On 6/7/2017 1:09 PM, Attila Kinali wrote:

Actually, the oven mass of the E1938A is much heavier than the
10811 and is made of copper to boot.

The crystal case is well connected to the oven mass and gets
heated by conduction.  I don't think radiation is a player.

Rick N6RK

Hi

The crystal enclosure may (or may not … who knows ..) be back filled with a low level
of helium. It does not take much of a backfill for conduction inside the crystal holder to
dominate the heat transfer vs radiation transfer.

Bob

Hoi Rick,

On Wed, 7 Jun 2017 14:21:52 -0700
"Richard (Rick) Karlquist" richard@karlquist.com wrote:

[...]

Hmm.. seems like I have misunderstood the design.
I think I should aquire one of the non-working E1938 and
pry the can open.

Does anyone have a E1938 that's not working anymore
and would be willing to offer it to science? :-)

		Attila Kinali

--
You know, the very powerful and the very stupid have one thing in common.
They don't alters their views to fit the facts, they alter the facts to
fit the views, which can be uncomfortable if you happen to be one of the
facts that needs altering.  -- The Doctor

Hi

Keep in mind that on a per cubic inch basis, copper is right up there
for creating a thermal mass ….

Bob

On Wed, 7 Jun 2017 14:21:52 -0700
"Richard (Rick) Karlquist" richard@karlquist.com wrote:

Do I interpret the papers correctly, that the oven mass is
a closed can, with the crystal holder "molded" into it?
Then wrapped around it are the face and the rim heaters,
and outside those comes the insulation and the outer can?

		Attila Kinali

--
You know, the very powerful and the very stupid have one thing in common.
They don't alters their views to fit the facts, they alter the facts to
fit the views, which can be uncomfortable if you happen to be one of the
facts that needs altering.  -- The Doctor

Yes, that's right.  The copper oven mass
has two pieces:  a main piece and a lid.
The main piece has a wall around the outside
into which the cover fits, using an O-ring.
In the center, there is a cylindrical cavity
into which the crystal mounts.  It is like
a 10811 crystal, except the height is reduced.
In retrospect, there is no reason why the
10811 crystal shouldn't have been that size.
It is potted into the cavity, therefore, it
doesn't use the famous flat head 4-40 screw
welded to the can as in the 10811.  Thermistors
are potted into holes drilled in the cavity
wall.  The lid has screws that screw into
the cavity wall.

The main piece holds a circular PC board
containing the bridge oscillator circuit.
It has a big hole in the middle for the
crystal cavity.  The production people
immediately named it the "donut board."
And the finished oscillator is the "hockey
puck".

Summarizing, the crystal is very well thermally
connected to the copper oven mass.

There are 3 flex circuit heaters.  One for
the lid, one for the other face of the
main piece, and one that goes around the
outside.  The two face heaters are operated
together as one heater.  The ratio of heat
to the faces vs the outside rim is adjusted
for maximum thermal gain.  The rim heater is
the difference between under 1,000 gain and
over 1,000,000 gain.

We were stuck under 1,000 for a long time
using only face heaters.  I still remember the
day that I rigged up the first crude rim header
by winding a piece of magnet wire around the rim
and holding it in place with 5 min epoxy.  I
didn't know what else to try.  This seemed like
a Hail Mary play at the time, until I
measured the gain.  We instantly went to gains
above 20,000.  It seemed high at the time but
it was only the beginning.

The exact insulation is relatively unimportant.
We even tried still air using a knife edge
cradle.  Didn't make much difference.

Rick N6RK

On 6/8/2017 1:27 PM, Attila Kinali wrote:

On Thu, 8 Jun 2017 16:17:20 -0700
"Richard (Rick) Karlquist" richard@karlquist.com wrote:

What is a knife edge cradle?

One big question remains: How did you set the ratio
between face and rim heater? Was it determined at design
time and then just set the same for all units or was
it part of the production test?

		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

On 6/11/2017 8:59 AM, Attila Kinali wrote:

We wanted to test still air as insulation.
We couldn't just replace the insulation with
"nothing".  There had to be some kind of mechanical
support to keep the oven mass suspended inside the
outer case.  Our ME built a skeleton framework
made of plastic to support it.  In order to minimize
conduction thru the plastic, he designed in knife
edges where the plastic came into contact with the
oven mass.

During proof of concept phase in R&D, I peaked up the
thermal gain on each unit by trial and error.  I could
usually get a run with positive thermal gain, then
increment the ratio, and get a run with negative
thermal gain.  I could then interpolate to get the
ratio that should give "infinite" gain.  Maybe one
or two more runs after that would get me to where
the gain passed through infinity at some ambient
temperature, and was well into the millions over the
whole range.  At extremely high thermal gain, the
gain is not constant over ambient temperature.

I collected data on a number of units and then used
the average ratio as the production setting.  I
checked production units from time to time and they
typically ran well in the 100's of thousands for gain.

This compromise was workable because we individually
programmed the oven set point to the exact turnover
temperature.  This is a lot easier because it doesn't
require environmental chamber runs.  The E1983A software that
I "leaked" to the time-nuts community I believe has a
command that can be used to search for the turnover.

Rick N6RK

Hi

I certainly saw the “positive gain at this setting” going to “negative gain at that setting” result on a lot
of OCXO designs. I never had the patience (or a stable enough system) to get into the millions
or even 100K’s on a single oven.  As a practical result, a gain of -500 is not really any better or worse
than a gain of +500. It can be a bit confusing to set up though ….

Bob