Topic: Klondike heatsink sourcing (Read 11751 times)

I ordered 250 heatsinks yesterday. They are 78x100mm. Just wide enough to cover the thermal pads and not covering the mounting holes. They cost about $4.50 each without shipping or tax.

I will have to mill out an area over the thru hole connector and drill/tap the mounting holes myself.

I've been busy sourcing all other parts and barely have had time to think about heatsinks. Seems there are a number of things to keep in mind:

First it needs to contact each of the reverse side of the 16 chips, of which there thermal vias to transfer heat out. Easy.

It should avoid the four side mounting holes. This is easily solved by using a 80x90mm size.

With a flat 80x90mm heatsink is there needs to be a small recessed area so not to short out the PCIe pins. You also need four mounting holes drilled through the heatsink for spring-loaded pins. This is where the cost begins to add up.

I've been floating the idea of using two separate, smaller heatsinks, 30x90mm. I don't know, that might be more expensive in the long run and you'd lose precious surface area. Bkk also gave me the idea of using a copper or aluminum shim to raise the heatsink a bit away from the PCIe pins. I'm just thinking out loud now, feel free to chime in with thoughts.

Any news on the heatsinks for the k16? If so, lets get a group buy going in the US. I am based in Texas and wouldn't mind ordering 1000pcs to get a group buy started. PM me.

Any news on this yet? I need to get heatsinks ordered for my assembly service ASAP.

I'm looking into it so as to add it as an option to my build services, but I don't have anything final yet.

http://mittenmining.com

Hello,
is somebody out there who is intending to produce a watercooling solution for BKKcoins K16?

user toolhead created a block for burnins bitburner.
it would be very nice if there would be a solution for the K16s too.

I didn't account for serrations but I think I can apply a surface area multiplier - this will improve heat-sink performance. The two raise channels will probably have little effect on the final result. The current assumption is that there is "perfect" contact between the sink and the board. I think I can add a compound on the interface but have not done so yet.

I will adjust the thickness of the base in the next run. As far as the thermal vias, I am still working on tweaking this. I currently am using copper lands which extend from the pad of the Avalon to the heat sink. Modeling is always a trade-off, it is important to get things as close as possible but sometimes too much detail is not worth the effort. These PCB/heat models can be very accurate (90-99%) so long as the main features are present and the mesh resolution is good enough.

As interesting as producing my own heatsinks would be, I currently have no immediate plans to sell heatsinks, but who knows how things will work out in the future. This was a rough model of a commercially available heatsink posted on another forum by Bicknellski.

Yes, heat-transfer is a game of surface-area. This is why insulating a pipe too much can actually have the opposite effect.

Well i Guess i will be following this tread ,

I have 6 K-16 orderd so I wil want to buy at least 6 heatsink ,
if a fan that goes with them also gets sourced would be verry great

I got a quote from Heatsink USA for 3.5"(90mm) x 3.5"(90mm) here.

Qty. 100, SKU I004, Cut to 3.500” x 3.500” Heatsinks @ $4.36 each = $436.00 (Retail Cost)

Qty. 100, Cut to length charge @ $.20 each = $20.00

Qty. 100, Machine Work @ $5.02 each = $502.00

Total Cost = $963.02 ---> $9.63 each

 

Qty. 400 (Retail Cost)

 

Qty. 416, SKU I004, Cut to 3.500” x 3.500” Heatsinks @ $3.92 each = $1,630.72 (10% discount from retail)

Qty. 416, Cut to length charge @ $.20 each = $83.20

Qty. 416,  Machine Work @ $ 3.90 each = $1,622.40

Total Cost = $3,336.32 --> $8.34 each  

 

Qty. 1,200 (12% discount)

 

Qty. 1,250, SKU I004, Cut to 3.500” x 3.500” Heatsinks @ $3.71 each = $4,637.50 (15% discount from retail)

Qty. 1,250, Cut to length charge @ $.20 each = $250.00

Qty. 1,250, Machine Work @ $3.64 each = $4,550.00

Total Cost = $9,437.50 --> $7.86 each  

Interesting what difference 10mm to 32mm makes. You have an interesting software there.

Im not really sure the best approach for cost effectiveness. Has anyone found a 100mmx100mm "standard" style aluminum heatsink that is worth bringing to a machine shop to get the holes drilled? Or is it best to just get them custom made? I live in the Boston area, and this is the best looking place so far I've found: http://www.qats.com/Request.aspx (Advanced Thermal Solutions). They have a plethora of heatsinks and designs pre-fab'd, and accept custom orders. I'll look into it.

I am currently trying to get prices for heat sinks locally (Tangerang) and in Jakarta. These links will be helpful for those in Indonesia.

www.superexagraha.com/

www.aluminiumextrusionsatria.com/

www.katalogalumunium.com/consumer-application/AlumuniumHeat%20Sink.html

www.alumas.co.id/product.html

www.pt-alexindo.com/product-range.html

www.alkomandiri.com/about_profile.html

Note to everyone...

You are best to find a fabrication / extruder site near you. Shipping even 1 of these heat sinks any distance doubles the cost. Find something local. In terms of design you need to develop that. I am hoping we will have something recommended and posted by BKKCoins once he is testing that phase in the GitHub. Start sourcing a local fabricator NOW!

I sent 4 heat sinks of various sizes to BKKCoins and others have also sent heat sinks. Marto sent his heat sink with the holes drilled and notch cut out etc... so follow along in the BKKCOins thread at some point in about 2 weeks we will probably see a heat sink posted there. FANS are needed as passive cooling will likely be a too much especially for those with higher ambient temps.

Vigil, are you taking orders for your heatsink when you've perfected it to your liking? If not, could you recommend the best place in the US to get them manufactured? I'm looking for 16.

The fins are slightly serrated as well or ribbed and there are two raised channels for holding fans or mounting, two fins in on either side.

But that is an amazing result. What is the heat compound simulated in the model?

Also the thermal vias will be filled with an epoxy, I think???, to prevent solder wicking... also the base of the heat sink is 5mm that is pictured pretty sure.

I have since run another model which is based directly from "Heat sink #4" posted by Bicknellski. The initial design above was based on it but I did not have the length of the fins correct. This model changes the overall height from 10mm to 32 mm. There is no airflow, just standing air with the heat-sink portion facing "up" so that the heat can rise away from the board and not get trapped. I accidentally took the picture rotated incorrectly, so please take that into account. The heat is rising not falling. Extending the fins has drastically improved the temperature, reducing it from 150C to 78C.


Heatsink #4 - Length 100.00 mm, Width 97.00 mm, Height 32.00 mm, Thickness taken to be 2mm.


Extending the fins by about 20 mm has reduced the temperature to 78C, approximately a 50% reduction in temperature. (Note: picture was taken upside down - I will update).

This is my first attempt at a simplified model of the K16 I've made using only the Avalon chips. I have them generating 2W each. They are modeled as QFN chips with most of the details and majority of the dimensions correct. I have modeled the lands that go through the board but they may not be meshing correctly so I will continue to look at that. I can add other chips and more detail if I can get the measurements and power output. The more exact I make the model, the more realistic the results will be.

This model uses an aluminum heat-sink which covers the entire bottom surface, the fins and base are 2mm thick - it is based off the heat-sink posted about 20 pages ago. I can alter this to any type of design: change fin size and number, change to fingers, etc. Additionally, this model is cooling only from the heat-sink, i.e., there are no fans or any air movement except that from buoyant convection. It is actually a transient scenario but here I am only using a steady-state model, but I can look at transient.

The temps seem a little high to me, 148 C (300 F), but of course this is without any air movement. However, many GPUs can get to 120 C or more if heat-sinks aren't attached or no air. This was just a first shot and I am going try some other things and recheck this solution, etc.


Avalon chips wire-frame.


Simplified board model with heat-sink.


Heat-transfer analysis showing temps of chips with no forced convective flow air exchange. Temps of chips reach approx. 150 C while operating at 2W with this heat-sink.

Wot?

I was kinda hoping for a 20% price reduction. A man can dream!

Yes .
We are going to output 2 versions:
1. With M3 thread on the 55x62 and recess
2. With trough hole 4mm dia and 8 mm dia blind for spring type clamping , with cutout for PCI-e pins

Glad the recess design worked out for you. I'll keep it in mind.