[time-nuts] Thunderbolt stability and ambient temperature

Ulrich Bangert df6jb at ulrich-bangert.de
Thu Jun 11 10:46:13 UTC 2009


> For PLL steered devices, you want your device enclosed by a 
> thermal mass which is again enclosed by a layer of thermal 
> isolation.  The goal is to filter/average all rapid (daily ?) 
> external temperature influences, only letting through such 
> slow variations (seasonal ?) which the PLL can comfortably cope with.

My impression with this thread is that two sligthly different things are
being discussed here: 

Some posters explain how a crystal oven shall be designed as Poul-Henning
does with his words above. Absolute correct, no doubt about it!

Nevertheless only a VERY few of us (Rick for example) have real experience
in designing ovens and most of us are confronted with a existent ready to go
OCXOs. That is why one can start a discussion about the question whether the
thermal properties of an existing OCXO can be improved which is what the
second group of posters has done starting with the Thunderbolt.

The situation with the ready to go OCXO is a bit different than starting
from scratch: The OCXO DOES already have a certain thermal mass inside and
it DOES already have a certain thermal insulation of this mass against the
ambient. 

BOTH of these parameters have been important properties for the designer of
the oven's temperature controller. The temperature controller's properties
must match EXACTLY the thermal mass as well as the amount of insulation in
order to work as expected. If one of you has personal experience with high
precision temperature controllers I am sure he can second this claim. The
situation gets the harder the better the thermal insulation is: You need to
avoid every kind of regulation overshot because it is only the high
insulation that allows energy to leak out of the thermal mass. Which in turn
requires high regulator time constants. 

By the way: That is also the reason for the heat sinks on rubidium
standards. They form a good coupling to the ambient (=small insulation)
making the design of the lamp's temperature controller much more easy and
much faster than with a good insulation.

With a ready OCXO in your hand like the 10811 or the FTS1200 or anything
else: You can't improve the beast by "simply better insulation" or by
"simply higher thermal mass", because every change in this will have worse
impacts on the action of the temperature controller.

However, there ARE ways to improve the thermal behavour of an ready to go
OCXO:

First rule: Allow the temperature controller to "see" the original thermal
mass, i.e. don't open the enclosure and don't change anything, just leave
everything as it is.

Second rule: Allow the temperature controller to "see" the original thermal
insulation, i.e. allow for some cubic centimeters of air around the OCXO's
outer enclosure. It was engineered with that surrounding in mind!

Third rule: Outside of the these few cubic centimeters of air around the
OCXO you are allowed ANYTHING.

You can for example put a big thermal mass like a massive aluminium
enclosure around the OCXO (but please don't forget the few cubic centimeters
of air between this and the surface of the OCXO). THIS thermal mass in
conjunction with the air around it will work as the thermal lowpass filter
that some posters talked about but WITHOUT worrying the OCXO's temperature
controller because it "sees" the original thermal mass and the original
thermal insulation. But it "sees" smaller ambient temperature variations
than before. 

This would be a passive solution where the aim were to get high time
constants for the combination of the outer enclosure and the surrounding air
making as much thermal mass as possible the prefered way. One of my 10811
resides in a 12 X 12 X 12 cm aluminium enclosure with 2 cm wall thickness.
This has given a thermal time constant in the order of precious few hours if
i remember correctly and was far from averaging daily temperature changes.
It might however be helpful for the discussed problem of Thunderbolt
reactions to short time temperature changes.

Or one can decide for a active solution. This would call for an outer
enclosure where not high thermal mass but high thermal conductivity were the
aim which is brought to a constant temperature by means of a second
temperature controller, the famous "double oven" principle if you like. As
in the passive case there must be enough air between OCXO and the outer
controller in order to avoid any interaction between the temperature
controllers.

Best regards
Ulrich Bangert

   

> -----Ursprungliche Nachricht-----
> Von: time-nuts-bounces at febo.com 
> [mailto:time-nuts-bounces at febo.com] Im Auftrag von Poul-Henning Kamp
> Gesendet: Donnerstag, 11. Juni 2009 00:43
> An: Discussion of precise time and frequency measurement
> Betreff: Re: [time-nuts] Thunderbolt stability and ambient temperature
> 
> 
> In message <4A3033FC.6070806 at xtra.co.nz>, Bruce Griffiths writes:
> >Hal Murray wrote:
> 
> >Thermal wave reflection at boundaries/interfaces does occur:
> 
> If your frequency or voltage standard is in a physical 
> environment where these effects are relevant, you have much 
> bigger problems to deal with before thermal reflection 
> becomes your number one priority :-)
> 
> Lets stay real here.
> 
> Summary:
> 
> For PLL steered devices, you want your device enclosed by a 
> thermal mass which is again enclosed by a layer of thermal 
> isolation.  The goal is to filter/average all rapid (daily ?) 
> external temperature influences, only letting through such 
> slow variations (seasonal ?) which the PLL can comfortably cope with.
> 
> Even if you use active temperature control, peltier or 
> otherwise, it is still a good idea to employ a thermal mass 
> to cope gracefully with power-failuers and other equipment glitches.
> 
> Poul-Henning
> 
> PS: The thermal mass need not be solid blocks of metal, 
> regular ceramic bricks or tiles work fine.
> 
> -- 
> Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
> phk at FreeBSD.ORG         | TCP/IP since RFC 956
> FreeBSD committer       | BSD since 4.3-tahoe    
> Never attribute to malice what can adequately be explained by 
> incompetence.
> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to 
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.




More information about the time-nuts mailing list