[mythtv-users] Quiet fans

Mike Perkins mikep at randomtraveller.org.uk
Tue Oct 21 13:40:42 UTC 2008

Darrin wrote:
> On Mon, Oct 20, 2008 at 4:01 AM, Peter Loron <peterl at standingwave.org>wrote:
>> On Oct 17, 2008, at 1:41 PM, Yan Seiner wrote:
>>> That way you could stick the temp sensors inside the raid array and
>>> run
>>> the fans based on the max temp in the array.
>>> (and a LCD showing the temp, usb interface, drivers... :-)  Just
>>> kidding
>>> on those.)
>> I've been tinkering with parts and spreadsheets. Looks like my cost
>> (in prototype quantities) for a unit that will programmatically
>> control 3 fans (via PWM), have 3 thermistors controlling them, and
>> have a 2x20 LCD screen, and USB connectivity would be around $26.
>> A "dumb" controller with direct temp controlled PWM driven fan headers
>> and 3 thermistors would be around $13. No microcontroller, no USB, no
>> display.
>> The fancy controller would have some kind of software that would allow
>> you to monitor the status of the fans, adjust the speed<-->temp ratio,
>> etc. The display would also have some programmability.
>> Thoughts? How much would *YOU* pay?  :-)
>> Basically if I can determine there is a market worth my time, I'll see
>> what I can do.
>> -Pete
> This has been an interesting thread.  I never thought I would be taking on
> an electronics project when I started playing with mythtv.
> I'm building something similar, but less ambitious.  One temperature sensor
> with a simple microcontroller and I'm not really planning to run more than 3
> fans (though the circuit could probably support more as long as the power
> supply is up to the task).  I want the flexibility to change the programming
> of the fan controller,  but that's where my ambitions end.  LCD's and direct
> monitoring of status does have a certain geeky attraction, but I don't
> really think I would use those features.  If it keeps the computer cool
> without being loud, it will be good enough.
> For my project, the parts aren't very expensive.  The only thing that's
> moderately expensive is getting the circuit board made.  Even then, I would
> have to have multiple boards manufactured just to use up the real estate of
> a standard board used in pcb manufacturing.  I think the end cost would be
> less than $10 each if I went with surface mount parts.
> I'll post my results.  I might have the programming/schematic finished in a
> week or two depending on my work schedule...
As someone who got his first soldering iron at age 11, I have been following 
this thread with some interest. Although I have nothing currently to contribute 
to the manufacture, schematics, etc. aspects of your projects, I do have some 
design/implementation questions.

If you are planning on using external sensors, rather than the ones on the 
motherboard, where do you expect to site them? I would naturally expect them to 
be somewhere near the items being monitored. If so, then there will be a certain 
level of decoupling from the fans which you are controlling, given that most 
fans are in the front, back or occasionally sides of the case, and not next to 
the items being cooled. How do you plan to handle the feedback delay? And how do 
you determine that the big fan you screwed into the back of the box is, 
actually, contributing to the cooling of the particular component you are 
monitoring? Airflow is not an exact science, after all[1].

These questions are not intended as criticisms, rather just to satisfy my curiosity.


Mike Perkins

[1] Aeronautical engineeers may disagree; however, just try flying your F-16 
around the inside of my server!

More information about the mythtv-users mailing list