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

[williamcll]

Well those nocturas are the industrial verison so I don't think a little splash will hurt anyone.

[bestbattle]

@AlexTheGreatish, TECs are mostly used for transferring heat fast from one side to the other.

That rating you find on it is usually their power consumption, not transfer capabilities.

For starting, if you get a 545W peltier and want to transfer 200W of heat from an Intel CPU, you need to have, on the hot side, a cooling system of AT LEAST 545+200W (for better results take in consideration a 30% margin of error, so around 970W of cooling is a minimum).

 

[WereCatf]

7 hours ago, EiiiTrone said:

Likely you guys know by now but the solid state relay you used can only switch AC loads. It needs to be at 0V for it to switch. You can get alternative relays that would work in DC applications but for what you wanted you could have moved it to the AC side of the power supply.

This is what I was going to comment! The reason it didn't work is because they used DC-current, and neither of them realized it.

 

_{*hisses under her breath* damn amateurs...}

[Regrets]

TEC can work. Its just that they have to be used correctly. Hammering screws in will work but not well.

I am an engineer that designs some very pricey science equipment cooled (and sometimes heated) by TECs. They are a pain in the ass and require a huge amount of tuning. But that's part of why the shit my team builds sells for $30k to $1m per.

i think you should next try cooling with a vortex cooler. But first, give us a tour of this new shop...

 

[AnonymousGuy]

It's pretty well known that TEC cooling never-ever works on a CPU directly.  You can use TEC's to cool down a reservoir of water and then watercool with chilled water, but it's more efficient to buy a water chiller than use 4000W of TECs to get 500W of cooling.  

[greenhorn]

Most of the comments I have have already been made in this thread.

I would also, as some suggested, use 2 water loops and use the Peltier to cool water from the CPU loop and have itself cooled by the second loop. I don't know the exact specs of this Peltier, but I am sure the element itself generates a lot of waste heat at that power, which is something that is not optimal to have that close to the CPU.
As for the PID controller, autotune made me cringe really hard. "blind tuning" (blind because the tuning method does not know anything about the underlying physics) seems quite risky to me and we were throoughly warned at university not to use anything of that kind.
Speaking of the PID controller, does it automatically reset the I part (integrator) for every new run? If not, this can become a potent source of stupid commands to the actuator.

[RasmusDC]

first of all, it is really nice to see you guys do something, that is actually interesting, even though failing..

 

this was a really great video, although i cringed on the safety aspect.. you would have been shut down as a work place in Denmark, with all the open electronics.

 

Peltier can work, we used them as part of the Compressor kits in the older days, when overclocking, that was in the early asetek days, where we regassed compressors, and used peltiers for heat transfer, into compressor cooling.

 

yes it was insane, but fun, actually i know we had CAD machines at B&O where i worked as an engineer that was asetek compressor cooled. 

 

remeber the old celerons 300mhz that just overclocked to insane amounts, on negative cooling.

 

But play a bit more with Alex, it is fun to watch, way more, than most other content that you have sought against.

 

you have the machines now, how about building a case, where the outer shell is like the passive older cases, where headpipes contacted the sides, but instead make it aluminium and make it a large watercooled heatsink, and then just a pump, no fans, and see how silent, and cold you actually can make such a system.

[Nup]

Great video guys. 

[TechyBen]

16 hours ago, samcool55 said:

@AlexTheGreatish Why didn't you just go the intel way?

Massive water chiller and off you go!

 

Also I still belive that a TEC cooler can work, but not like this.

What you can do is get 2 loops.

1 loop is connected to the cold side of multiple lower-powered TEC's (more info below) so you don't just have a cold baseplate but an actual cold sub-ambient loop with water that needs anti-freeze because you could easily pull the whole loop into negative C. And another loop that is connected to the hot side with that massive rad (maybe even more) to keep the hot side happy.

So, why get more TEC's? Because you were very likely hitting a thermal transfer limit with your current setup. If you use multiple TEC's that are lower power you can spread out the heat so it can be transfered easier from the TEC to the loops.

 

TL:DR do it again but instead of 1 TEC, use more of them but lower power and instead of whacking a baseplate on it and mounting it straight onto the CPU, make a loop of it.

Yeah, but you need 1000w or so of cooling for the Peltiers, then the Peltiers need to do about 500w cooling too. They cool a loop (as previously tried and Der8aur tried) and then you use a second loop to cool the Peltiers...

 

But sadly, you end up just under ambient temperatures, at like 1000s of watts of power use. An air conditioner is more efficient and workable.

 

Peltiers could work for keeping idle temps below ambient, and thus allow some boosting for small workload low latency work (say stock trading).

[Edit]

This could be used to cool a loop to sub ambient for boost benchmarking (-20 water, doing 5 min benches only). But all in all, the cost of LN2 is probably cheaper! XD

[samcool55]

17 minutes ago, TechyBen said:

Yeah, but you need 1000w or so of cooling for the Peltiers, then the Peltiers need to do about 500w cooling too. They cool a loop (as previously tried and Der8aur tried) and then you use a second loop to cool the Peltiers...

 

But sadly, you end up just under ambient temperatures, at like 1000s of watts of power use. An air conditioner is more efficient and workable.

 

Peltiers could work for keeping idle temps below ambient, and thus allow some boosting for small workload low latency work (say stock trading).

Well yea I know power usage will be absolutely stupid and insanely inefficient compared to a water chiller for an aquarium but hey, it's not completely impossible.

 

I was thinking of using 5-10 Peltiers with each using 300 watts so my guess for this working was between 1.5kw and 3kw. Yes that's absolutely insane I know. But my point is that it CAN work. But it's insanely inefficient and just a bad idea. But not impossible!

[The_Prycer]

I love these mad scientist episodes. 

 

I still want to see you use a 9590FX for zany cooling solutions because that thing is an absolute space heater.

[TechyBen]

1 hour ago, samcool55 said:

Well yea I know power usage will be absolutely stupid and insanely inefficient compared to a water chiller for an aquarium but hey, it's not completely impossible.

 

I was thinking of using 5-10 Peltiers with each using 300 watts so my guess for this working was between 1.5kw and 3kw. Yes that's absolutely insane I know. But my point is that it CAN work. But it's insanely inefficient and just a bad idea. But not impossible!

Or use their existing 300w one to cool a Raspberry Pi.

[SolarNova]

15 hours ago, AlexTheGreatish said:

SNIP~~~

in the future will just use a refrigerant based system for chilling.

Well if ur going to do that next I would suggest, and ask, that you do one along the lines of BUILDING the chiller from scratch, not re-purposing an existing one.

 

So acquire a compressor, condenser, a plate heat exchange, the copper tubing and capillary tubing, etc etc. And maybe build it designed for Propene (r1270) or Propane(r290). Go through the steps technicalities of building one.

That, with a sufficiently large res (like a repurposed  ~20 liter bear cooler/ picnic cooler /cooler box, which are pre-insulated) and coolant like Mayhems XT-1 in the correct mixture, would allow one to go subzero with a liquid loop.

 

However ull need to ensure your blocks are NOT plastic, and that the o-rings used within ur PC loop parts can withstand the temps that u could theoretically get (~ -40c)

 

I'd very much like to see this. I've seen water loops with rads immersed in ice water, but never seen a extensive video on a proper subzero liquid loop. It is possible however for sure.

[djmartrix]

not sure if its been said or suggested, but would it be feasible to jerryrig the tech cooler into the radiator to cool the liquid instead of directly on the CPU? you could save the headache of avoiding condensation (well sort of), it would end up being more like the fridge cooler from earlier this year or late last year.

[AlexTheGreatish]

10 hours ago, WereCatf said:

This is what I was going to comment! The reason it didn't work is because they used DC-current, and neither of them realized it.

 

_{*hisses under her breath* damn amateurs...}

We were using the SSR to switch the PSU on and off, so it was using AC

[RustyJames]

3 hours ago, AlexTheGreatish said:

We were using the SSR to switch the PSU on and off, so it was using AC

The whole point of PID is moot if you do on-off control (no Proportional nor Integral or differential control done)

Also, the PSU probably did not respond well to on/off switching

The TEC unit should be connected to a PWM type switch that can throttle the DC current using the PID as the driver

 

Would be cool if you tested the custom machined block as a water block (without the TEC, with/without the base heat spreader)

[pruegeL]

I think it's a stupid idea because it's too loud and complicated to be really useful but have you ever thought about using a vortex tube  and compressed air to cool a cpu?

Greetings from Germany

Andy

 

 

[Dav1d]

@samcool55 @AlexTheGreatish

If you want to actually do this wouldnt it be a lot smarter to do it the way i drew it i think thats also what sam meant in his message.

this way the radiator will cool the water somewhat and then the peltier will further cool the water (+ anti freeze maybe) to the max of its ability while not getting overwhelmed.

would love to see a video about this.

 

[alextulu]

On 9/15/2019 at 10:08 PM, Tamesh16 said:

Temperature gradient,

the temperature on the cpu die and the cold end of the TEC plate are not the same, so the thermocouple  would have a temperature in between those 

OK, but the CPU was 100 degrees, while the TEC was about 50. Why is the difference so big?

[Dav1d]

@alextulu

you have the chip (cpu temp) then thermal-paste , heat spreader, thermal-paste, copper block and then the other temp sensor touching the cold side of the peltier so there is a lot of material between the two temp sensors and one is right at  the heat source while the other is at a source of cold so youd expect some difference

[AlexTheGreatish]

9 hours ago, Dav1d said:

~snip~

We did this in the first video.  Basically unless you have thousands of watts worth of peltiers they will easily be overwhelmed.  Using a refrigerant based chiller is a much better idea (like this video).

[Bitter]

So that's what we'll be seeing next. Lol as well as Linus garnishing Alex's paycheck for the added cost of electrical bill.

 

 

Actually a bit off topic I'd enjoy a joint project between Alex and Anthony. Alex does the hardware and Anthony does the software. Maybe something for the machine shop?

 

[Domrockt]

Watercooling as it is wont be efficient enough.

My 9900k blasts away under full load.

 

When i cool my water with my chiller to 10°c and activate prime95 max load/of .. all Cores jump to 80°c and more .. so my CPU Block has a Delta T from core to water of 70°c ... Just Imagine THAT load plus the TEC... The concentratet heat of an CPU is just to much. 

 

There could be an improvement with an big ass waterblock Like the ones used for ln2. The surface area for transfering Heat to the water is waaaaaaay to low.

[Bitter]

1 hour ago, Domrockt said:

Watercooling as it is wont be efficient enough.

My 9900k blasts away under full load.

 

When i cool my water with my chiller to 10°c and activate prime95 max load/of .. all Cores jump to 80°c and more .. so my CPU Block has a Delta T from core to water of 70°c ... Just Imagine THAT load plus the TEC... The concentratet heat of an CPU is just to much. 

 

There could be an improvement with an big ass waterblock Like the ones used for ln2. The surface area for transfering Heat to the water is waaaaaaay to low.

Similar situation for me and my water cooled ye olde Xeon X3470. When I crank it up to 4Ghz or above the amount of heat it outputs is crazy, 350+W is no problem for that space heater and I do get that same instant jump in temps and will hit 80-90C on all 4 cores despite the water block being a very big lump of gold clad copper with a ton of surface area across it's WHOLE surface that water flows over back and forth in a S pattern. I still just can't get the heat out fast enough. I'm going to add a bunch more radiator because why not (and I have an old dehumidifier to hack up so it's free real estate) but I don't think it'll do anything for it lol. Just too much heat in a concentrated spot, I guess...still won't deter me from screwing around more!

[Sleeper0013]

The only way Thermo-electrical is going to work is if you use it as a stage in a normal loop. the problem is you are dealing with to many joules per second, the peltier is a huge bottle neck for transferring that kind of heat. for direct cpu cooling you would need a surface area probably the size of a dinner plate if not larger. But if you put a peltier block in the loop as a stage before the CPU block you will get a reduction in temps. this will probably not be supper efficient but im willing to bet that you will get a drop in CPU temp from a standard loop. 

 

You go from the CPU to the radiator, then you go from to the radiator to the pump, from the Pump to a Peltier cooled channel block then back into the CPU Block. So you have the CPU block and radiator doing their normal job, and the peltier block cooling the water below ambient temps before being injected into the CPU block. 

 

Another problem with video is that copper transfers heat well but it doesn't dissipate heat so well, its too dense. So cooling copper is going to have more resistance than heating. Aluminum on the other hand will transfer cold quickly, and water will cool exceeding fast. 

 

For your peltier block you use two peltiers sandwiching an aluminum plate with channels drilled parallel to the peltiers, (lots of small channels a few mm from the cooling plate). You cap the ends with etched acrylic to channel water to the drilled pipes, tap your fittings into the Ends of the acrylic. after that you use regular heat sink and fans clamped onto the hot sides of the peltier.