Crystal Oscillator issue

[Techman_7]

I am facing problem in RTC crystal oscillator circuit in automotive instrument cluster. The RTC works fine when date & time is set. The crystal oscillations are proper within frequency limits. But, sometimes the oscillations stops & the date & time gets reset.

Datasheet for crystal R38, RTC IC & circuit is attached. I have calculated the safety factor & circuit margin for crystal. These values are within limits. The Vdd supply is 5V. I2C resistors are 1K each.

What can be the root cause for it?

 

[andre_luis]

The crystal oscillations are proper within frequency limits. But, sometimes the oscillations stops & the date & time gets reset.

Review the RTC battery connection to the holder, to check if it is firmly tied.

 

[dick_freebird]

I'd look at 5V supply noise / dropouts, maybe the ground as
well. Self synchronous noise can "beat back" crystal oscillation
if you get the right, wrong phase (for example too much load
for a derived clock could bump the crystal through-the-ground
current loop).

Automotive environment has some nasty things like load dump
that can blow through slow voltage regulators' low HF PSRR.
Might look at what happens if (say) you interpose a common
mode choke w/ stiff output capacitor between the system
and the oscillator.

 

[rho-bot]

Recheck your crystal circuit.
The RTC already has a integrated load capacitance [CL(itg), 15-35pF], wit ext. C16/17 you are adding 7.5pF on top.
This will not only trim the crystal to get a clock deviation, but will also overstress the crystal. Osc.margin should also be considered for osc.startup.

As a first task, remove C16&C17, and recheck operation.

 

[Techman_7]

Recheck your crystal circuit.
The RTC already has a integrated load capacitance [CL(itg), 15-35pF], wit ext. C16/17 you are adding 7.5pF on top.
This will not only trim the crystal to get a clock deviation, but will also overstress the crystal. Osc.margin should also be considered for osc.startup.

As a first task, remove C16&C17, and recheck operation.

The load capacitance as per crystal datasheet is 10 pF typical & as per crystal datasheet (for R38 crystal) the load capacitance is 12.5 pF. Is it because of the difference in load capacitances the oscillations stops?

On the CLKOUT pin, it is open drain. I have connected pull up resistor of 2K & observed waveforms on the output. On some instances I do not get waveform, but when I touch the crystal leads the oscillator starts that is the crystal is triggered. Whys is it so? Is it because of load capacitance values 12.5 pF not within tolerance (15/2 + 5 pF stray capacitance = 12.5 pF). CL tolerance is not given in crystal & IC datasheets. Even the load capacitors are removed, the crystal works fine?

I don't understand the CL(itg) = (Cosci * Cosco) / (Cosci + Cosco); this is given on page 32 of RTC IC datasheet. In the RTC circuit Cosci = Cosco = 15 pF & it is mentioned in datasheet that CL(itg) = 25 pF (typical). For 25 pF CL (itg), Cosci should be equal to Cosco = 50 pF. This is bit confusing..

What can be the root cause for this?

- - - Updated - - -

The RTC battery connections are OK & Vbatt is also OK.

 

[rho-bot]

I don't understand the CL(itg) = (Cosci * Cosco) / (Cosci + Cosco); this is given on page 32 of RTC IC datasheet. In the RTC circuit Cosci = Cosco = 15 pF & it is mentioned in datasheet that CL(itg) = 25 pF (typical). For 25 pF CL (itg), Cosci should be equal to Cosco = 50 pF. This is bit confusing..

Yeah, the datasheet & especially the footnote for CL(itg) is confusing & misleading, it should be rewritten :!:

On the CLKOUT pin, it is open drain. I have connected pull up resistor of 2K & observed waveforms on the output. On some instances I do not get waveform, but when I touch the crystal leads the oscillator starts that is the crystal is triggered. Whys is it so?

Maybe best to check these NXP Application Notes:
https://www.nxp.com/docs/en/user-guide/UM10301.pdf
https://www.nxp.com/docs/en/application-note/AN10652.pdf

Cosco has 25pf built inside the RTC, so no need for C17. Remove it.
Keep C16 for Cosci. With 25pF, you should typ. reach 12,5pF load.
Considering StrayC, PinC, BoardC and CapTolerances, you may need to do finetuning on C16, to get the needed clock accuracy.
With this, your circuit should run.

If osc. still doesn't start, or drops, it may be an issue with Oscillator margin mismatch. For this case, come back here again.

 

[Techman_7]

Yeah, the datasheet & especially the footnote for CL(itg) is confusing & misleading, it should be rewritten :!:


Maybe best to check these NXP Application Notes:
https://www.nxp.com/docs/en/user-guide/UM10301.pdf
https://www.nxp.com/docs/en/application-note/AN10652.pdf

Cosco has 25pf built inside the RTC, so no need for C17. Remove it.
Keep C16 for Cosci. With 25pF, you should typ. reach 12,5pF load.
Considering StrayC, PinC, BoardC and CapTolerances, you may need to do finetuning on C16, to get the needed clock accuracy.
With this, your circuit should run.

If osc. still doesn't start, or drops, it may be an issue with Oscillator margin mismatch. For this case, come back here again.

The oscillator margin is within the limits. No problem for margin.

 

[KlausST]

Hi,

Mind: datasheets show the circuit how it should be.
Most datasheets are correct, thus - if the circuit doesn´t work - there may be a problem with your circuit.

Please post photos of your circuit, where we can see all the wiring.

Klaus

 

[Techman_7]

Hi,

Mind: datasheets show the circuit how it should be.
Most datasheets are correct, thus - if the circuit doesn´t work - there may be a problem with your circuit.

Please post photos of your circuit, where we can see all the wiring.

Klaus

Yeah, the datasheet & especially the footnote for CL(itg) is confusing & misleading, it should be rewritten :!:


Maybe best to check these NXP Application Notes:
https://www.nxp.com/docs/en/user-guide/UM10301.pdf
https://www.nxp.com/docs/en/application-note/AN10652.pdf

Cosco has 25pf built inside the RTC, so no need for C17. Remove it.
Keep C16 for Cosci. With 25pF, you should typ. reach 12,5pF load.
Considering StrayC, PinC, BoardC and CapTolerances, you may need to do finetuning on C16, to get the needed clock accuracy.
With this, your circuit should run.

If osc. still doesn't start, or drops, it may be an issue with Oscillator margin mismatch. For this case, come back here again.

Also, the oscillator starts when the crystal leads are touched by oscilloscope probe or by bare hand.
This is big surprising why it happens? The load capacitance value is as per the crystal datasheet (12.5 pF)

 

[rho-bot]

Please post your actual osc. circuit & pcb layout.
Otherwise further help not possible, as I am not aware, which of my& other suggestions you followed.

 

[Techman_7]

Please post your actual osc. circuit & pcb layout.
Otherwise further help not possible, as I am not aware, which of my& other suggestions you followed.

Please find the circuit.

 

[KlausST]

Hi,

S1J is a 600V 1A rectifier diode with standard voltage drop.

I recommend to use BAT54C or similar. (there are cheaper ones)
* smaller
* cheaper
* less voltage drop
* less part count
* less reverse current
* faster (less trr means you get a smaller negative voltage spike when 5V is switched OFF; only valid for fast falling 5V)

Please post your actual osc. circuit & pcb layout.

Klaus

 

[rho-bot]

Please find the circuit.

OK, so you still did not consider to implement my hints for improvement... Feels bad, man :sad:
Fix it first, and report back.
Besides, what's the battery voltage B1?
And give layout extract.