Re: [time-nuts] Symmetricom XPRO rubidium

[Matthias Welwarsky]

On Montag, 28. September 2020 04:19:12 CEST Stewart Cobb wrote:
> I have a Symmetricom* XPRO rubidium which appears to be reaching its end of
> life. The very sparse manual says that it sets a "service" flag when the
> lamp voltage reaches 600 mV. When I got it, that parameter was at about
> 540. Several months of continuous runtime later, it's down to about 510.  I
> assume this is a measure of light emitted by the lamp, but its label is
> "lamp voltage".

What is the operational range for that parameter? Are you sure about the 
scaling? 540 could be 5.4 Volts as well. In that case, nothing to worry about.

> I don't want to lose this Rb, because it seems to be the most stable
> reference in my lab (about 2x more stable with temperature than a PRS10, by
> eyeball).
> 
> Questions for the hive mind:
> 
> (1) Why and how is the lamp voltage falling?  What's the wear-out
> mechanism?

>From what I know there isn't really a failure mechanism for the lamp itself. 
It's not using up it's Rb like a Cesium beam standard, the only way for it to 
"loose" its Rubidium would be by embedding Rb ions into the glass.

> (2) Is there any hope of repair? Will the heat gun trick for the LPRO work
> on the XPRO? Could I replace the XPRO lamp bulb with one from a young LPRO?

Replacement: if the LPRO lamp fits mechanically, it's worth a try if the lamp 
is really faulty. But before butchering the lamp assembly, it's better to 
check other failure sources. I'm guessing that the overall design is similar 
to the LPRO, so there is an exciter oscillator that could be going bad (not 
pumping enough energy into the lamp), the lamp heater could be going bad (too 
low temperature) and there are a number of failure modes for both.

There doesn't seem to be much known about the innards of this device, but if 
there are patents mentioned in the manual, it's worth looking them up. The 
patents mentioned in the LPRO integration manual turned out to be very, very 
detailed and insightful. Especially those about the lamp exciter.

Regarding the "rejuvenation trick", it'll probably work as good or as bad as 
for the LPRO, provided you can get the lamp out.

HOWEVER: If you're thinking about doing "preventive maintenance", DON'T! You 
will likely impact the stability of the device. Wait for it to fail, then go 
through the common failure sources known for all Rb standards of that era. 

Regards,
Matthias



___
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to 
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Re: [time-nuts] Oscilloquartz BVA has been sold. Thank you all who expressed an interest.

[Mark Spencer]

Yes running being able to run at least some of my OCXO's from a DC battery 
system is my desired state of affairs for my home time lab.  It is nice to have 
two devices that have been running for a decade.  

Anyways thanks all for the comments about various ways to power my BVA and the 
trip down memory lane re large UPS systems, Telecom battery plants etc.  

It will be nice to have the BVA powered from the a DC battery system.   I will 
be curious to see how the BVA performs after being powered up for at least a 
year.  Currently after running for a month or so it seems to often outperform 
the other OCXO's  that have been running for a decade or so.   Between 
occasional power outages that exceed typical UPS run times and the need to 
change batteries on occasion in the consumer grade UPS systems I use in my home 
time lab, getting multi year un interrupted run times without using a DC 
battery system to power devices during AC power outages seems unlikely for me.



Mark Spencer
m...@alignedsolutions.com
604 762 4099

> On Sep 27, 2020, at 4:06 AM, Magnus Danielson  wrote:
> 
> Hi,
> 
>> On 2020-09-27 09:02, Poul-Henning Kamp wrote:
>> 
>> Bill Notfaded writes:
>> 
>>> Are you saying if the load is small it'll still run out pretty quickly?
>> Yes.
>> 
>> The constant loss of the inverter-stage will be nothing compared
>> to the full design load, but will totally swamp your light load.
>> 
>> This is why UPS vendors only publish hold-up times for full load.
> Which is why it is a good idea to run the load directly off the
> batteries. In telecom, that is -48 VDC (check out the ETSI EN 300 132-2
> spec) but there is also one for modern switch-mode supplies for more
> IT-infrastructure which is feeding the 230 VAC side the DC it achieves
> after rectifier (ETSI EN 300 132-3 spec), but that has not taken off as
> far as I know. Both avoids the inverter part. Running straight of the
> batteries for 24 VDC matches many of our devices. For instance, all the
> atomic clocks and many OCXOs I have essentially run off 24 VDC, so that
> will be my focus.
> 
> The voltage range for the -48 VDC is really -40 VDC to -60,5 VDC to
> match the out of charge to charging voltages of normal lead-acid batteries.
> 
> Sure, the 24 V or 48 V may not fit the needs of applications, but you
> can usually find DC/DC converters that can be decent enough loss-wise as
> they do switched mode drop-regulation. Choosing wisely amongst those,
> you can get better losses than a large supply which runs in a
> non-optimal mode.
> 
> Cheers,
> Magnus
> 
> 
> 
> ___
> time-nuts mailing list -- time-nuts@lists.febo.com
> 

___
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to 
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

[time-nuts] Symmetricom XPRO rubidium

[Stewart Cobb]

I have a Symmetricom* XPRO rubidium which appears to be reaching its end of
life. The very sparse manual says that it sets a "service" flag when the
lamp voltage reaches 600 mV. When I got it, that parameter was at about
540. Several months of continuous runtime later, it's down to about 510.  I
assume this is a measure of light emitted by the lamp, but its label is
"lamp voltage".

I don't want to lose this Rb, because it seems to be the most stable
reference in my lab (about 2x more stable with temperature than a PRS10, by
eyeball).

Questions for the hive mind:

(1) Why and how is the lamp voltage falling?  What's the wear-out
mechanism?

(2) Is there any hope of repair? Will the heat gun trick for the LPRO work
on the XPRO? Could I replace the XPRO lamp bulb with one from a young LPRO?

(3) Has anyone ever done surgery on an XPRO? Any hints or tips?

(4) Does anyone know anything more about the XPRO? I have data sheets
900-00542-000B and -000F, and manuals 098-00058-000 rev A and rev B. The
manuals are just barely sufficient to get the thing running. The device
itself displays parameters on its serial interface which appear to be
adjustable but which are not described in the manual.

Thanks in advance for any help.

*Symmetricom has been acquired at least twice since this device was built.
But the device label says Symmetricom, and the manual says Symmetricom, so
I'll stick with that name for now.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to 
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Re: [time-nuts] What do people use for measuring temperature?

[Bob kb8tq]

Hi

You would need to find a temperature sensor rather than a voltage reference…. :)

The AD590 still seems to be considered a pretty good device after all these 
years. If there is long
term data in the spec sheet, I did not spot it in a quick read. Based on what 
we saw back in the 1970’s
(which *is* a long time ago) mili-K / month is a bit ambitious for a 590.

Indeed a band gap device will have a much wider useful range than a thermistor 
mounted in a bridge circuit. The thermistor only “wins” if you are interested 
in a fairly
narrow temperature range. 

Bob

> On Sep 27, 2020, at 2:49 PM, John Moran, Scawby Design 
>  wrote:
> 
> Would not a band-gap temperature sensor such as the LT6657 be better than a 
> thermistor for precision, low drift applications?
> 
> The above device has 30 ppm/√kHr long-term drift. That should hold a mk for a 
> fair few years.
> 
> John 
> 
> https://www.analog.com/media/en/technical-documentation/data-sheets/6657fd.pdf
> 
> 
> 
> ___
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe, go to 
> http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.


___
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to 
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Re: [time-nuts] What do people use for measuring temperature?

[John Moran, Scawby Design]

Would not a band-gap temperature sensor such as the LT6657 be better than a 
thermistor for precision, low drift applications?

The above device has 30 ppm/√kHr long-term drift. That should hold a mk for a 
fair few years.

John 

https://www.analog.com/media/en/technical-documentation/data-sheets/6657fd.pdf



___
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to 
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Re: [time-nuts] What do people use for measuring temperature?

[Bob kb8tq]

Hi


> On Sep 27, 2020, at 11:59 AM, Jeremy Nichols  wrote:
> 
>> 
>> 
>> 
>>> We used three thermistors and averaged them.
>> 
>> I assume they were spread around in case one side of the package was
>> warmer than the other.
>> 
> 
>> Could one do an analog “average” by using a set of, for example, four
> ‘identical’ (there’s a risky word!)

They get closer the more you pay for them ….. :)

On an OCXO, you have a target range of temperatures. It may be as 
narrow as 10C it could be 30C. “Matching” outside that range is not 
going to help you any.

Indeed back in the “good old days” people sat and sorted thermostats 
to get ovens to match up with crystals. 

The Morion folks used to have a warehouse full of raw PTC material 
ready to dice up into thermistors for their “vacuum tube” OCXO’s. They 
matched it up / chopped it to size as needed …. The manufacturing process
on the stuff was not tight enough to deliver it to the accuracy they needed.

Bob

> thermistors in series-parallel?
> 
> Jeremy
> 
> -- 
> Jeremy Nichols
> Sent from my iPad 6.
> ___
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe, go to 
> http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.


___
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to 
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Re: [time-nuts] What do people use for meassuring temperature?

[Bob kb8tq]

Hi

There are indeed published works on the physics behind thermistors. The
ones I’ve read (many decades ago …) didn’t say a lot about aging mechanisms 
beyond the usual stuff you could guess at. Things like humidity can contaminate
the material, glass sealing is a good idea ( not a big surprise … ).

The bottom line is still that they get used at fairly high temperatures and do 
indeed hold quite good drift specs. Conditioning of the parts prior to use is
non-existant in an OCXO application. Mitsubishi makes some pretty good 
parts as do others …..

Bob

> On Sep 27, 2020, at 8:37 AM, John Ponsonby  wrote:
> 
> Bruce Griffith points to the note put out by Littelfuse. It is very meagre. 
> It says that '...thermistors can be produced with  typical drift of only 
> 0.001˚C to 0.002˚C per Year." It doesn't say that they are produced and still 
> less that Littelfuse produces them. Bruce also refers to thermistors being  
> 'Suitably conditioned at 25˚C' What is this conditioning process and what if 
> the intended working temperature is not 25˚C?. Surely more must be known 
> about this matter.
> Jeremy Nichols' questions are very apposite.
> I see that the Steinhart Hart Equation seems to be entirely empirical without 
> any underlying semiconductor theoretical foundation.
> John Ponsonby
> 
> 
> Do we know what this ?Long Term Aging Process? is or is it proprietary?
> 
> 
> 
> On Sat, Sep 26, 2020 at 5:03 PM Bruce Griffiths  >
> wrote:
> 
>> Drift ~1-2mK per year for suitably conditioned thermistors at 25C:
>> 
>> 
>> https://www.littelfuse.com/technical-resources/technical-centers/temperature-sensors/thermistor-info/thermistor-terminology/stability.aspx
>>  
>> 
>> 
>> 
>> 
>> Bruce
>> 
>>> On 27 September 2020 at 11:15 Bob kb8tq >> > wrote:
>> 
>>> 
>> 
>>> 
>> 
>>> Hi
>> 
>>> 
>> 
>>> Roughly speaking 99.9% of all OCXO?s use thermistors as temperature
>> sensors.
>> 
>>> The normal evaluation process on a new one *probably* would catch
>> something < 0.01C
>> 
>>> over a few months. You may do it a couple different ways depending on
>> the target
>> 
>>> OCXO. The net result is still in the ?golly gee wiz low? sort of range.
>> If you can detect a
>> 
>>> drift / shift, you disqualify that part and move on to another one. Very
>> few glass bead
>> 
>>> parts seem to get tossed out ?..
>> 
>>> 
>> 
>>> Bob
>> 
>>> 
>> 
>>> 
>> 
>>> 
>> 
 On Sep 26, 2020, at 4:23 PM, John Ponsonby >>> >
>> wrote:
>> 
 
>> 
 Have any time-nuts got any data on the long term stability or drift
>> rates/ageing characteristics of thermistors? I am concerned with  ability
>> of holding temperature constant at the milliK level for years. I reckon
>> that if one can measure it one can control it. Conversely if one can't
>> measure it, because of the instability of the sensor,  one can't control it.
>> 
> ___
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe, go to 
> http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.


___
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to 
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Re: [time-nuts] What do people use for meassuring temperature?

[Bruce Griffiths]

Here's a NIST paper on Thermistor stability:
https://ia800609.us.archive.org/2/items/jresv83n3p247/jresv83n3p247_A1b.pdf

Bruce
> On 28 September 2020 at 01:37 John Ponsonby  wrote:
> 
> 
> Bruce Griffith points to the note put out by Littelfuse. It is very meagre. 
> It says that '...thermistors can be produced with  typical drift of only 
> 0.001˚C to 0.002˚C per Year." It doesn't say that they are produced and still 
> less that Littelfuse produces them. Bruce also refers to thermistors being  
> 'Suitably conditioned at 25˚C' What is this conditioning process and what if 
> the intended working temperature is not 25˚C?. Surely more must be known 
> about this matter.
> Jeremy Nichols' questions are very apposite.
> I see that the Steinhart Hart Equation seems to be entirely empirical without 
> any underlying semiconductor theoretical foundation.
> John Ponsonby
> 
> 
> Do we know what this ?Long Term Aging Process? is or is it proprietary?
> 
> 
> 
> On Sat, Sep 26, 2020 at 5:03 PM Bruce Griffiths  >
> wrote:
> 
> > Drift ~1-2mK per year for suitably conditioned thermistors at 25C:
> > 
> > 
> > https://www.littelfuse.com/technical-resources/technical-centers/temperature-sensors/thermistor-info/thermistor-terminology/stability.aspx
> >  
> > 
> > 
> > 
> > 
> > Bruce
> > 
> >> On 27 September 2020 at 11:15 Bob kb8tq  >> > wrote:
> > 
> >> 
> > 
> >> 
> > 
> >> Hi
> > 
> >> 
> > 
> >> Roughly speaking 99.9% of all OCXO?s use thermistors as temperature
> > sensors.
> > 
> >> The normal evaluation process on a new one *probably* would catch
> > something < 0.01C
> > 
> >> over a few months. You may do it a couple different ways depending on
> > the target
> > 
> >> OCXO. The net result is still in the ?golly gee wiz low? sort of range.
> > If you can detect a
> > 
> >> drift / shift, you disqualify that part and move on to another one. Very
> > few glass bead
> > 
> >> parts seem to get tossed out ?..
> > 
> >> 
> > 
> >> Bob
> > 
> >> 
> > 
> >> 
> > 
> >> 
> > 
> >>> On Sep 26, 2020, at 4:23 PM, John Ponsonby  >>> >
> > wrote:
> > 
> >>> 
> > 
> >>> Have any time-nuts got any data on the long term stability or drift
> > rates/ageing characteristics of thermistors? I am concerned with  ability
> > of holding temperature constant at the milliK level for years. I reckon
> > that if one can measure it one can control it. Conversely if one can't
> > measure it, because of the instability of the sensor,  one can't control it.
> > 
> ___
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe, go to 
> http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.

___
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to 
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Re: [time-nuts] What do people use for measuring temperature?

[Jeremy Nichols]

>
> 
>
> > We used three thermistors and averaged them.
>
> I assume they were spread around in case one side of the package was
> warmer than the other.
>

> Could one do an analog “average” by using a set of, for example, four
‘identical’ (there’s a risky word!) thermistors in series-parallel?

Jeremy

-- 
Jeremy Nichols
Sent from my iPad 6.
___
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to 
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Re: [time-nuts] What do people use for meassuring temperature?

[John Ponsonby]

Bruce Griffith points to the note put out by Littelfuse. It is very meagre. It 
says that '...thermistors can be produced with  typical drift of only 0.001˚C 
to 0.002˚C per Year." It doesn't say that they are produced and still less that 
Littelfuse produces them. Bruce also refers to thermistors being  'Suitably 
conditioned at 25˚C' What is this conditioning process and what if the intended 
working temperature is not 25˚C?. Surely more must be known about this matter.
Jeremy Nichols' questions are very apposite.
I see that the Steinhart Hart Equation seems to be entirely empirical without 
any underlying semiconductor theoretical foundation.
John Ponsonby


Do we know what this ?Long Term Aging Process? is or is it proprietary?



On Sat, Sep 26, 2020 at 5:03 PM Bruce Griffiths mailto:bruce.griffi...@xtra.co.nz>>
wrote:

> Drift ~1-2mK per year for suitably conditioned thermistors at 25C:
> 
> 
> https://www.littelfuse.com/technical-resources/technical-centers/temperature-sensors/thermistor-info/thermistor-terminology/stability.aspx
>  
> 
> 
> 
> 
> Bruce
> 
>> On 27 September 2020 at 11:15 Bob kb8tq > > wrote:
> 
>> 
> 
>> 
> 
>> Hi
> 
>> 
> 
>> Roughly speaking 99.9% of all OCXO?s use thermistors as temperature
> sensors.
> 
>> The normal evaluation process on a new one *probably* would catch
> something < 0.01C
> 
>> over a few months. You may do it a couple different ways depending on
> the target
> 
>> OCXO. The net result is still in the ?golly gee wiz low? sort of range.
> If you can detect a
> 
>> drift / shift, you disqualify that part and move on to another one. Very
> few glass bead
> 
>> parts seem to get tossed out ?..
> 
>> 
> 
>> Bob
> 
>> 
> 
>> 
> 
>> 
> 
>>> On Sep 26, 2020, at 4:23 PM, John Ponsonby >> >
> wrote:
> 
>>> 
> 
>>> Have any time-nuts got any data on the long term stability or drift
> rates/ageing characteristics of thermistors? I am concerned with  ability
> of holding temperature constant at the milliK level for years. I reckon
> that if one can measure it one can control it. Conversely if one can't
> measure it, because of the instability of the sensor,  one can't control it.
> 
___
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to 
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Re: [time-nuts] Oscilloquartz BVA has been sold. Thank you all who expressed an interest.

[Magnus Danielson]

Hi,

On 2020-09-27 09:02, Poul-Henning Kamp wrote:
> 
> Bill Notfaded writes:
>
>> Are you saying if the load is small it'll still run out pretty quickly?
> Yes.
>
> The constant loss of the inverter-stage will be nothing compared
> to the full design load, but will totally swamp your light load.
>
> This is why UPS vendors only publish hold-up times for full load.
>
Which is why it is a good idea to run the load directly off the
batteries. In telecom, that is -48 VDC (check out the ETSI EN 300 132-2
spec) but there is also one for modern switch-mode supplies for more
IT-infrastructure which is feeding the 230 VAC side the DC it achieves
after rectifier (ETSI EN 300 132-3 spec), but that has not taken off as
far as I know. Both avoids the inverter part. Running straight of the
batteries for 24 VDC matches many of our devices. For instance, all the
atomic clocks and many OCXOs I have essentially run off 24 VDC, so that
will be my focus.

The voltage range for the -48 VDC is really -40 VDC to -60,5 VDC to
match the out of charge to charging voltages of normal lead-acid batteries.

Sure, the 24 V or 48 V may not fit the needs of applications, but you
can usually find DC/DC converters that can be decent enough loss-wise as
they do switched mode drop-regulation. Choosing wisely amongst those,
you can get better losses than a large supply which runs in a
non-optimal mode.

Cheers,
Magnus



___
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to 
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Re: [time-nuts] What do people use for measuring temperature?

[Hal Murray]

> The HP E1938A OCXO that I worked on had a feature where we could set the oven
> temperature individually on each oscillator to the exact turnover point of
> the crystal. 

How accurately did you set it?

> We used three thermistors and averaged them.

I assume they were spread around in case one side of the package was warmer 
than the other.

-- 
These are my opinions.  I hate spam.




___
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to 
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Re: [time-nuts] Oscilloquartz BVA has been sold. Thank you all who expressed an interest.

[Poul-Henning Kamp]

Bill Notfaded writes:

> Are you saying if the load is small it'll still run out pretty quickly?

Yes.

The constant loss of the inverter-stage will be nothing compared
to the full design load, but will totally swamp your light load.

This is why UPS vendors only publish hold-up times for full load.

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

___
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to 
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.