Hi,
this is a second attempt at an answer, the first seemed to evaporate.
Mercury-in -glass thermometers have formed the basis of a system of
thermostats
used constant temperature systems of very high performance.
A major producer of contact thermometers was Jumo (maybe  German).
These thermometers had a thin wire that went down the capillary and
contacted the mercury at the set temperature. The system could be
accurate to
0.005 degree. The control algorithm is alien to modern EEs but used a
large thermal mass and a fixed rate of heating to produce a slow
temperature ramp.
Fast response by the thermometer switching kept overshoot down to
millidegrees.
A cycle time of 5 or 10 seconds kept a very low amplitude temperature
ramp running up and down
with mean temperature held quite close.
Locating the thermometer close to the heater caused a little over
control which reduced overshoot and cycle amplitude.
The thermometers had up to 30 mA run through the wire, but more
modern units reduced this to 1mA.
If the tip is corroded on your thermometer contact, maybe a higher
resistance measuring circuit may
still operate reliably.
Good Luck with the unit,
Neville Michie

On 20/08/2008, at 9:17 AM, Mark Sims wrote:

Neville,
good answer. Perhaps the original poster could supply a photo of his switch,
perhaps more suggestions could be made. I think we had an old gr-1100 at one
time, it was a 100kc unit. Needless to say it was scrapped years ago and I'm
sure it's been melted down and sent back to the usa in tin cans or cars by
now! I'm curious as to how the switch failed. I'm assuming it was a sealed
unit.
Phil

----- Original Message -----
From: "Neville Michie" namichie@gmail.com
To: "Discussion of precise time and frequency measurement"
time-nuts@febo.com
Sent: Tuesday, August 19, 2008 7:46 PM
Subject: Re: [time-nuts] Restoring GR 1120-AB Frequency Standard

Hi,
this is a second attempt at an answer, the first seemed to evaporate.
Mercury-in -glass thermometers have formed the basis of a system of
thermostats
used constant temperature systems of very high performance.
A major producer of contact thermometers was Jumo (maybe  German).
These thermometers had a thin wire that went down the capillary and
contacted the mercury at the set temperature. The system could be
accurate to
0.005 degree. The control algorithm is alien to modern EEs but used a
large thermal mass and a fixed rate of heating to produce a slow
temperature ramp.
Fast response by the thermometer switching kept overshoot down to
millidegrees.
A cycle time of 5 or 10 seconds kept a very low amplitude temperature
ramp running up and down
with mean temperature held quite close.
Locating the thermometer close to the heater caused a little over
control which reduced overshoot and cycle amplitude.
The thermometers had up to 30 mA run through the wire, but more
modern units reduced this to 1mA.
If the tip is corroded on your thermometer contact, maybe a higher
resistance measuring circuit may
still operate reliably.
Good Luck with the unit,
Neville Michie

On 20/08/2008, at 9:17 AM, Mark Sims wrote:

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.

Phil,

The details of the thermoswitch failure are not completely clear.
Intermittently, the mercury column will pass the 78C point without
causing circuit closure. Instead, a pulsating open / close with an
internal resistance of 50K - 300K is observed. Circuit closure
eventually happens at a temperature above 82C. Once an external
failure, i.e. lead wiring, was ruled out, I used a borrowed microscope
to view the internal structure of the thermoswitch. Unfortunately, this
did not reveal the failure source.

Russ

-----Original Message-----
From: phil fortime@bellsouth.net
To: Discussion of precise time and frequency measurement
time-nuts@febo.com
Sent: Tue, 19 Aug 2008 9:33 pm
Subject: Re: [time-nuts] Restoring GR 1120-AB Frequency Standard

Neville,
good answer. Perhaps the original poster could supply a photo of his
switch,
perhaps more suggestions could be made. I think we had an old gr-1100
at one
time, it was a 100kc unit. Needless to say it was scrapped years ago
and I'm
sure it's been melted down and sent back to the usa in tin cans or cars
by
now! I'm curious as to how the switch failed. I'm assuming it was a
sealed
unit.
Phil

----- Original Message -----
From: "Neville Michie" namichie@gmail.com
To: "Discussion of precise time and frequency measurement"
time-nuts@febo.com
Sent: Tuesday, August 19, 2008 7:46 PM
Subject: Re: [time-nuts] Restoring GR 1120-AB Frequency Standard

Hi,
this is a second attempt at an answer, the first seemed to evaporate.
Mercury-in -glass thermometers have formed the basis of a system of
thermostats
used constant temperature systems of very high performance.
A major producer of contact thermometers was Jumo (maybe German).
These thermometers had a thin wire that went down the capillary and
contacted the mercury at the set temperature. The system could be
accurate to
0.005 degree. The control algorithm is alien to modern EEs but used a
large thermal mass and a fixed rate of heating to produce a slow
temperature ramp.
Fast response by the thermometer switching kept overshoot down to
millidegrees.
A cycle time of 5 or 10 seconds kept a very low amplitude temperature
ramp running up and down
with mean temperature held quite close.
Locating the thermometer close to the heater caused a little over
control which reduced overshoot and cycle amplitude.
The thermometers had up to 30 mA run through the wire, but more
modern units reduced this to 1mA.
If the tip is corroded on your thermometer contact, maybe a higher
resistance measuring circuit may
still operate reliably.
Good Luck with the unit,
Neville Michie

On 20/08/2008, at 9:17 AM, Mark Sims wrote:

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.

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.

Hi Neville,

Voltage across the thermoswitch contacts in less than 10VAC.  Current
is 6-8 mA.  Once in failure mode, the mercury column passes the 78C
point without causing circuit closure.  Instead, a rapidly pulsating
open / close with an internal resistance of 50K - 300K is observed.
Circuit closure eventually happens at a temperature above 82C. After
this overtemp condition, the thermoswitch will perform normally for
numerous cycles before going back into failure mode.

Once I detected the intermittent thermoswitch failure, I increased
voltage and current to see if this would restore reliability.  Failures
went from minutes and hours to about 12 hours and occasionally days but
not good enough to declare success. I haven't increased voltage and
current further as I feel that it better to replace the failing
component.

Thanks,

Russ

-----Original Message-----
From: Neville Michie namichie@gmail.com
To: Discussion of precise time and frequency measurement
time-nuts@febo.com
Sent: Tue, 19 Aug 2008 7:46 pm
Subject: Re: [time-nuts] Restoring GR 1120-AB Frequency Standard

Hi,
this is a second attempt at an answer, the first seemed to evaporate.
Mercury-in -glass thermometers have formed the basis of a system of
thermostats
used constant temperature systems of very high performance.
A major producer of contact thermometers was Jumo (maybe  German).
These thermometers had a thin wire that went down the capillary and
contacted the mercury at the set temperature. The system could be
accurate to
0.005 degree. The control algorithm is alien to modern EEs but used a
large thermal mass and a fixed rate of heating to produce a slow
temperature ramp.
Fast response by the thermometer switching kept overshoot down to
millidegrees.
A cycle time of 5 or 10 seconds kept a very low amplitude temperature
ramp running up and down
with mean temperature held quite close.
Locating the thermometer close to the heater caused a little over
control which reduced overshoot and cycle amplitude.
The thermometers had up to 30 mA run through the wire, but more
modern units reduced this to 1mA.
If the tip is corroded on your thermometer contact, maybe a higher
resistance measuring circuit may
still operate reliably.
Good Luck with the unit,
Neville Michie

On 20/08/2008, at 9:17 AM, Mark Sims wrote:

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.

Russ,
In other words, you sound like you are not 100% sure it's the switch. A
shame you couldn't graph voltage and current going to the switch as it
cycles. Is it possible a cap or something else is intermittently failing in
the circuit. Again I'm assuming with that resistance there is something
between it and the heater. Though I had one, I don't think I ever looked
inside the thing.

I have seen some similar units (I think what you have) that use an almost
paper thin column of mercury. I would guess that is how they can achieve the
fine resolution/precision.

Phil

----- Original Message -----
From: wa3frp@aol.com
To: time-nuts@febo.com
Cc: fortime@bellsouth.net
Sent: Wednesday, August 20, 2008 2:34 AM
Subject: Re: [time-nuts] Restoring GR 1120-AB Frequency Standard

Russ,
Also could it be a cold solder joint on the tubes connections, possibly look
good but have a high resistance. That could account for an intermittent as
you are describing. Sometimes we overlook the simplest of things.
Phil

----- Original Message -----
From: "phil" fortime@bellsouth.net
To: time-nuts@febo.com; wa3frp@aol.com
Sent: Wednesday, August 20, 2008 3:26 AM
Subject: Re: [time-nuts] Restoring GR 1120-AB Frequency Standard

Phil,

I didn't believe that the thermoswitch was the problem, at first,
chiefly because of the simplicity of operation. Eventually, after
checking wiring, a carbon resistor that is in series with the
thermoswitch, and components around the inner oven control circuitry, I
removed the thermoswitch to the bench.

After hooking up to a ohmmeter and using a 60 watt light bulb as the
heat source, I found that I could duplicate the
a pulsating open / close as before. I first focused on the bulb leads
and eventually completely removed the old leads and rebuilt each one
and did all new soldering under magnification. The problem remains the
same.

I'm ready to move on at this point noting that this component failure
has me stumped and that the fault is most likely internal to the
thermoswitch (as strange as this seems). Years ago, when I first saw
how internal temperature worked using the mercury thermometer switch, I
remarked that it was one component that would never fail. HA! That
statement came back to haunt me.

Best,

Russ

-----Original Message-----
From: phil fortime@bellsouth.net
To: Discussion of precise time and frequency measurement
time-nuts@febo.com
Sent: Wed, 20 Aug 2008 3:45 am
Subject: Re: [time-nuts] Restoring GR 1120-AB Frequency Standard

Russ,
Also could it be a cold solder joint on the tubes connections, possibly
look
good but have a high resistance. That could account for an intermittent
as
you are describing. Sometimes we overlook the simplest of things.
Phil

----- Original Message -----
From: "phil" fortime@bellsouth.net
To: time-nuts@febo.com; wa3frp@aol.com
Sent: Wednesday, August 20, 2008 3:26 AM
Subject: Re: [time-nuts] Restoring GR 1120-AB Frequency Standard

A

failing

something

looked

almost

achieve

happens

wiring,

gr-1100 at

and

cars

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.

wa3frp@aol.com wrote:

I have been following this thread with some interest, as I expect to
have similar  equipment  in the future. What is amazing  is  how you
discovered the problem!

Like you, I would not have believed a mercury switch could fail. But
a quick  search  showed  the contact can  oxidize, and  gave several
patents aimed at solving the problem:

1. Reduction of oxides in a fluid-based switch - US  Patent 7071432,
   07/04/2006

Often, oxides  may  form  within  the  switch  and  inhibit proper
functioning of  the switch. For example, the oxides may  increase or
decrease the surface tension of the liquid metal, which may increase
or decrease the energy required for the switch to change state.

Oxides can lead to poor switch performance, and even switch failure,
because they  lessen  or  prevent  a  switching  fluid  from wetting
surfaces it is supposed to wet.

2. Preventing corrosion degradation in a fluid-based switch - United
   States Patent 6781074, 08/24/2004

Liquid metal switches rely on the cleanness of the liquid  metal for
good performance.  If  the liquid metal forms oxide  films  or other
types of  corrosion  product buildup within the  switch,  the proper
functioning or  performance  of  the  switch  may  degrade  or be
inhibited.

For example, the oxide film or other corrosion products may increase
the surface  tension  of the liquid metal,  which  may  increase the
energy required  for the switch to change state. Films of  oxide and
other corrosion  products may increase the tendency  for  the liquid
metal to  wet  to  the substrate  between  switch  contacts, thereby
increasing undesirable  short circuits in  the  switching operation.
Build up of oxide and other corrosion products may also  degrade the
ability of  the  liquid  metal to wet to  the  switch  contacts, and
thereby may increase the probability of undesirable open circuits in
the switching operation.

This is very interesting. Thanks for posting your experience.

Best Regards,

Mike Monett

Russ
Do you have a photo, would it be possible to build? I have a few flasks of
mercury and a few 10's of thousands of mercury wetted relays that some could
be cannibalized for electrodes. The trick here is how thin is that column of
mercury. My guess the thinner the column, the finer the resolution or the
more it will travel with a small temp swing. Better yet, possible to cut
that thing open and fix it.
I know that's more time than it's worth.

Another thought, if the contacts have corroded in the mercury, perhaps from
contamination, would it be possible to "burn" the contamination off.
Thinking of excessive voltage/current. Perhaps making it arc internally.
Trick would be to limit current not to explode the thing, perhaps using a
charged capacitor. Just a thought.

Oh, do you have the whole rack, all dividers and the clock? I bet the old
clock alone is worth a thousand bucks today. Seems like it was driven by 1
KC. My old standard was complete, a 5 or 6 foot rack and came out of an old
Hamerland Radio plant.

phil

----- Original Message -----
From: wa3frp@aol.com
To: time-nuts@febo.com
Cc: fortime@bellsouth.net
Sent: Wednesday, August 20, 2008 4:16 AM
Subject: Re: [time-nuts] Restoring GR 1120-AB Frequency Standard