Hi

For further info on just why the crystal temperature is such a crazy thing to track, check out
Rick’s paper. At first glance it seems like it’s a trivial thing. In reality, gradients are a very
big deal.

Bob

Hi

That paper is the basis for the MCXO. It is an interesting way to do a TCXO.
The drift between the two modes makes it a difficult thing to master in an OCXO.
Plating a pair of electrodes (one pair per mode) is also an approach that has been
tried.

Bob

I have never been able to find a reference to them on the internet but
there was a similar product intended for TO-99 packages that could be
used with operational amplifiers.

On Mon, 5 Jun 2017 08:35:35 +0000 (UTC), you wrote:

Hi

For a while, a couple of outfits made TO-5 and TO-8 “cap heaters” with
PTC material. There are still a few obscure places that people do the same
sort of thing with a mini-pcb based design. In an OCXO design, the gotcha
is matching the PTC oven temperature to the crystal turn. You can do that if you
have a substantial inventory of material and custom fab the oven after the crystal
is bult.

Bob

In message A9E08DB0-2CB4-46EE-BC6F-3F3F9BC65931@n1k.org, Bob kb8tq writes:

Bob, you of all people must be able to answer this:

OCXO's have specified temperature ranges for instance
-40...+70°C for the heavy duty stuff.

But I cant imagine the ovens used are so perfect that they have the
same regulation performance at all temperatures.

I can choose the exterior temperature, which I should prefer ?

Disregard aging of electronics and materials, we all know that
stuff, what I'm interested in is at which exterior temperature
OCXO ovens work best ?

--
Poul-Henning Kamp      | UNIX since Zilog Zeus 3.20
phk@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.

On 6/6/17 4:26 AM, Poul-Henning Kamp wrote:

Here's my guess...

1. you want minimal gradients across the device for a variety of reasons
2. therefore you want least amount of heat from the heater
3. therefore somewhat below the setpoint of the heater.

How far below? I've no idea.

On Tue, 6 Jun 2017 05:52:56 -0700
jimlux jimlux@earthlink.net wrote:

I counter that guess! :-)

1. You want the control loop as stable as possible
2. Stability is directly related to controllability
3. The larger the heat flow, the better the controllability
4. therefore the outside temperature should be as low as possible

Do we have to battle now, to see who is right? :-)

			Attila Kinali

--
It is upon moral qualities that a society is ultimately founded. All
the prosperity and technological sophistication in the world is of no
use without that foundation.
-- Miss Matheson, The Diamond Age, Neil Stephenson

Poul-Henning wrote:

Jim replied:

It depends on what you mean by "best."  "Best" can mean "minimizes the
wander in oven-regulated temperature at a constant (or slowly-changing)
ambient temperature," or it can mean "fastest recovery when the ambient
temperature changes more rapidly."

Minimum heater power tends to favor the first, but be careful -- this
means a low available (maximum) power, not just using a high-powered
heater at a lower output.  This is because the rate of temperature
change for full-scale heater swings is proportional to maximum heater
output, not to the ambient-to-oven differential, and this "granularity"
of the heater control function is what determines the oven wander at a
constant (or slowly-changing) ambient temperature.

A higher differential between the set point and ambient, and higher
maximum heater output, are necessary for fast recovery from larger
and/or faster changes in ambient temperature (i.e., to achieve a higher
dTemp/dTime).

Note that the above is why, for an oven controller with decent loop
gain, it is not necessary to control the ambient temperature very
tightly -- it is only necessary to slow down the rate of change in
ambient temperature to the point that the loop can track it to the
required tolerance.  See previous list discussions of "cast aluminum
boxes" and "thermal capacitance."

Best regards,

Charles

On Tue, Jun 6, 2017 at 7:34 AM, Attila Kinali attila@kinali.ch wrote:

I think you are correct but within reason of course.  It is easy to see
that the extremes can't work. If the internal set point is very close to
ambient the oven is uncontrollable.  because you only use the first bit of
the DAC to control the heater and after a few seconds you have overshoot.
Moving the set point up lets us use the full range of current on the heater
can gives us 8 or 10 bits of control and the rate of change is slow enough
that we have time take thousands of samples and see a rate of change in
temperature.  The PID algorithm needs something that is slow to change
compared to the control loop cycle.  So you want a good size thermal mass
compared to the amount of heat.

At the other extreme, where the set point to far above ambient we would
need to run the heater full time and also loose control.  So I disagree
with #4 above.  The heater would have to run full-on at 100% duty cycle.
(In other words avoid using liquid nitrogen baths)

There is an optimum were it peaks but I don't know how to find it.    Look
at the specific heat of the thermal mass (likely you are using aluminum)
and multiply that by the mass and I think you want that to be large
compared to the heat from a full-on heater so that the rate of change looks
slow compared to your control loop cycle.

--

Chris Albertson
Redondo Beach, California

Hi

As you point out, there really is no answer that is obviously better than all the others.
If you keep the outside warm (say 40C) you will reduce the strain on the heater devices
and likely not degrade the MTBF of anything outside the OCXO by very much.

Bob