Dear Paul,

On 12/02/11 21:34, paul swed wrote:

I bet it is a thin layer of Rb... so yes.

I did not measure the temperature, but I was able to heat it up fairly
quickly with my small heat-gun, despite the fact that the removal tool
was cooling it off... maybe I was heating it for 5 min or so... but I
was able to have the dark shade evaporating.

It is well worth the effort, but the ability to heat it may vary.

Cheers,
Magnus

Well indeed Magnus the black stuff and silver blob were rb.
300 degrees F for 15 minutes made it all go away.
I was using a j thermocouple to measure the temp and right at the bulb.
The trick seems to be for a FRS c to get the blob to the lowest point of the
capsule thats observable. I did this with the normal RB power system heated
to 177 degrees. Gently tapped a few times to get it concentrated.
Then powered off and let it cool.
Next step hung the assembly from a lamp so the heat gun could be pointed
straight up to boil the RB off at 300 degrees for 10 minutes and let it go
for 5 more for good luck.
The dark area on the capsule is significantly reduced and no trace of the
silver blob.
During the heating I could actually see the silver blob shrink.
The RB lamp lights well and looks good.
Now to try reassembling everything. That will be a job.
Regards
Paul
WB8TSL

On Sat, Feb 12, 2011 at 5:07 PM, Magnus Danielson <
magnus@rubidium.dyndns.org> wrote:

Dear Paul,

On 13/02/11 00:02, paul swed wrote:

I had mine go greyish on the first heat-up, but then I cleared it on the
second run.

Good news!

Eager to hear of lock-up results!

It seems like the normal temperature of the bulb is good for keeping it
lit, but you need to go further up for the clean-up operation. Maybe
this is why the grey rubidium-coating didn't rub off for others... they
didn't go high enough. Still, no extreme temperatures is needed to get
the job done.

Cheers,
Magnus

Well pretty good news
The lamp voltage went from 1.83 volts a dead bulb to* 8.9 volts a new bulb*.
By adjusting the oscillator I can get to 9.6 volts but I know the the
oscillator will not start correctly at 82 Mhz its happy at 92 Mhz.
Its not locking but that can be due to several issues.
At least 3 grounds are not connected. These are completed by the boards
being mounted on the oven.
The +24 lamp ignition is hanging at +17. So again the grounds or I have a
problem in the lamp regulator. I suspect its the grounds actually. I
disconnected pin 1 into the oven and applied 24 volts when the lamp ignited
manually backed it down to 17volts.
So tomorrow will start to reassemble the boards for correct grounding.
Regards
Paul.

On Sat, Feb 12, 2011 at 7:57 PM, Magnus Danielson <
magnus@rubidium.dyndns.org> wrote:

By the way I can disconnect pin 1 on the lamp assembly because I made a 5"
jumper assembly so that the area can be easily worked on.

On Sat, Feb 12, 2011 at 10:27 PM, paul swed paulswedb@gmail.com wrote:

Paul,

On 13/02/11 04:27, paul swed wrote:

This is good news, for the level of re-assembly you where at.
The lamp is in the ball-park now...

Cheers,
Magnus

Magnus it indeed does look very good.

I have re-assembled the various boards and the system locked up as normal
after the warm up period. I do indeed have a failure in the lamp control
startup crkt.
So I pulled the fet gate Q2 to ground through a 1 K ohm resistor setting the
lamp to 24 volts. After ignition just let it float taking the lamp voltage
to 17 Volts.

The lamp, after ignition is 10 volts when its oven is still cool and has
dropped in brightness. This is normal for the warm up period. Still at
8.5044 volts after the lamp oven gets to 177 degrees.
VCO correction voltage is 6.023v. As I recall, this is mid range on the
voltage.

The reference after warm up and adjusting the c field is down in the 1x10-11
area. Anyhow I adjusted the c field and the systems not fully assembled nor
has it really been stabilized long enough. Was simply curious so I do not
think this is really a very accurate assessment. Especially since this is
time-nuts territory.

By the way I purchased the heat gun from a supplier thats on amazon.com. Its
the NTE or ECG HG300d for $19.95. Two heat ranges 250 degree c and 350
degree c. Small and easy to handle. I used the low range and adjusted the
height of the lamp to the heat gun tip approximately 4" distance to get 300
degrees F.

So next steps is to let the unit cook in or maybe I will further assemble it
and troubleshoot the startup crkt. Add the unlock counter. (A pedometer and
opto coupler) and let the system run for a long while perhaps a week or so
to see what really happens.

Thanks for your guidance and help Magnus. Who would have thought there was a
recovery technique for FRS-c RB references.

Regards
Paul
WB8TSL

On Sun, Feb 13, 2011 at 8:12 AM, Magnus Danielson <
magnus@rubidium.dyndns.org> wrote:

Dear Paul,

On 13/02/11 18:50, paul swed wrote:

My R&S does not have a startup circuit, so I can't help on that
particular detail. I'm sure you will debug it yourself.

At least it proves that you are up and running at least, modulus quirks.

Sounds like a plan.

Happy to help. Happy to learn of your progress! Happy to hear I contribute.

Cheers,
Magnus

Startup circuit looks like a bad op amp used as a comparator. U1 section a.
This is a LM 158. I have a LM 148 on hand suspect they are different in
temperature quality. So will look it up and see. This is a common op amp so
if the 148 does not work for some reason any numbers of more modern ones
will. Accept one issue its a T08 can. What pin was 8? ;-)
Regards

On Sun, Feb 13, 2011 at 1:24 PM, Magnus Danielson <
magnus@rubidium.dyndns.org> wrote:

paul swed wrote:

The LM148 (quad 741) and LM158 (dual opamp) have different input common
mode ranges, that of the LM158 extends to the negative supply it is also
rated for a large input differential voltage.
LM158 pin 8 is the positive supply pin.

Bruce

Who would have thought a seriously leaky cap c7 on the input of U1a pin 3.
Roughly 10K ohm.
System is running at 1.6 x10^11. See if it hangs in there. Lamp is actually
going higher in voltage at 8.2 V. All in all a good experiment.
Regards
Paul.

On Sun, Feb 13, 2011 at 1:46 PM, paul swed paulswedb@gmail.com wrote:

Thanks Bruce turned out to be C7 leaky. I actually used a LM358 dip with
wires kind of a ugly bug. Then realzed that really was not the issue. It was
c7.
I do not have any lm158s can or dip. Pretty sure they would be hard to come
by these days also.
Something to look for at hamfests this year.
Regards
Paul.

On Sun, Feb 13, 2011 at 2:01 PM, Bruce Griffiths <bruce.griffiths@xtra.co.nz

At 17:50 -0500 13-02-2011, paul swed wrote:

Should you need one, they are in stock @ Digi-Key, Newark/Farnell and
several other distributors.

JDB.
[courtesy of http://findchips.com/ ]

If you think the problem is bad now, just wait until we've solved it.
-- Arthur Kasspe

The LM 158 and LM358 are the same part, just different screening levels and packaging.
For commercial applications at normal temperatures, the LM358 is fine.

Didier KO4BB

Sent from my Verizon Wireless BlackBerry

-----Original Message-----
From: paul swed paulswedb@gmail.com
Sender: time-nuts-bounces@febo.com
Date: Sun, 13 Feb 2011 17:50:18
To: Discussion of precise time and frequency measurementtime-nuts@febo.com
Reply-To: Discussion of precise time and frequency measurement
time-nuts@febo.com
Subject: Re: [time-nuts] FRSc RB lamp experiment

Thanks Bruce turned out to be C7 leaky. I actually used a LM358 dip with
wires kind of a ugly bug. Then realzed that really was not the issue. It was
c7.
I do not have any lm158s can or dip. Pretty sure they would be hard to come
by these days also.
Something to look for at hamfests this year.
Regards
Paul.

On Sun, Feb 13, 2011 at 2:01 PM, Bruce Griffiths <bruce.griffiths@xtra.co.nz

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.

Two authors come to mind regarding crystal oscillators: Eric Vittoz and
Marvin Ferking. Eric Vittoz is the more modern of the two. His writings
tend towards long term stability of crystal oscillators. Basically, most
designs put too much energy into the crystal, which he claims wears it
out. I'd had to dig up his papers, but my recollection is the failures
were soft (error in frequency) rather than hard (total failure). Ferking
covers temperature stability and crystal pulling.

I haven't dealt with crystal manufacturers in a long time, but my
recollection is the crystal is "tuned" by metal deposition. As you
deposit metal on the crystal, the frequency lowers. Possibly today they
laser trim, i.e. remove metal. Anyway, I don't think opening up the case
and fiddling with the innards is a good idea.

In the dark ages, when I took a class in wafer fabrication, we would
sense the amount of metal sputtered on the wafer by measuring the
frequency shift of a crystal in the chamber. As you sputtered metal, the
crystal frequency would get lower.

Ferking's DSP book is supposedly the bible in software defined radios.

Back to crystal manufacturers, these companies tend to be pretty small.
when I was working on video chip designs, it was no problem talking to
the CEO or VP engineering. I think it is a capital intensive rather than
labor intensive business. They have a few gurus doing product design and
that's about it.

In the dark ages, these guys were the easiest to deal with for technical
info:
http://www.crovencrystals.com/
They have an impressive list of projects that they worked on:
http://www.crovencrystals.com/croven_pdf/heritageprograms.pdf

On 2/13/2011 7:26 PM, Bill Hawkins wrote:

On 2/13/11 8:15 PM, gary wrote:

metal deposition is still how it works.

The real unknown is what the frequency will be after aging.. you make a
bunch, run them for a while at higher than normal temperatures, and see
which ones look like decent candidates.

Anyway, I don't think opening up the case

Bliley is another crystal maker.

But you're right.. there's typically very few people who actually do the
stuff for high performance crystals/oscillators.  THere will be a few
techs who have the skills to put the crystal in the holder and seal it
up. A few who make the crystals (running the saws), etc.

On 14/02/11 04:26, Bill Hawkins wrote:

On the any help level:

Aging usually makes crystals go adrift in frequency. Other parameters
may also shift, especially if the DC connectivity is lost due to
oxidation or so. I think it is likely that this is what happend and all
of a sudden the loop conditions does not support oscillation (i.e. gain

hacks to be added to get the loop oscillating again and there might be
hacks to the crystal assembly, but swapping the crystal is probably a
good remedy in this case.

Cheers,
Magnus

Hi Bill I have never worked on quartz crystals, but I was attached to a
group that had made many of them for the British Post Office in the 1950s. I
also used, and replaced, a lot in commercial 2-way radios in the 1980s and
90s.

First are you really sure the crystal has changed it is more likely that the
capacitors may have changed in a 40 year old circuit. You really need to
measure the crystal ESR to be sure. I came across cases of "lazy" crystals
in the 1990s usually the cheaper variety that had to be swapped out of
two-way radios. Generally these were probably "off-angle" and of marginal
activity. a slight change in ESR would stop them working. Some makers were
notorious for these problems. I suppose a more vigourous drive circuit might
have activated them, but with less stablity. Good oscillator circuits are
usually designed with only just enough drive to start and maintain
oscillation. Any major change in ESR and they stop, or wont start but will
maintain oscillation if tapped or are mechanically shocked.

Older crystals say before about 1965 would probably be made from natural
quartz, synthetic quartz was very expensive at that time and only used for
expensive top notch standards. Whereas I believe now all balanks are
synthetic (grown) crystals. Older natural crystals have more crystal defects
and more contamination (and thus usually age faster I believe). High levels
of drive could I suppose mill-out crystal dislocations which could provide
traps for contamination. Many crystal drive circuits, particularly in older
gear,  like the Pierce, can drive the crystal very hard. Then you have
possible problems with the plated electrodes due to contamination (usually
only cause frequency changes) and also micro cracks because the
evapourated/sputtered plating is quite thin.

After all they are passive but they are a "moving part" :-))

Alan G3NYK
----- Original Message -----
From: "Bill Hawkins" bill@iaxs.net
To: "'Discussion of precise time and frequency measurement'"
time-nuts@febo.com
Cc: "'Scott Robinson'" spr@earthlink.net; "'Roy Morgan'"
k1lky@earthlink.net; "'Jim Garland'" 4cx250b@muohio.edu
Sent: Monday, February 14, 2011 3:26 AM
Subject: [time-nuts] Why do crystals go bad?

and

onto

Oscillator

https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts

Hi

From a number of radios I've torn crystals out of and looked at:

1. The blank is broken / cracked. Most likely the radio got dropped somewhere along the line. Just like that stone hit on your windshield (on two cars at the moment), the crack can start small and propagate. Time from impact to failure might be years.

2. Cement fractures. Except in pressure mounts, the blank is held in the holder (clips) with cement. Normally this both holds the part and makes electrical connection. Some epoxies shrink with time eventually they shrink enough to crack. Big enough crack at the wrong place = no electrical contact.

3. Plating adhesion. Metal is deposited on the blank using a thin film process. If everything is nice and clean, and it's the right metal it sticks very nicely. If it's the wrong metal or stuff is dirty - not so much. Add to that (possibly high) drive and the metal comes off. Loose enough metal and it stops working.

4. Air in the package. On high performance / low frequency crystals (think 5 MHz 3rd), the parts run in vacuum. Let in air and the resistance goes up a lot. High resistance = no works.

5. Broken mounts. If you put enough vibration on a crystal for long enough, at the right frequency, you break the mounts. Often this is right where the mount hits the post coming up from the base. Happens mostly in mobile gear. You can do it with random vibe, but much easier to do with a sine at the right frequency.

6. Contamination. Put a fine layer of crud on the surface of a crystal and you can stop it dead. It acts as a mechanical damper, especially at low drive levels (like startup).

Stuff that you see in papers:

7. Micro fractures. The machining process that makes the blank bashes away at the blank on a microscopic level. Stone to windshield on a micro scale.

That's no where near a complete list, but it's a start. I have seen no convincing evidence of a wear out mechanism in a quartz resonator. Quartz is pretty sturdy stuff. There are lots of examples of parts older than I am still out there running. They do age on and on, they don't seem to die except when they are deliberately killed.

Bob

On Feb 13, 2011, at 10:26 PM, Bill Hawkins wrote:

Very interesting Bob Thanks.

It brings to mind an annoying issue I run into from time to time.

VCXO 12.8 Mhz used as a reference in communications gear.  Most of the gear is roughly 10 years old and of similar make, both mobile and portable styles.  They all are dropping in operating frequency. Many as far as 3000 hz at 850 MHz but it does vary with most in the 1300 hz area.  The trim adjustment is made through a programming interface and in some cases we are running out of adjustment.

My question is what is the process in the aging if the crystals that causes this drop in frequency?  I understand it may be mechanical but the level if vibration the radios are exposed to varies greatly and there does not seem to be a correlation with the frequency shift we are seeing.

Just thought I start a discussion.  I thinks still soy of on topic?

Shawn

Sent from my iPhone

On Feb 14, 2011, at 5:02, Bob Camp lists@rtty.us wrote:

And sorry about the typos!  Still getting used to this touch screen!

Sent from my iPhone

On Feb 14, 2011, at 19:47, Shawn Tayler shawn@xmtservices.net wrote:

On Mon, Feb 14, 2011 at 7:47 PM, Shawn Tayler shawn@xmtservices.net wrote:

All crystals change frequency as they age.  Some of this is that the
crystal absorbs mass as gas is defused into the structure or looses
mass by outgassing or the mounting metal defuses into the outer layer
or the metal oxidized.  All these are chemical changes

I've also read about microscopic cracks or defects in the crystal can
grow over time

In a radio the oscillator also has other components like caps and
resistor that can age.

Most crystal oscillators that you buy come with a spec for aging and
also an adjustment and I find the engineers have build maybe only a 20
year worst case life into the adjustment range

=====
Chris Albertson
Redondo Beach, California

On 2/14/11 7:47 PM, Shawn Tayler wrote:

I think aging of crystals is sufficiently time-nutty..

I had heard that one theory on aging is that the (microscopic)
vibrations are throwing off mass one molecule at a time (which would
raise the frequency).  And on the other hand, it's microfractures which
reduce the stiffness (lowering the frequency)...

I think the short answer is "nobody knows" except in pathological cases
(cheap crystal oscillators with aluminized plates that are oxidizing, or
something like that)

Hi

Drop in frequency = mass increase. Mass is moving on to the crystal as time goes on.

Since the package (hopefully) is sealed, there are a fairly limited number of ways to do that.

1. The case of the crystal was "more dirty" than the blank when things were sealed, equilibrium is to move to the blank.
2. The part is a welded seal (resistance weld) and they got a bit more energy in the weld than they needed. The spray from the weld is the mass headed to the crystal.
3. They didn't quite get them sealed and they are leaking. That's only going to work if they were vacuum sealed to start with (not likely).
4. The fill gas (assuming no vacuum) is reacting with the plating on the blank (also not likely).
5. The crystals may have a teflon anti-shock sleeve in them. Teflon is good stuff, but it does outgas. Net would be frequency down.

Stepping back a bit, let's look at what they are doing. At 850 MHz 3000 Hz is just about 3.5 ppm. Your 1300 Hz shift is about 1.5 ppm. If the crystals are 10 years old, they are running at an aging rate of 0.15 to 0.35 ppm / year. That's pretty good for a radio crystal that likely was specified to be at 0.5 ppm per year. What ever is happening to your crystals, it's very much in the "expected to happen" category.

Of course if it's 3,000 Hz past the end of the adjustment range - that's a lot more ppm per year. Back in the days I designed that sort of stuff adjustment range was spec'd as greater than +/- 15 ppm. That's a lot of range.

Bob

On Feb 14, 2011, at 10:47 PM, Shawn Tayler wrote: