PDA

View Full Version : The Crystal problem.



Loopy Byteloose
07-26-2006, 10:21 PM
When I look in Mouser or Digikey, I see a capacitance rating included with the crystals frequency.

But, when I buy crystals elsewhere, no one knows what this number is.
I just got some 40Mhz crystals locally·and need to figure how to work around this obstacle.·

Does one have to have an oscilloscope?·
Can a frequency counter help [I can easily use my Propeller as a frequency counter]?
Is there an easy 'Rule of Thumb' or a standard proceedure?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"If you want more fiber, eat the package.· Not enough?· Eat the manual."········


···················· Tropical regards,····· G. Herzog [·黃鶴 ]·in Taiwan

Bean
07-26-2006, 10:30 PM
At my work we use a S&A 250B crystal impedence meter to measure the load capacitance.
If no finish point is given, it's posible they are finished at series resonance.

Bean.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Cheap 4-digit LED display with driver IC·www.hc4led.com (http://www.hc4led.com)

Low power SD Data Logger www.sddatalogger.com (http://www.sddatalogger.com)

"The meaning of life, is to give life meaning" Unknown.
·

Loopy Byteloose
07-26-2006, 10:58 PM
Bean,
Is a crystal impedence meter a huge expense?
I was trying to figure a what to just work with trial and error.
I am also wondering if this in another 'goodie' that the Propeller can be set up to duplicate?

That would be grand.

What is the 'finish point' and how is it given?

I understand the 'load capacitance' to be the combined capacitance of the circuit board and the two parallel capactors. Is that corrected theory?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"If you want more fiber, eat the package.· Not enough?· Eat the manual."········


···················· Tropical regards,····· G. Herzog [·黃鶴 ]·in Taiwan

Loopy Byteloose
07-27-2006, 07:48 PM
Well I did some reading.
I can't afford a crystal impedance meter and I certainly don't want to be sending anyone a bag of crystals to check.

Correct me, if I am wrong.
Wikipedia has a good intro to crystal oscillators and I think I understand that the 'fiinish point' is the second reading from the crystal impedance meter. The first reading would be the series resonance and the second reading [or finish point] would be the parallel resonance. There are only two.

As an alternative, low cost method [trial and error], I was thinking of setting up an SX28 with the 40Mhz crystal and having the unit toggle one pin at a rate divided by the 40Mhz.

If I used a Propeller to measure as a frequency counter, I could they measure in I am getting the crystal to start and I have the correct capacitance for the crystal's design. If not, I could change the capacitors until I get something acceptible [withing 30PPM is what most crystal are sold as]. My only dilemma is that I need to calibrate the Propeller's crystal or the error may just be randomly magnified or cancelled out.

I guess I could look around for a TXO [accurate to 10x-7] as a calibation reference.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"If you want more fiber, eat the package.· Not enough?· Eat the manual."········


···················· Tropical regards,····· G. Herzog [·黃鶴 ]·in Taiwan

Bean
07-27-2006, 09:32 PM
Kramer,
Yes that will work, but like to say you will need an accurate reference frequency.

A 0.1ppM TCXO will not be easy to find. TCXOs are usually sold with the tolerance being the change over a certain temperature range. Because of crystal aging the TCXO will not hold 0.1ppM for long (maybe a couple months) after being calibrated.

What accuracy do you require from these 40Mhz crystals ?

P.S. At my work we are designing are own GPS locked frequency standard to take the place of our Rubidium standard. It's is working well and we have discovered that our Rubidium standard is 5*10x-10 parts low. The design is using an ultra stable OCXO, but we will probably try a TCXO in the future just to see what stability we can achieve. Our current design can lock to within 1*10x-8 in less than one minute. Of couse the design uses an SX chip for the microcontroller ;)

Bean.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Cheap 4-digit LED display with driver IC·www.hc4led.com (http://www.hc4led.com)

Low power SD Data Logger www.sddatalogger.com (http://www.sddatalogger.com)

"The meaning of life, is to give life meaning" Unknown.


Post Edited (Bean (Hitt Consulting)) : 7/27/2006 2:43:19 PM GMT

Loopy Byteloose
07-28-2006, 12:54 AM
Hi Bean,
From what I read, I need 1.7% for a CANbus device. I want to build about 12 nodes that will work well together.
They have a very sophisticated system of adjusting for drift over long transmission, so their tolerance is a bit tighter than the usual.
It seems crystals are necessary rather than resonators.

From what I understand, resonators are roughly within 3% tolerance [or 3000PPM] which is acceptable for most RS-232 interface. On the other hand, crystals are mostly sold at the 30PPM [which seems to me to be .003% if I got my zeros right].

I am not getting into anything as tight as a clock or GPS. So I may try a different approach, which would be to simply attempt to locate the parallel resonance via a frequency counter and tweaking capacitors.

There should be either one low and one high spot that the crystal wants to remain at OR as you mention there may in some cases only be one.
I just don't want to get the wrong resonance and a communication failure.

From there I guess I could presume that the actual number is trivial as the crystal is within its natural resonance and +/- 30PPM of the package markings.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"If you want more fiber, eat the package.· Not enough?· Eat the manual."········


···················· Tropical regards,····· G. Herzog [·黃鶴 ]·in Taiwan

Bean
07-28-2006, 01:21 AM
Kramer,
· I don't think resonators are that bad...
· I just looked at the datasheet for the Parallax resonator.
· It states 0.5% initial setting.
· And a total of < 1% over that whole temperature range.

Bean.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Cheap 4-digit LED display with driver IC·www.hc4led.com (http://www.hc4led.com)

Low power SD Data Logger www.sddatalogger.com (http://www.sddatalogger.com)

"The meaning of life, is to give life meaning" Unknown.
·

pjv
07-28-2006, 09:34 AM
Actually Kramer, you do have your zeros wrong;

3% tolerance is 30,000 PPM

Also, there are fairly simple ways to have an SX VP UART track an incoming signalling rate. Outgoing of course would have no such ability unless it received a sample reference first.

Cheers,

Peter (pjv)

Loopy Byteloose
07-28-2006, 05:43 PM
Thanks pjv and Bean,

Well, one can get off the shelf resonators with the capacitor adjustment included.
If the resonators are within 0.5%, it seems that crystals merely create an unneeded engineering problem for us little guys.

I had read the 3% figure on the net and that starting me thinking that I needed to commit to crystals if I wanted tighter accuracy.
But, I my applications are mundane communications. 500PPM or 0.5% is quite adequate.

BTW pjv,
I have more time for study. So I am about to have an operating copy of your RTOS design. I am just completing the hardware setup with an SX-28. Using the hidden instructions of the SXes still is very appealing to me and I absolutely must master multiplexing the device in order for me to fully understand it.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"If you want more fiber, eat the package.· Not enough?· Eat the manual."········


···················· Tropical regards,····· G. Herzog [·黃鶴 ]·in Taiwan