1. ## Can Cooler 4000

With all the people that run through my cave bar every week you would think one of them would put another case of soda in the refrigerator when they take the last cold can. But no, invariably, I am stuck with a warm can of diet coke. . . Kids!

So, just for fun, we are working on an all electric machine that can very quickly cool a beer or coke beverage right in the can. The
"Can Cooler 4000" will, we hope, be able to take a room temperature canned beverage to a very cold temperature in just a few minutes. It will consume up to 4000 watts of power, stand just over 4 feet tall, and weigh close to 300lbs with 8 peltier cooling chips and 8 500W power supplies. Here is a video of a peltier chip in action.

Stay tuned and stay thirsty my friends.

Cheers
Dave

2. ## Re: Can Cooler 4000

I hope you'll reconsider the use of peltiers.

Cooling 500ml of water from 23C to 1C requires 46k joules of cooling or about 13 watthours of energy to be moved. Doing so in 3 minutes means moving 260 watts of heat from the can for those 3 minutes. Peltiers are only maybe 5% efficient in cooling so it'll require pumping at least 1300 watts into the Peltier. And of course providing a huge amount of heat sinks and cooling and fans to dissipate that 1300 watts. Of course you've already done this math.

A conventional compressor based cooler is more like 300% efficient. If you are scratching your head at how "over 100%" is possible: refrigeration isn't about making heat or making cold. It is about moving heat. A gasoline tank trunk can move thousands of gallons of fuel while only consuming a few gallons itself. Air conditioners and refrigerators and heat pumps all work by this principle. Side note: If you use electric base board heating and replace that with a heat pump you can expect your heating bills to go down by like 70%.

Back to the beer cooler. In practice it gets worse because the peltier is only 5% efficient when both sides of the plate are at the same temperature. Efficiency drops to 0% as the temperature difference increases. As the can cools the Peltier can't cool as efficiently and so you need to pump even more power into it.

A conventional cooler could move those 260 watts with under 100W of power. Well within the range of a cheap mini fridge. The main difference is that the fridge's cooling coils would be spiral wrapped around a tube that the can drops into. Much cheaper to build and to operate. Dramatically quieter and won't pump out as much heat as a hair drier. Possibly faster since it can get colder than the peltier.

I think the planned can cooler design will solve the problem just because people are going to be so terrified of the giant box that sounds like an angry jet engine that they'll keep the fridge stocked to avoid having to use the machine.

3. ## Re: Can Cooler 4000

Hell, you'll have to be careful not to freeze the soda on the outside edge of the can with a small refrigerant based device, a TEC capable of that would be ridiculously large and obnoxious, let alone inefficient.

4. ## Re: Can Cooler 4000

paramettrek, FRITZHID,

Thanks for your input, I appreciate your perspective and as for efficiency, I agree with you. At first glance, your math looks correct to me too. I was hoping there would be some interest in this crazy project and it sounds like you guys have some experience with this stuff.

I know enough about heat pumps and the refrigeration cycle to get myself into trouble, LOL. I just recharged my 5 ton cave unit yesterday because it was the first warm enough day here for me to use the subcooling method. I installed that unit myself a few years ago and I now know it has a small leak. My equipment is simple, just a set of analog yellow jacket gauges and a temp probe. Do either of you do HVAC work?

The purpose of this Can Cooler build project is to make a fun toy, so to speak, it is not practical in any way. It will end up setting in my bar next to my refrigerator. The goal is to see just how fast I can cool a can with peltier chips, just one can in one shot.

BTW, my rig will not make any noise except for the switching power supply fans and they are almost silent setting inside the base in my tests. I will post a pic of the base ASAP. And yes it is large and obnoxious.

Cheers
Dave

5. ## Re: Can Cooler 4000

This is a rendered computer model image of what we think the final project will look like. Also below that is, a photo of the base with me holding onto it.

Cheers
Dave

6. ## Re: Can Cooler 4000

My experience with Peltier devices used in coolers is they typically have a Coefficient of Performance of about .1 in a complete system, in other words for each calorie of heat transferred, 10 calories of energy will be required/consumed. Cooling a 300ml can of a water based liquid from 25 C to 5 C will then require the transfer of 6000 calories, and doing so will require 60,000 calories or roughly 200,000 joules! Doing so in 3 minutes doesn't alter the energy requirements, but merely says the devices must consume about 65,000 joules per minute or 1.1 kilowatts of power for 3 minutes. This is an absurd amount of heat to be dissipated . A mechanical refrigerator typically has a Coefficient of Performance of about 3. I gave up on Peltier refrigerators and bought an small Engel refrigerator. Because Peltier Devices have great difficulty pumping energy across a temperature gradient of more than about 20 C, they cannot make ice in a room at 25 C. A 20 liter Engel refrigerator can actually make ice easily and can easily achieve a 45 C temperature differential.

7. ## Re: Can Cooler 4000

Originally Posted by mattheww50
My experience with Peltier devices used in coolers is they typically have a Coefficient of Performance of about .1 in a complete system, in other words for each calorie of heat transferred, 10 calories of energy will be required/consumed. Cooling a 300ml can of a water based liquid from 25 C to 5 C will then require the transfer of 6000 calories, and doing so will require 60,000 calories or roughly 200,000 joules! Doing so in 3 minutes doesn't alter the energy requirements, but merely says the devices must consume about 65,000 joules per minute or 1.1 kilowatts of power for 3 minutes. This is an absurd amount of heat to be dissipated . A mechanical refrigerator typically has a Coefficient of Performance of about 3. I gave up on Peltier refrigerators and bought an small Engel refrigerator. Because Peltier Devices have great difficulty pumping energy across a temperature gradient of more than about 20 C, they cannot make ice in a room at 25 C. A 20 liter Engel refrigerator can actually make ice easily and can easily achieve a 45 C temperature differential.

I agree, it will not work as well if it is out in the sun on a hot day.

The starting conditions that I care about for my device are the temperatures typical of my cave. That is usually about 67 degrees F ( 20 deg C ). There are lots of details surrounding the thermodynamics of the temperature difference between the hot and cold side of the Peltier chip, the thermal conductivity of the hole system including the water/ice and lastly but not least the specific heat of the hot side.

As you will see very shortly, my machine does not need to dissipate this "absurd" amount of power during the 4 minute cooling run. I will post a video of the cooling engine assembly later today so you all can see how the chips are mounted.

Cheers
Dave

8. ## Re: Can Cooler 4000

Originally Posted by Data
BTW, my rig will not make any noise except for the switching power supply fans .... As you will see very shortly, my machine does not need to dissipate this "absurd" amount of power during the 4 minute cooling run.
You're skipping all of the active cooling and using a thermal sink? So 30k joules from the can and another 300k joules from the peltier. Those 330k joules if pumped into 10 kg of aluminum would raise the temperature of the sink by 29C. Since the peltier can't handle a gradient that large it would end up heating the can by around 10C.

9. ## Re: Can Cooler 4000

Can Cooler 4000 engine parts video and build video condensed to 3 minutes.

Cheers
Dave

10. ## Re: Can Cooler 4000

Originally Posted by parametrek
You're skipping all of the active cooling and using a thermal sink? So 30k joules from the can and another 300k joules from the peltier. Those 330k joules if pumped into 10 kg of aluminum would raise the temperature of the sink by 29C. Since the peltier can't handle a gradient that large it would end up heating the can by around 10C.

You are zeroing in, but the sink is almost 60Kg of copper.

11. ## Re: Can Cooler 4000

The eight power supplies mounted in a rail that slides into the base and all the fan holes, displays and control knob holes line up.

12. ## Re: Can Cooler 4000

Originally Posted by Data
the sink is almost 60Kg of copper.
Copper has a pretty terrible specific heat though. Aluminum is more than twice as good. 330kJ would still raise that sink by 14C and that eats most of the thermal gradient. So the can will probably be all of 6C cooler.

And it will only be able to do that trick once. The 2nd can through would heat up unless the machine was given a while to cool off. This is why everyone thought there would be fans. You need active cooling to be able to cool more than one can. Though next you're probably going to say that the copper sinks are removable and replaceable.

edit: Your plates could use some more work. There really shouldn't be any visible machining on the junctions and that much thermal paste should be unnecessary. I'm also surprised you didn't opt for the latest and greatest carbon nanotube thermal material.

very late edit: The big noisy DC power supplies could be eliminated. A \$1 bridge rectifier could feed the peltiers from the wall directly if they are put into a series/parallel combination totaling 170V. This will require 30% more peltiers for the same cooling but that is probably a worthwhile tradeoff given their relative prices.

13. ## Re: Can Cooler 4000

Seems like a shame to do stuff this way, since the power supplies still have moving parts, and this thing will make so much heat I'd need another drink to cool off. And another shame to use metal as a heat reservoir instead of something like water. An absorption cycle would be cool.

14. ## Re: Can Cooler 4000

Well, there are those here who are amazed and interested in this project. Bold and offbeat. Thanks for sharing it and hoping to learn more. Thanks!

15. ## Re: Can Cooler 4000

Nothing less than an incredible Coolfall creation! Congrats for the hard development and machining work Dave. I, as one of many, am waiting for the utilization of the Coolfall Can Cooler 4000! I'll take one!!

Karl

16. ## Re: Can Cooler 4000

Thanks for the kind words guys.

Here is a pic of the saw cutting 8" aluminum for the wheel hub.

Cheers
Dave

17. ## Re: Can Cooler 4000

Originally Posted by parametrek
Copper has a pretty terrible specific heat though. Aluminum is more than twice as good. 330kJ would still raise that sink by 14C and that eats most of the thermal gradient. So the can will probably be all of 6C cooler.

And it will only be able to do that trick once. The 2nd can through would heat up unless the machine was given a while to cool off. This is why everyone thought there would be fans. You need active cooling to be able to cool more than one can. Though next you're probably going to say that the copper sinks are removable and replaceable.

edit: Your plates could use some more work. There really shouldn't be any visible machining on the junctions and that much thermal paste should be unnecessary. I'm also surprised you didn't opt for the latest and greatest carbon nanotube thermal material.

very late edit: The big noisy DC power supplies could be eliminated. A \$1 bridge rectifier could feed the peltiers from the wall directly if they are put into a series/parallel combination totaling 170V. This will require 30% more peltiers for the same cooling but that is probably a worthwhile tradeoff given their relative prices.
The specific heat of copper is a function of its mass. I wanted the unit to be as compact as possible. For the same size of block (2" x 4" x 12") the copper can hold more heat than aluminum. But that is only half the story as copper has a very high conductivity too. As a heat sink for high end CPU coolers and flashlight emitters, it is the king. I use it as a heat sink on both my emitter and my Texas Instruments STFu power IC in my SPY flashlights.

It is a one trick pony. It will take hours to cool off before it can be run again. This is part of my design and it is necessary for it to be quiet and compact. This is not the kind of toy you would want to run several times when there is a fridge setting next to it.

The cold plates show machining marks but they have a very smooth flat surface. It just looks that way because that is how an end-mill marks a part when it is end-on machining. I did not measure the surface finish but when you have been CNC machining as long as I have, you know from experience what feed, speed and step-over is needed to get a particular finish. You are right, I later took most of that silver paste back off of it. The thermal conductivity is very good between the two with the silver. Where do you purchase a small quantity of nanotube material? Arctic Silver does not have it.

The power supplies are big but they make almost no noise. Keep in mind, I picked power supplies with variable speed automatic fans and with the short runs, the fans never ramp up. They stay so low that you really can not hear much at all. A bridge rectifier would not give me the ability to tune the output of the system. These power supplies are variable output voltage with voltage displays that show through the windows on the base. The voltage*amp power through the Peltier chips is critical to maximize the cooling of the system. If it is set too high it will actually have an adverse effect on cooling.

I need to dig in the attic and find my first Peltier can cooler, I built it back in the 80's or 90's, I do not remember exactly when. It kind of worked but it was very low power and only intended to keep a can cool. I remember having it in my 75 Impala!

Cheers
Dave

18. ## Re: Can Cooler 4000

Originally Posted by spaceminions
Seems like a shame to do stuff this way, since the power supplies still have moving parts, and this thing will make so much heat I'd need another drink to cool off. And another shame to use metal as a heat reservoir instead of something like water. An absorption cycle would be cool.
Water is very good at storing heat but it is not good for thermal conductivity. You have to mix it or agitate it during the process. This is actually part of the Can Cooler 4000 design. Hang in there, more details are coming.

Cheers
Dave

19. ## Re: Can Cooler 4000

With CPU coolers, where there's also a demand for low thermal delta, heatsink faces are often polished to a near-mirror finish with as flat a surface as possible, so that even on a microscopic level the amount of space the thermal paste occupies is minimized. Then some people use gallium based liquid metals, though of course that's only with some metals. Visible markings may feel smooth but everything adds up, esp with multiple interfaces in a row. It might be a way to gain a little performance, maybe not. I think there's a small 40mx40mm thermal pad that's carbon based; IC graphite or something. I'd just use the paste.

20. ## Re: Can Cooler 4000

Sorry I misspoke about the details. It is a graphite cooling pad that was only released in April. Here are the specs on it:

There have been several reviews that confirm it does work better than thermal paste. A 3cm x 3cm chunk is \$15 on Amazon. Expensive but it is zero mess and reusable.

I was talking about this with some buddies who are wiser than me and they pointed out 2 things that I wasn't aware of. The fastest COTS way to cool a can (1 minute) goes to the \$25 "Chill-O-Matic." It required being loaded with ice and agitates the can for faster cooling. The least expensive and quietest would be to keep a bucket of salt water in the freezer. Plonk the can in the bucket and it will cool extremely quickly. Have to be careful because leaving it in too long might freeze the can.

21. ## Re: Can Cooler 4000

Originally Posted by parametrek
Sorry I misspoke about the details. It is a graphite cooling pad that was only released in April. Here are the specs on it:

There have been several reviews that confirm it does work better than thermal paste. A 3cm x 3cm chunk is \$15 on Amazon. Expensive but it is zero mess and reusable.

I was talking about this with some buddies who are wiser than me and they pointed out 2 things that I wasn't aware of. The fastest COTS way to cool a can (1 minute) goes to the \$25 "Chill-O-Matic." It required being loaded with ice and agitates the can for faster cooling. The least expensive and quietest would be to keep a bucket of salt water in the freezer. Plonk the can in the bucket and it will cool extremely quickly. Have to be careful because leaving it in too long might freeze the can.

Yea, I have see the chill-o-matic before too. Cool toy.

The bucket is faster if you can agitate the can some way.

22. ## Re: Can Cooler 4000

Originally Posted by parametrek
Sorry I misspoke about the details. It is a graphite cooling pad that was only released in April. Here are the specs on it:

. . .

If the pad is 1mm thick and the Arctic Silver compound fills a .003" gap then the pad has to be 13 times better to break even. I think Arctic silver 5 is between 0.4 and 8 W/mK depending on its age.

Cheers
Dave

23. ## Re: Can Cooler 4000

It more than breaks even. So far every independent review has shown a measurable improvement in CPU temperatures under load.

24. ## Re: Can Cooler 4000

Originally Posted by parametrek
It more than breaks even. So far every independent review has shown a measurable improvement in CPU temperatures under load.
I have looked and I cannot find any reviews comparing to a hi quality paste like Arctic Silver 5 where the reviewer has any idea what they are talking about. Most of these people are confused by the simplest thermodynamics science like comparing thermal conductivity and not understanding it has to do with the thickness of the intersection. It looks like the entire subject is driven by the manufacturers trying to make a buck or user reviewers looking for YouTube likes. This is a very slippery slope when you look close.

https://www.tomshardware.com/reviews...ison,5108.html

25. ## Re: Can Cooler 4000

There's LTT; what did you think of their video? https://www.youtube.com/watch?v=YpphKzmDiJM

26. ## Re: Can Cooler 4000

Originally Posted by spaceminions
There's LTT; what did you think of their video? https://www.youtube.com/watch?v=YpphKzmDiJM
Thanks for popping that video up here, it was great.

I like those guys and have watched countless of their videos over the years because I am a PC builder gear head all the way back to my first PC, a 286. They do not understand the science of thermal conductivity either, as per his comment at 3:37, LOL. What he left out is the square meter of surface area that the wattage is pouring through. But they sure know how to build and over clock. Some of the comments on the video trash this test as too unscientific. I do not agree, I think this is a fair test. However having said that, everybody needs to keep in mind that this is a sample of one. There are many factors that they are not considering like simply switching A and B and seeing if the results switch, and yes I know they tested on this rig before. Also, and even bigger oversight is the specific spreading application of the diamond paste. Then finally, can the test be repeated by another lab? The devil is in the details.

My take on it is the graphite pad is amazing given its thickness. Clearly it is a competing technology.

Cheers
Dave

27. ## Re: Can Cooler 4000

The mecanum wheels have 12 rollers each. Below is how we are making the hub of the wheel.

There are six holes and three of them are used for bolting to the fixture below. This way we can index the hub as we machine it by taking out the bolts and rotating it 60 degrees each time. There are dowel pins in the other three holes to keep it perfectly aligned.

Here the fixture is tilted 45 degrees in the vise so the locations for the roller axles can be machined.

28. ## Re: Can Cooler 4000

Crazy! Awesome!

29. ## Re: Can Cooler 4000

Cool stuff Dave, hope to see a video of this thing in action working

30. ## Re: Can Cooler 4000

Originally Posted by Data
Thanks for popping that video up here, it was great.

I like those guys and have watched countless of their videos over the years because I am a PC builder gear head all the way back to my first PC, a 286. They do not understand the science of thermal conductivity either, as per his comment at 3:37, LOL. What he left out is the square meter of surface area that the wattage is pouring through. But they sure know how to build and over clock. Some of the comments on the video trash this test as too unscientific. I do not agree, I think this is a fair test. However having said that, everybody needs to keep in mind that this is a sample of one. There are many factors that they are not considering like simply switching A and B and seeing if the results switch, and yes I know they tested on this rig before. Also, and even bigger oversight is the specific spreading application of the diamond paste. Then finally, can the test be repeated by another lab? The devil is in the details.

My take on it is the graphite pad is amazing given its thickness. Clearly it is a competing technology.

Cheers
Dave
Yeah, I've seen plenty of times before that they didn't have any sense about some things; once they did a huge cooling loop but didn't do very good plumbing and didn't realize they had to do something to keep stuff from growing in the coolant they used. And they have been painfully unfamiliar with all sorts of things before.

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts
•