RTC crystal frequency measurement problem

Thread Starter

Electronic_Maniac

Joined Oct 26, 2017
248
Hi all,

I am measuring RTC crystal frequency of 32.768kHz using a 300MHz scope.

I am connecting the scope ground to my board ground. And using one channel I am probing the crystal pins.

For the first two times, I measured and I got the sine wave of the mentioned frequency.
But aftersometime, I tried to measure, I am not getting the sine wave. I am just getting a DC level shift of 100mV when I probe the crystal pins.

What could be the potential problem? Am I doing something wrong while measuring?

Please help.

Thanks.
 

MrChips

Joined Oct 2, 2009
19,420
You cannot measure the frequency of oscillation of a quartz crystal at its pins.
The crystal does not oscillate on its own. You need to put it into an oscillatory circuit.
Once you get it oscillating, you still should not measure the frequency at the crystal pins. The probe has enough capacitance that will shift the frequency. You need to feed the signal into a buffer stage then you can measure the frequency at the output of the buffer.
 

Thread Starter

Electronic_Maniac

Joined Oct 26, 2017
248
You cannot measure the frequency of oscillation of a quartz crystal at its pins.
The crystal does not oscillate on its own. You need to put it into an oscillatory circuit.
Once you get it oscillating, you still should not measure the frequency at the crystal pins. The probe has enough capacitance that will shift the frequency. You need to feed the signal into a buffer stage before you can measure the frequency at the output of the buffer.

Thank you MrChips.

I understand. But for the first two times I measured the frequency at the crystal pins, I can say that I got a perfect sine wave with my 32.768kHz frequency. But on the third and fourth time, I got a DC level shift only.

Can you please tell me how I got it for the first few times and failed in the next time?

Thanks.
 

danadak

Joined Mar 10, 2018
3,627
The crystal has a specific C loading spec, so scope probe probably
adding too much C and killing the oscillator. When probing, if
you do not have a FET low C probe, at least use a 10 x probe,
that might be just low enough to keep OSC running. Also
can decouple probe C by using a series R between probe tip
and pin, 10's of K ohms. Note that alters what you see on scope,
if you are trying to make measurements that will kill accuracy, but
you will be able to observe.

Or use a cmos inverter, connect that to the pin, and monitor its output.
Pay attention to layout, short leads, and minimize stray C.

Regards, Dana.
 

DickCappels

Joined Aug 21, 2008
5,950
Do you suppose that your crystal might be a crystal oscillator? If it is a crystal oscillator (with active components) then if should oscillate when an appropriate power supply is connected, some older ones require a pull-down resistor on the output. A crystal resonator is merely a (high tech) slap of quartz with electrodes.

upload_2018-12-7_21-43-57.png
Typical crystal

upload_2018-12-7_21-46-20.png
Crystal oscillator

upload_2018-12-7_21-47-39.png
Crystal Oscillator
 

Thread Starter

Electronic_Maniac

Joined Oct 26, 2017
248
The crystal has a specific C loading spec, so scope probe probably
adding too much C and killing the oscillator. When probing, if
you do not have a FET low C probe, at least use a 10 x probe,
that might be just low enough to keep OSC running. Also
can decouple probe C by using a series R between probe tip
and pin, 10's of K ohms. Note that alters what you see on scope,
if you are trying to make measurements that will kill accuracy, but
you will be able to observe.

Or use a cmos inverter, connect that to the pin, and monitor its output.
Pay attention to layout, short leads, and minimize stray C.

Regards, Dana.
Hi Dana,

I forgot to mention. The scope probe had 2 settings, 1x and 10x. While Measuring I made sure that I was using the 10x setting in the scope.

And what do you mean by killing the oscillator? Does it mean that the crystal will cease to work on the board? The board will malfunction

Thanks.
 

Thread Starter

Electronic_Maniac

Joined Oct 26, 2017
248
Do you suppose that your crystal might be a crystal oscillator? If it is a crystal oscillator (with active components) then if should oscillate when an appropriate power supply is connected, some older ones require a pull-down resistor on the output. A crystal resonator is merely a (high tech) slap of quartz with electrodes.

View attachment 165298
Typical crystal

View attachment 165299
Crystal oscillator

View attachment 165300
Crystal Oscillator
Hi DickChappels,

It is a crystal only. Not an oscillator.

Thanks.
 

danadak

Joined Mar 10, 2018
3,627
Killing means stop the oscillator. Should have used better terminology.
Its where the additional C causes feedback network to shift its phase
away from 180 degrees (180 typical, the other 180 comes from the amp
which is typically inverting).

Weather or not a stopped oscillator will damage something on board
depends on what else the circuit design depends on using the oscillator.
The 10X probing should not hurt the osc itself or the xtal.

Regards, Dana.
 

AlbertHall

Joined Jun 4, 2014
8,544
For the first two times, I measured and I got the sine wave of the mentioned frequency.
But aftersometime, I tried to measure, I am not getting the sine wave. I am just getting a DC level shift of 100mV when I probe the crystal pins.
Commonly, one of the two crystal pins connects to the input of an amplifier and the other pin connects to the output.
When connecting a 'scope probe you will see the oscillation on the output pin but connecting the 'scope to the input will stop the oscillator.
 

ebp

Joined Feb 8, 2018
2,332
The entire circuit for a real time clock is designed to be ultra low power. Older designs usually required two "loading" capacitors for the crystal, one of which was typically a trimmer cap to allow adjusting the frequency. The capacitors are very small, often less than the tip capacitance of a good scope probe at 10:1. Adding the capacitance of the probe, even on the oscillator output, will "pull" the frequency, so the measured frequency will not be the same as with the probe out of circuit. DC loading due to probe resistance may be sufficient to stop oscillation because the circuit is designed to be high impedance. Many of the more modern designs have internal trimming circuitry and the crystal is connected directly to the pins with no other capacitance.

Other crystal oscillators, such as the main oscillator on a microcontroller, usually have a low impedance output pin which can be probed with care and even used to drive external circuitry. No so for many RTCC chips. Trimming of the timing must be done by using the interrupt output to observe the frequency.

Although not the most convenient because of the need for satellite reception, using a voltage controlled crystal oscillator in a phase locked loop with the timing reference from a GPS receiver is a reasonably inexpensive way of getting an extremely precise reference frequency suitable for calibration of RTCC oscillators. With the newer RTCC chips with built-in frequency trimming and with the right bit of programming, you can hook the reference to the microcontroller and push GO and have the clock trimmed automatically.

32 kHz crystals for RTCCs generally have a nasty temperature dependence (non-monotonic with a fairly sharp peak) that can be a problem in devices that need to operate over a wide temperature range.
 

Thread Starter

Electronic_Maniac

Joined Oct 26, 2017
248
Killing means stop the oscillator. Should have used better terminology.
Its where the additional C causes feedback network to shift its phase
away from 180 degrees (180 typical, the other 180 comes from the amp
which is typically inverting).

Weather or not a stopped oscillator will damage something on board
depends on what else the circuit design depends on using the oscillator.
The 10X probing should not hurt the osc itself or the xtal.

Regards, Dana.
Thanks
 

Thread Starter

Electronic_Maniac

Joined Oct 26, 2017
248
Commonly, one of the two crystal pins connects to the input of an amplifier and the other pin connects to the output.
When connecting a 'scope probe you will see the oscillation on the output pin but connecting the 'scope to the input will stop the oscillator.
Hi AlbertHall,

By mistake, I tried probing on the two sides of the crystal, I got the sine wave on both the ends. But on one end, It was slightly distorted.
 

Thread Starter

Electronic_Maniac

Joined Oct 26, 2017
248
The entire circuit for a real time clock is designed to be ultra low power. Older designs usually required two "loading" capacitors for the crystal, one of which was typically a trimmer cap to allow adjusting the frequency. The capacitors are very small, often less than the tip capacitance of a good scope probe at 10:1. Adding the capacitance of the probe, even on the oscillator output, will "pull" the frequency, so the measured frequency will not be the same as with the probe out of circuit. DC loading due to probe resistance may be sufficient to stop oscillation because the circuit is designed to be high impedance. Many of the more modern designs have internal trimming circuitry and the crystal is connected directly to the pins with no other capacitance.

Other crystal oscillators, such as the main oscillator on a microcontroller, usually have a low impedance output pin which can be probed with care and even used to drive external circuitry. No so for many RTCC chips. Trimming of the timing must be done by using the interrupt output to observe the frequency.

Although not the most convenient because of the need for satellite reception, using a voltage controlled crystal oscillator in a phase locked loop with the timing reference from a GPS receiver is a reasonably inexpensive way of getting an extremely precise reference frequency suitable for calibration of RTCC oscillators. With the newer RTCC chips with built-in frequency trimming and with the right bit of programming, you can hook the reference to the microcontroller and push GO and have the clock trimmed automatically.

32 kHz crystals for RTCCs generally have a nasty temperature dependence (non-monotonic with a fairly sharp peak) that can be a problem in devices that need to operate over a wide temperature range.

Thanks ebp for the detailed answer.
 

Thread Starter

Electronic_Maniac

Joined Oct 26, 2017
248
But now you don't see oscillation on either of the crystal pins?
No.. Actually, I am testing my board and got to this crystal section.

Having lot of queries now and googling to find the answer.

1. Amplitude at both ends of the crystal is a little different in the order of millivolts. have to see why
2. how the oscillation starts when no input is given.

please share some light if you can. thanks
 

Thread Starter

Electronic_Maniac

Joined Oct 26, 2017
248
Basically a high G amp sits at a bias point in its linear region. Along comes
some noise, gained up, fed back to the input, and amped up again, fed
back to its input........


Regards, Dana.
So, it the noise that starts the oscillations. Thanks Dana. So, the amplitude at the ends of the crystal differ?
 
Top