Crystal frequency changes from bread board to final circuit

Thread Starter

dansawyer

Joined Jun 3, 2018
7
The circuit is from an oscillator in a Stabilock 4040. I am trying to repair the instrument. The original oscillator stopped working, after checking multiple parts I identified the xtal as the culprit. The project is to replace the xtal. I found a 10 MHz crystal and tested it in a simple 74ls00 circuit. The circuit oscillated at a about 40Hz below 10 MHz. (the original xtal failed in this circuit)
I proceed to replace the xtal in the attached circuit, please forgive the image quality. All went well on the replacement. However under bench testing the circuit oscillated at 3 k Hz above 10 MHz. (Bench testing was done by stubbing power to the board, this seems to work without issue ) This is way outside the adjustment range.
I removed the voltage control capacitor and tried replacing it with various values. The closest I could get was a large value cap, .1 uf to pull the frequency down to about 10,001,200 or 1200 Hz too high. I checked the fixed parallel cap and it is 18 pf as per the circuit.
(The issue is not frequency measurement. The time base is calibrated from a GPS synced oscillator)
Questions:
1. What makes the attached circuit pull the frequency up by over 3 kc? (the xtal is in an oven of sorts which raised the temperature from 70 F to about 140 F )
2. Is there a way to pull this back down to 10 MHz?
3. Why would the crystal frequency change from a breadboard circuit to the oscillator circuit?
Thanks, Dan
 

Attachments

be80be

Joined Jul 5, 2008
2,072
The bread board loads the crystal down a little never really seen how much but you have capacitance on the pins from how a bread board is made. You could add 2 small caps from each side of the crystal
 

Thread Starter

dansawyer

Joined Jun 3, 2018
7
The bread board loads the crystal down a little never really seen how much but you have capacitance on the pins from how a bread board is made. You could add 2 small caps from each side of the crystal
Thank you. Do you mean from the crystal to ground?
 

AlbertHall

Joined Jun 4, 2014
12,344
Is your 10MHz crystal frequency specified at parallel or series resonance and which type is the circuit intended for? The two resonances happen at slightly different frequencies.
 

ebp

Joined Feb 8, 2018
2,332
There is someone at AAC who I believe is quite expert on crystals. Unfortunately, I can never remember his name or nym when I want to - begins with b and ends with ov. With luck, he'll notice your post and have some useful input. Someone else is sure to know who I mean and may ping him for you.
 

SLK001

Joined Nov 29, 2011
1,549
You answered your own questions. The original xtal was in an OVEN!!! To get proper frequency operation of the xtal, the blank with an S-curve that is flat at 140° is selected and ground to 10MHz. You MUST replace the bad xtal with one with the same S-curve, or your frequency will never be close.
 

MrChips

Joined Oct 2, 2009
30,706
There is someone at AAC who I believe is quite expert on crystals. Unfortunately, I can never remember his name or nym when I want to - begins with b and ends with ov. With luck, he'll notice your post and have some useful input. Someone else is sure to know who I mean and may ping him for you.
Is it @Bordodynov you are thinking of?
 

Thread Starter

dansawyer

Joined Jun 3, 2018
7
Thank you all for input. This input led to creating a second, series test circuit. This circuit is named Pierce-IC-close. The original crystal oscillated at about 10,002,000 Hz in the instrument.
I have a collection of 10 MHz crystals, I tried one from a second lot. That one oscillates at about 9,998,000.
I don't know if the instrument circuit is parallel or series. That said is it reasonable that the crystal from the second set that oscillates in the series test circuit at 9,998,000 can be tuned to 10 MHz?
Again thank you.
 

MisterBill2

Joined Jan 23, 2018
18,167
One additional consideration is the case of the crystal, which may, or not, be tied to the circuit common. That can have some effect on the frequency as well. But not as much as an oven.
 

Thread Starter

dansawyer

Joined Jun 3, 2018
7
One additional consideration is the case of the crystal, which may, or not, be tied to the circuit common. That can have some effect on the frequency as well. But not as much as an oven.
How can this be tested? Is it as simple as measuring the voltage of the case or temporarily shorting it to ground?
 

MisterBill2

Joined Jan 23, 2018
18,167
How can this be tested? Is it as simple as measuring the voltage of the case or temporarily shorting it to ground?
Observing the original crystal installation should make it obvious if the original crystal case was grounded. But an ohm meter check would provide an answer in the case of a clamp holding the crystal. On a breadboard setup a short and direct to the oscillator circuit ground would be the way to determine what the effect on the frequency would be. The amount of change will depend on the particular circuit as well as the physical arrangement of the components.
 

recklessrog

Joined May 23, 2013
985
Although rather difficult to see in the diagram, the feedback control circuit shows a variable capacitance diode in parallel with the capacitor marked * which could mean "selected on test." These need to be at the same temperature as the Xtal for the frequency to be "pulled" correctly, Testing of the frequency can only really be done once everything is correctly housed in the oven and sufficient time allowed for stabilisation.
The 1k preset r141 is the frequency set pot and should allow a small adjustment to the frequency.
Failing that, you Could fit a small value variable ceramic trim capacitor in place of the capacitor marked *, set the 1k pot half way, adjust the capacitor until the frequency is close, then final trim with the 1k pot.
The temperature coefficient of of the capacitor also has to be considered. But, as I said earlier, all this needs to be done with everything in the oven at a stabilised temperature and may well take you several attempts before it is correct.
One other thing to remember is crystal aging. When you replace a Xtal in a circuit, it can take many hours for the crystal to age and reduce any frequency drift.
Cheap xtals are a nightmare as usually they drift appallingly with temperature and time. Buy from a supplier who can give you a calibration sheet with the xtal for critical applications.
 

Thread Starter

dansawyer

Joined Jun 3, 2018
7
Success. Thank you all. The first key was to find a test circuit that mimicked the actual circuit. That produced a close match. It required replacing the * capacitor with a smaller one. Once that was done the adjustment went without incident.
Dan
 

Bordodynov

Joined May 20, 2015
3,177
So I figured out how the additional capacitors and inductors affect. On the left, I have presented diagrams of simplified oscillators. The oscillator works at maximum.CirQ.png PlotQ.png
 

Thread Starter

dansawyer

Joined Jun 3, 2018
7
The crystal was chosen from a set I had on hand. Based on the experience with the first attempt I chose one that was below the desired frequency by about the difference between the test board and the instrument. At ambient, no oven, in a simple series ls7400 oscillator the crystal is about 2000 Hz low. That 2k is consistent with the 2k Bordodynov modeled above. In circuit it came in a few hundred Hz high. In order to put the variable cap diode in the mid range of the adjustment required replacing a 18 pf parallel capacitor with a 4 pf one. After that the only tricky part was the variable cap diode bias resister. It seemed to have flat spots in the adjustment. The final oscillator consistently measures +/- 2 Hz.
 

MisterBill2

Joined Jan 23, 2018
18,167
The crystal was chosen from a set I had on hand. Based on the experience with the first attempt I chose one that was below the desired frequency by about the difference between the test board and the instrument. At ambient, no oven, in a simple series ls7400 oscillator the crystal is about 2000 Hz low. That 2k is consistent with the 2k Bordodynov modeled above. In circuit it came in a few hundred Hz high. In order to put the variable cap diode in the mid range of the adjustment required replacing a 18 pf parallel capacitor with a 4 pf one. After that the only tricky part was the variable cap diode bias resister. It seemed to have flat spots in the adjustment. The final oscillator consistently measures +/- 2 Hz.
That accuracy is very impressive. What did you use to measure the frequency ? Was there a warm-up drift period? Of course I realize that the performance of a circuit is quite dependent on the crystal.
 

Bordodynov

Joined May 20, 2015
3,177
Dan.
I chose the quartz parameters so as to get (approximately) the frequencies observed by you.Look at the results of my calculations carefully.It turned out that when the capacitor is shortened to 18 pF and 150 pF capacitors are replaced by 360-390 pF, 10,000 kHz can be obtained.It may be necessary to select the mode of the transistor (because of oscillation failure).
Draft2tranR.png
 
Top