[time-nuts] Thunderbolt stability and ambient temperature

Mark Sims holrum at hotmail.com
Tue Jun 9 18:32:06 UTC 2009

Well,  it's air conditioning season again and I have been studying the effects of the varying temperature caused by AC cycling on the Thunderbolt output (as reported by the unit's internal monitoring messages).  The attached image shows the effects of the temperature cycling on the Tbolt outputs when the unit is in free air.

Each cycle of the AC produces around a 0.6 degree C change in the Thunderbolt temperature when the Tbolt and power supplies are in free air.  This causes a 6.2 ns RMS shift in the PPS output and 0.046 ppb shift in the oscillator with each temperature cycle.

Covering the Tbolt and the power supply with corrugated cardboard boxes (I used separate boxes for the Tbolt and the power supply) reduces the temperature swing to around 0.17C.  The PPS output shift reduces to around 2.7 ns and the oscillator error to around 0.030 ppb.  The boxes increase the Tbolt internal temperature around 3 degrees C.

Placing the Tbolt in a corrugated cardboard box that was lined with waffled foam rubber reduced the temperature swings to around 0.060 degrees C (and increased the internal temperature by 10 degrees C).   The PPS error reduced to around 2.3 ns and the osc error to around 0.027 ppb.

By separately exposing the power supply to free air and insulating it,  it appears the power supply contributes about 35 percent of the total error.

A well insulated Thunderbolt and power supply produces the lowest amount of error at the expense of increased internal operating temperature (it went over 50C).   A modest amount of insulation (placing the units in uninsulated corrugated boxs) produced almost as much improvement with only a 3C increase in electronics temperature.   

Lauren found her dream laptop. Find the PC that’s right for you.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: cycles.gif
Type: image/gif
Size: 36381 bytes
Desc: not available
URL: <http://www.febo.com/pipermail/time-nuts/attachments/20090609/9845e241/attachment-0001.gif>

More information about the time-nuts mailing list