Since I need a better thermal solution than the heatsink I have for my LM4780s, I snagged some Xeon Wind Tunnel heatsinks off eBay.
Check these suckers out...
This is the package you get with a Xeon processor - the heatsink in a little acrylic shipping box (save the box - it's a great tiny-parts bin!), an 80mm 12VDC fan with a tach lead (standard 3-pin motherboard header), a base cradle, a tunnel-like cover, a fan bracket, and a bag o' hardware.
Surprisingly enough, the bag o' hardware contains a small syringe which in turn contains a little bit of micronized-silver thermal goo. Don't know how it compares against the king of thermal compounds - Artic Silver - but hey, I'll use it.
The heatsink itself is a thin-fin design, with the fins inserted into slots cut into the base. The base itself is about 3/8" thick, and hefty - a good pound or so. The center fins are, as the pic shows, copper.
The basic idea is to mount the cradle onto the motherboard's heatsink holes (which generally are aligned with studs on the case) and set the heatsink into it and on top of the processor. Two metal clamps secure heatsink into cradle.
The top cover then snaps into place around and over the heatsink.
Once the cover's in place the reason for the "wind tunnel" moniker is obvious. It has a detachable top opening that can be used to mount the fan over the heatsink in case you need to blow air in from the top and out both sides, but that gets left in place if you use the next part...
The kit includes a small plastic part that the fan attaches to, and this in turn snaps onto one end of the cover to form the wind tunnel. This kind of arrangement seems like it'd lend itself VERY well to a ducted internal-heatsink design for an amplifier - low profile and active cooling in one package.
Once the fan is attached to the end, you have your wind tunnel.
My only real complaint about the unit is the milling on the heatsink - it's rough milled and has grooving that a fingernail can feel. For use as a processor heatsink it'd really work better if lapped with some fine sandpaper proir to installation. The same can be said for use as a chip-amp heatsink.
Another view of the heatsink's base finish, before I stuck the second one into the cradle.
And finally, here's what I intend to do with the ones I have: bolt paralleled LM4780 amp modules to 'em!
The design of the cradle makes it possible to mount the heatsink by its plastic cradle and let the heatsink itself float electrically, and this makes it possible to secure the heatsink somewhere and mount one heat-generating part to it without having to worry about insulating it. Better thermal transfer = longer component life and better operation under heavy loads!
Once I've had a chance to bolt the 4780s to these and run some tests I'll post my thoughts on whether and how well these work...
oO
Check these suckers out...
An externally hosted image should be here but it was not working when we last tested it.
This is the package you get with a Xeon processor - the heatsink in a little acrylic shipping box (save the box - it's a great tiny-parts bin!), an 80mm 12VDC fan with a tach lead (standard 3-pin motherboard header), a base cradle, a tunnel-like cover, a fan bracket, and a bag o' hardware.
Surprisingly enough, the bag o' hardware contains a small syringe which in turn contains a little bit of micronized-silver thermal goo. Don't know how it compares against the king of thermal compounds - Artic Silver - but hey, I'll use it.
An externally hosted image should be here but it was not working when we last tested it.
The heatsink itself is a thin-fin design, with the fins inserted into slots cut into the base. The base itself is about 3/8" thick, and hefty - a good pound or so. The center fins are, as the pic shows, copper.
An externally hosted image should be here but it was not working when we last tested it.
The basic idea is to mount the cradle onto the motherboard's heatsink holes (which generally are aligned with studs on the case) and set the heatsink into it and on top of the processor. Two metal clamps secure heatsink into cradle.
An externally hosted image should be here but it was not working when we last tested it.
The top cover then snaps into place around and over the heatsink.
An externally hosted image should be here but it was not working when we last tested it.
Once the cover's in place the reason for the "wind tunnel" moniker is obvious. It has a detachable top opening that can be used to mount the fan over the heatsink in case you need to blow air in from the top and out both sides, but that gets left in place if you use the next part...
An externally hosted image should be here but it was not working when we last tested it.
The kit includes a small plastic part that the fan attaches to, and this in turn snaps onto one end of the cover to form the wind tunnel. This kind of arrangement seems like it'd lend itself VERY well to a ducted internal-heatsink design for an amplifier - low profile and active cooling in one package.
An externally hosted image should be here but it was not working when we last tested it.
Once the fan is attached to the end, you have your wind tunnel.
An externally hosted image should be here but it was not working when we last tested it.
My only real complaint about the unit is the milling on the heatsink - it's rough milled and has grooving that a fingernail can feel. For use as a processor heatsink it'd really work better if lapped with some fine sandpaper proir to installation. The same can be said for use as a chip-amp heatsink.
An externally hosted image should be here but it was not working when we last tested it.
Another view of the heatsink's base finish, before I stuck the second one into the cradle.
An externally hosted image should be here but it was not working when we last tested it.
And finally, here's what I intend to do with the ones I have: bolt paralleled LM4780 amp modules to 'em!
The design of the cradle makes it possible to mount the heatsink by its plastic cradle and let the heatsink itself float electrically, and this makes it possible to secure the heatsink somewhere and mount one heat-generating part to it without having to worry about insulating it. Better thermal transfer = longer component life and better operation under heavy loads!
Once I've had a chance to bolt the 4780s to these and run some tests I'll post my thoughts on whether and how well these work...
oO
be mindful that you won't need more than 2 oC/W at the max for one LM3875 -- even in the TF (insulated) package.
i think that it's a very nice solution, however -- if you can source the C/W ratings then perhaps you could use multi-LM3886's on one heat sink.
in fact, the best solutions seem to come from either your experience or surplus or end-of-the-run sinks for big SCR's.
i think that it's a very nice solution, however -- if you can source the C/W ratings then perhaps you could use multi-LM3886's on one heat sink.
in fact, the best solutions seem to come from either your experience or surplus or end-of-the-run sinks for big SCR's.
I spent a couple hours lapping the heatsink base with 600-grit and 1200-grit paper with a chaser of Brasso.
The finish is a LOT better now...
It's not mirror finish, but a mirror finish is overkill in this application. Lapping it this much, however, should be good for a detectable decrease in thermal resistance.
oO
The finish is a LOT better now...
An externally hosted image should be here but it was not working when we last tested it.
It's not mirror finish, but a mirror finish is overkill in this application. Lapping it this much, however, should be good for a detectable decrease in thermal resistance.
oO
The heatsinks I´ve shown I believe are manufactured by Fischer Elektronik.
You can download PDF catalogues from their site.
I got the heatsink from ELV and I think they´re the same. I´ll compare and get back hopefully with C/W-ratings as ELV doesn´t mention them.
(They just say the small one can take 100W and the bigger one 250W with full throttle of the fans;-))
If you go to ELV-page you´d have to search for "LK 75" or "LK 40" in the textfield on the bottom.
For the small one you need two halfs costing 10.50€ each.
The bigger one will cost 15.50€ each.
So it´s not particularly cheap but still better than directly from Fischer-Elektronik.
Another problem is the groove to hold the screws.
Your IC won´t have contact in places and the base is also not that thick.
Going for a high power GC and mounting multiple IC´s one the sink you´ll definitely want to install a fat aluminium bar or whatever to improve contact.
But yes if you include the heatsink properly in your amp you can make a nice compact design.
You can also mount it vertically like in one of Peter´s amp with the copper pipes but that not the most flat amp then.
(BTW: Searching for the LK´s you´ll find a gainclone PCB in the shop which is the worst design ever.
It doesn´t include input cap, resistor path to ground from input.
I´ve tried it, if the pot comes lose you´ll have a substantial DC offset at the ouput which a friend of mine burned his speakers with. I emailed them about it but they´re rather stupid)
You can download PDF catalogues from their site.
I got the heatsink from ELV and I think they´re the same. I´ll compare and get back hopefully with C/W-ratings as ELV doesn´t mention them.
(They just say the small one can take 100W and the bigger one 250W with full throttle of the fans;-))
If you go to ELV-page you´d have to search for "LK 75" or "LK 40" in the textfield on the bottom.
For the small one you need two halfs costing 10.50€ each.
The bigger one will cost 15.50€ each.
So it´s not particularly cheap but still better than directly from Fischer-Elektronik.
Another problem is the groove to hold the screws.
Your IC won´t have contact in places and the base is also not that thick.
Going for a high power GC and mounting multiple IC´s one the sink you´ll definitely want to install a fat aluminium bar or whatever to improve contact.
But yes if you include the heatsink properly in your amp you can make a nice compact design.
You can also mount it vertically like in one of Peter´s amp with the copper pipes but that not the most flat amp then.
(BTW: Searching for the LK´s you´ll find a gainclone PCB in the shop which is the worst design ever.
It doesn´t include input cap, resistor path to ground from input.
I´ve tried it, if the pot comes lose you´ll have a substantial DC offset at the ouput which a friend of mine burned his speakers with. I emailed them about it but they´re rather stupid)
neutron7 said:fans on a gainclone???? SACRILIGE! its supposed to be quiet, its not like you are trying to heatsink a 500W amp or something.
use the cool heatsinks by all means but not the fan!
just turn it upwards in your desighn and let lovely silent convection do its work
ya i am going to use the heatzinc the fan is ugly
pinkmouse said:
That has to be the ugliest piece of industrial design I have ever seen! Burn it! Burn it, now!
yes it is and it hardly fits in a pc case without hiting the psu bad design by coolmaster
Banned
Joined 2002
karma said:good thing i saved this one now i wana try it
Always thought those Coolermaster Jets were oogly.
neutron7 said:fans on a gainclone???? SACRILIGE! its supposed to be quiet, its not like you are trying to heatsink a 500W amp or something.
use the cool heatsinks by all means but not the fan!
just turn it upwards in your desighn and let lovely silent convection do its work
Well, I do plan to mount them vertically for convective operation, but also plan to go one better...
I've got some TC622s coming. They're 5-pin TO-220 temperature-operated switches. I'll throw in a power MOSFET and bolt a 622 to each heatsink. Then, once the temp reaches whatever I set as the trip point the fan will activate and cool the 'sink down to a few degrees below the trip point. Thus there'll be no unsavory fan noise unless and until the amp actually needs it, and even then only on whichever channel needs it at the time.
Besides, my goods are going into a portable amp that'll be lugged around for DJing, etc. so fan noise is a non-issue in that application. (Practically all sound-reinforcement amps are fan-cooled.)
pinkmouse said:That has to be the ugliest piece of industrial design I have ever seen! Burn it! Burn it, now!
I have no idea if they even work. Personally I prefer Thermalright's SLK series for cooling processors.
joensd said:
Ooh, thanks for the info! I still must possess a few of those for other projects - besides, I could always get longer segments by length and bandsaw them into smaller segments for other projects.
karma said:
ya i am going to use the heatzinc the fan is ugly
yes it is and it hardly fits in a pc case without hiting the psu bad design by coolmaster
Like I said, the Thermalright SLK series owns. Swiftec has some awesome HSFs as well but they're pricey.
JasonL said:
A lot of these end up being tossed or sold on eBay, etc. as many builders are using 1U rackmount enclosures, and at about 3" tall the wind tunnel won't fit in a 1U.
And yes, these are the HSFs that come with the 604-pin retail P4 Xeons. Fans even have the Intel hologram labels.
oO
Got the Wind Tunnels mounted, and amp boards mounted to them...
Note the DMM, which is reading both the AC driving the speaker and the DC in the output - it's reading 0.0107VDC, and I didn't see any DC over three hundredths of a volt during the low-power tests. This is a Very Good Thing™.
Here's the business end - two of my amp modules bolted to Xeon Wind Tunnel heatsinks. With the fans running they don't get above ambient until I run a 4-ohm load at 10 watts of music power for about half an hour. They kick that much ***. I zapped one with an IR thermometer and it read 64 degrees F with the fan on and 1W RMS into a 4 ohm load.
Close-up of one of the modules and its attachment to the heatsink. The amber-colored thing is a 3-mil thick layer of Kapton MT 100, DuPont's thermally conductive but electrically insulative heatsink material.
These heatsinks definitely work. The fans are relatively loud though, so if you're not happy with fans you might want to cook up a thermally controlled fan circuit so they'll only be activated when necessary. In my incredibly scientific and precise tests the fan pulled the heatsink down to below ambient from warm to the touch in thirty seconds.
oO
An externally hosted image should be here but it was not working when we last tested it.
Note the DMM, which is reading both the AC driving the speaker and the DC in the output - it's reading 0.0107VDC, and I didn't see any DC over three hundredths of a volt during the low-power tests. This is a Very Good Thing™.
An externally hosted image should be here but it was not working when we last tested it.
Here's the business end - two of my amp modules bolted to Xeon Wind Tunnel heatsinks. With the fans running they don't get above ambient until I run a 4-ohm load at 10 watts of music power for about half an hour. They kick that much ***. I zapped one with an IR thermometer and it read 64 degrees F with the fan on and 1W RMS into a 4 ohm load.
An externally hosted image should be here but it was not working when we last tested it.
Close-up of one of the modules and its attachment to the heatsink. The amber-colored thing is a 3-mil thick layer of Kapton MT 100, DuPont's thermally conductive but electrically insulative heatsink material.
These heatsinks definitely work. The fans are relatively loud though, so if you're not happy with fans you might want to cook up a thermally controlled fan circuit so they'll only be activated when necessary. In my incredibly scientific and precise tests the fan pulled the heatsink down to below ambient from warm to the touch in thirty seconds.
oO
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.