Making a HP 10811 and similar types better in the real world Part I
GM List,
Basic commandments of precision grade Crystal Oscillators. (1) It is both an art and a science. (2) The devil is in many details. Unbelievers suffer.
Here is the framework of methods that really work. (Assuming a used unit)
-
Make sure your power supplies are stable AND meet the low noise requirements stated in the data sheet. This means using something based on the buried zener principle for both the heater, oscillator and EFC supply using three separate circuits. Hint: filter, filter, and more filters.
-
The EFC voltage should be derived from something like a REF-10.
-
The voltage references of all the supplies should be in a temperature regulated housing.
-
Wrap the oscillator in fiberglass and insert into a wide mouth thermos bottle with the opening downward.
-
Wrap that assembly in another 6 inches or so of fiberglass insulation and still keeping the fiberglass filled mouth downward, place into a wooden box of choice.
-
Place box of choice on concrete floor.
-
Fire up oscillator. Let it cook for at least two weeks.
8 After burn-in period use the EFC voltage to set the frequency to 10MHz (An old Bourns 10 turn linear pot with a counter dial is ideal) using your GPS or Rubidium standard using your favorite comparator. Note dial reading and record.
-
Wait one week and repeat step 8.
-
Wait another week and repeat step 8.
-
After another week reset the frequency and compare the DIFFERENCES in your recorded dial readings. After the differences become close to the same (EFC is non-linear) week after week one has a very stable oscillator that can start to be useful.
Why this works
High stability oscillators do not want to be bumped or have their factory set temperature changed. They are not like a car engine that needs a radiator. The ovens are proportionally controlled. On start-up all ovens, proportional or not, will have over-shoot. Some more, some less. An inescapable fact of life.
Cooling the exterior case of an ovenized crystal oscillator only causes the oven(s) to come on more frequently. When the heater current(s) go to a minimum level, that is when the oven has reached its factory set-point and is a happy camper.
The vacuum flask minimizes heat losses. The interior and fiberglass insulation prevents air current flow and add more insolation.
In part two I’ll expand on the 10811 temperature problem and explain my precision oscillator credentials,
Perry Sandeen wrote:
GM List,
Basic commandments of precision grade Crystal Oscillators. (1) It is both
an art and a science. (2) The devil is in many details. Unbelievers
suffer.
Here is the framework of methods that really work. (Assuming a used unit)
Doing what you describe will result in a very sensitive
humidity sensor, having eliminated the thermometer effect.
In any event, none of this affects crystal aging or
frequency jumps. This is what limits the E1938A which
is hermetic and has a thermal gain 1000 times better than
the 10811.
Rick Karlquist N6RK
Hi
Just how good is the thermos bottle in this case? (as in degrees / watt).
You can get some very good vacuum ones and some pretty poor Styrofoam ones.
Bob
-----Original Message-----
From: time-nuts-bounces@febo.com [mailto:time-nuts-bounces@febo.com] On
Behalf Of Perry Sandeen
Sent: Wednesday, September 21, 2011 4:20 PM
To: time-nuts@febo.com
Subject: [time-nuts] Making a HP 10811 and similar types better in the
realworld Part I
GM List,
Basic commandments of precision grade Crystal Oscillators. (1) It is both an
art and a science. (2) The devil is in many details. Unbelievers suffer.
Here is the framework of methods that really work. (Assuming a used unit)
-
Make sure your power supplies are stable AND meet the low noise
requirements stated in the data sheet. This means using something based on
the buried zener principle for both the heater, oscillator and EFC supply
using three separate circuits. Hint: filter, filter, and more filters. -
The EFC voltage should be derived from something like a REF-10.
-
The voltage references of all the supplies should be in a temperature
regulated housing. -
Wrap the oscillator in fiberglass and insert into a wide mouth thermos
bottle with the opening downward. -
Wrap that assembly in another 6 inches or so of fiberglass insulation
and still keeping the fiberglass filled mouth downward, place into a wooden
box of choice. -
Place box of choice on concrete floor.
-
Fire up oscillator. Let it cook for at least two weeks.
8 After burn-in period use the EFC voltage to set the frequency to 10MHz
(An old Bourns 10 turn linear pot with a counter dial is ideal) using your
GPS or Rubidium standard using your favorite comparator. Note dial reading
and record.
-
Wait one week and repeat step 8.
-
Wait another week and repeat step 8.
-
After another week reset the frequency and compare the DIFFERENCES in
your recorded dial readings. After the differences become close to the same
(EFC is non-linear) week after week one has a very stable oscillator that
can start to be useful.
Why this works
High stability oscillators do not want to be bumped or have their factory
set temperature changed. They are not like a car engine that needs a
radiator. The ovens are proportionally controlled. On start-up all ovens,
proportional or not, will have over-shoot. Some more, some less. An
inescapable fact of life.
Cooling the exterior case of an ovenized crystal oscillator only causes the
oven(s) to come on more frequently. When the heater current(s) go to a
minimum level, that is when the oven has reached its factory set-point and
is a happy camper.
The vacuum flask minimizes heat losses. The interior and fiberglass
insulation prevents air current flow and add more insolation.
In part two I'll expand on the 10811 temperature problem and explain my
precision oscillator credentials,
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Perry Sandeen wrote:
GM List,
..............................................................................The
ovens are proportionally controlled. On start-up all ovens, proportional or not,
will have over-shoot. Some more, some less. An inescapable fact of life.
Imagine that the set point is variable, and can be set below the desired
temperature. Then imagine that the set point can approach the desired
temperature more closely as it gets closer to the desired temperature.
... And you will have discovered (100 years late) the PID controller.
PID controllers do not have to overshoot the desired temperature. It is
not an inescapable fact of life. 30 years ago I was designing PID controllers,
with a little microprocessor magic, that could quickly arrive at the set point
temperature and never, I repeat, never, exceed that temperature. Someone's
internal organs would have become toast if it did.
-Chuck Harris
On 09/22/2011 12:30 AM, Chuck Harris wrote:
Perry Sandeen wrote:
GM List,
..............................................................................The
ovens are proportionally controlled. On start-up all ovens,
proportional or not,
will have over-shoot. Some more, some less. An inescapable fact of life.
Imagine that the set point is variable, and can be set below the desired
temperature. Then imagine that the set point can approach the desired
temperature more closely as it gets closer to the desired temperature.
... And you will have discovered (100 years late) the PID controller.
PID controllers do not have to overshoot the desired temperature. It is
not an inescapable fact of life. 30 years ago I was designing PID
controllers,
with a little microprocessor magic, that could quickly arrive at the set
point
temperature and never, I repeat, never, exceed that temperature. Someone's
internal organs would have become toast if it did.
Overshot is fairly easy to avoid for a well controlled PID loop simply
by setting the damping factor properly.
PIDs is nice in that you can control loop bandwidth and damping factor
fairly well. Overshot properties vs. damping factor is a well researched
field and already tabulated before I was born.
Cheers,
Magnus
Hi
Actually, overshoot is pretty easy to eliminate on a conventional OCXO by picking a good location for the thermistor. The heater will always run "hot", but the rest of the stuff does not have to.
Bob
On Sep 21, 2011, at 6:37 PM, Magnus Danielson wrote:
On 09/22/2011 12:30 AM, Chuck Harris wrote:
Perry Sandeen wrote:
GM List,
..............................................................................The
ovens are proportionally controlled. On start-up all ovens,
proportional or not,
will have over-shoot. Some more, some less. An inescapable fact of life.
Imagine that the set point is variable, and can be set below the desired
temperature. Then imagine that the set point can approach the desired
temperature more closely as it gets closer to the desired temperature.
... And you will have discovered (100 years late) the PID controller.
PID controllers do not have to overshoot the desired temperature. It is
not an inescapable fact of life. 30 years ago I was designing PID
controllers,
with a little microprocessor magic, that could quickly arrive at the set
point
temperature and never, I repeat, never, exceed that temperature. Someone's
internal organs would have become toast if it did.
Overshot is fairly easy to avoid for a well controlled PID loop simply by setting the damping factor properly.
PIDs is nice in that you can control loop bandwidth and damping factor fairly well. Overshot properties vs. damping factor is a well researched field and already tabulated before I was born.
Cheers,
Magnus
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
That is not always as easy as it sounds.
The thermal equivalent of a "rigid body" does not exist. If you apply heat
to a block of metal at one end, it takes a while for it to propagate to
the other end. In fact, a long thin rod looks a lot more like a
transmission line than an isothermal block.
This matters because if you try and increase the loop gain, the wrap-up of
the phase shift soon reaches 180 degrees, and the thing becomes unstable
as negative FB at low frequencies becomes positive FB at higher
frequencies.
-John
===============
Hi
Actually, overshoot is pretty easy to eliminate on a conventional OCXO by
picking a good location for the thermistor. The heater will always run
"hot", but the rest of the stuff does not have to.
Bob
On Sep 21, 2011, at 6:37 PM, Magnus Danielson wrote:
On 09/22/2011 12:30 AM, Chuck Harris wrote:
Perry Sandeen wrote:
GM List,
..............................................................................The
ovens are proportionally controlled. On start-up all ovens,
proportional or not,
will have over-shoot. Some more, some less. An inescapable fact of
life.
Imagine that the set point is variable, and can be set below the
desired
temperature. Then imagine that the set point can approach the desired
temperature more closely as it gets closer to the desired temperature.
... And you will have discovered (100 years late) the PID controller.
PID controllers do not have to overshoot the desired temperature. It is
not an inescapable fact of life. 30 years ago I was designing PID
controllers,
with a little microprocessor magic, that could quickly arrive at the
set
point
temperature and never, I repeat, never, exceed that temperature.
Someone's
internal organs would have become toast if it did.
Overshot is fairly easy to avoid for a well controlled PID loop simply
by setting the damping factor properly.
PIDs is nice in that you can control loop bandwidth and damping factor
fairly well. Overshot properties vs. damping factor is a well researched
field and already tabulated before I was born.
Cheers,
Magnus
Hi
If the thermistor is on top of the heater, the heat will cut back before anything inside the oven overshoots on warmup.
Bob
On Sep 21, 2011, at 7:02 PM, J. Forster wrote:
That is not always as easy as it sounds.
The thermal equivalent of a "rigid body" does not exist. If you apply heat
to a block of metal at one end, it takes a while for it to propagate to
the other end. In fact, a long thin rod looks a lot more like a
transmission line than an isothermal block.
This matters because if you try and increase the loop gain, the wrap-up of
the phase shift soon reaches 180 degrees, and the thing becomes unstable
as negative FB at low frequencies becomes positive FB at higher
frequencies.
-John
===============
Hi
Actually, overshoot is pretty easy to eliminate on a conventional OCXO by
picking a good location for the thermistor. The heater will always run
"hot", but the rest of the stuff does not have to.
Bob
On Sep 21, 2011, at 6:37 PM, Magnus Danielson wrote:
On 09/22/2011 12:30 AM, Chuck Harris wrote:
Perry Sandeen wrote:
GM List,
..............................................................................The
ovens are proportionally controlled. On start-up all ovens,
proportional or not,
will have over-shoot. Some more, some less. An inescapable fact of
life.
Imagine that the set point is variable, and can be set below the
desired
temperature. Then imagine that the set point can approach the desired
temperature more closely as it gets closer to the desired temperature.
... And you will have discovered (100 years late) the PID controller.
PID controllers do not have to overshoot the desired temperature. It is
not an inescapable fact of life. 30 years ago I was designing PID
controllers,
with a little microprocessor magic, that could quickly arrive at the
set
point
temperature and never, I repeat, never, exceed that temperature.
Someone's
internal organs would have become toast if it did.
Overshot is fairly easy to avoid for a well controlled PID loop simply
by setting the damping factor properly.
PIDs is nice in that you can control loop bandwidth and damping factor
fairly well. Overshot properties vs. damping factor is a well researched
field and already tabulated before I was born.
Cheers,
Magnus
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.