Bitcoin Forum

First off, I'm a big fan of what Vorta has been doing on his thread - if you haven't checked it out, definitely go there and check it out:

https://bitcointalk.org/index.php?topic=1072176.0

I've been experimenting a fair bit with different case designs as well - in particular trying to find the best balance of cooling + sound + ease of build.  As Vorta mentions, the thermal sensor on the blades is really a poor design - it doesn't really give you a good sense of what the board is actually doing, as I've I've seen it report 70c when it reality it was 90c+.  I have a thermal camera (BTW, you can mod the E4 to have all the features of the E8, awesome!) that I've been using to analyse the heat dissipation and what is actual happening to the boards as I change things around - and try to use things less subjective, like anemometer to measure actual airflow, etc.

http://www.analogx.com/images/lasers5v1/dscf3149s.jpg

It's important to note on the thermal side of things it's a bit tricky measuring it - because things like acrylic are not thermally transparent, so they need to be removed in order to measure temperatures.  At the same time, this affects the airflow, so in general I try to let it run for a couple minutes assembled, then pop off the side or whatever and image it so I get something that's closer to what reality is.

The first thing I did was replace the sides - personally I like the look of the circuit boards, so transparent acrylic works well for it in this case.  I was also bored and decided to etch a BTC logo in it to be fancy:

http://www.analogx.com/images/lasers5v1/dscf3162s.jpg

I went through several design iterations, but ended up with this as the best general configuration - the sides are 2mm acrylic and the top and bottom are 5.6mm, but only because I used up all my 2mm and only had 5.6mm left.  I think 2mm all around is what I'll use for my finished version.  I also took the hint from Vorta's on ducting, and put a 5.6mm spacer between the fan and the enclosure - this actually made a big difference no matter what you're doing, so if you're looking for a super simple mod to do - do this.  Here's the assembled verison:

http://www.analogx.com/images/lasers5v1/dscf3163s.jpg

It's worth pointing out the only the sides and bottom are screwed in - the top just rests on the PCB and sides.  I found this to be the most convenient, and the cables and everything else hold it in place really well, so there really isn't a big need for anything else.  You can see here how it attaches to the bottom:

http://www.analogx.com/images/lasers5v1/dscf3167s.jpg

Vorta filled in the bottom area with his design - which I think is technically better, but most of the fins on the bottom of the heatsink on my S5's look like a dogs lunch, so this works in all situations, and gets the bulk of the gains.  You can see here a closeup on the top, and how it's edged around the Beaglebone:

http://www.analogx.com/images/lasers5v1/dscf3168s.jpg

In some of my earlier designs, I basically screwed into the top much like I did on the bottom of the heatsink, and then made my own mounting adapter for the Beaglebone - but this didn't make any difference in my measured results, and was actually a huge PIA to install - whereas this one just drops on and you're good to go.

Now, the only thing better than not generating heat is getting it the F away from whatever is making it, so ducting is also a big consideration for me - here was one of my first experiments with this design and a 4" duct:

http://www.analogx.com/images/lasers5v1/dscf3169s.jpg

Unfortunately, a 4" duct doesn't allow a large enough volume of air through, so it hurts the thermal performance too much for my tastes.  I'm in the process of changing the flange that connects to the case, and moving up to the larger 6" ducts common to HVAC.  I know I could have gone to 5", but in the US at least they are hard to find at Lowes or Home Depot, so better to have too much air volume than too little.

Once I was happy with the general design, I started playing with fan configurations - push, push/pull, and just pull.  While I love the performance of the stock fan, I don't like the sound level - I've literally bought every 120mm fan that Amazon has to offer, and I've settled on this one:

http://amzn.to/1S6nDAV

I've probably bought about 50+ in the past few weeks, so they're probably wondering what's going on!  ;)  It doesn't have nearly the performance that the stock fan does, but it has the best balance of the ones I've used.  One downside is that it's not PWM controlled, but it's fixed speed of around 3k is more than acceptable for my purposes, since all my S5's ever do is spin up to full speed anyway - and personally I'd rather have the fan fully working the whole time anyway.  Back to the case, without the enclosure, push is definitely the way to go - with there being a slight advantage to push/pull, but generally not worth the extra cost IMO.  But, with the enclosure, pull works significantly better, and push/pull gains even less than the non-enclosure situation.  Now, keep in mind, I'm basing this off my thermal readings of the PCB, not the temp reported - although in general they tell the same story.  So here's one of them in action:

http://www.analogx.com/images/lasers5v1/dscf3171s.jpg

I think I may do a couple more minor tweaks, but I'm pretty happy with this design, and I'm more than likely going to do a big batch of them so I can upgrade all my existing cases.  I wish I had one of the new S5's with the heatsinks on the side to test with, but even though I've bought a ton of S5's in the last month, I haven't been able to score one with them.

If anyone has any suggestions or ideas, I'd love to hear them... Since I'm in the mode of still tweaking it, now's the time!

And if there's enough interest, then I may make them for other people when I do my big batch - one nice thing is it's super easy to do, you can install it in under 5 mins and it's good to go.

Very nicely done!  I love the clear of still being able to see the circuit boards. 

Another very well done mod!

Nice.This is how I like it: simple,cheap and does the job. Vorta's design is cool but for me it's little too HIFI.Well done  :).

Good idea, because heat dissipation is bad on this models.

Very nice indeed! Good work with all that! I see you also took the Scythe Ultra Kaze fans. They really are the best choice.

What I would suggest is try getting some mini heatsinks to glue them on the outside and see if that drops the temps. In your case they should drop them more than they do in open case versions. On my first miner, which came without external heatsinks, I used these: www.ebay.com/itm/171764372281. Glued them on with this thermal adhesive: http://www.ebay.com/itm/111172816819
I got a price of $30.00 for 60 heatsinks, so if you're buying them, make sure you click "Make offer" instead of just buying them at stated price.

I would definitely prefer to have acrylic side panels on my miners. You have me a very nice idea. :D

Hey Vorta, thanks for the heads up on the heat sinks - I was searching around for something good like that, but hadn't found anything close with nice big fins like that.  I just submitted an offer, so we'll see how it goes - I'd love to update all my S5's to use then, but first I'll see how much of a PIA doing one of them is.  ;)

One thing I was wondering about was the placement - if you see on the image with my thermal camera, those heat dissipation squares on the PCB are actually fairly cool, it's the surrounding area that really heats up.  Here's a closeup  of that area on one of the newer PCB's (the one in the test case is an older S5):

http://www.analogx.com/images/laserS5v1/flir0069.jpg

The actual thermal squares are running a cool 105-115f range, whereas the area immediately surrounding it is in the 160-180f.  I'll be interested to see what kinds of difference they make - I'll try some with just thermal paste first and see what they do.  I can't tell if those regions are cooler because they're more effective at dissipating the heat (I'd be surprised if it were that pronounced), or if there were a heat sink there if it would absorb and radiate some of that surrounding heat.

With your ducts, did you play around at all with sizing the ports that you're using to the sides?  I know you have one with and one without the sides, but just wondering if you played with larger vs smaller openings.  Also, did you play around with different lengths?  Mine is 5.6mm but only because that's the thickest material I had around - I could easily just use multiple ones to get more length, but wasn't sure if the gain between 5.6 and 25mm would be worth it (I read somewhere that you reached peak duct performance at 25mm for case fans, but who knows).  ;)

Nice I like the heat view!  ;D

Great job.

Could you give us a screenshot of the hashrate and temp with and without the acrylic?

I really like that clear case design!! 

I like the S1 as the boards are open.  THe s3 is awesome and all but I like the rugged look of the S1.

MarkAz, do you by any chance have schematics of this? i have access to a laser cutter and would like to use your design. if not, all good. just thought i'd ask.

Can you talk a little bit about the advantages of this?  Will you be selling parts to do these mods?

Vorta mentioned the ducting benefit on his thread, and I was familiar with it from the PC modding side of things - but basically if you put some clean flowing air between the fan and the device, you can get better efficiency.  For example, currently there are the two pillars (from the head sinks) that split the intake into effectively 3 parts - when the fan is pushing or pulling air against this, it's less effective because it's so close - less opportunity to allow adjacent air to flow around it.  This isn't technically the most accurate way to describe it, but it gets the idea across - so by adding a clean space for the air to move through, it can more easily move around those obstacles would undo turbulence.   In practical terms, if you just throw even one of my 5.6mm spacers inbetween the stock fan and the case, you'll see about a 2c drop in indicated temperature.  But once again I have to stress, the indicated temperature isn't worth a whole lot, but it is interesting.

The reason I did it was more because my pull fan would hit the screws on the back of the case, so I wanted some spacing.  I believe what I've read is that around 1 inch (for case fans) is ideal length, anything more doesn't buy you anything - but I was surprised to see such a small space made such a big difference.

As far as selling parts, when it's all said and done, if there's enough interest than I probably will... For small guys, it's a super easy mod to do, totally reversible so you can undo it if you don't like it, and it makes for a much more house-friendly S5.  For the bigger boys, I think the ducted solution will be awesome, as my hope is to eliminate the need for any fans at all on the S5's, and instead have a plenum and single large fan.  I've seen similar configurations before in manufacturing, and I spoke to an HVAC designer I know and he confirmed that it should be doable.  Greater reliability, and massive noise reduction - plus depending on how large a deployment you're doing, it probably could even save a couple amps when it's all said and done. 

One other maintenance plus of this design over the push design is that the heat sinks are exposed on the front, and if you've ever taken a fan off of an S5 that's been running for a while, all kinds of junk builds up behind the fan - and this really affects the cooling.  With the pull design, that's not something you need to worry about because it's obvious when buildup happens.

There was previous S5 fan noise thread quite a while back. One member mentioned that he used a dead 120mm fan as a "spacer" between the regular S5 fan and the metal plate of the S5. . He cut out the motor, blades, motor support, and just retained the 120mm shroud of the fan. He noticed a reduction in noise just using the stock fan. While it looked almost like two fans stacked at the one end, it supposedly worked well at reducing the noise.

This was essentially the spacer mentioned above, and about 25mm "thick". I think the idea is that the air flow is a bit more "organized" by the time it actually hits the heatsink. That's my simplistic understanding.

I don't want to say about aesthetics, it's a matter of taste, but this mod is good for cooling.  ;)

Why don't you stick a little radiators to those S5s on which this is possible, like on this:

http://www.analogx.com/images/lasers5v1/dscf3163s.jpg

The nice thing about acrylic is that it's in pretty much every color, so instead of clear, it could be black or whatever else - it really doesn't matter.  Opaque black would look very much like the older miner...  I'm with you on taste - and I'm not saying I made some piece of art, I'm just saying I prefer this look over the flimsy black plastic sides - they make an otherwise solid device feel a bit cheap IMO.

As far as heat sinks, I ordered some up, but I'm going to experiment with where I place them as the thermal image shows heat dispersion that's different than I would have expected.

And that idea of hacking a fan for a duct - I saw that also, and actually bought these to test it out:

http://amzn.to/1HxzCoi

It's pretty much the cheapest 120mm on Amazon (basically $4 each prime)... I can't speak to the quality of the fans, but the enclosure works well.  I didn't see much difference between my cut spacers and these, so I just stuck with the ones I made - but if you want a quick and easy way to do them yourself, this way definitely works.

That really nice. I did try yesterday to put the S3 cover on a S5 and was surprise to realise that this doesnt work since the fan plug are on the top and block the cover and because the hole are not at the same place.  :-\

Very very nice!
Only thing I would suggest changing is that for fire safety, use polycarbonate (Lexan) vs acrylic. Acrylic does not self-extinguish whereas polycarb does.
Only problem though is esthetics as polycarb gives a bit bubbly & brown edge when laser cut.

From an aesthetics standpoint, I actually think that laser cut Lexan looks cool - I've done it a couple times, but nothing large.

The problem with Lexan is that it outgasses some nasty stuff that ends up damaging the laser - that's why most places won't laser cut it - only places with rocking air extraction (and while I do have outdoor power venting, I do not have air extraction). In all of my temperature tracking on both the PSU's and the miners, I don't think I've seen them exceed 200f at the point of heat generation.  Now, that is right at the bottom temp that acrylic will start to soften but the melt point is north of 300f, so it shouldn't be an issue.  Plus even at the screw to heat sink points (at the bottom of the case), and where I would expect to see the highest direct temps, about the max I'm seeing is 115f.  The big advantage is the airflow does a good job of keeping the acrylic cool...

If you have any other material suggestions, let me know!  I was more concerned about the temperature issues on the PSU's, as the definitely hit higher temps - and I was looking for a thin material I could use as a buffer layer, but hadn't found anything that was super cost effective yet.

I've been buying a bunch of used S5's lately, and once I get and test them, I'm converting them over to my case - and this is a prime example of why I prefer pull vs push fans (warning, graphic imagery):

http://www.analogx.com/images/lasers5v1/dscf3177s.jpg

With pull, you always have buildup behind the fan, which works as basically insulation... That's not to say that pull isn't without it's issues, you basically have buildup where the air comes into the case.  The difference is that with pull its usually very easy to see and clean up - whereas pull is just sort of lurking behind the scenes.

Anyway, I'm in the process of converting most if not all of my machines to the 2mm version, and we'll see what kind of difference that makes overall in airflow with my boxes.  Unfortunately I've also up in forced air at the same time, so I can't really compare apples to apples, but I think the ability to focus the exhaust air in one basic direction is going to be great.

I hope they'll work for you as well. You're looking at ~1h of sticking all 60 of them. :D They are quite small but they seem to be working quite well in my case. Anything else that I found to fit on the PCB had much shorter fins.

Your thermal camera is very useful. I wish I had one while playing with my miners. This temperature find is quite interesting. Since the area is in direct contact with the chip it doesn't make sense to be so cold and you can get a burn on touch. Can the material on the surface affect camera's thermal readout?

When I was developing the duct I played around with several parameters. I tried different lengths and went for the one that is short enough for a 40mm screw to hold along with the fan. I didn't go with smaller as this is really minimum. I couldn't make it smaller without affecting the airflow too much. I also tried different openings for the sides and realized that the most effective size is the one that just makes the air move instead of standing there, but nothing more. Outside heatsinks are completely ineffective compared to internal heatsinks, so you don't want to waste too much air on them.

I have my miners running for days, overclocked at 400MHz, in a room which reached temperatures of 45°C during the day this week, without overheating or a jump in HW errors. Hotter miner reached 79°C with hottest chips at 93°C actual temperature. So I can definitely say for myself that the duct properly serves the purpose I made it for. ;D Default miner is not certified for ambient temperatures above 35°C at stock clock.

http://www.pohrani.com/f/22/13A/mloCMMC/s5-black.png

http://www.pohrani.com/f/34/13F/1EKOB0IU/s5-green.png



Why not cutting your own back of the case? :)

For those that dont have access to a laser cutter it can be done with regular tools.
 I am pretty handy with acrylic using a table saw with a plastic cutting blade and a router and also plastic cutting drill bits. ( I have built lots of fish tanks and sumps and various parts even dispaly cabinets and PC cases ). all of them look like you could have bought them in the store.. I certainly cant etch the bitcoin symbol but the rest of it is easy enough.

I thought about doing this.. but I am currently running mine in a vertical stand versus the horizontal placement.

do the Sides really add anything to the cooling process... that the original sides did not already provide.. yes I know the original sides were very thin plastic but if it just for air flow does it matter.

I am not sure if your laser cutter friend will crank these out for free or not.. but I have so much acrylic laying around. I can make a template and crank out 100 of these with just the scrap I have from previous jobs.

I use google sketchup for all my design projects what do you use ??

Yes (http://thermal-imaging-blog.com/index.php/2010/09/01/emissivity-understanding-how-it-affects-your-thermal-images/), definitely (http://www.arl.army.mil/arlreports/2005/ARL-TR-3690.pdf)! And bare metal is among the worst affected materials -- it reads _much_ cooler than actual. You can coat it with black spray paint or spray-on boron nitride to bring the surface emissivity closer to what the imager assumes. Or, if you have an imager with adjustable emissivity assumption, set it to 0.02 or so and look only at the gold plated surfaces.

Hey Vorta, in regards to why not printing the back of the case - I use the metal front and back as well as heat sinks for the screw mount points.  Otherwise with acrylic the only thing I've found to work is captured screws, but they can be a PIA to work with.

With your temps, what I've seen in fiddling around - and I generally don't let it go as high as yours, is that the hotter it gets, the less the ambient room temp seems to affect it.  Meaning if it's 35c in the room, and the miner is at 75, the room could go to 45c and the miner might only increase by a couple degrees.  It seems like airflow affects the miners temps more than ambient air temp, but I haven't really looked into it seriously.

innerchaos, the sides are critical - if it were open, then there would be no point to this mod.  Don't forget I'm enclosing the top and bottom - look closely at the pictures, it goes all the way across and overlaps the sides.

E, definitely true - the emissiveness is what dictates it, and things like aluminum (heat sink) are notoriously bad - the coated part of the PCB normally are what I focus on.  I've never used any kind of paint or coating (but I know you can use it) - typically I just use electrical tape, as that is what I was originally shown as a really good general purpose solution.

Around what cost do you have in acrylic to make 1 kit?    Looks very nice just wondering cost of this vs pla plastic type.

And keep up the cool looking mods.  Really does look great.

I bought these a while back but haven't used them yet. They also have thermal paste on there already sealed by like a sticker you need peel off. Press it on the chip and you are done! Remember, the thinnest layer of any thermal paste/grease is best for dissipating heat. That's why these are great, since the thermal "stuff" on these already is paper thin.

http://www.ebay.com/itm/100pcs-Aluminum-Computer-VGA-Card-Xbox360-DDR-RAM-Memory-Cooling-Cooler-Heatsink-/121331886647?pt=LH_DefaultDomain_0&hash=item1c3ff1aa37

Not mine, from a photo I grabbed a while ago.

http://www.photopiks.com/s5heatsinkallover.png

Temp sensor on the S5. Hard to find.

http://photopiks.com/tempsensor1.jpg
http://photopiks.com/tempsensor2.jpg
http://photopiks.com/tempsensor3.jpg

Just throwing another picture out. These are some of my miners. I have also purchased about 20 different fans and tried MANY different setups to keep these guys cool. Surrounding the miner, like putting a case around it did nothing. Temps stayed the same. The stock fans were a little too loud so I'm using 2 Noctura Industrial 3000's in a push/pull with no extra boxes or ducts. Temps keep stable and consistent at 57-60. I do have air conditioning, so of course that is a huge factor in temps. The air conditioning would be powered on regardless if the miners were there or not, so it is no extra expense for me. Although I do have to say around 3pm EST USA if the temps outside hit 95+ the room does heat up and temps will climb to a max of 63 (only with door closed), soon as I open the room door the heat just shoots out of the room into the rest of the house, which is also air conditioned and I would imagine the extra heat is expelled by the AC system. I can keep the room door closed %100 of the time, but I do not like 63c. I did think of just ducting the heat right out the window, but not good figuring out such a complicated duct system with the 6 miners in that room. How to expel the individual heat flow from 6 miners out the window would be something interesting to see work. I wish I could come up with something but haven't figured it out. So things will stay the way they are for now, they are working good and temps in check. If no air conditioning in the house, the miners would probably have blown up by now.

http://photopiks.com/s5minersinroom.jpg

try a pad under the miner

 http://www.amazon.com/Silicone-Baking-Mat-Non-stick-Measurements/dp/B00CVQENUO/ref=pd_sim_79_4?

your miners base vibrates a bit and leaks air.  the pad above will lower sound and seal the miner to the table.

this forces air to go out the pull fan and some extra air goes up at the controller passing over the heat sensor.

put a little heat sink on the heat sensor .   I was able to get my s-5's to do freq 412  with decent errors  and decent sound. 

I used the silverstone  141 as a pull and a delta 120 x 38 as a push.

some where I posted a thread  on my results. around Jan or Feb.

I already tried all that. I tinker with fixing watches and I had some of those semi-soft, malleable mats that a watch maker would work on. I sat a couple miners on it and since it was semi-soft it filled up any small cracks or holes. Temps stayed the same, didn't matter. I have also tried closing off the entire bottom and top with some thin aluminum I had. Sealed just about everywhere, and there was no difference.

NOTE: When the miners all had stock fans, doing the seal up did have a small affect when ambient temps increased a little. So I can't say it doesn't work at all, but when using two of the Noctura ippc 3000 fans no matter what I did the temps were always the same regardless of what I did. I have a feeling the stock fan just blows loud and all over the place and the Noctura fans are very directional and don't need such a setup. When watching TV in just the next room, even with the door open I don't hear anything. If I had the blasters (stock fans) on them I would always hear that hum/whine. I'm sticking with my expensive Noctura's. Never saw such good performing fans. I also bought two of those Sythe fans from Amazon and even though they are 38mm thick, they couldn't keep up with the Noctura's.

I have lots of fans. I have 10 or 15 brand new STOCK S5 fans that I received from Bitmain a while ago, plus another 6 STOCK S5 fans that I used for about 2 weeks. If anyone in the US interested in S5 STOCK fans, let me know.

opentoe, unless you full enclosed it like I did, with almost no airgaps, then you'll get very different results.  I had tried the Noctura's as well, but they were not even in the running compared to the Kaze - but since you're running in an air conditioned situation, then that's probably your biggest variable. 

On the heat sinks you linked to - they look nice, but you mention they have thermal paste already on them - are you sure they don't just have thermal stickers?  If it's paste, that's great, although I would probably just recommend using thermal adhesive and being done with it.  Also, contrary to what I've seen written on here, it is NOT a good idea putting a heat sink on the temperature IC - this makes an already marginal measurement almost completely worthless, and will increase the likelihood you're overheat your hardware substantially.

I've been converting all of my test rigs over to the cases - but I'm gone right now, so when I get back next week I'll finish it up.  Since I'm in AZ, I'm testing the worse case situation, where ambient temps surpass 110f... And what I'm interested in designing is something that works with the device using just airflow in that situation.  Right now if you have enough CFM for the room forced air, you can run machines using the stock fans just fine... The issue I have with that is how loud it is.   With my configuration so far I'm seeing temps that are a fair bit higher than stock in this super-heated situation, so let's say 5c above stock as reported by their IC, but the actual miner's heat pattern looks more even than stock.  One other plus is that this 5c temperature increase is regardless of spacing between machines - whereas the stock configuration gets hotter if you put them right next to eachother (whereas mine doesn't) - this is just really because stock it's venting hot air in all directions.  That's why ducting is interesting to me - if I can completely get it out of the mix, then I think that's a great gain.

notlist3d, hard to say what it will cost when done, since the thickness of the materials affects things a fair bit - but as it stands right now with 2mm bodywork and 5.6mm fan spacers, it's under $20/machine, and shipping should be cheap since it basically all becomes flat...

I would love to get 6 of those stock fans I don't mind the noise.. I just want to be able to keep em cool and having six extra will allow me to do  quick swap out for cleaning

I've started converting over my test S5 setup at my house, and I'm leaving one of them stock, so I can use as a comparison.  Right now I'm running them at basically ambient air temperatures, which are brutal here right now, so let's say between 44c and 46c during the day - so if you're interested in how the Antminer does under high temps, then you'll be interested in this.  ;)

The stock unit (with spacer, I don't run any without it, so figure it's about 1c lower than without) runs out at about 67/64 in the temperatures mentioned above.  I also tried for the heck of it just doing on in pull configuration, but the unit started sounding the thermal warning after a couple minutes (80c), so that ends any of my tests.

With my enclosure (plus spacer), in the same push configuration, it runs about 71/65, so about +4c on average.  In the pull configuration (my normal one), it runs 69/63, so about +2c on average.  Now while this may sound worse, I argue that it's actually better - when I do the thermal imaging on the boards, the heat tends to be more even across the board than in the stock configuration.  One other advantage is that point it's the hottest is towards the fan side (in pull), which I suspect means I could create some inlets in order to allow channeling some cold air in later in the path - the trick being not to make these holes so large that they starve the main intake.

After mounting them on 20+ S5's, I've refined the screw hole size/positions, as there is a fair amount of variation between units.  When I originally put together my test rig, I bought all of them used so I could get a nice cross-section of devices, but there was alot more variation than I expected.

I'll have more this week, I'm finishing up a logging and management application for the S5's, so I can get a better sense of their status over time - and then combine it with the networked temperature monitors I built.

I added the heat sinks that Vorta linked to earlier - and as Tony the Tiger would say, "They're Great!"

I used Arctic thermal adhesive - not thermal paste:

http://amzn.to/1dYPRyw

I figured that I would leave them on either way, so might as well stick them on for good... The adhesive is the two part kind, so a bit of a pain but nothing major.  I cleaned the surface first with Isopropyl alcohol - I prefer these wipes because the wipe doesn't leave any fibers, it's 99%, and they don't skimp on the alcohol:

http://amzn.to/1Gh8xRH

I applied the adhesive to the base of the heat sink, and then just stuck it to the miner - you don't need a thick layer of the stuff, just enough to cover the bottom, so I imagine the thermal paste kit above would be enough to do two miners (or at least that's what I would probably be able to manage).  Here's what they look like:

http://www.analogx.com/images/laserS5v1/dscf3180s.jpg

So I have a bit less envy about those lucky enough to get factory ones.  I would point out that it took me about 30 minutes to apply them all, so not super long, but not fast either.

As I mentioned above, I'm running these miners in what would best be described as a 'harsh' environment - so it was hot when I test it.  This particular miner runs typically about 73/73 - this is with the case mod and the Kaze fan in pull configuration.  I got it up and running again and let it roll for about an hour, then checked how the temps were doing, and it was a solid 67/66, so about 6c decrease in temp - of course that's just what the internal sensor says, let's take a look at what Mr Flir has to say:

http://www.analogx.com/images/laserS5v1/flir0019.jpg

They're definitely running hot, but if you look at the surrounding areas, they're running much cooler, so they definitely seem to be pulling heat from the board.  I would guess that the impact on a case without the airflow being as controlled as mine may not be as dramatic - but with mine I'm forcing air over the sides somewhat evenly, so probably a more pronounced effect.

Either way, if you've wondered if and how much of an effect heat sinks on that side can make, now you know.  I would classify the ones I got as somewhere inbetween the nice ones that Bitmain put on some of there, and definitely better than those small silver ones most people put on them.

These heatsinks are pretty good. They use the high quality Sony T4000 thermal tape, so you don't need an extra cost for thermal grease/paste/tape.


http://www.ebay.com/itm/Bitcoin-Miner-Antminer-ASIC-BTC-180GH-s-Cooling-kit-72pcs-DIY-Aluminum-Heatsink-/121331886591?pt=LH_DefaultDomain_0&hash=item1c3ff1a9ff

I'm using thermal adhesive, as I don't really see much of a reason to ever remove them - get them done right once and not have to worry about them falling off.

As far as those, I've purchased some in the past, but as you pointed out the ones you linked to have much better thermal tape (the ones I got had crap tape, next to useless).  One of the used miners I bought actually had those applied with presumably thermal adhesive as well, although whoever did it installed them perpendicular to airflow, which sucks.  This particular unit doesn't perform as well as the other heatsinks mentioned, but it's hard to say how much off it is because of the wonky orientation. 

All things being equal, if I'm going to spend the time to mount these on a machine, I'm going to use the best heat sink I can lay my hands on.  I really like the ones that Bitmain put on the rare few machines (I have two), but I think those long finned ones are the next best thing - I can't really see much of a difference between my machines with them vs the Bitmain ones, but I haven't done really extensive testing yet.

Here's a shot of my test setup prior to mounting all the heat sinks, and I think I still had one or two with the stock case - now they're all running my custom case + kaze fan + spacer + heat sinks - so nothing else I can really think of to keep them cool mod-wise.

http://www.analogx.com/images/lasers5v1/DSCF3186s.jpg

how do you get rid of the heat in your garage. I had mine in the garage with a 1500btu AC unit and it was still so hot I had to underclock them just to keep them under 65c.

temperature here is 105F day high 85F overnight low

I have forced air (coming via an 18" duct from outside air) coming in on the 'cold' side if you can all it that in AZ, and on the hot side I have a big exhaust fan that dumps the air into the attic.  The whole situation works pretty well, although it definitely does get hot in there... I've written an app that changes the speed of the S5's based on the ambient temperatures, so that keeps them within range at all times.  I have Environsense modules on all the PDU's, plus I built my own WiFi temperature sensors that I scattered around to also get a feel of what outside, garage, attic, etc temps were.

With my case mods, the airflow is much more controllable - although ultimately I'd like to duct them into a plenum, and then exhaust that separately.  As it stands now, it's much more about airflow than air temperature - at least from what I've experienced.  I've considered doing some sort of cooling, but doing AC just wouldn't pencil out - I've been looking more at direct-indirect evaporator cooling.  The big advantage with these kinds of systems is you get the cheap price of evap, but also don't raise the humidity - at the cost of a bit more complexity, and slightly less effective cooling.  There are some of these kind of coolers that are up around 40 SEER, which is outstanding.  They're not going to work in all environments, but in a desert setting, they rock.

Is there anyway I could get the dxf files or drawing for the enclosure? I run a sign shop with a couple CNC's and just bought 2 x S5's and possibly a few more in the near future. Enclosing the whole thing with heatsinks over the chips ducting the air across them seems much more sensible.

Thanks

Gerald  

absolutely beautiful! Have you thought of selling these once youve got the design down? I know i would be interested, and likely know a few others who would aswell!

This is the best fan,  better than ultra kazes...

http://www.enermax.co.uk/fans/twisterstorm/


up to 154.53 cfm at 38 decibels ...  3500 rpm

stop spamming old threads with your company's link for fans

I SECOND THAT MOTION!!!

Vorta or MarkAz, did any of the miners that you purchased have a hashing board v.1.3 that have different heatsink pads than the later version say 1.7 that MarkAz had pictured? The reason that I ask is I purchased some of the heatsinks you have both tried out and I am not sure how or if it is recommended to attach heatsinks on the 1.3 boards based on the configuration.  Here are pictures of the 1.3 version board.

www.personaljohn.com/images/IMG_7058.JPG (http://www.personaljohn.com/images/IMG_7058.JPG)

www.personaljohn.com/images/IMG_7055.JPG (http://www.personaljohn.com/images/IMG_7055.JPG)

Any Suggestions?

Uhg, yeah, I had some of those earlier S5's that were that configuration, and you're right, the heatsink we were using don't work on those pads.  I actually found some head sinks that were small enough, about 1/3rd the size of the ones I used, just small enough to fit inside the copper region.  Overall I didn't notice enough of a difference to do it for more than the two I had originally bought parts for - the acrylic enclosure made way more of a difference than anything else.  So unfortunately for that particular one, I don't think there's much to do.

Sorry no better news - maybe Vorta found a better solution than I did...

MarkAz,  what is your power setup that you are running?  Any idea why the price of miners has dropped so much?  I am seeing S5's for $260 and S7's for $1,155?

For power I'm running either DPS-2000BB's or the IBM 2880W PSU's, both are server power supplies that require you to also get a breakout board, but there's several forum members who sell those plus high quality PCIe cables - and you get WAY more PSU for the dollar than PC power supplies.  One thing I don't ever skimp on is buying power gear - you can use it from generation to generation of machine, making it a good investment, and you can pretty much resell them from what you paid, as they've already really been depreciated when we buy them (as all miners are buying the PSU's used).

As far as pricing, that's easy - miner prices are typically influenced by a combination of the price of bitcoin, the current network difficult, and the power efficiency of the miner.  In your above price example, the S7 is roughly 4x the price, but 4x the hashpower - so you might think those prices are close, but the reality is they aren't.  The S7 is almost 2x the efficiency, so your power cost will be 1/2 of what it would be if you had the same hashpower in S5's.  So, depending on your power prices, you may not even be able to mine an S5 profitably, which is why the price has dropped.

On the difficulty side, hash power has been on a rampage, and the total network hashrate has more than doubled since October, which is pretty amazing:

https://bitcoinwisdom.com/bitcoin/difficulty

Less thrilling is that means you'd be making 1/2 as much as you were back in October with the same machine.  ;)

Generally speaking, you're almost always better off going with current gen hardware, although sometimes you might find special situations where it makes sense to get something one gen back (like an S5).

MarkAz do you do any or are you considering doing any solo mining?  Do you mine any other currencies?

so if i box in the s1's i have i will get better temperatures cause of there open box design ? i thought about doing this a while ago till i seen this from you btw very cool with the plexi

Bitmain has reduced the price of the latest batch S7 miner. The Batch 10 S7 is priced at 2.1 BTC, which is less than $800. The biggest drawback is that you won't get it until roughly February 20th at the earliest. Chinese Spring festival, so Bitmain is on "vacation" until that's over. That will also push down the price of a used S5 as well. See:

https://enshop.bitmain.com/productDetail.htm?pid=00020160129054250700R0OMI3KI061A