Years ago I had all sorts of accuracy issues with the PCF8563 RTC despite trying different crystals. The issue was exactly as you described, inaccurate calculation of the stray capacitance. Great tip about indirect near field measurement 👍 Awesome video as always
To be honest, I placed a 10p cap in that circuit purely by accident; it wasn't really a measured or calculated value - while making this video I was really surprised just how accurate the circuit was
This information is sooo good; thank you! I use 20MHz pill can crystal oscillators with x2 22pF ceramic caps for PIC18F4550 projects. I have noticed that, when proofing these oscillators on my DIY Colpitts oscillator test circuit, before installing them, they all seem to run a little slow... and about the same amount. Even though the MCU has never complained, now, I know why the oscillator is slow and what to do about it. Of course, any correction should be based on the conditions when the oscillator is mounted within the MCU circuit, not in the Colpitts tester. FesZ, we really appreciate you!
Thanks. Great video as always. Just a small note; the “capacity” of the capacitor is referred to as “capacitance” thats just to keep it in the same form of resistance and inductance.
I salvage lots of crystals from old electronics and I've been looking for a way to reuse them in the absence of a datasheet. Your video is the best source of information on this that I've been able to find, so thanks!
Here is another additional rule of thumb to use when designing and validating an oscillator design: Check the frequenzy at the extreme ends of the temperatures and voltages of your device under test (DUT). Measure at the maximal voltage and lowest temperature specified by your DUT. Measure at minimal voltages and highest temperature specified by your DUT. This way you will stay within specification even if voltages and temperatures changes.
My crystal osc wouldn't start in cold. MicroChip engineers suggested to use one 20pF and one 22pF capacitors instead of 2x20pF for a 4MHz xtal. Worked fine after that.
When you lower the temperature, the gain of the amplifier increases; you may also have a drive level/distortion issue. Anyway, I did read that having different values will help with the phase shift of the network and is recommended when startup is an issue.
I would go with the high impedance buffer connected permanently to the crystal circuit. Characteristic impedance of the circuit will continuously change if you move beside it anyways, not to mention the other variables like pressure, humidity, and temperature. Also there is the option of using mismatched loading caps. I've seen it used in few circuits.
I'm not sure if there is a practical buffer that has a lower added capacity than a piece of coax floating at ~1cm from the circuit; and leaving the buffer there after the measurements is not really practical either. Pressure/humidity/temp will have an impact, but its not like anything can be done about this... I mean, the final circuit will see these parameters vary unless the whole circuit is put into an airtight temperature controlled box (like a TCXO). I did mention using the mismatched capacitors, keeping the large one on XO, I think...
@@FesZElectronics your videos are a great inspiration. Since we never draw any practical power from the Oscillator circuit, we do keep the buffer as first line of signal extrapolation due to high input impedance. Therefore, that's why we keep the buffer in circuit. Please don't be intimidated, and keep up the good work. I love your videos because they fringe on the black magic design. I miss this all too much.
I guess you are referring to the case where the oscillator is built from discrete components and there is some sort of interface between this circuit and the next one where the signal is used - then of course a buffer will be needed and that can be used for the measurements; In the cases that I measured (the RTC and the USB codec) the oscillator and subsequent circuits where all inside of the same IC package; only the crystal and capacitors where external so the oscillator clock didn't have to go anywhere else (outside of the first IC) - that is why I didn't use a buffer based circuit
Indeed, that is also a valid way of measuring, as long as the oscilloscope has a good enough counting resolution. I noticed that not all oscilloscopes have a good enough measurement resolution to achieve the multiple digits needed for quartz measurements.
hello my friend I need help! I wanna to make a circuit that increases milivolt changes to volts I seen some circuits like exponential converters but they are noisy are there any circuit replacement for that thanks....
i have an idea for the video :D for example how to match a desired frequency from a transmitter by using LC tanks (and how to chose values to have a perfect match) with a receiver :D (no ICs or crystals etc.)
@@SaihoS1 Well if you don't add any capacitors you will be at the series resonant frequency - some amount of +ppm; and if you add an infinite amount of load capacity you end up at the parallel resonant frequency - again some amount of -ppm; the crystal specifies a load capacitance - if you add less, you have positive ppm, if you add more, you have negative ppm (in reference to the desired value)
Years ago I had all sorts of accuracy issues with the PCF8563 RTC despite trying different crystals. The issue was exactly as you described, inaccurate calculation of the stray capacitance. Great tip about indirect near field measurement 👍 Awesome video as always
To be honest, I placed a 10p cap in that circuit purely by accident; it wasn't really a measured or calculated value - while making this video I was really surprised just how accurate the circuit was
Boy, do you make this complicated!!! I love your work.
This information is sooo good; thank you! I use 20MHz pill can crystal oscillators with x2 22pF ceramic caps for PIC18F4550 projects. I have noticed that, when proofing these oscillators on my DIY Colpitts oscillator test circuit, before installing them, they all seem to run a little slow... and about the same amount. Even though the MCU has never complained, now, I know why the oscillator is slow and what to do about it. Of course, any correction should be based on the conditions when the oscillator is mounted within the MCU circuit, not in the Colpitts tester. FesZ, we really appreciate you!
Thanks. Great video as always. Just a small note; the “capacity” of the capacitor is referred to as “capacitance” thats just to keep it in the same form of resistance and inductance.
Another great video 👍we often overlook the devil in the detail and just place data sheet caps with scant understanding of why .
I salvage lots of crystals from old electronics and I've been looking for a way to reuse them in the absence of a datasheet. Your video is the best source of information on this that I've been able to find, so thanks!
Great video and to the point ! Thanks a lot !
Here is another additional rule of thumb to use when designing and validating an oscillator design:
Check the frequenzy at the extreme ends of the temperatures and voltages of your device under test (DUT).
Measure at the maximal voltage and lowest temperature specified by your DUT.
Measure at minimal voltages and highest temperature specified by your DUT.
This way you will stay within specification even if voltages and temperatures changes.
Great video, another detailed topic explained very clearly 🙂
My crystal osc wouldn't start in cold.
MicroChip engineers suggested to use one 20pF and one 22pF capacitors instead of 2x20pF for a 4MHz xtal.
Worked fine after that.
When you lower the temperature, the gain of the amplifier increases; you may also have a drive level/distortion issue. Anyway, I did read that having different values will help with the phase shift of the network and is recommended when startup is an issue.
thank you
Learned a lot, thanks.
clear & concise. Cheers!
Thank you very much!
Nice video..!! Thanks..!!
I would go with the high impedance buffer connected permanently to the crystal circuit. Characteristic impedance of the circuit will continuously change if you move beside it anyways, not to mention the other variables like pressure, humidity, and temperature. Also there is the option of using mismatched loading caps. I've seen it used in few circuits.
I'm not sure if there is a practical buffer that has a lower added capacity than a piece of coax floating at ~1cm from the circuit; and leaving the buffer there after the measurements is not really practical either. Pressure/humidity/temp will have an impact, but its not like anything can be done about this... I mean, the final circuit will see these parameters vary unless the whole circuit is put into an airtight temperature controlled box (like a TCXO).
I did mention using the mismatched capacitors, keeping the large one on XO, I think...
@@FesZElectronics your videos are a great inspiration. Since we never draw any practical power from the Oscillator circuit, we do keep the buffer as first line of signal extrapolation due to high input impedance. Therefore, that's why we keep the buffer in circuit. Please don't be intimidated, and keep up the good work. I love your videos because they fringe on the black magic design. I miss this all too much.
I guess you are referring to the case where the oscillator is built from discrete components and there is some sort of interface between this circuit and the next one where the signal is used - then of course a buffer will be needed and that can be used for the measurements; In the cases that I measured (the RTC and the USB codec) the oscillator and subsequent circuits where all inside of the same IC package; only the crystal and capacitors where external so the oscillator clock didn't have to go anywhere else (outside of the first IC) - that is why I didn't use a buffer based circuit
Hello how can I measure quartz watches crystal outside of the watch?
Thank you
You could also use oscilloscope frequency counter
Indeed, that is also a valid way of measuring, as long as the oscilloscope has a good enough counting resolution. I noticed that not all oscilloscopes have a good enough measurement resolution to achieve the multiple digits needed for quartz measurements.
Nice video, thanks :)
hello my friend I need help!
I wanna to make a circuit that increases milivolt changes to volts I seen some circuits like exponential converters but they are noisy are there any circuit replacement for that thanks....
Discrete or IC instrumentation amplifier will do the job
i have an idea for the video :D for example how to match a desired frequency from a transmitter by using LC tanks (and how to chose values to have a perfect match) with a receiver :D (no ICs or crystals etc.)
so actually from my 4 MHz quartz crystal i could get for example 27 MHz ? There is no specificiation for my crystal because it is from Aliexpress ...
Not really - the frequency that you can adjust by the added capacitors is usually ~+/-200ppm - for 4MHz that is +/-800Hz;
@@FesZElectronics I understand how to get -X ppm, but how to get +X ppm? Need negative capacitance?
@@SaihoS1 Well if you don't add any capacitors you will be at the series resonant frequency - some amount of +ppm; and if you add an infinite amount of load capacity you end up at the parallel resonant frequency - again some amount of -ppm; the crystal specifies a load capacitance - if you add less, you have positive ppm, if you add more, you have negative ppm (in reference to the desired value)
@@FesZElectronics Now it is clear. You consider 0 ppm @ the recommended Cload.