B3 Air Cooling Modification

How are you using your Bubba Two or Excito B3? Got pictures? Share here!
plbrandon
Posts: 7
Joined: 15 Nov 2011, 06:40

B3 Air Cooling Modification

Post by plbrandon » 24 Nov 2011, 11:21

In discussing another matter with Johannes at Excito, I included some pictures of the fan modification I did on the B3 and he asked me to post them here. Semiconductor life is dramatically improved with cooling and I am always concerned about this and willing to spend some time and effort if needed. In my case, I had about 15W inside a closed case (the B3) and this was more than I was comfortable with. The standard Excito design is quite good. Excito tells me that they have a failure rate that is very low, confirming that for most applications, the design is appropriate. The air cooling I have done here is beyond normal requirements but may be considered if you are a cooling nut (like me), planning very heavy use or want to assure maximum system life. The approach I took was influenced by a desire to minimize modifications to the B3 and minimize the time required while achieving a high level of cooling capability that would accommodate any disk changes or high use of the system that I might want to do in the future. My modification consisted of:

Remove the front panel (it is glued on) and drill it with 2 holes to mount it on the fan enclosure.
Remove the back panel and cut 2 slits in it for the escape of cooling air. This is the most severe part of the modification. I cut these slits with a Dremel using a metal cutting disk (carefully from the back). I cut it on top and bottom for both disk and motherboard airflow, being careful to leave all original labeling.
Carefully remove the disk and circuit board, tap 4 holes (4x40) in the B3 case front (in existing case channels like the ones Excito tapped for the screws holding the rear panel on).
Cut and drill the fan enclosure for the fan, exhaust hole and mounting to the B3 case, and other apparatus.
Optically redirect light from the status LED. In my case, I decided to mount the B3 front panel on the side of the fan enclosure and carefully mount, fit and adjust a small piece of fiber optic cable to redirect the light to the original B3 plastic light “window”. (http://www.jameco.com #171272 10' fiber optic cable for U$5.95). This allows me to slip the enclosure cover on and off with the LED “window” stopping in front of the fiber optic and take the motherboard in and out with the status LED stopping in front of the fiber optic.
Cut and fit the EMI screening and add a removable fan air filter.
Fit everything in the modification, replace the disk and circuit board and assemble.

To make mounting screw access easier, I mounted the fan enclosure on the B3 front with 1/2” below the bottom of the B3 case, then added 1/2” “feet” to the back end of the B3. To maintain EMI shielding, I used some galvanized steel screen (1/8” square grid) between the fan and the fan enclosure and shaped and slip-fit into the air exhaust slits.

The simplest method you could use to address heat would be to place an external fan blowing on the case. This would have a signifficant effect. If you are concerned about noise and fan life, you can add a 1-3W potentiometer (depending on the wattage of your fan) in the 12v side of your fan wiring. See the following for one discussion on this:
http://www.bit-tech.net/modding/2001/12 ... _control/1
In my approach I used a linear voltage regulator which has become sort of a standard for me in doing this type of thing. Another simple method is the use of a single bipolar or FET transistor for the voltage control. Search Google for examples. If you want a simple off-the-shelf controller, search Google shopping for things like “12v dc motor speed controller” or “12v fan speed controller” for many fairly inexpensive products. Here is an example of a simple 6-12v (U$6.99) fan controller, this one designed to mount in a standard PC case slot, but I presume it could be easily modified to include in your controller enclosure. I expect it is a single FET and is biased to force a minimum 6v on the low-end: http://www.xoxide.com/sunbeam-pci-fan-controller.html

In my approach, I added two 12v power jacks (add another if you want for an external eSATA disk), an on-off switch, an 80mm fan and filter, a linear voltage regulator and potentiometer for fan speed control, and a pair of jacks for voltage measurement of the fan voltage, all in a standard 6” x 4” x 5” aluminum enclosure. On the front of the fan enclosure I added an LED and trim pot (for LED intensity adjust) for “Power ON” status. I used a 1k ohm resister into a 5k ohm trim pot then through an LED for 2ma to 12ma adjustment for different ambient light conditions.

I considered smaller fans (like 1-3 30mm or 1-2 40mm) to better fit the size of the B3 case. This would have required more miniaturization, possibly building a custom enclosure, possibly more difficulty in redirecting the status LED, reduced air flow (or more noise) from smaller fans the lack of a air chamber, so I decided upon the larger case approach. Using an external case for the electronics is also possible but I did not want to deal with a second case and I had the space with my placement needs so I went with the larger fan and enclosure incorporating all electronics etc. into one package that was fairly easy to build.

Any amount of air flow would be very helpful. You could consider one (or two) 40mm fan(s), mounting all electronics (switch, speed control, etc.) in an external enclosure, with one power jack from the control to the fan(s). This would require making a B3 front plate replacement that would accommodate the fan(s), power jack (on a stand-out bracket) and drilling a hole for LED visibility. The plate would need to be thick enough to allow for countersinking the fan mounting screws (to not protrude into the B3 case) and should be conductive for EMI shielding. The existing front plate would be ideal. This plate is glued on and the “B3” lettering is also glued on and can be popped of with light pressure using a chisel. You could carefully drill the plate for screws to go into holes you tap (4x40) into the existing case channels. Cut a hole(s) for the fan(s), drill and countersink for fan mounting screws and the power jack bracket (that you must make). This would allow for the use of the existing status LED “window”. A fan air filter and EMI shielding should be provided. With a quick look I found 40mm air and EMI filters at http://www.newark.com (parts 21M7155 and 56P2964). There are others, but this should get your thinking started. I expect that this approach would be fairly simple, clean looking and allow for a smaller “footprint” that may be better for your placement needs for the B3 (but requires an external case for speed control and switch, etc. or doing clever miniaturization and construction to get everything on the front plate).

A word about power. I took the fan power from the B3 power supply. I added a switch to turn on the power for the fan and the B3. I do not know the sensitivity of the power needs for the B3, however, it is important to add a ceramic capacitor (0.1 uF or greater) across the switch to reduce switch contact bounce possibly causing confusion to the B3 motherboard. By-the-way, the following jack and plug parts are compatible with the B3 jack and power module plug: http://www.jameco.com parts 28760 and 151555 (standard power connectors, 2.1mm center pin, 5.5mm barrel diameter, 10mm barrel length).

The voltage regulator I used was an LM2941. This unit has a very low dropout voltage (0.1 volt at low current), however it is only good for 1A (very adequate for this need). You could use a standard LM317T (1.5A), LM350T (3A) or LM338T (5A) but these give you a dropout voltage of from 1 to 2.5v depending on the unit and the current (maximum fan voltage of 10.5v to 11v which should be adequate for this project unless you are using small fans, then I would recommend the LM2941). One or two fans will not be a problem, but if you use higher current in some other project, you must be concerned with heat-sink needs. I use the LM2941 in such fan speed applications because it allows me to crank the fan voltage up to 11.8v or 11.9v with a 12v input, great on computer case fans where I may have a need for the higher speed. For resistor values on the voltage regulator, I always use values that will prevent adjusting the low-end voltage below 5v or 6v for fans (I run my B3 fan at about 6v which is very quiet and with an 80mm fan gives very adequate air flow). This is important since you may adjust it below 5v and the fan will not start (some fans will not start below 6v), so you think it is nice and quiet (which it is) because it is not on (this could be bad - ruin your whole day). So, here is a set of resistances that will work for all the voltage regulators I have mentioned. Use 1k ohm from the adjust pin to ground, 3.3k ohm from the output to a 5k ohm potentiometer and the output of the 5k ohm potentiometer to the adjust pin. This will give you a calculated fan voltage range of 5.5-12v (with a 12.1v input). The final range will depend on variation in resistor actual values. For the LM2941, the 1k ohm from adjust to ground is required, it can be different for the other regulators (where you recalculate the remaining resistances from the specification formula). Just for completeness, (you can get this from the voltage regulator specifications) the remaining parts needed are 0.1uF (50v or more) from input to ground and 100uF (25v or more) from output to ground.

Below are the pretty pictures of what I built (3 here and 3 in the next post).

Happy cooling,

Paul
B3_fan_external_600.jpg
B3_fan_external_600.jpg (55.65 KiB) Viewed 13648 times
B3_fan_front_600.jpg
B3_fan_front_600.jpg (52.57 KiB) Viewed 13648 times
B3_fan_rear_600.jpg
B3_fan_rear_600.jpg (55.2 KiB) Viewed 13648 times
Last edited by plbrandon on 25 Nov 2011, 07:54, edited 1 time in total.

6feet5
Posts: 269
Joined: 13 Apr 2007, 17:32
Location: Gnesta, Sweden
Contact:

Re: B3 Air Cooling Modification

Post by 6feet5 » 24 Nov 2011, 15:17

Is this for real? :lol:

A lot of text and I must admit I haven't read all of it, but I couldn't find any figures on noice or temperature compared to an unmodified unit. Was it worth it (not that I'm considering this kind of mod)?

Thanks for sharing though. It's always nice to see what others do with their bubbas.

/Johan

Ubi
Posts: 1547
Joined: 17 Jul 2007, 09:01

Re: B3 Air Cooling Modification

Post by Ubi » 24 Nov 2011, 15:25

It's definitely an interesting mod. The term "monster truck" does come to mind though :)

johannes
Posts: 1467
Joined: 31 Dec 2006, 07:12
Location: Sweden
Contact:

Re: B3 Air Cooling Modification

Post by johannes » 24 Nov 2011, 16:00

I've already told Paul but I think this is a piece of art. Thanks for sharing! :)
/Johannes (Excito co-founder a long time ago, but now I'm just Johannes)

plbrandon
Posts: 7
Joined: 15 Nov 2011, 06:40

Re: B3 Air Cooling Modification

Post by plbrandon » 25 Nov 2011, 05:36

Here are the other 3 pictures:
B3_fan_rear2_600.jpg
B3_fan_rear2_600.jpg (72.98 KiB) Viewed 13603 times
B3_fan_mtgs2_600.jpg
B3_fan_mtgs2_600.jpg (55.25 KiB) Viewed 13603 times
B3_fan_elec2_600.jpg
B3_fan_elec2_600.jpg (67.7 KiB) Viewed 13603 times

plbrandon
Posts: 7
Joined: 15 Nov 2011, 06:40

Re: B3 Air Cooling Modification

Post by plbrandon » 25 Nov 2011, 05:56

Johan -

Yes there is a lot of text. It was my effort to be helpful. I did not do any measurements on noise or temperature, nor will I. With an 80mm fan at 6v it is inaudible. The temperature will depend on how hard you are running the system and the fan speed. Any airflow will have a significant cooling impact. As I mentioned, cooling the B3 is beyond normal requirements but it will extend system life. It is important in my application to minimize any possible failures and maximize flexibility in use. Now it meets my needs, that was worth the little time it took.

Ubi -

It is not a monster truck, it is a simple cooling fan, very simple.

Johannes -

Thank you.

Take care,

Paul

DanielM
Posts: 637
Joined: 28 Mar 2008, 06:37
Location: Sweden

Re: B3 Air Cooling Modification

Post by DanielM » 25 Nov 2011, 12:43

Hahaha... ROTFL! This is one of the weirdest projects I've seen. Buy the smallest, most quiet and most power efficient server you can find and then buy a huge fan and then put them together. It really forfeits the entire purpose of buying an Excito server. But thanks for a nice guide :wink:

/Daniel

Ubi
Posts: 1547
Joined: 17 Jul 2007, 09:01

Re: B3 Air Cooling Modification

Post by Ubi » 25 Nov 2011, 13:41

A monster truck, in its essence, is also simple: It's a car with some specific parts completely oversized, while no real improvements to the primary purpose of the machine are made and ability to park is decreased. I'd say the analogy is pretty adequate.

6feet5
Posts: 269
Joined: 13 Apr 2007, 17:32
Location: Gnesta, Sweden
Contact:

Re: B3 Air Cooling Modification

Post by 6feet5 » 25 Nov 2011, 14:38

Paul,

"A lot of text" wasn't a complaint, I think it's great you took your time sharing this. What I meant to say was I couldn't find the information I'm interested in, and admitting I was only skimming the text.

I can't stand equipment that make a lot of noise (apart from my stereo and I'm a metal head so I guess some (including my neighbours) would consider this a greater source of noise than any NAS could make, modified or not).

My (unmodified) bubbas are close to dead silent most of the time, and you mentioned yours is inaudible, which is great, so I guess this is a tie.

How about temperature? I expect this is where your mod will outperform the rest of us using unmodified units (and I realize workload will affect the figures). My B3 (HDD actually) is reporting 42 degC at a workload of 0.63 last 15 minutes, in open air (21 degC) (using 'hddtemp' and 'uptime' (all part of a 'vanilla' B3) and a room thermometer).

The temperature on my units will usually rise a bit during summer time (close to 50 degC), which makes me a bit uneasy, though they will probably last for the lifetime I expect and, I admit, a failure would not affect me that much. I have no important data (and it's backed up) on my units apart from my music and por.., hrmm, film collection ;-)

/Johan

PS Monster truck is a great analogy

Puma
Posts: 223
Joined: 29 Sep 2008, 06:30

Re: B3 Air Cooling Modification

Post by Puma » 30 Nov 2011, 06:37

This is what I use,

Just 2 lose cool plates and a small fan beneath.

Results are great 4 or 5 degrees colder. normal operation 43 degrees

Puma.



Image
Image
Image
Linux is like a wigwam - no windows, no gates, apache inside!

su_root
Posts: 38
Joined: 29 Nov 2011, 06:27

Re: B3 Air Cooling Modification

Post by su_root » 30 Nov 2011, 12:59

@puma;

how hot is your HDD after this "mod"? Is the fan running all the time?

I'm having similar ideas for my (hopefully coming) B3 but with a laptop cooler.

edit: what is that "temp control" you have there next to the fan?

plbrandon
Posts: 7
Joined: 15 Nov 2011, 06:40

Re: B3 Air Cooling Modification

Post by plbrandon » 02 Dec 2011, 09:24

Puma -

Nice work and nice pictures. I have a suggestion: remove the cool plates and just let the fan blow on the B3 units. The B3 case IS the heat spreader and the heatsink. The cool plates introduce additional thermal resistance between the source (B3) and the sink (air flow). Removing the plates will allow cooling air to hit the bottom, sweep the sides and even allow eddy currents to sweep the top of the B3, removing more heat. I expect removing the plates will more than double your cooling. If you try this, let us know how it works out.

Paul

plbrandon
Posts: 7
Joined: 15 Nov 2011, 06:40

Re: B3 Air Cooling Modification

Post by plbrandon » 02 Dec 2011, 09:34

Because of some forum comments, I am left with the impression that information I did not take time to present would be useful in understanding why cooling is important and the effects of air flow.

Semiconductor life is directly related to temperature. A doping material (like Boron) is diffused into the silicon lattice in an oven at about 1,200 degC with a dopant atmosphere (like Boron). The Boron migrates into the Silicon until the desired geometry is reached, in minutes to hours depending on the device design needs. Then the Boron continues to migrate through the Silicon lattice, even at room temperature, though very slowly. Eventually, the geometry of the semiconductor region will no longer meet specifications and the circuit will fail. In the graph below I used the Arrhenius equation which is the definitive equation for determining semiconductor life. I adjusted the calculation to show a value of 1 at 20 degC. Normally, you would use the manufacturer's specifications for Mean Time to Failure (MTTF) at a specific junction temperature then use the equation to adjust for thermal impact on device life. I have used 1 here for clarity. Notice in the plot that if you are running at 50 degC, the failure rate will be 20 times higher than at 20 degC. Today I find that some Chinese manufacturers are providing poor design, manufacturing techniques and quality control as well as falsifying specifications. Also, the Chinese are engaged in practices of fraudulent labeling and selling components under a reputable manufacturer's name. Additionally, you can never know the source of components in a product that you buy today. This means that we are subject to more and more marginal components and we can never know what the actual MTTF is. What we can know for a fact is that whatever the MTTF is, it will get worse at higher temperatures. That is why I am a “cooling nut” when I want a product to last.
silicon.jpg
silicon.jpg (194.7 KiB) Viewed 13474 times
In air flow, I wanted to share with you that a little means a lot. Calculating cooling from air flow is complex. I have developed a basic capability (an equation) for a previous project need that incorporates an assumption for a very good heat sink. I have used that equation for the example below. Other systems, like the B3, will not have the same heat sink capability, but the plot from this equation is instructive for the effects of air flow. While the Delta T for Air In vs. Air Out is too high for the B3 heatsink (and I do not reflect only convective cooling), note in the plot, however, that 4 CFM will DOUBLE the energy transfer of 2 CFM. Any air flow will provide a dramatic effect vs. half that (or none). Provide SOME air-flow and you will have a major cooling effect.
air_flow.jpg
air_flow.jpg (172.64 KiB) Viewed 13474 times
In my first post, I commented upon an air chamber behind a fan, but I did not elaborate, so I thought I should. I have found (subjectively) that if I mount a fan about an inch from the object to be cooled it produces about twice the air flow (at the same RPM) vs. mounting the fan closer. Also, I have found that if I use an air chamber between the fan and the object to be cooled it further enhances air flow at the same RPM, though I do not even have a subjective estimate on how much, it does depend upon the specific configuration. So, when noise is a concern, I provide an air gap and a chamber. This has been very effective for me.

I must say again that the B3 design seems to be very adequate for most needs. Cooling the B3 is beyond normal requirements but it will extend system life. It is important in my application to minimize any possible failures and maximize flexibility in use while providing a Linux capability with a very low power consumption. What I have now achieves those objectives.

The information here is not definitive or complete, I did not take time to research it or focus more on the specific configuration of a B3. This is simply my quick sharing of knowledge and experience I have acquired. One of my degrees is in Physics and I worked in the semiconductor industry for a brief time, doing discrete device design and laboratory research in thermal effects. I have designed and managed several large, real-world projects and have worked as a consultant for many years, including clients in Asia. I am certainly not an expert is the areas I have commented upon here, but I am knowledgeable. I cannot take the time to pursue this further, but I thought that I would take the time to share what I have learned that impacts my choices. I do hope that it may be useful to some of you.

Paul

su_root
Posts: 38
Joined: 29 Nov 2011, 06:27

Re: B3 Air Cooling Modification

Post by su_root » 02 Dec 2011, 10:48

Nice update, thanks!

Actually I think that the problem should be addressed and a different approach towards the problem could be taken.

The CPU @ full speed consumes 2W and the HDD is between 4 - 8W depending on use. In other words, it is the HDD that generates the heat and not the CPU.

In this thread: http://forum.excito.net/viewtopic.php?f=9&t=3473 I was thinking on starting active cooling (ie a fan) when the temperature rises over x degrees. I'm still thinking about this but it is not top priority anymore.

I want to address the problem and get rid of what is causing the heat and not just taking care of the heat once generated. There are several ways this could be accomplished but at the moment I think the best way is to have OS and storage separated.

1) Boot OS from a flash drive
2) tmpfs for /var/log & /tmp (only to save the flash memory)
3) minimal swap (due to usb flash memory speed and to save read/writes)
4) internal HDD for storage

Depending on usage, I estimate that the internal HDD in above case idle >90% of the time (2 movies, 1 backup / week = ~6h. Heavy torrent downloading/seeding etc needs different approach) and thus it can be set to sleep after ~20min idle & suspend after 60min idle. This would save much power and generate >50% less heat over 90% of the time.

If B3 idle temperature is now between 35 - 45 degrees, I guess that it can run @ room temperature when HDD is suspended = no need for fans (which also consumes power).

edit: I have no degrees in this subject what so ever. I just try to think logically and do what seems right. I love linux and what can be done with it. Figure out new ways to do things is part of what I get paid for.

Puma
Posts: 223
Joined: 29 Sep 2008, 06:30

Re: B3 Air Cooling Modification

Post by Puma » 02 Dec 2011, 15:04

Hello su_root,

My setup is working fine.

fan rpm controlled (but not needed)

Power supply just simple 230VAC plug supply.

Normal operation temperature 43-44 degrees C

Puma

Image
Linux is like a wigwam - no windows, no gates, apache inside!

Post Reply