Crystal oscillator circuit insulation

Thread Starter

k1ng 1337

Joined Sep 11, 2020
940
Hi,

I'm building an analog/digital clock based on a quartz crystal pierce oscillator. A goal of this project is to learn how each part comes together and eventually produce a high quality prototype (lack of better term) mounted inside a wooden enclosure. I am happy with the circuit design aside from failing to include temperature control. I don't want to get too crazy with controlling temperature because feedback is currently beyond my understanding which I can return to later.

Alternatively I am considering methods and materials to soften the temperature gradient experienced by the enclosure when exposed to sudden changes such as a door being left open for 10 seconds in the winter, here in Canada the breeze is a nice -40c at times! I haven't done enough testing to know how rapidly temperature changes a quartz circuit.

Temperature and fire suppression are aspects I'm investigating as long term fidelity of this project is important for my understanding. Any information on the topic is appreciated.

Regards,
 

MrChips

Joined Oct 2, 2009
30,708
Assuming that you want to design and build a clock that keeps accurate time, then your first choice for a time-base ought to be the 60Hz line frequency. It may come as a surprise but this is more accurate than a quartz crystal.

Your quartz crystal may be stable to 5-10 ppm. But if it is off by 5ppm your time keeping ability will drift further away over the long term.

(A stopped clock is more accurate. It is correct twice each day.):D

The short term accuracy of line frequency is nowhere as good as a quartz crystal but by the end of the day it’s back on track.

For a more sophisticated time keeper you may want to consider capturing the time from a reliable source such as NIST, GPS, or the web.
 

sparky 1

Joined Nov 3, 2018
756
Insulation should be fire proof.
An oven controller can be made with thermistor and a comparator like the LM393
The computer fan controller works the same. The fan is replaced with a low ohm resistor.
The selection of the resistor value can make a very small heater. On a hot day in the shade
the temperature inside the enclosure will be warm.. So on a cold day the temperature should also be warm.

An OCXO (oven controlled xtal osc) with good temperature stability and low phase noise will out perform HC49

Cement mixed with foam has been popular with home insulation, it is very fireproof and is low cost.
A double oven is used when precision gets crazy. Try a 15 Ohm 2 W resistor and a variable power supply.
An LM35 is also convenient can be placed inside as a temperature indicator.

 
Last edited:

Thread Starter

k1ng 1337

Joined Sep 11, 2020
940
Assuming that you want to design and build a clock that keeps accurate time, then your first choice for a time-base ought to be the 60Hz line frequency. It may come as a surprise but this is more accurate than a quartz crystal.

Your quartz crystal may be stable to 5-10 ppm. But if it is off by 5ppm your time keeping ability will drift further away over the long term.

(A stopped clock is more accurate. It is correct twice each day.):D

The short term accuracy of line frequency is nowhere as good as a quartz crystal but by the end of the day it’s back on track.

For a more sophisticated time keeper you may want to consider capturing the time from a reliable source such as NIST, GPS, or the web.
My fascination with clocks is their direct variation with the Earth's rotation. I'm in the process of producing the schematic of my circuit for verification which has 11 CD4017s dividing the signal down to 12 hours. I don't have much use for all the 4017's so this project is also about using up spare parts in addition to making best use of those parts. I'm satisfied with the time drift of my circuit on the breadboard so I'm looking at controlling the parameters more easily correctable such as temperature rather than getting extremely precise time. I have no problem resetting the clock once a month and it will be driving LEDs so I expect significant noise on the line after all the switching is said and done.

Insulation should be fire proof.
An oven controller can be made with thermistor and a comparator like the LM393
The computer fan controller works the same. The fan is replaced with a low ohm resistor.
The selection of the resistor value can make a very small heater. On a hot day in the shade
the temperature inside the enclosure will be warm.. So on a cold day the temperature should also be warm.

An OCXO (oven controlled xtal osc) with good temperature stability and low phase noise will out perform HC49

Cement mixed with foam has been popular with home insulation, it is very fireproof and is low cost.
A double oven is used when precision gets crazy. Try a 15 Ohm 2 W resistor and a variable power supply.
An LM35 is also convenient can be placed inside as a temperature indicator.

The cement and foam is similar to a basic corn starch fire retardant often used in home made forges.

I'm going to try the resistor idea. The clock will be indoors at 23c and I don't expect much variation. I found this excerpt:

"All resistors have the same efficiency, they convert the dissipated power to heat. The answer depends on whether you want to heat air or a surface. A resistor that is mounted to the chassis with screws will transfer the heat to a surface more efficiently. A resistor with poor contact to the surface, will mostly heat the air."

Thus I could achieve fair temperature control by always having the enclosed temperature a few degrees above ambient via passing current through a power resistor. Hysteresis will be an issue or is the circuit you described fine being in unstable equilibrium without feedback? I have played with the TMP36 and I've been meaning to pick up some thermistors, thermocouples and power resistors to start some experiments. I have 0.1 10W on hand but I imagine that will be hard to control.

Cool stuff.
 

MisterBill2

Joined Jan 23, 2018
18,167
First question is just exactly HOW ACCURATE you want to make that clock? Withing 1 second per week? or one second per year? Or a lot closer than that?? And how many digits of resolution? Second, milliseconds?, or microseconds?
Complexity and cost increase much faster than accuracy. So it becomes a trade-off.

To keep a crystal oscillator at a constant frequency independent of short term temperature changes a well insulated enclosure inside an insulated enclosure will be good, and not so expensive. And having a stable and well regulated power supply will also help. Putting the oscillation system in the same enclosure and keeping the supplied power constant will also keep the frequency stable. Using a metal shielding insulated enclosure will avoid needing extra heaters and the hazards perceived by the fearful. Well filtered and regulated power supply voltages, and constant temperatures, will keep the delays in the logic string at stable times. Actual ovens have a much greater power consumption and add more expense than adequate insulation and shielding.

Beyond that come the cesium vapor clocks and other similar devices, at a much higher price.
 

Reloadron

Joined Jan 15, 2015
7,501
Roman Black is an Australian electronics enthusiast who was an active member here on AAC with the screen name THE_RB.

https://www.romanblack.com/xoven.htm
I remember Roman, he devised some pretty cool DIY projects for home enthusiast.

I also have memories of the early TCXO used in early HP frequency counters. Somewhere I may still have one, an octal base with everything in a can. The can actually, in the US, used 120 VAC power and gave a stable 10 MHz output. Roman's approach is actually a pretty slick approach.

Today we can buy off the shelf small package crystal oscillators inexpensive. I still have a few Sale (Now Vishay / Dale) crystal oscillators like the XO-52B and the XO-43B in the 1.0 MHz flavor. Stable between 0 to 70 C. They have come a long way.

Ron
 

sparky 1

Joined Nov 3, 2018
756
Picture of crystal oven, a finite drive level was determined if the temperature settles enough the frequency almost level off slightly this was the most stable point. This is called the crystals turnover point. The oven sits on top of the impedance test unit. By averaging the low drive levels just above the thermal noise floor, the parameters L and C are derived shown fig 21 pdf below shows the record which describes what was known at finite low level in 2004. All the motional parameters are very accurate however we should not expect to come this close at home. We can start with a higher frequency crystal stabilize it, divide it down and let a microcontroller adjust the error when comparing to a known reference standard.


crystal oven.JPG

This reference comes from institut Femto-ST found at research gate:
https://www.researchgate.net/public...ystal_Resonators_at_Low_Drive_Levels_A_Review
 
Last edited:

MrChips

Joined Oct 2, 2009
30,708
I have this NIST traceable stopwatch.
I just checked the time against current internet time and it is 2 seconds off since I last set it 6 months ago.
That translates to 130ppb or 0.13ppm.

1641948559479.png
 

MisterBill2

Joined Jan 23, 2018
18,167
I still have not sen the description of just how accurate the system need to be, That affects how much it costs and how much power it consumes.
 

MrSalts

Joined Apr 2, 2020
2,767
I still have not sen the description of just how accurate the system need to be, That affects how much it costs and how much power it consumes.
The last similar thread on another site, that OP asked for a schematic for a "perfectly accurate" clock. I can't wait for this OP's answer.
 

MisterBill2

Joined Jan 23, 2018
18,167
The closest that can reasonably be achieved is to synchronize with the NBS standard clock using a phase-locked loop scheme, adding a bit of intelligence to correct the basic setting as any phase differences are detected. That would be adjusting the cesium vapor time base. Fairly complicated by most standards of consideration.
And all of that effort just to avoid missing coffee breaks?
Actually, the one second time signals from the GNSS satellites are quite accurate. But I have not seen how those pulses are used to stabilize the ten megahertz signal coming from the box. What sort of loop filter is used for a PLL with a 1: 1x101**7 divider to reach the 1 PP reference frequency??
 

Thread Starter

k1ng 1337

Joined Sep 11, 2020
940
Interesting replies, phase locked loops and global time standards are a challenge for another day. This project is about applying common engineering steps and thus some aspect of temperature "control" and fire prevention are among the steps. Once everything is all up and running I will be conducting tests against benchmarks. The performance and reliability of any one competent is not as important as being able to identify and log a change in it's operation over lifetime.

I am producing the schematic so please don't lose your shirts just yet haha. The resolution to the user is in minutes and I've incorporated variable capacitor in the pierce design for incremental tuning over time.

As suggested I think an enclosure inside an enclosure will be sufficient as ambient temperature is stable. I'll be making a trip to the store to try out some simple resistor heating methods. The timing circuit is entirely CMOS and will be on battery. I'll then use an optocoupler at the point in the circuit where driving loads begin (wall power) as there will be external terminals where I can connect any output (think of a projected custom 7 segment display). The circuit is going to be a big lamp / light show running around the whole house so I'll probably update to a GPS (or better) time eventually.
 

Reloadron

Joined Jan 15, 2015
7,501
The last similar thread on another site, that OP asked for a schematic for a "perfectly accurate" clock. I can't wait for this OP's answer.
Well OK. Might find a good deal on an old slightly used but not abused HP 5060A and as I recall just under 60 Lbs but a fine clock. Likely can find a used model for a few grand. Then too a wide range of GPS receivers out there can now be had for under about 5 grand. Actually time and frequency domain has gotten less and less expensive.

Now in the interest of keeping things simple and inexpensive I have as mentioned earlier several Half Size Clock Oscillator Enable/Disable laying here. These are originally Dale manufacture and now Vishay Dale Model 52B. They are the 1.0 MHz version and 5.0 Volt in gives a nice clean 5.0 Volt output square wave. If the thread starter is local to me here in the US I will be happy to drop one in the mail on my dime as in free to a good home. I'll never use half the stuff I have piled up before I am dead and most ends up in a dumpster anyway. Just shoot me a PM with a US mailing address and it's yours. That or roll your own.

Ron
 

MisterBill2

Joined Jan 23, 2018
18,167
The really interesting clock project here, quite a while back, was a teacher wanted to have a clock project not using digital IC devices,although I guess the display was digital. How to do division without flipflops registers, or counters? I suggested stairstep dividers and they saw that wold work, and so a digital clock with analog logic was born. A project that school kids could build and actually learn some real electronics from.
 

MrChips

Joined Oct 2, 2009
30,708
The really interesting clock project here, quite a while back, was a teacher wanted to have a clock project not using digital IC devices,although I guess the display was digital. How to do division without flipflops registers, or counters? I suggested stairstep dividers and they saw that wold work, and so a digital clock with analog logic was born. A project that school kids could build and actually learn some real electronics from.
Here is that thread.

https://forum.allaboutcircuits.com/threads/counter-circuits-for-all-transistor-digital-clock.156950/
 

ThePanMan

Joined Mar 13, 2020
773
produce a high quality prototype (lack of better term) mounted inside a wooden enclosure. I am happy with the circuit design aside from failing to include temperature control.
High quality without temperature compensation? That's like calling someone a stupid genius, or a smart idiot. And no - I'm not calling you that - just for the record.
your first choice for a time-base ought to be the 60Hz line frequency. It may come as a surprise but this is more accurate than a quartz crystal.
Not in my house. I have a few clocks that gain 2 minutes a month.
(A stopped clock is more accurate. It is correct twice each day.):D
Awe gees - I was gonna say that.
 
Top