Someone help me see if I have this right...

[Liam_G]

hahaha, thought you'd like that. and yes i noticed you like HIGH VOLTAGE, haha. my head hurts after writing that though :P

[Ultrasonic2]

hahaha, thought you'd like that. and yes i noticed you like HIGH VOLTAGE, haha. my head hurts after writing that though :P

3 reasons why 24v is better than 12v

1)
24 /3 = 8v or at 27/3 = 9
12v /2 = 6 (which is a bit low for 4 to 6 tec's total )
12/ 3 = 4 to low

2) a 24v psu is more efficient than a 12v one. the drop from 110/220AC is less to 24v than 12v so the PSU is more efficient meaning less complete watts used for the same DC wattage. so that means the system costs less to run and heats the room less

3) For the same DC wattage a system running at 24v will have 1/2 the current flowing than at 12V system. This leads to less resistive losses of all the wiring and mosfets and what not. because of this by dropping the current you make a mosfet / low pass filter more likely to work and not catch fire. as mentioned in the other thread there is a difference between theoretical electronics and reality. That mostfet could pass some crazy number of amps like 327amp. however that is not reality. it's legs ( wires ) were 1.5mm of aluminium. the resistance of 1.5mm of aluminium is FAR to great to allow a constant 327amps to flow. if you tried the legs would be vaporized.

[Ultrasonic2]

4 in series with 12V input would pull 13Amps (each TEC would get 3V/13A)


NOPE you still haven't grasped the original ohms law point that i made.

No point answering the rest since the primes is incorrect

lets assume the resistance is constant and of 0.5 ohms

1 TEC with 12v applied to at resistance of 0.5 ohms = a current flow of 24 amps so total wattage is 24*12 = 288watts

2 TEC's you now have two TEC's in series so from the PSU's point of view you have doubled the resistance to 1ohm so 12 volts applied to 1ohms = a current of 12 amps so total draw is 12*12 = 144watts

you could express the same maths like this 12 volts imput / the number of TECs 2 = 6Volts at each TEC now each TEC has a resistance of 0.5 ohms so the current = 12amps and the draw PER tec is 72watts so total system draw is twice that( cos you have 2 tec's) or 144watts



Just a random point i will put out there regarding COP.

Phase change units will always tell you their POTENTIAL COP not their actual running COP. This is because the manufactures have no idea what load you will apply to their unit. So when comparing phase to TEC cooling it is fair to compare both at potential COP ( Apples with Apples)

[Ultrasonic2]

right time for a real example of TEC's in series one 24v V TEC's in parallel at 12 volts

im my case i have 4 15.4 Umax 245Qmax tecs

4 in parallel @ 12v would result 952watts or 78 amps !!!!!!

it is my opinion that you couldn't pass 78 amps which becomes 108 amps at start up through a single loop that is practical to fit in your PC.I would certainly burst out laughing if you told me you were going to stuff it all down 1.5mm of aluminium (the mosfet)

Where as i have 4 in series at 27.5 volts = 8.5 amps


by the way Liam intends on have 2 parallel pairs of 3 TEC's in series at about 24 volts. however it would be more efficient to a have all 6 in series and apply 48 volts

[Liam_G]

quick question ultra. how much noise do meanwell PSU's make? Say the SE600-24 or something similar, that's a concern for me if they are noisier than say Seasonic Platinum PSU's. If they are pretty well built and not too noisy then great! I was starting to have doubts about squeezing one into my case.

and yes i know all the benefits and options of higher input voltage and them being more efficient, most of the Meanwell's seem to be 90+ rated. my head hurt from writing that long drawn out spiel :P

yes if running at 12v i think you would need to have a separate TEC driver/mosfet for each tec in parallel, otherwise things get a bit risky.

the 2x3 setup is one of my plans, still tossing some ideas around.

[Ultrasonic2]

well i dont have a 600watt one so can't comment but all the others i have a silent . i added a 80x15mm fan to them to keep them cool

[bds71]

2 TEC's you now have two TEC's in series so from the PSU's point of view you have doubled the resistance to 1ohm so 12 volts applied to 1ohms = a current of 12 amps so total draw is 12*12 = 144watts

yea - i figured that out yesterday, and spent a bit of time recalculating my figures. i was completely forgetting the proper usage of Mr. Ohm's Laws. i was not taking total resistance of the serial leg into account and was assuming if volts were maxed, then so was current. obviously, not the case. my numbers now are coming up with .45 to .83 CoP's now - much more reaslistic (however, much more dissapointing as well lol). i want to do a bit more with my numbers to see if any of the TECs can put out any decent numbers (still using 12v for the numbers for now, but definately open to ideas down the road).

anyway, thanks for the help once again in figuring this stuff out. as always, you two have been great - and extremely helpfull!! my son came up yesterday, so i prob won't be as active on the boards for a while. again - thanks for everything, both of you!!

[Ultrasonic2]

my pleasure

[bds71]

real quick:

in my calculations i'm seeing that CoP remains constant across parallel branches - Qmax and Pinput go up at the same rate for each added leg. is that right? (ie: 2 legs doubles power input, but also doubles Qmax)

note: 2 equal legs of course

[Ultrasonic2]

if your calculating potential COP then yes
if your calculating actual COP then no

[bds71]

i calculated at dT 20 - which gave CoP of 1 with 2 legs of 3 each for 192 Qmax and power draw. i guess you might have inefficiencies in the power supply and whatnot, so actual might be "slightly" off - is that what you mean?

[Ultrasonic2]

i calculated at dT 20 - which gave CoP of 1 with 2 legs of 3 each for 192 Qmax and power draw. i guess you might have inefficiencies in the power supply and whatnot, so actual might be "slightly" off - is that what you mean?

That wasn't what i was referring to but they are valid points to be added in

1 245 TEC at 100%umax while moving 0 = a potential COP of 0.6192893401015228
2 245 TEC's at 100%umax while moving 0 = a potential COP of 0.6192893401015228

However
1 245 TEC at 100%umax while moving 150 = an actual COP of 0.3235
2 245 TECs at 100%umax while moving 150 = an actual COP of 0.1747

However
1 245 TEC at 100%umax while moving 150 and maintaining a delta of 28.5c = an actual COP of 0.3235
2 245 TEC's at 48%umax while moving 150 and maintaining a delta of 29.2c = an actual COP of 0.7989

By the way it is NOT possible to achieve an actual COP of 1 or greater while moving 150watts of heat to a delta of 29c NO MATTER THE NUMBER OF TEC'S OR INPUT VOLTAGE
The only way to achieve a COP like that is to reduce the delta

[bds71]

that actually kind of answered another question i had: i noticed that parallel branches maintained the same CoP across branches (maintaining the same serial circuits of course)

again: thanks for all the help!! time to hang with my boy

[Ultrasonic2]

all good

[mindchill]

Just go look at the performance charts. you can get all the information you need off them

[Kayin]

All right, so I'm back. My radiator is actually a showerhead (so full of pinholes) and I'm back to liking this idea. I notice that I didn't clarify myself properly to begin with.

I have no care if the temps are only as good as the loop that was planned for it. In fact, I'd be perfectly fine with it. That's all I was aiming for. So am I completely daft, or am I closer to getting a workable solution that way. From what I could tell, getting enough oomph to cool a i5-3570K isn't impossible anyway. It's how far and what delta. Well, sure, subzero would be cool, but I'm honestly more interested in simple stable operation than I am a ridiculous temp number.

I also happen to keep a 1x120 rad on hand at all times for research work, would cooling the hot side with that for multiple small TECs be worth it or just stick with an aluminum extrusion?

[Ultrasonic2]

im not sure i understood any of that

[Kayin]

I basically stated that I didn't care if its best temps achieved weren't any better than water. I am not looking for subambient, I'm looking for a way to fit the cooling in the space I have. I have a drop-out bracket that fits a radiator in my case, but if I install one, it hits the vent motor for the case. Swapping to an idea like this would allow me to build it to fit exactly, and then I wouldn't have to worry about damaging the rest of my case. Some things are only done because I can.

I have a 750W 12V PSU I can drag out, should I look at implementing it instead?

[Ultrasonic2]

i still dont know what you are proposing

Are you trying to cool your CPU with TEC's that are air cooled cos you can't fit the radiator in Or are you trying to say you're trying to cool your cpu with TEC's and one 120 rad ?

[Kayin]

All right. I can see where I've been cryptic.

My idea was to cool the CPU with air cooled pelts. I can fab a block so water can be chilled by the pelts and I can screw a large aluminum extrusion to the other side for a sink, and focus 3 120s on it for cooling. I found a 750W 12v PSU I can mount to take the load off my 1100W if I need it. I'm just wondering if I can make a setup like that match regular watercooling on an i5-3570K. That's all I'm asking of it. Just to match watercooling. Is it doable?

[Ultrasonic2]

if you have enough Qmax and enough surface area on the Extrusion then yes. However it ends up being cheaper and performs better if you watercool the hot side too

[Kayin]

I was wondering about that. I'll look into that, watercooling it doesn't seem impossible. I could also possibly look at 1U heatsinks or something instead of extrusions for the increase in surface area.