Richard Jones' Log: Shuttle HTPC ... continued

Tue, 27 Jun 2006

In my attempts to get the Shuttle HTPC working again, I tried the official "fix the product you sold me" channels. After contacting the retailer, Shuttle and finally the wholesaler, I got the message that the fan problem I'm having is "common" and that I'd need to ship the motherboard off to Taiwan to be fixed. The fix would cost me about $15-$20 (plus shipping) and would take about 6-8 weeks.

So I played around with the system fan, PSU, some wire and some resistors. I found that a serial resistance of about 55Ω got the fan running at a good speed (not too noisy). Unfortunately, it seemed to be drawing about 300mW or so, and at that power the little 1/4W resistors let out all their magic smoke.

So I bought a 5W 56Ω resistor ($0.48), soldered it in and the HTPC is now back up and running!

Comment by Alan Green on Wed, 28 Jun 2006

Congratulations. It's always great when you fix something yourself.

There is also the old Department of Fair Trading not-fit-for-purpose routine, but that can take quite some time.

Comment by David Eads on Mon, 28 Aug 2006

I have a Shuttle SN95Gv3, which is a gorgeous looking little case, but the big fan blows like a vacuum because (I believe) the temperature sensors are inaccurately reporting the CPU temperature. I've done a variety of tests, and the BIOS and lmsensors report a CPU temp of between 120 and 200 degrees. I made a little test matrix (fan connected / fan disconnected X cpu < 10%/cpu ~ 50%/cpu ~100% for both minimum and maximum scaling frequencies controlled by cpudynd, gpu throttled down as much as possible for all tests) and in about 90% of all cases, the temperature reading stayed between 180 and 200 degrees celsius. This seems impossible to me, particularly because I did some qualitative/subjective tests for each case, touching the CPU heatsink and out system case. The variance in what I felt was significant, with case and CPU being too hot to touch after the 100% cpu usage and cool to the touch after the low load tests.

Anyway, I'm relaying all this because I relayed all this to Shuttle tech support, who said that my tests "couldn't be right" and that an earlier test in which I changed the clock multiplier (to force a lower clock speed via hardware) voided my warranty. The conversation literally concluded with the tech guy shouting at me about how I was forever disqualified from support for that box.

This is exceptionally frustrating -- Shuttle's designs seem very well engineered, only to be hobbled by crappy components and actively hostile support. I'm a board member and volunteer staff for FreeGeek Chicago, and we just got a couple of Shuttle cases donated that only run at the lowest multiplier. The donor had similar problems with Shuttle tech support. It really is too bad, because I want a small, quiet, visually appealing case, but I won't buy another Shuttle until I'm pretty certain they've resolved these issues.