UK Vintage Radio Repair and Restoration Discussion Forum

UK Vintage Radio Repair and Restoration Discussion Forum (https://www.vintage-radio.net/forum/index.php)
-   Components and Circuits (https://www.vintage-radio.net/forum/forumdisplay.php?f=40)
-   -   Quartz crystal accuracy over time (https://www.vintage-radio.net/forum/showthread.php?t=142932)

Philips210 11th Jan 2018 6:00 pm

Quartz crystal accuracy over time
 
Hi.

It's probably a basic question but do quartz crystals deteriorate over time even if not used? I appreciate that they tend to resonate at a lower frequency if used over some time which I take to be due to mechanical wear through usage. I have had to replace a couple in some digital clocks where the time was running slow.
So if I have a brand new crystal that's kept in it's packaging for say 10 years, then will it perform as if it was a brand new crystal or is some deterioration to be expected?

Regards
Symon.

MrBungle 11th Jan 2018 6:10 pm

Re: Quartz crystal accuracy over time
 
Yes they do age over time. The ones in digital clocks seem to be terrible. They are cut differently to other crystals as well apparently. I have a Casio watch from the 1980s that loses about 10 minutes a week.

Based on my observation of a large bag of 40m calling frequency crystals (7030KHz) which were dated early 1980s, they have drifted down by quite a bit with the correct load capacitance.

Usually you can add some capacitance to the circuit to tune them rather than replace the crystals however.

Philips210 11th Jan 2018 6:22 pm

Re: Quartz crystal accuracy over time
 
Ah, thanks for that MrBungle. I wonder what the reason for the drift is even if not used.

Regards
Symon.

MrBungle 11th Jan 2018 6:40 pm

Re: Quartz crystal accuracy over time
 
Have a look here - probably explains it better than I can :)

https://en.wikipedia.org/wiki/Crysta...lity_and_aging

G6Tanuki 11th Jan 2018 6:41 pm

Re: Quartz crystal accuracy over time
 
It depends on the 'cut' of the crystal. http://www.radio-electronics.com/inf...s-at-sc-ct.php Some cuts are more-stable than others.

As an example, I recently dragged out of retirement a Pye W15FMB Westminster two-way VHF radio that I'd used for 145MHz mobile in the 1980s. This uses xtals on 4.mumble MHz whose output is successively multiplied-up to 145MHz. To my amazement after 20 years in hibernation in the attic, all the channels were still within 1KHz.

Philips210 11th Jan 2018 6:44 pm

Re: Quartz crystal accuracy over time
 
Interesting stuff, I'll take a look through the above links later. Thanks.

Regards
Symon.

Radio Wrangler 11th Jan 2018 7:09 pm

Re: Quartz crystal accuracy over time
 
Cheapo crystals in plastic or soldered-together cases can get shifted due to muck getting onto the surface of the crystal.

Once we get away from contamination, qood quality quartz crystals have a logarithmic ageing characteristic. When they are made, their frequency is off of the frequency they will finally settle to. In their first week from manufacture they will drift so many parts per million in frequency. In the next ten weeks, they'll drift pretty much the same number of PPM. In the next hundred weeks, the same again, and then the next thousand weeks. and so on.

So if you order a crystal with tight specs from a reputable manufacturer, they make it, and then they keep it while it 'ages in' usually in a temperature-cycling oven to help it along.

So, your old crystals will be more stable rather than less

David

CambridgeWorks 15th Jan 2018 10:58 am

Re: Quartz crystal accuracy over time
 
Pye Telecom used soldered can HC6/U for lowest spec (eg. mobiles), Glass for base station and cold weld (cw) metal cans in latter days for base station. Glass was the most expensive.
Rob

MrBungle 15th Jan 2018 11:05 am

Re: Quartz crystal accuracy over time
 
I'm quite impressed with at least one HC6/U crystal I have. It is a 1MHz one dated 1969. Has the load capacitance on the can. Plopped it in an oscillator circuit with the right capacitance and hey presto 1.00001 MHz on my counter (!)

G0HZU_JMR 14th Feb 2018 2:45 am

Re: Quartz crystal accuracy over time
 
Sorry to dig up a fairly old thread but I knew I had some old/decent 10MHz xtals here somewhere and I found the old sample bag today. It's always the same... I find 'lost' things when I'm looking for something else that is lost.

Back at work in 1993 I designed an RF converter that needed a fairly decent (but cheap!) crystal reference. A TCXO or OCXO was way too expensive and I got a bag of samples from IQD. The spec for these was:

CRYSTAL 20/10/20/30 10000KHz IQD

I think this is , 20ppm initial calibration, 10ppm over -20 to +70degC and 30pF load capacitance.

Sadly, I've only got one unused one left in the IQD sample bag, there were originally 10 of them. The bag is dated July 1993 by IQD.

I stuck the crystal on my E5071 VNA this evening and took a two port model of it. Note that my VNA has the snazzy 1E5 OCXO option fitted and so it can measure frequency very accurately. This measurement also required a narrow RBW and some averaging on the VNA but the exported VNA model showed on a simulator that with a 30pF loading it would be resonant at 9.999936MHz. The VNA reference plane was set right at the crystal legs.

This is only out by 64Hz and is well within spec even today. I haven't tried it in an oscillator but the VNA method should be the most accurate. I could try resonating it with a 30pF capacitor and measure it again on the VNA but I think this would be slightly less accurate due to strays and the accuracy of the 30pF test cap at 10MHz.

Back then this was a fairly serious 'standard' crystal because of the initial ppm spec and the spec over temperature. But it looks like it is still in spec today for initial calibration.

Radio Wrangler 14th Feb 2018 8:41 am

Re: Quartz crystal accuracy over time
 
Typical specs I used for good quality crystals were:

AT cut, series mode, 30pf load.
+/-10ppm initial manufacturing tolerance
+/-10ppm over 0-70C
+/-10ppm ageing in the first year

Then there were fancier ones with parabolic freq/temp curves centred on +80C for ovened use.

Evacuated glass housings were best (valve packaging technology!)
Glass HC6U were good
Cold-welded metal cans showed about double the ageing of glass HC6U.
Soldered metal HC6U were at least ten times worse than glass HC6U.

All crystals have both series and parallel modes, and a full set of overtones of both. When a crystal is described as one particular mode and one particular overtone, it simply means that that is the resonance the manufacturer planted as accurately as possible on the required frequency. The other modes (overtones and fundamentals) are still there and can be used. Just remember that overtone frequencies are not precise integer ratios of the fundamental. This can help when you're desperate for a crystal close to some particular frequency.

David

G8HQP Dave 14th Feb 2018 1:17 pm

Re: Quartz crystal accuracy over time
 
I seem to recall reading somewhere (RSGB Handbook?) that crystals intended for overtone use are cut or treated in a way which discourages but does not fully inhibit the fundamental mode. Is this true or a myth? I suspect it is a myth, and crystals will more happily oscillate at their fundamental mode if given the chance.

MrBungle 14th Feb 2018 1:25 pm

Re: Quartz crystal accuracy over time
 
I've tried overtone crystals in my crystal tester and they quite happily oscillate at the fundamental only. I think they are made purely so that the overtone frequency is pretty accurate as the harmonics aren't exactly 3, 5x etc the fundamental.

Radio Wrangler 14th Feb 2018 1:34 pm

Re: Quartz crystal accuracy over time
 
Surface contouring is used by better crystal manufacturers to improve activity on chosen modes, and to worsen that on other modes but the effect is slight. If you really want a specific mode it is safest to have an LC resonator in the oscillator to force it. So you can choose modes other than the intended one with a bit of care, and that care is a good protection even if you want the intended mode. There is no guarantee that the intended mode is the most active

Crystal oscillators with no other tuned circuits in them are common, save money, but are risky.

David

G0HZU_JMR 14th Feb 2018 4:39 pm

Re: Quartz crystal accuracy over time
 
One other way to operate on an overtone is to starve the circuit of negative resistance at the fundamental. If you plot the negative resistance vs frequency looking into the base of the transistor it typically has an abrupt response where the transition into negative resistance occurs at a sharp knee point. Below this the resistance is positive all the way down to LF. It should be possible to get adequate negative resistance at the overtone but have positive resistance at the fundamental. This can be achieved by controlling the feedback and the resistive loading at the base of the transistor. The circuit would look the same as a classic Colpitts crystal oscillator but with different values for the resistors and capacitors. Some care would be needed to avoid having spurious responses at higher/unwanted overtones but in theory this approach would work. I'm not sure I recommend it though!

G0HZU_JMR 14th Feb 2018 7:05 pm

Re: Quartz crystal accuracy over time
 
I had a look through my stash of old UART/timing crystals and found some old/decent/unused IQD crystals that still have a known spec with them.

These are 4.032MHz crystals and they would have been intended for use with a modem IC some time around 1996. There's no date on the bag but I remember that these were intended for use in a little low power module that had to survive being buried in frozen ground or in hot sand and could be deployed anywhere around the world.

The temp spec is over -40deg to +85degC and the initial cal tolerance is 15ppm and the load capacitance is 30pF. I'll stick one on the VNA later tonight and maybe also knock up a test osc circuit and compare all of them for how much variation there is in frequency between each crystal. I've still got 8 of them in the bag and they are about 20 years old and these have never been used :)

G8HQP Dave 14th Feb 2018 9:31 pm

Re: Quartz crystal accuracy over time
 
Quote:

Surface contouring is used by better crystal manufacturers to improve activity on chosen modes, and to worsen that on other modes but the effect is slight.
Thanks for clarifying that. So the information I saw is not a myth, but it is not always true and best not relied on even when it is true. I would generally expect to put some LC with the crystal in an overtone oscillator.

Radio Wrangler 15th Feb 2018 12:28 am

Re: Quartz crystal accuracy over time
 
I've seen some fundamental oscillators which erratically jump onto overtones and back, so an LC in a fundamental oscillator can be seen as insurance!

The main modes of a crystal can be viewed as a cluster of modes with extremely small frequency offsets between them. Oscillation seems to wander around these sub modes. They're close enough together that this may seem like angels dancing on a pinhead, if it wasn't for bits of energy seem to be left behind by mode umps, and then reabsorbed with a bit of phase drift when the mode jumps back. This seems to fit some of the small but abrupt phase umps a running crystal oscillator can do. They can run quietly for a good while, and then exhibit a group of little phase umps. Rather like the horrible popcorn noise that used to plague certain opamps. Don't worry about this unles you need to go into ultra low phase-noise oscillators at offsets in the few hertz region.

And crystals are gravity sensitive as well... turn them over and the frequency shifts, turn themback over and it shifts back. They act as microphones for rumbly bearings in cooling fans!

David

G0HZU_JMR 15th Feb 2018 12:59 am

Re: Quartz crystal accuracy over time
 
It's definitely worth examining a crystal on an analyser to assess the risks of adjacent spurious modes. I've seen the nearby spurious modes on a VNA on plenty of crystals but I've never seen any problems caused by them even though they can look quite alarming. Maybe I'm not looking close enough though... I rarely measure oscillators below about a 10Hz offset. The spurious modes tend to be several kHz away on a typical VNA sweep. Maybe they can get excited when the waveform is at a certain part of the cycle but I've never witnessed any adverse effects so can't really comment.


Quote:

I would generally expect to put some LC with the crystal in an overtone oscillator.
I think you can also exploit the RC loading of the transistor to cause a very large positive base resistance at the fundamental so there would be no need to add an LC circuit to prevent oscillation at the fundamental. I have no experience of doing this but I tried it on a couple of simulations of typical transistor circuits and it isn't difficult to achieve this goal and it should allow a 3rd OT oscillation to happen at the desired frequency F without anything happening at the fundamental. I'd expect a fundamental crystal at F/3 to oscillate here as well (at F = the 3rd OT) but it will probably have lower amplitude and the frequency would presumably be a bit off compared to a properly designed/optimised 3rd OT crystal.

But this approach is risky and not very efficient and I'd much prefer to use an LC circuit somewhere. I've not designed many OT oscillators over the years at work, just a handful maybe and that was many years ago. They all used an LC tank somewhere.

G0HZU_JMR 15th Feb 2018 1:11 am

Re: Quartz crystal accuracy over time
 
I only had time to look at one of the 20 year old 4.032MHz crystals this evening on the VNA but the results were interesting. This crystal has a very high motional inductance and it is therefore quite 'stiff' in terms of its resonant frequency when the parallel capacitance is tweaked away from 30pF. But with 30pF loading on a simulator it was resonant at 4.032025MHz so was just 25Hz high. Changing the 30pF to 20pF on the simulator only shifted the frequency up 150Hz. Changing the loading to 40pF only shifted it down by about 30Hz.

I'll therefore make up a decent 30pF test cap tomorrow and test all 8 on the VNA for resonance. This will be better than trying them in an oscillator. I'm testing the crystals at a low drive level on the VNA to prevent any funny effects. I suspect that all 8 of these crystals will perform well in this test but I need to make up a socket based test jig for them and I'll do this tomorrow sometime.


All times are GMT +1. The time now is 7:01 pm.

Powered by vBulletin®
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.