Topic: Swedish ASIC miner company kncminer.com - page 314. (Read 3049515 times)

The VRM heat goes from VRM -> foam -> plate ->  some radiates off the plate some may transfer down from the plate into ASIC heatsink but that would depend on temperature of each part.
The heat only goes to cooler places, if the base of heatsink is hotter than the plate(not VRM temp BTW) it will warm it(plate) too.

(second law of thermodynamics??)

If the 'heatsink base' is cooler than the plate, adding a dab of grease between plate and heatsink would help.
Else the ASIC will contribute to VRM temp issues.

Would need to attach thermocouple to a small probe  and measure plate and base of heatsink to 'know'.

The plate is WELL MANUFACTURED.
They are almost perfectly flat.
Turns out IMneverHO that flat is not ideal.
I bend the ends of the outer legs down for better contact of plate/foam/VRM.

Easy no parts needed improvement in VRM temps!
You can look at imprint in foam and determine where it needs more pressure.

It can be done without unsticking heatsink from ASIC if you bend the fan back and carefully pry plate up from VRM's.
The foam acts like doublesided tape.
It's more like clay or plumbers putty than foam.

I also squeezed the foam/clay a bit to make it taller before reinstall of plate.

YMMV


EDIT there is also foam below PCB under same challenged VRM's.
Raising cubes up a bit will help bottom of case radiate some of the heat.

The whole foam thing is an OK idea but poorly implimented.

Looks like a time-machine.

Im not too impressed with the airflow on that Noctua.  Its only about 20% more airflow than the stock fan that comes with the Neptune.  I require at least 200+ CFM.

A 140mm fan would need to spin higher than 2000 rpm.  There are not many out there.

thank you for the reply! do you have any idea how much cpm the stock 120mm fan has?

Its 240 CFM. 

http://www.newegg.com/Product/Product.aspx?Item=N82E16835213001&cm_re=delta_120mm_fan-_-35-213-001-_-Product

There stock 120mm fan was mounted on the inside but I removed that and had to mount the delta fan on the outside because the delta fan is thicker and wont fit on the inside.  Also I had to use a 120mm to 140mm fan adapter which also adds thickness.  Thats why it looks so bulky in the pic.  It blows air into Neptune.

He mentioned a 90 degree cooling system fitting that failed in an earlier post.

A slight misfortune has visited tolips secret watercooling lair.

First WB install went almost perfect.
Slight problem with ASIC temp. Not bad just no improvement.
Turns out there is sand in the vaselene that came with waterblocks.
Was prepared with other grease anyways.
I had hoped stock grease was adequate for testing, guess not.
So far trivial expected issues.

Move on to second waterblock install and lady luck exits left stage.
One of the 90's I attached to the waterblock FAILED.
It did not seperate completely, I would have had water everywhere.
(yes I was working on it with loop running and other ASICs hashing away)
I have so much flow I have to run only one pump out of two else air bubbles do not collect in resivour.
Woulda been like turning on a garden hose if it had seperated completely.

After I replaced the failed part my PCB only indicated 5 VRMs.
If you ever had a hard disk failure on a PC you may know how I was feeling at the time.
I did not panic, just got out the vacuume and PC crevace tool.
It took a few tries to get the VRMs to go again.
I also let the working VRM's help boil away the water.

All is well now and I'll (much more carefully) install the other 3 today sometime.

Come monday it's time to throw more money at it and get better 90's for all the waterblocks.
I'll also try and extract a pound of flesh from vendor for selling such shit.

Second time I had an issue with 90's from this vendor.
First issue was restricted flow in otherwise bulletproof  cool lookin 90's.
With the leak I am done playing nice, the shit is too fragile for the application.

What's interesting about this adventure is that it is same PCB that challenged me installing 12AWG power wires onto.
I ended up bypassing the PCB 12V copper and running wires directly to each VRM.
The good news is that it was not my mods failed only a temorary shit new part/rain delay.

All in all, for me, it's fun to visit the bleeding edge.

YMMV

true, somewhat, if a bit backwards......  the 8 VRM Jupiters were the "early" ones.  
Mostly, people got the 4 VRM Jupiters, which "overclocked" just fine tho.
The "lucky" few people who got the early 8VRM Jupiters normally had dead VRM's... unless you did some tweaking.
Now, with Neptune's power requirements, all come with 8 VRM's, with all 8 nearly "maxed out".

trivial reasons, not needed.

Can stack PCBs 1 per 80-100mm

I stacked with straight barbs but not as dense as possible.
https://i.imgur.com/xoHsaiJ.jpg
https://i.imgur.com/GXvNz3g.jpg

With 90's I could stack the Hoopiter on the left and the Nep on the right, remote the radiator and replace with single large fan.
Not that I would. but I want it to be just big enough eventually. whatever I end up with for an ASIC case.

YMMV


EDIT My hoopiter configuration has a potential flaw if I get a big air bubble in inverted waterblocks while running flat.
Designed for sitting on it side though. (that density thing again)

FTFY

just curious, any reason to use 90^o fittings versus straight in fittings to the waterblock?

Are you in this to get a little money back or just fooling around and tinkering like a hobby?