Overclock.net banner

1 - 15 of 15 Posts

·
Registered
Joined
·
521 Posts
Discussion Starter #1
Hi all,

Can someone guide me to a simple diagram or maybe a place where I can buy a DC-DC converter / PWM booster usable for a TEC chiller?

Right now I have a 1.5KW PSU (Enermax), and 4 TECs from analog technologies (ATE1-288-20A). The Tecs are watercooled and are also cooling so my setup is a simple chiller.

1) I want to be able to adjust voltage to tecs, between 2v-20v or something, based on the 12v rail I got. I can pull 125 Amps sustained at that rail, If I am not wrong.

2) It would be nice if the DC-DC converter can handle around 1000w, which I think could be the needed with 20v, at all 4 tecs. Correct me if I am wrong there. Otherwise I might need to make converters for each Tec.

3) Would be really nice, if I could get something that can be adjusted according to temperature, so the tecs are feeded according to watertemps on the cold side.

Today, my chiller can chill the water to 4C, where the hot side is 32C (old 120.2 radiator with scythe 3000 rpm fans), and ambient around 23C. Thats using the 12v rail on all tecs. I think I can get it to around 0C, just by isolating, but I am still waiting for all the Neoprene stuff to arrive.
 

·
Registered
Joined
·
628 Posts
I don't think such a step up/down booster is available. The ones I have seen cannot handle such high amperage. Also note that all power supplies have loss, so the efficiency will not be higher than 90%, which means there will be a lot of waste heat.

Because the current is very high for each TEC, it would be better if you connect them in series so that the current on the bus is reduced. In that case you will need a higher input voltage. I would recommend to connect your 4 TECs in 2//2 pattern ( 2 in series and then parallel to the other 2), and make two boosters to power each 1+1 module. It means you need to convert 12V to 36V (maximum safe voltage for human).

A simple inductive booster will do the step-up job. The voltage can be varied from 12V and up. But it cannot lower the voltage than 12V.

Inductive-Boost DC-DC Converter.pdf 106k .pdf file

If you want the whole range of tunability (1V-36V), the circuit will be more complicated, which basically combine step-up and step down converters.

With that much said, I really think it would be better to use a dedicated PSU.
smile.gif
 

Attachments

·
Registered
Joined
·
521 Posts
Discussion Starter #3
Quote:
Originally Posted by foxrena View Post

I don't think such a step up/down booster is available. The ones I have seen cannot handle such high amperage. Also note that all power supplies have loss, so the efficiency will not be higher than 90%, which means there will be a lot of waste heat.
Because the current is very high for each TEC, it would be better if you connect them in series so that the current on the bus is reduced. In that case you will need a higher input voltage. I would recommend to connect your 4 TECs in 2//2 pattern ( 2 in series and then parallel to the other 2), and make two boosters to power each 1+1 module. It means you need to convert 12V to 36V (maximum safe voltage for human).
A simple inductive booster will do the step-up job. The voltage can be varied from 12V and up. But it cannot lower the voltage than 12V.

Inductive-Boost DC-DC Converter.pdf 106k .pdf file

If you want the whole range of tunability (1V-36V), the circuit will be more complicated, which basically combine step-up and step down converters.
With that much said, I really think it would be better to use a dedicated PSU.
smile.gif
+1

According to the 90% effeciency, I can see a major problem that I wasnt aware of. Does that rule also count for inductive booster or is it for power supplies only?

If it does count for inductive booster, I see that even if I step up the voltage from 12v to 36v and make 2 sets of tecs, where each set is running in serial so each tec will get from 6v to 18v, I will end up with like 360-400w under full load on each step up converter, + the inefficiency that I guess/hope will be a place between 80-90%. Putting in another 35-70w of heat to cool on each of the two converters. Thats a lot, and maybe too much to make sence.

I would really love to use my everyday PSU for this. I am trying to hold all parts of this inside a miditower case.
 

·
Registered
Joined
·
628 Posts
Quote:
Originally Posted by qwwwizx View Post

+1
According to the 90% effeciency, I can see a major problem that I wasnt aware of. Does that rule also count for inductive booster or is it for power supplies only?
If it does count for inductive booster, I see that even if I step up the voltage from 12v to 36v and make 2 sets of tecs, where each set is running in serial so each tec will get from 6v to 18v, I will end up with like 360-400w under full load on each step up converter, + the inefficiency that I guess/hope will be a place between 80-90%. Putting in another 35-70w of heat to cool on each of the two converters. Thats a lot, and maybe too much to make sence.
I would really love to use my everyday PSU for this. I am trying to hold all parts of this inside a miditower case.
Yes, that 90% figure applies to almost all power supplies/converters. Often the efficiency is lower.
Another problem I noticed is that your triple radiator may not be enough to cool all the heat when you bump up the voltage.
Time to upgrade to a larger case/bench
biggrin.gif
 

·
Registered
Joined
·
521 Posts
Discussion Starter #5
Quote:
Originally Posted by foxrena View Post

Yes, that 90% figure applies to almost all power supplies/converters. Often the efficiency is lower.
Another problem I noticed is that your triple radiator may not be enough to cool all the heat when you bump up the voltage.
Time to upgrade to a larger case/bench
biggrin.gif
I am actually on a dual radiator on that system, and its a dual 120mm rad. My other system/from profil pic has a triple 140mm radiator. I can imagine a TEC setup to use something more extreme, like a 9x140 or something.

450
(just to give an idea)
 

·
Registered
Joined
·
521 Posts
Discussion Starter #7
Quote:
Originally Posted by chaotic View Post

You are in my mind...I found this on ebay http://www.ebay.com/itm/DC-12V-to-24V-10A-240W-Step-Up-Power-DC-DC-Converter-/260999484459?pt=LH_DefaultDomain_0&hash=item3cc4c8142b#ht_3356wt_1392
i dont know if you can use two in series (even better three)...So what did you say foxrena?
Thank you for helping me here, but the thing is that 24V and 10A wont do it. 24V is far too much and will pull a lot more than 10 Amps. I need something stepping up to 18-20V, or double that like 36-40V, to be used for Tecs in series. Again a huge amounts of amps will run through and a lot of heat will be generated due to bad efficiency.

I am pretty sure that Foxrena is correct about his advice on getting a dedicate 36V PSU for this. Trying to convert/stepup will be too inefficient.

In my initial build I will have to run the Tecs with 12v directly from PSU. They are just cooling the water in the system, and I will later on find a way to make them work with a thermostat. They will then be ON until a target temperature is reached, adjusted on the thermostat, and then OFF until the temps are like 10C above the target temp. It wont be as hardcore cooling as I hoped, but it will work until I get my hands on some other solution/18V-36V PSU...

My next question will be, if there are any efficient way to smoothly (not in big steps) adjust 36V down to like 6V, or if that will be the same 10-20% of waste energy running through?
 

·
Registered
Joined
·
628 Posts
Quote:
Originally Posted by chaotic View Post

You are in my mind...I found this on ebay http://www.ebay.com/itm/DC-12V-to-24V-10A-240W-Step-Up-Power-DC-DC-Converter-/260999484459?pt=LH_DefaultDomain_0&hash=item3cc4c8142b#ht_3356wt_1392
i dont know if you can use two in series (even better three)...So what did you say foxrena?
Good find ! But I would be very careful to put those in series. I'm not sure what kind of circuit it uses, but possibly the DC booster link the ground of output and input together, so putting them in series will actually cause short circuit.
Use them (or any power supplies) in parallel is also a bad idea. It may appear the current output capacity is increased, but actually one will be over-loaded and the other will be minimally loaded due to the slight output voltage difference. Unless there is some power distribution circuits built in, paralleling will eventually kill both power supplies.
Quote:
Originally Posted by qwwwizx View Post

My next question will be, if there are any efficient way to smoothly (not in big steps) adjust 36V down to like 6V, or if that will be the same 10-20% of waste energy running through?
Inductive DC Step down booster will do that. There is always inefficiency. Adding heatsinks(and fans) usually solve the problem.
In your case I think two 36V 10A power supplies from Ebay will do best. Each power supply powers two TECs in series, giving 18V max to each TEC.
 

·
Registered
Joined
·
521 Posts
Discussion Starter #9
Well, I got my neoprene isolation stuff now, and temps looks fine. I havent tried out 4x 12v yet, but at 4x6v my water is around 14C, ambient 24C and CPU is idling around 16-17C. I upgraded my old 120.2 to a hwlabs gtx360. The idle temps is making me a bit nervous about condensation.

For now I have been testing it on an Intel Pentium 4 930D, and a Celeron 360 D. At 5Ghz (vcore1.65v), both of them reach temps around 45C. I will later on try to put all tecs on 12v.

450

I think the 6v and 12v are just fine. I would like a method of adjusting voltage between the values and I would also love to be able to push it to 18v (50%), but I wont get another PSU just for that.
 

·
Registered
Joined
·
628 Posts
Quote:
Originally Posted by qwwwizx View Post

Well, I got my neoprene isolation stuff now, and temps looks fine. I havent tried out 4x 12v yet, but at 4x6v my water is around 14C, ambient 24C and CPU is idling around 16-17C. I upgraded my old 120.2 to a hwlabs gtx360. The idle temps is making me a bit nervous about condensation.
For now I have been testing it on an Intel Pentium 4 930D, and a Celeron 360 D. At 5Ghz (vcore1.65v), both of them reach temps around 45C. I will later on try to put all tecs on 12v.
450
I think the 6v and 12v are just fine. I would like a method of adjusting voltage between the values and I would also love to be able to push it to 18v (50%), but I wont get another PSU just for that.
You can use these to adjust TEC
http://www.ebay.com/sch/i.html?_odkw=pwm+motor+controller+50A&_osacat=0&_trksid=p2045573.m570.l1313&_nkw=pwm+motor+controller+30A&_sacat=0
 

·
Registered
Joined
·
303 Posts

·
Registered
Joined
·
628 Posts
Quote:
Originally Posted by Mindchi|l View Post

Nah.. those don't work real well.. the pwm frequency is normally to low. I tried something like that a few years ago, it was $50 and went straight into the trash can. Save your money imo.
Why it didn't work? The circuit is very simple and functional.
I have two of those, one switching at 3KHz, the other at 12KHz. There are some switching at much higher frequencies (e.g. 30KHz).
 

·
Registered
Joined
·
521 Posts
Discussion Starter #13
Well, converting DC-DC with the overhead etc. seems to inefficient and problematic in my opinion, and I cant have another PSU inside my chassis, so what I am thinking of right now is a more simple solution.

Since I have the settings 6v and 12v to play with, I will have to live with that for now. What I am considering to do is to make a simple adjustable thermostat. The thermostat will switch a relay, adjusting 6v / 12v. I will then use the 12v to chill the cold water temp down to fx 15C, and then switch the system to 6v until the water reach the temp of fx 19C. In most normal use scenarios 6v on the 4 tecs, seems to be fine for holding the water temp below 18C.

Yesterday I measured the system power-usage:

Code:

Code:
4x12v, [email protected] 
Benching ~ 600W measured on the AC plug. 
Idle ~ 530W.  
CPU temps maxed out around 52C*

4x6v, [email protected] 
Benching ~ 300W measured on the AC plug. 
Idle ~ 230W.  
CPU temps maxed out around 57C
Thats with an idle Nvidia 7300LE (low power, passive cooled)

So I definitly need the 6v setting for everyday usage. 2kwh costs $1 USD where I live.

*Unfortunately I did a big bummer buying fans "Enermax Cluster UCCL12 PWM" for my new hwlabs GTX360 radiator, so I cant cool the hot water (hot water temps 41C). I have to shift those fans to Scythe 3000rpm or something similar. I had far better cooling performance on my old 120.2 (hot water temps 35C) with scythe fans. They really mean it when they say that the GTX360 radiator needs high speed fans with high static pressure. - Another lesson learnt

My Phobya thermometer on the cold-water died. Its on, but I the digits are gone, so I have to RMA it. Pretty annoying when I cant measure the cold water temp.

Just to learn something more, I will consider buying one of those ebay pwm motor controllers. My intuition tells me that it is fail economy - like Mindchill writes, but sometimes things just need to be tested.
 
1 - 15 of 15 Posts
Top