Author Topic: DC load using a CPU cooler  (Read 57242 times)

0 Members and 1 Guest are viewing this topic.

Offline Jeroen3

  • Super Contributor
  • ***
  • Posts: 4209
  • Country: nl
  • Embedded Engineer
    • jeroen3.nl
Re: DC load using a CPU cooler
« Reply #125 on: October 04, 2014, 10:31:39 am »
@Jeroen3, unless I'm wrongly estimating the size, that heatsink won't get close to 300W.
Correct, 300 Watt is an enormous amount of energy. I'm not designing for that kind of currents or voltages.
For typical use, a 35-40 Volt OR (not and) 10 Amps should be enough. If more is required, I'd need to buy a better heatsink. Which will be heavier, bigger and more expensive.
The specification received from the OEM say: 0.25 °C /W @ 3000 RPM.
 

Offline microbugTopic starter

  • Frequent Contributor
  • **
  • Posts: 568
  • Country: gb
  • Electronics Enthusiast
Re: DC load using a CPU cooler
« Reply #126 on: October 04, 2014, 06:01:45 pm »
I finally got round to resending the sample request email to IXYS. I think the last one didn't get through, as I sent it from my website's domain so it may have been spam filtered.

@Jeroen3, I think I'll stick with the Freezer A30 as it's more widely available if it fails. It's not too expensive and will easily fit two TO-247s.
 

Offline microbugTopic starter

  • Frequent Contributor
  • **
  • Posts: 568
  • Country: gb
  • Electronics Enthusiast
Re: DC load using a CPU cooler
« Reply #127 on: October 04, 2014, 06:04:04 pm »
@timb Aargh! I just realised my PM to you didn't send! I've sent another; let me know if it gets to you.
 

Offline Mr.B

  • Supporter
  • ****
  • Posts: 1249
  • Country: nz
Re: DC load using a CPU cooler
« Reply #128 on: October 08, 2014, 03:32:39 am »
I am also designing a DC Load.
Seems to be project of the year here.

I am considering using a Corsair H60 watercooler. A bit more expensive than a Freezer A30, but should work well.
The copper plate is removable from the waterblock/pump assembly.
This way I should be able to directly reflow solder the FETs to the copper heat exchanger face.
I have emailed Corsair Tech Support asking for more detalis:
What is the thermal spec for the unit in deg C per W?
What is the maximum power dissipation of the unit?
Can the 'O' ring on the copper plate be easily removed and put back after soldering?
I eagerly await their response... I hope their Tech Support is quicker than Zelman Coolers.

One disadvantage of using the watercooler is that I don't know the electrical conductivity of the coolant.
This means that the fan forced radiator needs to be kept isolated from everything including the chassis.

Any thoughts as to whether this is a good or bad idea...?
« Last Edit: October 08, 2014, 03:34:57 am by Mr.B »
Where are we going, and why are we in a handbasket?
 

Offline SeanB

  • Super Contributor
  • ***
  • Posts: 16385
  • Country: za
Re: DC load using a CPU cooler
« Reply #129 on: October 08, 2014, 04:11:15 am »
The coolant will be water based, and likely quite conductive.
 

Offline NiHaoMike

  • Super Contributor
  • ***
  • Posts: 9323
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
Re: DC load using a CPU cooler
« Reply #130 on: October 08, 2014, 04:26:36 am »
Those packaged water cooling kits don't perform that much better than good heatsinks, certainly not enough to offset their cost. They're really designed for those who want water cooling for the "cool" factor and compact machines that don't have the space for a big heatsink.

Something that really works in our favor is that power MOSFETs withstand higher temperatures than CPUs, and power resistors and SiC FETs even higher. Just beware that for the heatpipe coolers, beyond a certain temperature (somewhere above 100C, well above what most CPUs are rated for), the vapor pressure will get very high and could cause the tubing to burst.
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 

Offline Mr.B

  • Supporter
  • ****
  • Posts: 1249
  • Country: nz
Re: DC load using a CPU cooler
« Reply #131 on: October 08, 2014, 06:07:33 am »
@SeanB - Yip, aware of that. Thats why I would maintain electrical isolation for the radiator.

@NiHaoMike - Thanks and yes, I would not want the waterblock exceeding about 90 degrees C.
The main reason for using a watercooler is to reduce the size of the enclosure that is required as you mentioned. The A30 is huge.

It would be nice if TiN noticed this thread. He appears to have loads of experience with cooling... being a (somewhat) professional overclocker.
(I am not rude enough to PM him)

« Last Edit: October 08, 2014, 06:38:18 am by Mr.B »
Where are we going, and why are we in a handbasket?
 

Offline Jeroen3

  • Super Contributor
  • ***
  • Posts: 4209
  • Country: nl
  • Embedded Engineer
    • jeroen3.nl
Re: DC load using a CPU cooler
« Reply #132 on: October 08, 2014, 12:26:24 pm »
Those water coolers don't perform much better. Instead, they excel in a transporting the generated heat over a distance.
Where air heatsinks transport the energy a few centimeters away before convection to air. Watercoolers can do tens of centimeters.
Very useful when you're putting your project in a desktop case.

Unfortunately electrical isolation pads directly on the fets severely reduce thermal performance. Especially short peak performance or impulse load changes.
 

Offline Mr.B

  • Supporter
  • ****
  • Posts: 1249
  • Country: nz
Re: DC load using a CPU cooler
« Reply #133 on: October 08, 2014, 06:25:43 pm »
Yes, i have seen the performance specs on various 'review' sites.
I am not expecting the watercoolers to be much better than a good quality standard fan cooled unit.

My DC Load will be in a desktop case and the water cooler is slightly more compact than something like an A30.
The thermal efficiency between FET case and cooler should be better as I intend to solder the FETS directly to the cooler. (As I said above)
The cooler face is flat copper. Fairly easy reflow job.
Where are we going, and why are we in a handbasket?
 

Offline NiHaoMike

  • Super Contributor
  • ***
  • Posts: 9323
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
Re: DC load using a CPU cooler
« Reply #134 on: October 08, 2014, 06:42:52 pm »
If you're looking for great cooling in a small heatsink, look at rack server heatsinks.
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 

Offline Mr.B

  • Supporter
  • ****
  • Posts: 1249
  • Country: nz
Re: DC load using a CPU cooler
« Reply #135 on: October 08, 2014, 06:54:29 pm »
Thanks, I will investigate them as an option.
I was after something with a solid copper face so that I can solder the FETs directly to it.
The removable copper face on the water cooler is attractive because it will have lower thermal mass while soldering.
Where are we going, and why are we in a handbasket?
 

Offline eneuro

  • Super Contributor
  • ***
  • Posts: 1562
  • Country: 00
Re: DC load using a CPU cooler
« Reply #136 on: October 08, 2014, 07:31:25 pm »
@eneuro and the other guy talking about welding/brazing:

I would discourage anyone from spot welding copper for thermal purposes, and even more so actual IC packages to copper. 
Heating those IC units by classic soldering or putting them to ovens is thermaly better?  :-DD
Haven't got a chance to test spot welding IC taps with brazing yet, probably next week while my buffy spot welder should be ready and tested this week after upgrade to 25A SCRs in pararell on its primary and just developing its software to add support for brazing mode :-/O

12oV4dWZCAia7vXBzQzBF9wAt1U3JWZkpk
“Let the future tell the truth, and evaluate each one according to his work and accomplishments. The present is theirs; the future, for which I have really worked, is mine”  - Nikola Tesla
-||-|-
 

Offline mrflibble

  • Super Contributor
  • ***
  • Posts: 2051
  • Country: nl
Re: DC load using a CPU cooler
« Reply #137 on: October 09, 2014, 03:21:22 am »
@eneuro and the other guy talking about welding/brazing:

I would discourage anyone from spot welding copper for thermal purposes, and even more so actual IC packages to copper. 
Heating those IC units by classic soldering or putting them to ovens is thermaly better?  :-DD
The issue is the cross section of the resulting thermal interface. Spot weld ==> smaller cross section (compared to soldering) ==> poorer thermal conductivity across that thermal interface. Anyways, it has already been explained pretty well earlier on in the thread.
 

Offline Jeroen3

  • Super Contributor
  • ***
  • Posts: 4209
  • Country: nl
  • Embedded Engineer
    • jeroen3.nl
Re: DC load using a CPU cooler
« Reply #138 on: October 09, 2014, 07:14:05 am »
If you're looking for great cooling in a small heatsink, look at rack server heatsinks.
There are plenty for sale on eBay of recycled servers.
 

Offline eneuro

  • Super Contributor
  • ***
  • Posts: 1562
  • Country: 00
Re: DC load using a CPU cooler
« Reply #139 on: October 09, 2014, 06:33:48 pm »
The issue is the cross section of the resulting thermal interface. Spot weld ==> smaller cross section (compared to soldering) ==> poorer thermal conductivity across that thermal interface.
This is not in the case when brazing method is implemented.
Spot welder is only used to create huge short pulses to join copper areas by brazing, so it is simply source of high energy released in short times and those two surfaces should get soldered, but not by heating everything around including semiconductors chips.
Withput brazing it could be simply not possible join together copper metals without huge currents while it has very low resistance and very good thermal conductivity.
I hope it will work, then will tell more after experiments are made, but I do not dismiss it until tests are done.
Without brazing it couldn't have any sense to play and melt even with 1mm copper TO220 taps.

Tips for Soldering and Brazing Copper Tubing

12oV4dWZCAia7vXBzQzBF9wAt1U3JWZkpk
“Let the future tell the truth, and evaluate each one according to his work and accomplishments. The present is theirs; the future, for which I have really worked, is mine”  - Nikola Tesla
-||-|-
 

Offline mrflibble

  • Super Contributor
  • ***
  • Posts: 2051
  • Country: nl
Re: DC load using a CPU cooler
« Reply #140 on: October 10, 2014, 03:42:10 am »
The issue is the cross section of the resulting thermal interface. Spot weld ==> smaller cross section (compared to soldering) ==> poorer thermal conductivity across that thermal interface.
This is not in the case when brazing method is implemented.
Well, your remark (complete with that bit you bolded for emphasis) and my reply to your remark were about spot welding. No remarks from me about brazing, since brazing is just soldering with an attitude. And as you point out, with brazing you don't get a poor cross section as with spot welding. What with it being the same kind of filler action as with boring old lower temperature soldering.
« Last Edit: October 10, 2014, 04:13:57 am by mrflibble »
 

Offline rob77

  • Super Contributor
  • ***
  • Posts: 2087
  • Country: sk
Re: DC load using a CPU cooler
« Reply #141 on: October 11, 2014, 05:34:05 am »
The issue is the cross section of the resulting thermal interface. Spot weld ==> smaller cross section (compared to soldering) ==> poorer thermal conductivity across that thermal interface.
This is not in the case when brazing method is implemented.
Well, your remark (complete with that bit you bolded for emphasis) and my reply to your remark were about spot welding. No remarks from me about brazing, since brazing is just soldering with an attitude. And as you point out, with brazing you don't get a poor cross section as with spot welding. What with it being the same kind of filler action as with boring old lower temperature soldering.

and actually brazing would kill the mosfets easily - the max soldering temperature for most of the components is around 300 degree celsius while brazing happens well above 400 degree celsius. so brazing is a big big NO-GO for mosfets.
 

Offline eneuro

  • Super Contributor
  • ***
  • Posts: 1562
  • Country: 00
Re: DC load using a CPU cooler
« Reply #142 on: October 11, 2014, 06:51:39 am »
so brazing is a big big NO-GO for mosfets.
Did you tested it with your own spot welder?  ???

By brazing using spot welder and short high power pulses I mean something different than that  :-DD
« Last Edit: October 11, 2014, 06:53:26 am by eneuro »
12oV4dWZCAia7vXBzQzBF9wAt1U3JWZkpk
“Let the future tell the truth, and evaluate each one according to his work and accomplishments. The present is theirs; the future, for which I have really worked, is mine”  - Nikola Tesla
-||-|-
 

Offline mrflibble

  • Super Contributor
  • ***
  • Posts: 2051
  • Country: nl
Re: DC load using a CPU cooler
« Reply #143 on: October 11, 2014, 07:14:22 am »
@eneuro:
Just to avoid confusion, what do you mean when you say "brazing"? Essentially the same thing as soldering, but with higher temperature melting point for the filler material? Or something different that I am not aware of?

Because you mention "brazing using spot welder"? Wut? How does that work?

At any rate, as far as I am concerned for a direct join of the metal tab of say a mosfet to a heatsink:
- spot weld ==> noooope, crap cross section ==> poor conductivity
- brazing ==> noooope, high work temperature ==> high risk of killing semiconductor
- soldering ==> yup

That is not to say that I would pick soldering over boring old clamping, but it is to say that I would pick soldering as the prefered method out of the above three choices.

Unless I am missing something, in which case please explain.
 

Offline eneuro

  • Super Contributor
  • ***
  • Posts: 1562
  • Country: 00
Re: DC load using a CPU cooler
« Reply #144 on: October 11, 2014, 06:08:47 pm »
Unless I am missing something, in which case please explain.
I wrote about those instructions in post #118  ;)
I found there  How to Spot Weld Copper those instructions, and applying brazing in 1st step is a key to do it.

Quote
"Instructions:
1 Apply brazing paste in between the two points to which you will apply the electrodes of the spot welder.
2 Lower the electrodes of your spot welder and pinch the two pieces of copper you will spot weld.
3 Activate your spot welder for quick bursts, turning the welder off and removing the electrodes after each burst."

Didn't tested it yet, but makes sense.
Brazing pase between two much better conductors should behave like this steel ball, but it will be diffrent thanks to point #3 of those instructions above and I have much more power full spot welder I think.
Paste should be melted quickly and since datasheet of one of mosfets I used says about 300*C for 10 seconds it is a lot of time to glue two pieces of copper.

Probably at the end of next week I will make some experiments-using spot welder with brazing pase to join together good conductors like copper and aluminium.  :-/O


12oV4dWZCAia7vXBzQzBF9wAt1U3JWZkpk
“Let the future tell the truth, and evaluate each one according to his work and accomplishments. The present is theirs; the future, for which I have really worked, is mine”  - Nikola Tesla
-||-|-
 

Offline SeanB

  • Super Contributor
  • ***
  • Posts: 16385
  • Country: za
Re: DC load using a CPU cooler
« Reply #145 on: October 11, 2014, 06:14:35 pm »
300C for 10 seconds on the LEADS, not the case. While you probably could get away with reflowing them at 250C in an IR oven to reflow it to a copper clad board you would be severely stressing the package attachment.  Brazing at 300C will probably cook the case off.
 

Offline mrflibble

  • Super Contributor
  • ***
  • Posts: 2051
  • Country: nl
Re: DC load using a CPU cooler
« Reply #146 on: October 12, 2014, 09:23:15 am »
Unless I am missing something, in which case please explain.
I wrote about those instructions in post #118  ;)
I found there  How to Spot Weld Copper those instructions, and applying brazing in 1st step is a key to do it.
I thought you might have meant that, but better to be sure. Reason I asked is because even that article you linked to refers to the joining action as spot welding, as opposed to brazing. Anyways, that's semantics, now I know what surface joining action you mean.

I suspect many molten mosfets lay in your path to enlightenment, but we'll await your experiment results. ;D
 

Offline eneuro

  • Super Contributor
  • ***
  • Posts: 1562
  • Country: 00
Re: DC load using a CPU cooler
« Reply #147 on: October 12, 2014, 01:52:24 pm »
300C for 10 seconds on the LEADS, not the case.
300*C on TO220 tap was taken only as comparision to its leads solder temperature 1.6mm from case, since I'm not sure if there are tap soldering temperatures and time in datasheet ;)
12oV4dWZCAia7vXBzQzBF9wAt1U3JWZkpk
“Let the future tell the truth, and evaluate each one according to his work and accomplishments. The present is theirs; the future, for which I have really worked, is mine”  - Nikola Tesla
-||-|-
 

Offline Jeroen3

  • Super Contributor
  • ***
  • Posts: 4209
  • Country: nl
  • Embedded Engineer
    • jeroen3.nl
Re: DC load using a CPU cooler
« Reply #148 on: October 13, 2014, 06:46:02 pm »
Earlier in this topic there was talk about an IXYS mosfet. Which one if I may ask?
IXYS seems be manufacturing quite fancy stuff. But the seemingly most effective is an depletion mode, which equals to a short without a control supply.

Those TO-247's and 264's are great for mounting underneath cpu coolers. Don't fool around with TO-220.  ;)
« Last Edit: October 13, 2014, 06:47:35 pm by Jeroen3 »
 

Offline microbugTopic starter

  • Frequent Contributor
  • **
  • Posts: 568
  • Country: gb
  • Electronics Enthusiast
Re: DC load using a CPU cooler
« Reply #149 on: October 13, 2014, 06:50:28 pm »
I just sent an email to an IXYS sales representative in the UK asking for two IXTH75N10L2 N-channel enhancement-mode MOSFETs. The first sales rep (h-rep) turned me down, and I can't find the MOSFETs online in minimum quantities less than 30. I might go for a different one if I continue to have no luck.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf