cesium clocks..
Jim,
If you're searching Ebay, just make sure you don't accidentally buy a
Nike brand of shoe...
Jim Palfreyman
Quoting Jim Palfreyman jim77742@gmail.com, on Thu 19 Jun 2008
09:27:12 PM PDT:
Jim,
If you're searching Ebay, just make sure you don't accidentally buy a
Nike brand of shoe...
Jim Palfreyman
Well... in fact, that's sort of the question I'm asking. if one does
search Ebay, what's an effective search term?
Leaving aside the $60 Nike Cesiums...
Maybe try
hp (5071a,5061a,5061b)
... to start with.
The 5071as seem more desirable but they're also usually more expensive, and
(the really scary part, true of all of them): how do you tell how much tube
life is left?
-- john, KE5FX
-----Original Message-----
From: time-nuts-bounces@febo.com [mailto:time-nuts-bounces@febo.com]On
Behalf Of Jim Lux
Sent: Thursday, June 19, 2008 9:44 PM
To: Discussion of precise time and frequency measurement; Jim Palfreyman
Cc: time-nuts@febo.com
Subject: Re: [time-nuts] cesium clocks..
Quoting Jim Palfreyman jim77742@gmail.com, on Thu 19 Jun 2008
09:27:12 PM PDT:
Jim,
If you're searching Ebay, just make sure you don't accidentally buy a
Nike brand of shoe...
Jim Palfreyman
Well... in fact, that's sort of the question I'm asking. if one does
search Ebay, what's an effective search term?
Leaving aside the $60 Nike Cesiums...
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.
In message: PKEGJHPHLLBACEOICCBJOECPINAB.jmiles@pop.net
"John Miles" jmiles@pop.net writes:
: Maybe try
:
: hp (5071a,5061a,5061b)
:
: ... to start with.
:
: The 5071as seem more desirable but they're also usually more expensive, and
: (the really scary part, true of all of them): how do you tell how much tube
: life is left?
Beam Tube Voltage?
Warner
Well... in fact, that's sort of the question I'm asking. if one does
search Ebay, what's an effective search term?
Leaving aside the $60 Nike Cesiums...
Were you aware that you can add terms in the search box preceded by a
negative sign to exclude items that match that? ie "cesium -nike"
(without the quotes) should eliminate the multiple shoes listings.
The British spelling adds an A (caesium), so you may want to include that.
Choosing just the HP numbers (5071, 5061) would exclude other
manufacturers like FTS and Oscilloquartz, so perhaps a generic
cesium/cesium is better.
As an aside, and if I may be permitted a small advertisement, I have
an FTS4060/S24 which I would like to sell. I picked it up off ebay a
couple of years ago, and whilst it works, the tube is very near the
end of its life. It really is built like a battleship - very heavy.
If anyone is interested, please contact me off list
(pvince@theiet.org).
Regards,
Peter Vince (G8ZZR, London, England)
Having just gone through the process of finding, acquiring, and fixing
an HP 5061A, here are a few pointers for qualifying one for purchase.
The biggest problem you're likely to face is a beam tube that's at
end-of-life. If there is an electronics failure (like mine, read on),
these clocks are really fairly simple to troubleshoot and fix; almost
everything is discrete components; the main circuitry is mostly analog.
If the seller says the unit locks and goes into continuous operation
mode, the quickest check you can ask them for is to report the beam
current. It should be above 15. However, note that this is a relative
measurement. There is a meter adjustment control that sets the meter
sensitivity, and it's possible it's just not properly set. So, a low
beam current isn't an absolute failure indication. With low beam
current, if it locks, then you can still have a working clock. The
primary impact of low current is more noise in the signal, which leads
to greater short-term frequency variation. Even with that, it's still
going to be in the 10e-10 or 10e-11 range. If you average, over time the
accuracy will be about as good as a newer tube.
If it goes into continuous operation for a while but then loses lock,
you're taking your chances. The tube could be so depleted that it can't
maintain lock, or the clock could just not be adjusted properly, or you
could have an electronics failure. A quick check is to ask the seller
for the ion pump current reading. If it's not less than 10, then the
problem could just be that the unit has been sitting around for too long
without the ion pump having been run. This is curable just by having it
run for a few days, or in extreme cases, using an external 3500v 5 ma
supply to run the pump more energetically than the clock itself can.
In any case, if you're feeling ambitious and can get a good price ($500
or less?), give it a shot.
Finally, if it won't lock at all, then either the tube is gone, there is
an electronics problem, the clock is way out of alignment, or the ion
pump hasn't run in a long time. Have the seller report the ion pump
current. If it's over 10, then you might want to take your chances if
you get a good price. If it's less than 10, buy it if you like a
challenge and can get a good deal.
I got mine for $300. It wouldn't lock. The ion pump current was high, so
I decided to give it a shot. I almost got lucky. After running for a few
days, the pump current went to zero, which is good. But, the clock would
only lock for a second or so, then lose lock. After a bit of testing, I
found that the crystal oven had fried itself and some wiring inside the
can. (The design is really stupid; can't imagine why it was packaged the
way it was) Anyway, I rebuilt the oven, fired it up, and now have a
nicely-working clock that locks, stays locked, gives a nice 20 reading
on beam current, and has a high-output tube. BTW, the tube is 25 years
old! ('82).
Finally, if you do need to troubleshoot and align the clock, you can
easily get by with a good ac/dc DVM (10 meg or higher impedance) a 100
Mhz scope, and a reasonably good counter, one that can reliably read 12
Mhz to 1ppm).
If anyone wants any more tips or info, feel free to ask, and good luck!
Bill
--
Bill Ezell
They said 'Windows or better'
so I used Linux.
Careful how you toss around the 'stupid' when discussing oscillator design.
Some folks still have a close relationship to those old 00105-6xxx
oscillators.
Dave
----- Original Message -----
From: "wje" wje@quackers.net
To: time-nuts@febo.com
Sent: Friday, June 27, 2008 4:32 AM
Subject: [time-nuts] cesium clocks..
Having just gone through the process of finding, acquiring, and fixing
an HP 5061A, here are a few pointers for qualifying one for purchase.
The biggest problem you're likely to face is a beam tube that's at
end-of-life. If there is an electronics failure (like mine, read on),
these clocks are really fairly simple to troubleshoot and fix; almost
everything is discrete components; the main circuitry is mostly analog.
If the seller says the unit locks and goes into continuous operation
mode, the quickest check you can ask them for is to report the beam
current. It should be above 15. However, note that this is a relative
measurement. There is a meter adjustment control that sets the meter
sensitivity, and it's possible it's just not properly set. So, a low
beam current isn't an absolute failure indication. With low beam
current, if it locks, then you can still have a working clock. The
primary impact of low current is more noise in the signal, which leads
to greater short-term frequency variation. Even with that, it's still
going to be in the 10e-10 or 10e-11 range. If you average, over time the
accuracy will be about as good as a newer tube.
If it goes into continuous operation for a while but then loses lock,
you're taking your chances. The tube could be so depleted that it can't
maintain lock, or the clock could just not be adjusted properly, or you
could have an electronics failure. A quick check is to ask the seller
for the ion pump current reading. If it's not less than 10, then the
problem could just be that the unit has been sitting around for too long
without the ion pump having been run. This is curable just by having it
run for a few days, or in extreme cases, using an external 3500v 5 ma
supply to run the pump more energetically than the clock itself can.
In any case, if you're feeling ambitious and can get a good price ($500
or less?), give it a shot.
Finally, if it won't lock at all, then either the tube is gone, there is
an electronics problem, the clock is way out of alignment, or the ion
pump hasn't run in a long time. Have the seller report the ion pump
current. If it's over 10, then you might want to take your chances if
you get a good price. If it's less than 10, buy it if you like a
challenge and can get a good deal.
I got mine for $300. It wouldn't lock. The ion pump current was high, so
I decided to give it a shot. I almost got lucky. After running for a few
days, the pump current went to zero, which is good. But, the clock would
only lock for a second or so, then lose lock. After a bit of testing, I
found that the crystal oven had fried itself and some wiring inside the
can. (The design is really stupid; can't imagine why it was packaged the
way it was) Anyway, I rebuilt the oven, fired it up, and now have a
nicely-working clock that locks, stays locked, gives a nice 20 reading
on beam current, and has a high-output tube. BTW, the tube is 25 years
old! ('82).
Finally, if you do need to troubleshoot and align the clock, you can
easily get by with a good ac/dc DVM (10 meg or higher impedance) a 100
Mhz scope, and a reasonably good counter, one that can reliably read 12
Mhz to 1ppm).
If anyone wants any more tips or info, feel free to ask, and good luck!
Bill
--
Bill Ezell
They said 'Windows or better'
so I used Linux.
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.
Yes, but my comment is rather specific... my oscillator failed because
the heater pass transistor shorted.
This sent the oven heater into full-on. The overtemp sensor is far
removed from the heater. There's a ribbon cable between the driver board
and the temp bridge sensor board that runs directly over the heater, not
outside the oven insulation. Net result, transistor shorts, cable fries
before overtemp fuse opens. There's no reason the cable should be inside
the oven insulation.
My rebuild fix - (after removing all the carbonized foam insulation)
refoam the oven, replacing the cable and transistor, and moving it
outside the foam!
While I respect in general the brilliance of HP engineers, my classmate
was one, this isn't one of the more intelligent decisions.
Bill Ezell
They said 'Windows or better'
so I used Linux.
Dave 'SqueezeBox' Carlson wrote:
Careful how you toss around the 'stupid' when discussing oscillator design.
Some folks still have a close relationship to those old 00105-6xxx
oscillators.
Dave
----- Original Message -----
From: "wje" wje@quackers.net
To: time-nuts@febo.com
Sent: Friday, June 27, 2008 4:32 AM
Subject: [time-nuts] cesium clocks..
Having just gone through the process of finding, acquiring, and fixing
an HP 5061A, here are a few pointers for qualifying one for purchase.
The biggest problem you're likely to face is a beam tube that's at
end-of-life. If there is an electronics failure (like mine, read on),
these clocks are really fairly simple to troubleshoot and fix; almost
everything is discrete components; the main circuitry is mostly analog.
If the seller says the unit locks and goes into continuous operation
mode, the quickest check you can ask them for is to report the beam
current. It should be above 15. However, note that this is a relative
measurement. There is a meter adjustment control that sets the meter
sensitivity, and it's possible it's just not properly set. So, a low
beam current isn't an absolute failure indication. With low beam
current, if it locks, then you can still have a working clock. The
primary impact of low current is more noise in the signal, which leads
to greater short-term frequency variation. Even with that, it's still
going to be in the 10e-10 or 10e-11 range. If you average, over time the
accuracy will be about as good as a newer tube.
If it goes into continuous operation for a while but then loses lock,
you're taking your chances. The tube could be so depleted that it can't
maintain lock, or the clock could just not be adjusted properly, or you
could have an electronics failure. A quick check is to ask the seller
for the ion pump current reading. If it's not less than 10, then the
problem could just be that the unit has been sitting around for too long
without the ion pump having been run. This is curable just by having it
run for a few days, or in extreme cases, using an external 3500v 5 ma
supply to run the pump more energetically than the clock itself can.
In any case, if you're feeling ambitious and can get a good price ($500
or less?), give it a shot.
Finally, if it won't lock at all, then either the tube is gone, there is
an electronics problem, the clock is way out of alignment, or the ion
pump hasn't run in a long time. Have the seller report the ion pump
current. If it's over 10, then you might want to take your chances if
you get a good price. If it's less than 10, buy it if you like a
challenge and can get a good deal.
I got mine for $300. It wouldn't lock. The ion pump current was high, so
I decided to give it a shot. I almost got lucky. After running for a few
days, the pump current went to zero, which is good. But, the clock would
only lock for a second or so, then lose lock. After a bit of testing, I
found that the crystal oven had fried itself and some wiring inside the
can. (The design is really stupid; can't imagine why it was packaged the
way it was) Anyway, I rebuilt the oven, fired it up, and now have a
nicely-working clock that locks, stays locked, gives a nice 20 reading
on beam current, and has a high-output tube. BTW, the tube is 25 years
old! ('82).
Finally, if you do need to troubleshoot and align the clock, you can
easily get by with a good ac/dc DVM (10 meg or higher impedance) a 100
Mhz scope, and a reasonably good counter, one that can reliably read 12
Mhz to 1ppm).
If anyone wants any more tips or info, feel free to ask, and good luck!
Bill
wje wrote:
Yes, but my comment is rather specific... my oscillator failed because
the heater pass transistor shorted.
This sent the oven heater into full-on. The overtemp sensor is far
removed from the heater. There's a ribbon cable between the driver board
and the temp bridge sensor board that runs directly over the heater, not
outside the oven insulation. Net result, transistor shorts, cable fries
before overtemp fuse opens. There's no reason the cable should be inside
the oven insulation.
My rebuild fix - (after removing all the carbonized foam insulation)
refoam the oven, replacing the cable and transistor, and moving it
outside the foam!
While I respect in general the brilliance of HP engineers, my classmate
was one, this isn't one of the more intelligent decisions.
Bill Ezell
Bill
The usual reason for running temperature sensor leads over the oven
inside the insulation is to thermally shunt them to the oven reducing
heat transfer to the temperature sensor via the wiring.
Your modification may reduce the oven temperature stability significantly.
Bruce
In this case, the temp thermistor bridge is outside the oven cavity
itself. The cable only passes power and the already-processed bridge
delta to the heater power amp. So, there's no particular benefit from
having the cable stuck to the heater wrap. (at least, I think so; my
basic failure was because the cable fried and shorted power to ground)
Bill Ezell
They said 'Windows or better'
so I used Linux.
Bruce Griffiths wrote:
wje wrote:
Yes, but my comment is rather specific... my oscillator failed because
the heater pass transistor shorted.
This sent the oven heater into full-on. The overtemp sensor is far
removed from the heater. There's a ribbon cable between the driver board
and the temp bridge sensor board that runs directly over the heater, not
outside the oven insulation. Net result, transistor shorts, cable fries
before overtemp fuse opens. There's no reason the cable should be inside
the oven insulation.
My rebuild fix - (after removing all the carbonized foam insulation)
refoam the oven, replacing the cable and transistor, and moving it
outside the foam!
While I respect in general the brilliance of HP engineers, my classmate
was one, this isn't one of the more intelligent decisions.
Bill Ezell
Bill
The usual reason for running temperature sensor leads over the oven
inside the insulation is to thermally shunt them to the oven reducing
heat transfer to the temperature sensor via the wiring.
Your modification may reduce the oven temperature stability significantly.
Bruce
time-nuts mailing list -- [1]time-nuts@febo.com
To unsubscribe, go to [2]https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
References
wje wrote:
In this case, the temp thermistor bridge is outside the oven cavity
itself. The cable only passes power and the already-processed bridge
delta to the heater power amp. So, there's no particular benefit from
having the cable stuck to the heater wrap. (at least, I think so; my
basic failure was because the cable fried and shorted power to ground)
Bill Ezell
Bill
If the temperature bridge is outside the oven cavity then its critical
that the temperature sensor leads are thermally shunted to the oven.
Substituting constantan wire for copper wire also helps as the thermal
conductivity of constantant is significantly lower than that of copper.
For example if the temperature sensor has a thermal resistance of 1K/W
to the oven and the leads have a thermal resistance of 100K/W to ambient
then ambient temperature fluctuations of 10 K will induce temperature
sensor temperature variations of 0.1K which the oven controller will
correct by varying the oven temperature by 0.1K.
The lead themal resistance would have to be > 1E4K/W to maintain oven
temperature fluctuations below 1mK when the ambient temperature varies
by 10K.
Such a high thermal resistance is difficult to achieve.
The thermal resistance of a length of wire can be estimated by measuring
its electrical resistance and dividing it by the product of the thermal
conductivity and electrical resistivity.
For copper wire thermal resistance ~ 1.6E5 x Electrical resistance
For 10 cm of 20swg Cu wire the electrical resistance (at 20C) is about
4.4 milliohms and the corresponding thermal resistance is about 700 W/K.
Bruce
Hi Bruce,
You would have to be careful with your constantin wire as there is a
thermocouple
junction of 40 microvolts/K at each copper/constantin connection.
If the pairs of junctions are kept together thermally all cancels out.
Stainless steel is also very non-conductive of heat, but would have
similar
thermocouple problems.
cheers, Neville Michie
On 28/06/2008, at 12:26 PM, Bruce Griffiths wrote:
wje wrote:
In this case, the temp thermistor bridge is outside the oven
cavity
itself. The cable only passes power and the already-processed
bridge
delta to the heater power amp. So, there's no particular
benefit from
having the cable stuck to the heater wrap. (at least, I think
so; my
basic failure was because the cable fried and shorted power to
ground)
Bill Ezell
Bill
If the temperature bridge is outside the oven cavity then its critical
that the temperature sensor leads are thermally shunted to the oven.
Substituting constantan wire for copper wire also helps as the thermal
conductivity of constantant is significantly lower than that of
copper.
For example if the temperature sensor has a thermal resistance of 1K/W
to the oven and the leads have a thermal resistance of 100K/W to
ambient
then ambient temperature fluctuations of 10 K will induce temperature
sensor temperature variations of 0.1K which the oven controller will
correct by varying the oven temperature by 0.1K.
The lead themal resistance would have to be > 1E4K/W to maintain oven
temperature fluctuations below 1mK when the ambient temperature varies
by 10K.
Such a high thermal resistance is difficult to achieve.
The thermal resistance of a length of wire can be estimated by
measuring
its electrical resistance and dividing it by the product of the
thermal
conductivity and electrical resistivity.
For copper wire thermal resistance ~ 1.6E5 x Electrical resistance
For 10 cm of 20swg Cu wire the electrical resistance (at 20C) is about
4.4 milliohms and the corresponding thermal resistance is about 700
W/K.
Bruce
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.
Neville Michie wrote:
Hi Bruce,
You would have to be careful with your constantin wire as there is a
thermocouple
junction of 40 microvolts/K at each copper/constantin connection.
If the pairs of junctions are kept together thermally all cancels out.
Stainless steel is also very non-conductive of heat, but would have
similar
thermocouple problems.
cheers, Neville Michie
Neville
Yes, not a great problem if used in matching pairs.
The room temperature junctions will be most problematic as their
temperature swing is much wider than those at the oven temperature
(provided that the oven controller is working properly)
Stainless is trickier to solder than constantan.
Welding may be preferable.
Using an AC bridge (preferably a transformer bridge) or equivalently a
dc bridge using periodic bridge supply reversal eliminates the effect of
slowly varying thermojunction voltages.
Some of the sigma delta instrumentation ADC's allow this, however one
has to be careful to ensure that the sensor self heating doesn't vary
too much when reversing the bridge excitation polarity.
Bruce
The following paper indicates how to optimise the voltage drops in
electrical leads for both cryogenic and other applications:
http://arxiv.org/pdf/cond-mat/0508697.pdf
Bruce
On 28/06/2008, at 1:14 PM, Bruce Griffiths wrote:
Stainless is trickier to solder than constantan.
Welding may be preferable.
A hint for soft soldering stainless steel, iron, nickel, chromium,
copper, brass, nichrome etc. but not aluminium.
Apply a very small amount of phosphoric acid to the clean metal when
COLD, then solder with a soldering iron and pure solder.
The metal will then solder as easily as clean copper.
The flux residue is very water soluble and washes off.
The dilute acid is not a problem but the hot concentrated flux
residue (mainly phosphorous pentoxide) is corrosive to fingers and
clothing and should be treated with respect until dissolved in water.
If you apply the phosphoric acid and heat the metal for a minute or
so before soldering then some metals ( mainly ferrous) will develop a
passive
phosphate coating as in the Parkerizing process and you will never
solder it. You will need to clean the metal
with abrasive paper until bright.
Quite often I "tin" the item with the phosphoric acid as a flux, then
wash it clean, then later it can
be soldered with resin cored solder with other components. The
phosphoric acid and resin will
not act together as a flux but make a mess.
Not recommended for children.
Just in case you were interested,
Neville Michie
G'day fellow time-nuts.
I was just going through some old projects and found my old home-made
mass spectrometer. This was a project based on a design from either
Scientific American or The Amateur Scientist back in the '60s. I was
wondering if anyone here had heard of a similar project for a cesium
clock? Although my home built mass-spec wasn't in the same league as a
bought one, it did work and did get me a HD in physics! How hard is it
to make a cesium clock? Just thinking about it and wondering what is
actually inside the physics package.
Tim.
This is a chicken vs egg sort of problem. When I took apart the oven
on my 5065A rubidium, it looked to me as though the enamel insulated
nichrome wire developed a short about 50% into the coil, and wiped out
the transistor. It also heated the rubidium lamp hot enough to reflow
the solder on its circuit board and thoroughly char the PCB's epoxy.
I am not sure how the quartz ovens are wound (I suspect they are the
same), but with the rubidium ovens, the nichrome wire is wound as a
bifilar loop. This is done for two reasons, 1) to cancel the magnetic
fields, and 2) to make the start and finish wires of the winding happen
at the end of the oven where the terminals are.
-Chuck Harris
wje wrote:
Yes, but my comment is rather specific... my oscillator failed because
the heater pass transistor shorted.
This sent the oven heater into full-on. The overtemp sensor is far
removed from the heater. There's a ribbon cable between the driver board
and the temp bridge sensor board that runs directly over the heater, not
outside the oven insulation. Net result, transistor shorts, cable fries
before overtemp fuse opens. There's no reason the cable should be inside
the oven insulation.
My rebuild fix - (after removing all the carbonized foam insulation)
refoam the oven, replacing the cable and transistor, and moving it
outside the foam!
While I respect in general the brilliance of HP engineers, my classmate
was one, this isn't one of the more intelligent decisions.
Bruce Griffiths wrote:
The following paper indicates how to optimise the voltage drops in
electrical leads for both cryogenic and other applications:
http://arxiv.org/pdf/cond-mat/0508697.pdf
Bruce
Equation 4 in the above reference should be divided by e (electron charge).
The extension to the case where the cold temperature isnt negligible
compared to the hot temperature is straighforward.
For a cold temperature of 300K and a hot temperature of 400K the voltage
drop in each lead should be around 0.1V.
Bruce
swingbyte wrote:
G'day fellow time-nuts.
I was just going through some old projects and found my old home-made
mass spectrometer. This was a project based on a design from either
Scientific American or The Amateur Scientist back in the '60s. I was
wondering if anyone here had heard of a similar project for a cesium
clock? Although my home built mass-spec wasn't in the same league as a
bought one, it did work and did get me a HD in physics! How hard is it
to make a cesium clock? Just thinking about it and wondering what is
actually inside the physics package.
Tim.
Not a great deal:
A Caesium oven
A mechanical collimator for the Caesium beam
A pair of state selection magnets
A Caesium detector
A pair of Microwave cavities
Bruce
Chuck Harris wrote:
This is a chicken vs egg sort of problem. When I took apart the oven
on my 5065A rubidium, it looked to me as though the enamel insulated
nichrome wire developed a short about 50% into the coil, and wiped out
the transistor. It also heated the rubidium lamp hot enough to reflow
the solder on its circuit board and thoroughly char the PCB's epoxy.
I am not sure how the quartz ovens are wound (I suspect they are the
same), but with the rubidium ovens, the nichrome wire is wound as a
bifilar loop. This is done for two reasons, 1) to cancel the magnetic
fields, and 2) to make the start and finish wires of the winding happen
at the end of the oven where the terminals are.
-Chuck Harris
An Ayrton-Perry style winding will also have low magnetic field.
Bruce