This Cooler DRAWS 545W!!? Bad Cooling Ideas #2



Linus Tech Tips makes entertaining videos about technology, including tech reviews, showcases and other content.

35 Comments on “This Cooler DRAWS 545W!!? Bad Cooling Ideas #2”

  1. the more power you put in the more heat the water cooler has to remove. If the hot side gets significantly hotter than the cold the efficiency drops like crazy. TECs are better for constant gentle cooling like mini fridges, but are still the least efficient cooling technology.

  2. Doing the peltier control like that is a bit idiotic… you know there are thermally adaptive resistors… to analog control the input voltage for peltier and doing so cool it less or more if needed. Or you could do it with logic aswell but bit expensive for use of that nature.

    Edit: But Peltier device will allways have more heating on hot side than cooling on cold side.. just the nature of the beast… so as an cooler like that.. it will never work… would be more effective if attached to radiator and vent the hot air outside… if you need extreme cooling… need anti freeze on cooling liquid tho… + your pc will act as aprox 2000w heater at that point.. so well come back to fx 8350 running at 4.9 GHz …

  3. How about cooling the water in a water loop i feel that would stabilize the temperature would make it easier for the Peltier module to cool the cpu and probably would need a lower amount of power

  4. Have you ever tried using ammonia instead of water, it has better thermal properties. However it is poisonous, but as long as the system is closed (like in a industrial freezer) it is good. And instead of a peltier cooler, you use a vacum pump. That would get the temperature down to around -200C degrees if you desire.

  5. What if you use the TEC to cool the water in a water loop. You know water coolers work. you know a TEC can cool stuff down to real cool. The issue I think you are hitting is thermal transfer between the block. So can the TEC cool the water in the loop. So you can have a smaller radiator, on the water CPU side. Large radiator on the hot side of the TEC.

  6. so the thermocouple on the TEC is reading 43C whilst the pc is in the 80s and throttling? this seem more like a problem with your home built components than the TEC being over whelmed. very much an episode of sketchy heatsink

  7. Ran a TEC way back in 2004/5… still have the thing kicking around somewhere. Had a 350W secondary power supply just for the cooler. Issue was I was running it in a mid-tower with only a dual 80mm rad, so if I left it running for 2-3 days the heat buildup inside the case would cause the hoses to loosen and start to pull back off the barbs and coolant would leak into the case… many worm drive clamps and much silicone sealant was spent trying to resolve that issue. I could get the thing to run sub-zero temps pretty regularly, but not for more than a couple of hours at a time. Also had to conformally coat the back of my motherboard and use di-electric grease in my CPU socket to prevent condensation from forming.

    In hindsight, a full tower case with better airflow and a much bigger rad probably would have solved it. Just went back to aircooling after that, until very lately getting into AIO watercooling (for the RGB, of course!).

  8. TECs can work, but not like this. First you need a proper TEC controller like this from Oven industries https://www.ovenind.com/pdf/datasheets/DS-5R7-388.pdf , then you need to provide a ‘buffer’ between the cold side of the TEC and the device your trying to cool. The best option is have the cold side of the TEC chilling a cold water loop with thick tubing like the EK 16mm OD 10mm ID black hose to help with insulation. Swiftech used to make a great off the shelf TEC water chiller MCW-chill-452 many years ago. You never know they may still a few kicking around the warehouse. You probably want a couple of these to really get the effect your after so it’s going to get expensive on TEC controllers.

  9. I have a retarded Idea about this tech cooling thingi. Soo I was thinking as these peltier devices cannot cool a cpu which produces alot of heat. So what we can try is not waste 500 watts on cooling this way, but do it the way that you did before, but this time introduce a radiator before letting the water go into the perltier device cooling modue. This will help most of the heat from the water to escape and then these peltier devices might be able to cool the liquid more efficiently. and as far as I know if it is a 360mm rad it might take only 3 watts for the 3 fans. Soo not alot of more wattage will be used in this way, and I do believe that this might make the whole concept of peltier cooling worth it. But the temp drop might just be 2 to 3 degrees. SMH

  10. I've done peltier in various projects. I've noticed that peltiers drops in performance a lot over time. within 3-4 months of use, they where less than 50% of their initial performance, and after 6 months they've lost over 80%

  11. To be clear, TECs are not useless. The usefulness of a TEC is not its ability to remove large thermal loads (in fact, all a TEC does is MOVE heat [you still need a radiator to remove the heat], and the TEC GENERATES heat during this process).

    The usefulness of a TEC is in its superior control; which makes it especially well suited for use in temperature stabilization (servoing) systems. The TEC's superior control comes from two important properties of TECs: Its ability to heat or cool depending on the direction that current flows through the Peltier; And, the fact that you can control the rate of heat transfer by adjusting the current flowing through the Peltier.

    The application explored in this video did not take advantage the unique properties of TECs. Moreover, the PID servo used in this video did not take advantage of the superior control offered by TECs. Rather than adjusting the current flowing through the Peltier, the PID used in this video ran the TEC at full current or zero current. Cheap PID servos (like the one used in this video), run on a duty cycle loop. Basically every cycle (usually several seconds), the PID circuit determines the fraction of this cycle time the TEC should be running, and the fraction of this time the TEC should be off. The servo then uses a relay to turn the TEC off or on. While duty-cycle PID loops are cheap, and offer sufficient temperature stability for most applications, they do suffer from temperature ripples (caused by the TEC being either fully on, or fully off).

    A more sophisticated PID circuit would servo the current (both magnitude and direction) running through the Peltier. This removes the temperature ripples, and allows for a much better temperature stabilization. In my line of work (experimental physics), we use these types of temperature servos to stabilize temperatures to within a few milliKelvin.

  12. I bet that Linus used too powerful of a tec and the tec overheated due to not enough cooling with the water loop. Try a much less powerful tec that your loop can cool that won't overheat.

  13. You clearly did something wrong because I have personally used Peltier cooling with a large air cooler and it gave excellent temps under load under 10c and perfectly stable. And I didn't have all that crap you were using, just a 12v peltier plugged directly into a standard PC power supply with a closed cell foam condensation barrier around the Peltier.

Have a comment? Type it below!