Series Resonant Crystal Oscillator

[shanil]

Hi,
Why is it that people go for series resonant crystal oscillator in high frequencies? What are the advantages and disadvantages of series resonant oscillator when compared to parallel resonant crystal oscillator? Is there a good resource for understanding design of series resonant oscillator(Like vittoz paper for parallel resonance)?

 

[goldsmith]

Hi Shanil
Who told that most of the times HF oscillators are based on series resonant mode ?
I think you'd better to refer to this fantastic book for crystal oscillator design :
Crystal oscillator circuit written by Robert J Matthys . it will answer all of your questions about crystal and crystal oscillators as well .
Anywhere , you can see my blog , for more informations too . take a look below , please :
https://www.edaboard.com/blog/1788/

About the specifications of series mode of resonant for a crystal , we can mention :
1- phase shift = zero degree
2- it will pass our desired frequency .
Hence we can easily use it in feedback path . ( for oscillators in positive feedback path )
Best Wishes
Goldsmith

 

[shanil]

Hi Shanil
Who told that most of the times HF oscillators are based on series resonant mode ?
I think you'd better to refer to this fantastic book for crystal oscillator design :
Crystal oscillator circuit written by Robert J Matthys . it will answer all of your questions about crystal and crystal oscillators as well .
Anywhere , you can see my blog , for more informations too . take a look below , please :
https://www.edaboard.com/blog/1788/

About the specifications of series mode of resonant for a crystal , we can mention :
1- phase shift = zero degree
2- it will pass our desired frequency .
Hence we can easily use it in feedback path . ( for oscillators in positive feedback path )
Best Wishes
Goldsmith

Hi Goldsmith,
Thanks a ton for replying. This was my first ever post in edaboard and your reply has inspired a lot.
1. Who told that most of the times HF oscillators are based on series resonant mode ?
1.a )Wikipedia link on this:- https://en.wikipedia.org/wiki/Crystal_oscillator#Resonance_modes
1.b )An ieee paper :- https://ieeexplore.ieee.org/xpl/art...&queryText=high+frequency+crystal+oscillators
The abstract mentions that it is a well known fact that above 30MHz, designers go for series resonant mode crystals.

2. I read that crystals behave as a simple resistor when used in series resonant mode. Is it true that we have to use an extra inductor to cancel the shunt cap of the crystal? Or is it that vendor will caliberate the motional cap to include shunt cap in the 1/sqrt(LC) term?

3. Why exactly we need to use series resonant crystal oscillator when pierce oscillator gives us extremely good performance? I am asking this to get an idea of real life design trade offs when going to series resonant crystal oscillators.

Once again thanks a lot for your replies.

--Regards,
Shanil.

 

[RCinFLA]

If you are referring to overtone crystal oscillators (AT cut above about 30 MHz), yes they are designed for series resonance. It is very difficult to avoid lower order modes and other spurious mode oscillation for an overtone mode oscillator operated in anti-resonant mode.

The primary attraction to anti-resonant (parallel), fundamental mode is no coils necessary in the oscillator design. Other problem at higher frequency would be parallel load capacitance would need to be small making PCB strays a greater factor in frequency accuracy.

Other words used for series versus parallel mode is (series) resonant versus anti-resonant modes.

 

[shanil]

Thank you RCinFLA for your kind interest. Can you clarify why the parallel load capacitance need to be small for high frequency crystals.?
My problem statement is to build a crystal oscillator for frequencies till 200MHz. As I was not getting fundamental crystals in this range, I started exploring overtone crystals. For fundamental mode crystals, ESR kept on decreasing as frequency increased but for overtone crystals ESR started increasing. Due to this, I was not getting a good range of gm_critical to work with(gm_critical_max/gm_critical_min was coming to be less than 3).
So working with parallel(anti-resonant) crystals was not feasible for high frequency crystals. I am stuck at two things mainly for series resonant oscillators:-
1. Is the shunt capacitance(Packaging capacitance) calibrated in the motional capacitance itself?
2. Do we need to add extra inductor for compensating the shunt cap?

 

[RCinFLA]

A 200 MHz crystal oscillator would be 5th or 7th overtone and not a good first project to learn about crystal oscillator design.

As the overtone goes up the Rs (ESR) gets higher and the number of close in spurious responses increase. The spurious response Rs to desired overtone Rs gets closer as overtone number goes up.

All of this makes it tougher to hit and run with solid operation at overtone goes up. The loop gain must be enough at desired overtone but not too much or the oscillator make skip to close by spurious responses. There has to be filtering in the close loop to kill the loop gain at unwanted fundament and overtone responses.

Most manufacturers do not specify Cm or Lm of their crystal. Modern measurement is usually an impedance analyser like Agilent 4294A which if calibrated properly will account for Co shunt capacitance caused by resonator coupling spots and package strays. But the system is only good to about 100 MHz with proper fixturing.

Rs for a good 3rd OT is in the 20 to 40 ohm range, 5th OT in 30 to 60 ohm range, 7th OT in the 50 to 100 ohm range for AT cut crystals.

Attached is a solid circuit for 3rd OT oscillator. Warp coil should be selected for operation -10 to -50 ppm below true Fs of crystal.

Running simulation on crystal oscillators can be very time consuming. It can take 1 or 2 msec for a high Q AT cut quartz crystal oscillator to startup. In computer circuit simulation time, this can be a very long calculation run. You can 'tickle' them with an injection source lightly coupled into circuit model to speed the simulation startup process.

For 200 MHz I would use a 3rd overtone 66.7 MHz design with a 200 MHz parallel resonant tank in the collector of attached circuit to pick off the third harmonic of the oscillator. Any resistance/reactance at the 3rd OT frequency in the collector path will hurt loop gain. The tank should be a high C, low L used for resonance at 200 MHz to minimize 3rd OT freq path loss.

 

[shanil]

Thanks many tons RCinFLA. I went and read many things on high frequency crystals.
The book suggested by goldsmith says better design for 20MHz to 200MHz overtone is the Buttler oscillator. For lower frequencies, the author pierce is the one to use. Anyone here has experience in Butler crystal oscillator design? Are there good mathematical sources for understanding the same?

Regards,
Shanil.

 

[RCinFLA]

Many folks got their name famous by grounding an oscillator at different points.

Actually, a Butler oscillator is a little different in that it uses an emitter follower to provide a low output impedance drive to a crystal. A resonant L-C tank circuit is used between common base and common emitter gain stages to address the loop gain problems at unwanted resonances. The tank circuit tuning effects the loop gain at oscillation and has variable phase shift that changes oscillator frequency somewhat. A tank circuit that has a temperature variance in resonance will cause the oscillator freq. to drift over temp.

 

[shanil]

I have to now fix the architecture for my crystal oscillator project. Since the book my R.J Matthys is suggesting the use of emitter follower Butler oscillator, I am planning to go with that. This is the quote from the book.
"The Butler emitter follower is a VHF harmonic oscillator, operating at or
near series resonance; a basic schematic is shown in Fig. 5.31. The circuit
is useful mostly at frequencies between 20 and 200 MHz. The amplifier
has a gain of one; the oscillator’s loop gain is provided by L, and
C, operating near series resonance. Tapping the crystal into the capacitive
side of the LC tank gives a phase lead, and the crystal then oscillates
either at or slightly above series resonance"
As it is with any project, if I get face issues at latter stages, I will be doing irreparable damage. Similar to the temperature drift problem mentioned by RCinFLA, could there be any other issues that I need to take care with Butler oscillator.? Problems that a first hand designer is possibly not aware of?