The P.Multipliers were also off the shelf. (*) I still kick myself for not getting the schematic which they offered
Really interested in hearing what you're doing - if you can tell us.
thanks
- Howard
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Post Edited (CounterRotatingProps) : 8/10/2009 4:29:13 PM GMT
I agree with every word of kevin101. As a practical advice,I should quickly add that do not lick the hot side, too. There·are some other rule of thumb's or simple facts that helped me to construct, use and sell Propeller controlled programmable laboratory crystallizers, without moving parts and without liquid or gas coolants.
Peltier elements have very low efficiency. They will consume more power than they transport! Actual peltier elements may consume twice as much energy (in the form of electricity) as they transport (in the form of heat). So, if you are using a peltier element, the heatsink it is used with must be much more powerful than a heatsink used for cooling a heat source without peltier element. Take 3x as minimal first approximation.
Do not confuse the maximum amount of power a peltier element can transport with the maximum amount of power usage of the peltier element. Retailers sell "140W peltier element", without stating what this value actually means. This is misleading. I use 140W High-end Peltier elements that can pump a max of 40W in form of heat. See attachment.
You may ask, that if we have to use more powerful heatsink with peltier, what is it good for?
Short answer: A more powerful heatsink is always better.
Detailed answer (With many practical approximations):
Imagine that a chip generates heat with 10W. If that heat/sec is not conducted away, the tiny chip gets hot very soon, and burns. (And the magic smoke that drives electricity, gets out.)
Solution with a small, but not enough, heat sink and unforced air dissipater keeps the chip's temp at 90, but design goal is 80. For example
Chip surface temp.····· Heatsink outer surface temp.········· ·· Room temp.
===================================================
····· 90····
heat goes--->· 60·····
heat goes--->······ · 25
····················· · deltaT=30····················· · deltaT=35
·········· ···· carried heat prop. w.·········· removed heat prop. w.
············ ··· deltaT * Therm.res············ surface * deltaT^1.5
·············
This 60 degree of celsius of the radiator does not burn your hands for a short time and keeps your cat warm, but the chip is too hot.
So, you buy a 140W peltier and insert it between the chip and the original, heat sink. Congratulations. Although, calculation is a little bit more complicated. You see from the diagram, that to remove 10W heat/sec you can have a 55 C drop in temperature, but for the cost of· 20W joule heat that you have to inject. You have to remove now 3 times as much heat from the system! When that 30W has to be removed with the original heat sink/dissipater, values settle approximately as
Chip surface temp.······ ·· Peltier······················· ·Heatsink outer surface temp.·················· Room temp.
================================================================================
······ 133 ----heat goes---> || 188········
heat goes--->··· 98····
heat goes--->······ ·25
·························· ···· (55 drops············ ····· deltaT=90······················· · deltaT=73
························· ····· see diag.)········ · carried heat prop. w.··········· · removed heat prop. w.
······················································ · deltaT * Therm.res············· · surface * deltaT^1.5
·
The magic smoke goes off, and the cat will be angry with you for days.··············
OK, that was only an experiment. Air the room and build a new heat sink/dissipater with a 3 times less heat resistance between the surface of the chip and the fresh air, and with 3 times the surface to the air. To remove 10W+20W the following values will be settled.
Chip surface temp.······· Peltier····················· ·· Heatsink outer surface temp.·················· Room temp.
================================================================================
······ 35· -heat goes->···· || 90········
heat goes--->··· 60····
heat goes--->············ 25
······················ ··· ··· (55 drops··············· deltaT=30···················· ·· deltaT=35
························ ····· see diag.)········ carried heat prop. w.······ ··· removed heat prop. w.
··················································· · deltaT * Therm.res········· ·· surface * deltaT^1.5··
The cat is murring again and the chip is nice cool. But what if, when you do not want to use so much current? As an experiment, you apply the new, large capacity heat sink/radiator without a peltier/power supply/...
Chip surface temp.····· Heatsink outer surface temp.·············· · Room temp.
=========================================================
······ 52····
heat goes--->··· 42····
heat goes--->······ ·25
···················· ·deltaT=10························ · deltaT=17
··········· ·· carried heat prop. w.········ · removed heat prop. w.
·········· ··· deltaT * Therm.res········· ···· surface * deltaT^1.5
·············
Well, the IC is a little bit warmer, than with the peltier, but you spared about 3-4 amps and so decreased your carbon footprint. Although, the radiator is not warm enough for a healthy cat.
You see, everything boils down to the design of effective heat sink, radiator, and that's tricky.You need a good thermal conductor (large cross area, small ther.res.) to remove away the heat from the device, and you need a good· thermal radiator (large surface, HIGH surface temp.(!!!) remember that 1.5power) to remove the heat from the whole thing.
Thermal conductivities in W/(m*K)
Silver·········· ·· 430
Copper··········· 401
Gold············ ·· 318
Aluminum········ 237
Brass··········· · 109
Iron················ 80
Stainless Steel· 16
Bi2Te3 (aniz.)··· ·0.5 - 1.5
Wood··············· 0.05- 0.2
Paper··············· 0.05
Glass Wool········ 0.04
Polystyrene exp.·0.033
Air··················· 0.024··· (but transparent to radiated heat)
Urethane Foam·· 0.02
For thermal conductivities, silver, copper and gold are the candidates. Silver is the best, but it impractical because it can oxidate by air, and holds H2S (hydrogen sulfide or sulphide). The second one is copper. It better needs to be covered when used for electronical purposes. Zinked, golded, painted, silked, etc..., because after some time, the smell of copper may lead to arsenic increase in the body, especially noticeable in the human hair. Chuck Yeager said -I trust Dr. Beer. He knows what he is doing. In Gold we can trust, too. It suits us best chemically, maybe not as·good as·beer, but women·have an other opinion. (In fact, diamond has the highest heat conductivity of all.) At some drugstores you may face with colloidal gold (and Ag) for drinking. It is good also in your liquid cooler tank if you can afford it. (4oz is at least $25.) Is is good, when your buying a half a million dollar car, and you dont want to feel ripped off. Car makers, however, don't use gold on the street versions, so you have to take care of it that case.
For the heat radiator aluminum is the champion. Place a soldering iron on one end of 2 identical size pieces of aluminum and silver, and check the speed of temperature rise.· Aluminum beats silver (copper, whatever) every time and by a huge margin. What? Something is wrong than with that heat conductivity table! No. The physics charts all relating to thermal properties are by molecular mass not relative size. A piece of aluminum of a 1 inch cube has about 20 % of the mass of a cube of silver the same, hence the reason it rises its temperature about 3 times faster. (Only 3 times, as aluminum has about 1.7 times specific heat than silver.)
So, in summary
Silver:··· ·best conductor······ ·(expensive)
Copper:·· 2nd best conductor
Gold:······ 3rd conductor······· ·(more expensive, but healthy to have a lot)
Aluminum: best radiator········· (quickly on high temp)
If you have enough surface on the heat source, use aluminum for conduction and radiation. Use copper to conduct away heat in large quantities from a RESTRICTED area, and distribute it to a 3-4 times larger area within the copper body. The use aluminum on this larger area to remove the heat from the copper. Then dissipate heat toward the environment on an even larger surface area of the aluminum.
That sounds good, but keep in mind that a copper embedding is only useful if it is tightly bonded to the aluminum part for good thermal transfer. This is not always the case, especially not with inexpensive coolers. If the thermal transfer between the copper and the aluminum is poor, the copper embedding may do more harm than good.
Thermoconductive glue or sticky tape should only be used in situations where mounting with firm screws or tight clips isn't possible. When necessary use additional silver paste on untight connections. It's just silver dust mixed with some kind of thermal paste. (Gold would be better but it's too expensive so it's not made. Diamond is the best, but they don't make that either.)
With sandwiches of peltiers you can reach -50 -80 degrees of celsius. Than the thermal insulation of the cooled volume begins to play a role. See second part of the table.
Happy overclocking!
Cheers,
Istvan
Hi Howard,
The on-going project is for detection of bioluminescence taking place at cold temperatures. The biological part of it I don't really understand, being above my pay grade and so forth. The electronics is way above my pay grade, too, but so far I've been able to fake my way through it. My system counts photons and, as you might know, the warmer a photomultiplier tube (PMT) gets, the more electrons get ejected from its active surfaces and so the photon count appears to go up. By cooling the units, the thermionic (noise) electrons are greatly reduced and the signal to noise ratio gets better. Also, since the refrigeration systems used for these microbes are expensive, I thought I'd use the Propeller+Peltier to cool both the PMTs and the little plate systems upon which the microbes do their thing (at temperatures near 0 degrees F). The whole idea is to save money this way, so I hope the Peltier systems can do the trick.
Also, I use a plastic scintillator material to do sanity checks on my system, measuring cosmic rays and background radiations, so I sometimes ponder how to convert my system into a gamma ray spectrometer - someday, maybe, when I get money for a NaI crystal.
Occasionally, I also milk the forum for information concerning acoustic emission amplifiers and so forth because when cold things (such as ice) crack, they sometimes emit light, and I'm probably going to have to filter out those pulses from the actual biological signal as well.
Without this forum and the incredibly intelligent people who answer my questions, I would have gotten nowhere with this project. I might add, too, that this is one of the most polite forums I've ever seen. Too bad other forums aren't so courteous and kind.
thanks, Howard, for your inputs and interest,
Mark
Istvan,
yes, this is the cruel reality that is now setting in for me. The advertised value for a Peltier has little to do with what you can actually pump with it, especially if the target temperatures are really low. Luckily, I'm not using this to cool a CPU, so I've got room to make a pelt of Peltiers and use liquids for cooling, too.
Thanks for your detailed and resourceful insights!
Mark
Won't be on much until 20th
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5
Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.
Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
My conclusion: Water cooling may be necessary for high power Peltier devices (>300 watts), but for lower power stuff, a large heat sink is simpler and cheaper. Water cooling adds many levels of complexity, and many more opportunities for system failure.
To get anywhere near the rated 230 watt power consumption that my device is rated for, I need to power it with 15 volts. At 12 volts, it will only draw 90 watts of power. All I know is that the more power these things draw, the more heat they move and generate. 15 volt power supplies are hard to find for cheap, especially at the high currents Peltier devices operate at.
Does anybody know if Peltier junctions can be put in series? If they can, I was thinking it would be easy to buy 8, 60 watt units and put them in series for 120 volt operation. All that is need is a rectifier and they can operate off an AC line. This saves lots of money on power supplies and reduces complexity. It does make controlling the devices a bit more complicated tho...
Finally, does anybody know where to buy GIANT heat sinks? I was thinking of buying a fairly large chunk of aluminum off of ebay or from a scrap yard, but the lack of fins makes it harder to cool with fans.
evKni?
I have had very good success with LED drivers for peltiers, where PWM dimming was possible at over 5kHz, and the switching speed was >50kHz.· With a series capacitor across the peltier and the sense resistor,·it kept the power fairly clean, and with other added features of the driver (soft-start, enable pins, etc.) it was very easy to control.· I used a CPLD to handle the control, not a Propeller, but a prop could have easily done it.· Once you break the 200Watt mark, you probably need high-volume water cooling, not just a heatsink (the heatsink would start to get a little big...).· Just putting a pool of water over the top of the hot-side isn't enough - the convection currents aren't fast enough, and without a radiation surface the heat won't really go anywhere.
Try using a PC water cooling system to suck away the heat from the junction, a "Core 2"·Intel CPU cooler should be enough, they were around 200W or so.· If you need to drop the temps even further, try running water from the cold side to the item to be cooled then back to the hot side, then to a radiator, tank, pump and back to the cool side.· To get even better cooling go from the cold side of the first cooler to the cold side of the second cooler, then to the target and back to the hot of the second cooler, then the hot of the first cooler... etc.· See below picture.
-T
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
PG
The problem with line AC is that it is running at 60 Hz, so even if it is rectified, the Peltier would be cooling and reheating along with that low frequency cycle. From what I've read, you need a cycle somewhere above 2 kHz otherwise that kind of thermal cycling induces mechanical fatigue - cracks the elements and/or their connections inside. Mind you, this is only from things I've read. I have yet to start experimenting on my own with these things.
And yes, water cooling would require a pumping system of some sort. That's a whole challenge in itself, but I'm hoping to drive these things to fairly low temperatures.
thanks, y'all
Mark
Hey, a very interesting idea. Has anyone worked out the thermodynamic details of this? In some respects I can see how it would work great, but I'm a little concerned about the overall efficiency of it, too. Anybody know for sure?
thanks,
Mark
http://www.kidwind.org/xcart/product.php?productid=24&cat=71&page=1
I'm sure if you looked around you could find some small metal tubing to connect to it and run it at the bottom of the peltier.
As for increasing the efficiency, how about instead of just having the hot water go back over to cool again you used two systems, one specifically for the cold, and the other for the hot. Once you got the liquid on the cool side cold enough, you could turn off the peltiers and just let the cool liquid run over the device to cool it. Once the liquid warmed up again, the peltiers turned on saving them from having to run constantly.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
PG
Thanks, Pi Guy. That little pump is certainly in my price range. Wish me luck.
Mark
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
PG
pmrobert,
would you mind telling me which PID control program you based your tests upon? I presume you used a Propeller for this, and I've noticed several versions of PIDs out there for the Propeller. One on the OBEX named "PID routines" seems to have some grouchy reviews. It's been a long time since I've had to do anything with PIDs and relearning about PIDs while struggling with gnarly code would not make me happy. So I was wondering if you (or anyone else) has a suggestion and/or code I could leach off of.
thanks,
Mark