BJT for the lowest phase noise XTAL oscillator

biff44 said:

Interesting points. I would like to learn more about crystal ordering. Could you please post a "typical" specification for a crystal resonator for us to learn from. Thanks.

Click to expand...

**broken link removed**

I agree with Dr. Drew - Croven is another reputable supplier. As indeed are most of the companies who manufacture SC crystals.

But I would add:
a) Typically, if you are ordering a small quantity of high-performance crystals you will do best to speak to the prospective supplier and follow the recommendations. The background is that different manufacturers have slightly different production processes, and the designs of the best-performing crystals therefore vary somewhat between suppliers.
b) Following on from a) If you are lucky the manufacturer may have stock of something you can use. With some manufacturer's this could be product manufactured specifically for stock. In other cases it could be "over-makes" from previous orders - well-aged crystals that were manufactured for another user.
c) The generally-published specifications are by no means complete. In reality the manufacturing process (including many more items than material grade, surface finish, plating schedule, and mounting arrangements) forms part of the specification. In addition, manufacturers will have their own internal test specifications for items such as "second level drive" (the series resistance at intermediate drive levels).

Hi,
I will outline a typical crystal specification I use. Just a note first that I am a radio technician, (to component level) by trade and I also order crystals for personal use. Sometimes I have to work out the crystal specification from measurements on an original crystal, but thats another thread.
The specifications are pretty basic, but one doesnt need to over specify.
I will add some notes where applicable
(1)Frequency to 10 Hz. (Or 1 Hz at HF)
(2)Type of Cut. The most common is probably AT cut. For very high specification
oscillators such as OCXO, then SC cut. SC cut is normally more expensive and typically operated in an oven at elevated temperature
(3) Fundamental or overtone. Most generic type crystals for use up to 35 MHz or so are fundamental, but at higher frequencies odd order vibrational harmonics (overtone) are selected by the oscillator design. These may extend up to 9th in some cases but the crystal overtone must be specified. Typically for 100 MHz crystals the fifth overtone is selected. Note also that some high spec crystals at HF use an overtone mode
(4)The type of holder, and connections need to be specified. This can be downloaded from most crystal manufacturers websites. For wire-in crystals of ~12mm height, this is the HC49/U holder. The most common method of sealing these is resistance-weld. Although cold-weld is probably still available if one hunts around. Solder weld should not be used.
SMD packages are available, and are commonly used in modern equipment.
(4) Temperature range. This is typically -10 to +60C for room temperature use, with calibration done at ~ 25C. If using this type of crystal then one must specify the crystal frequency tolerance over the temperature range. This is normally in +/-ppm (Parts per million) A value of +/-10 ppm is typical, and one can select that from crystal manufacturers info.
If one is selecting a crystal for oven use, then an elevated temperature is selected depending on the application. For amateur type designs 60C is quite good (This elevated temperature is normally noted as the turnover temperature). Note that cutting of crystals for these exact temeperatures is very difficult, so some adjustment is normally available in the oven temperature of the OCXO to fine tune the turnover.
(5) The calibration accuracy is a specification where the crystal will sit when operating at its nominal temperature and correct load. Again in +/- ppm. 5 or 10 ppm is typical
(6) Load. For most HF crystals they are specified with a capacitive load such as 32 pF. The oscillator must present the correct load to crystal, for the crystal to operate on the correct frequency.
A crystal can operate in a series type circuit with, or without load capacitance. (Specify C-load value or Series resonance) Most overtone crystals are specified as series resonance, however they can be specified also with load capacity depending on the circuit
(7) Preaged. Typically specified in +/- ppm/year. About +/- 1 ppm is acheivable at low cost, using accelerated aging techniques by the crystal manufacturers. Otherwise wait for a year or more, to slow down. If you have a leaking crystal it doesnt slow down. (Been there, done that)
(8) ESR. (Equivalent series resistance) For 5th overtone crystals around 100 MHz, 30 to 35 ohms is acheivable and 3rd overtone are around 10 to 20 ohms.

I dont usually order HF crystals, but one can sometimes specify approximate motional parameters if needed fo specific applications.

(Google search Cliftonlaboratories and crystal and he shows some basic tests on crystals with which motional parameters can be derived)

The crystals I get, also have some of the motional parameter info, in the data when received, and typical unloaded Q factors of 70,000-100,000 for overtone VHF crystals have been calculated. For HF crystals this can be significantly higher and values of just below 1 Million have been seen calculated from high quality German KVG crystals around 2.5 MHz. Note it is very easy to throw away such performance in simple circuits.

I think I covered most items with a general note. When ordering I normally create a list of these parameters for each crystal and put each item of info in. Double check the frequency as it easy to put numbers in the wrong order, or miscalculate

If you are unsure when ordering a crystal, talk to (or email) the crystal manufacturer or ask questions of others and research. One can never stop learning.

Hope that is of some help

Kevin ZL1UJG

zl1ujg said:

Hi,
I will outline a typical crystal specification I use. ...
Kevin ZL1UJG

Click to expand...

Yes, this is a good starting point. For "standard" crystals where you simply want reasonable performance the only additional data needed is that the crystal conforms to a recognised international standard (as these already include related limits on second level drive). But there is nothing in this specification that relates to high-performance applications - indeed the specification of maximum ESR without also constraining motional capacitance tends to encourage the use of large-area crystals - which is not generally advantageous. A separate specification of "Q" is useful - albeit the manufacturer will translate this to different parameters for measurement.

By way of guidance, the ultimate "Q" limit for AT quartz is
Q <~ 16,000,000/Frequency(in MHz). I think that the limit for SC is somewhat higher
Impurity content means that this limit cannot reliably be reached with commercially available quartz, and other features (e.g. surface finish, plating, mounting resistance) typically increase the ESR. Nevertheless, you should expect the Q of a low-noise (high-performance) crystal to be within about 30% of this. ["Cooking crystal" Q is typically only 100k at 10-MHz - a factor of 16 lower]

Additional factors required for high performance (e.g. burst noise and phase noise) can be measured as described in Dr. Drew's reference (the Croven catalogue/application-note). However, interpretation of these is somewhat specialised, so the way forward for most of us will be to select a trusted manufacturer and take advice.

One aspect not specifically addressed in the Croven catalogue/application-note is temperature-localised frequency deviations due to activity dips. These are most obviously of importance if you intend to provide frequency-temperature compensation; however, they can also have a direct impact on short-term stability, due to the combination of reduced Q and of coupling to mountings and to edges of the crystal (where surface finish is likely to be degraded).

Re the activity jumps ,a year or more back I was testing a homebuilt VHF crystal oscillator at home, doing thermal runs and noticed extreme frequency variations over a small temperature range. It was repeatable. The source of crystals I was using then, also showed some severe frequency jumps (permanently). No longer use that company.
Where I get crystals now, the parameters appear to be checked, both at 25C and at the temperature I specified in my order.
Crystals are of course only a small (but significant) part in the concept of a high performance oscillator... There are many more chapters to be learnt

Hi-

TonyBo said:

Hi Kevin,

Do you have contact info on the Czech company which made SC cut crystals for you?

Thanks,

Tony

Click to expand...

If you prefer an european source you might consider KVG (KVG GmbH | Englische Startseite).

**broken link removed**

Q

Hi-Q said:

Hi-
If you prefer an European source ....
Q

Click to expand...

If you want a crystal ex-stock (or near) and you have a specific frequency in mind, it's worth talking a few reputable SC cut suppliers to see who might have something ex stock

Western sources for SC crystals that I have reason to believe produce good product include (country given is location of manufacture - ownership may be different):
Bliley (USA)
Croven (USA)
Krystaly (Czech republic)
KVG (Germany)
Nofech (Israel)
Rakon (France)
Vectron (Germany, possibly also USA)

There are certainly others - omission from this list does not necessarily have negative implications

PAoSU said:

This is what I mean. My FET-oscillator (with 7 components) is only 10 dB worse than the rather complicated one of Ulrich.
BTW I know the theory already for about 50 years.....;-)
Regards,
Herb.

Click to expand...

Are you using your oscillator for audio application?
I'm developing a Butler oscillator for audio at 11.2896MHz but I'm curious about other circuit, can you publish the schematics?
Can you explain better your phase noise measurement tool?

andrea_mori said:

Are you using your oscillator for audio application?
I'm developing a Butler oscillator for audio at 11.2896MHz but I'm curious about other circuit, can you publish the schematics?
Can you explain better your phase noise measurement tool?

Click to expand...

10 dB is a big deal, you know!!!

andrea_mori said:

I'm developing a Butler oscillator for audio at 11.2896MHz

Click to expand...

Audio means audible. I guess not even bats on Mars cannot hear a 11.2896MHz frequency :-D

biff44 said:

10 dB is a big deal, you know!!!

Click to expand...

Watching the red plot of the published image that oscillator does not seem too bad, I read -100dBc@1Hz and -130dBc@10Hz, typical of a good OCXO (I own an OCXO with such that performance).
I don't know if PAoSU is using an SC-Cut crystal, since is not simple to reach that performance with AT-cut type.

- - - Updated - - -

vfone said:

Audio means audible. I guess not even bats on Mars cannot hear a 11.2896MHz frequency :-D

Click to expand...

I meant audio application, master clock for digital to analog conversion.

Re: butler crystal oscillator 5th overtone

Hi,
look at https://by-rutgers.nl/PDFiles/Reproducible Low Noise Oscillators.pdf
At the end the Xtal oscillator has been treated.
73 de Herbert.

- - - Updated - - -

PAoSU said:

Hi,
look at https://by-rutgers.nl/PDFiles/Reproducible Low Noise Oscillators.pdf
At the end the Xtal oscillator has been treated.
QUOTE]

I forgot about the results. Look at the Dutch article: http://by-rutgers.nl/PDFiles/De%20Beste%20Xtal-oscillator.pdf for thr pictures.

Click to expand...