[time-nuts] 10811 foam enclosure

Magnus Danielson magnus at rubidium.dyndns.org
Mon Jan 10 01:43:05 UTC 2011

On 10/01/11 00:59, Tom Van Baak wrote:
>> Being a Shera Fan I finally broke down and bought a Tbolt. I
>> experimented  with a foam enclosure with roughly 1 centimeter on all
>> sides. What I found that it only increased the board temperature by 19
>> C. It was still as sensitive to ambient. Thanks to Lady Heater it even
>> shows an increase of 40 mC when I walk up to it with one minute
>> delay.. Changing to an aluminum extrusion the increase is only 7 C and
>> the response is much slower. In its final assembly it will be inside a
>> larger metal mass, I do not think that the higher temperature of 56 C
>> is conducive to longer life. Opinion: putting foam around a 10811 will
>> only give you a warm feeling unless you make it much larger than shown
>> in the picture of of KH6GRT.
> Nice set of experiments. Thanks for doing them and reporting
> the results. I also found that foam insulated enclosures do not
> help with ~hourly HVAC cycling. I reasoned that over the span
> of half an hour there is more than enough time for the delta T
> of the outside world to infect the inside temperature; thus the
> foam does little or no good. If the temperature changes are
> more rapid however the foam might start showing advantage.

This matches my experience. I don't use foam thought, I just use a 
convection barrier of some sort. The oscillator becomes much quieter 
from effects of fairly quick shifts in temperature. This relates to 
temperature gradients of the crystal (which Rick keeps pointing out, to 
seemingly death ears), as the outer shell time-constant now becomes much 
quicker than the time-constant into the outer shell and hence the 
gradient is "shorted" as the heat conduction equalizes the temperature 
quicker than the change occurs. Hence will the gradient over the crystal 
become less distinct and hence gradient shifts occurs. Also, the rate of 
temperature changes comes into the loop bandwidth and less remaining 
error occurs. However, for building heat systems, HVAC and similar a 
simple card-board box does no good by its own.

So, such a simple scheme as a simple low-conductive box can help to 
clean up transients but doesn't do magic for 1000 s and longer periods.

>> Based on the above test results I feel mass is more important. So I
>> took a aluminum enclosed dewar weighing 943 gr. and did first put a
>> resistor in it and heated it up with 2 W. Temp. did rise to 70 C. Next
>> I disassembled my worse 10811, which is very easy, since it is a nice
>> compact unit and inserted it in the dewar. Monitoring temp with my YSI
>> it shows 65.2 C with 24 C ambient. Power dissipation is 1.575 W. Will
>> track it over time before I replace the 10811 with one of my better
>> ones for frequency tests.
>> Questions to our experts: A) will Removing the foam mess with the
>> temp. control loop
> It seems this would change the time constant so the loop may
> no longer be optimized. But by how much is hard to say.

It changes the time-constant for which changes enters the oscillator, 
but doesn't do too much to the actual loop.


More information about the time-nuts mailing list