RTC stops on LPC1768

[lpcware]

Content originally posted in LPCWare by Karlsberg on Thu Sep 13 07:18:09 MST 2012
Dear Sir/Madam,

for the first time we use the RTC on the LPC1768 in a new product.

XTAL1 and XTAL2 pins are connected to a quartz crystal 32.768KHz.
Additionally each pin has one 22pF capacitor to ground.
Tuning Fork Crystal Specifications:
Series resistance: max. 40k Ohm 
Load capacitance: 12.5pF
A CR2032 lithium battery is used for alternative power supply for the RTC.

Unfortunately, the RTC often stops when the 3.3V main power supply was turned on.
Likewise, it is possible to "blow out the RTC", even without halitosis. ;-)

Replacing the quartz has not helped.
Does anyone have any idea what is wrong?
Is there a recommendation for the quartz?
What's about the parasitic capacitance of the chip?

Thank you for your trouble.

Yours faithfully,

Karlsberg

[lpcware]

Content originally posted in LPCWare by hparracho on Thu Feb 05 04:22:14 MST 2015
You can config the CLKOUT pin to show the RTC frequency and use the scope on that pin without affecting the RTC input circuit.

[lpcware]

Content originally posted in LPCWare by rocketdawg on Wed Feb 04 12:50:09 MST 2015

---

Quote: probatista

Eg. using an oscilloscope to measure RTCX1 signal,

---

sometimes, but since we are talking about a few pf of capacitance, the scope probe itself may disturb the circuit (by adding stray capacitance or resistance).  Sometimes one can wave a probe near the area of the crystal and see a waveform.

[lpcware]

Content originally posted in LPCWare by probatista on Wed Feb 04 08:24:31 MST 2015
Isn't there an empirical method to check if the RTC clock signal is as it should be?
Eg. using an oscilloscope to measure RTCX1 signal, how should it look like? Sinusoid, 32.768 KHz, what else? What mininimum and maximum voltages? Wrong capacitor would cause lower amplitude?

[lpcware]

Content originally posted in LPCWare by Karlsberg on Tue Jan 08 08:02:47 MST 2013
Hi!

Now we use the following combination:

Watch Crystal: Q-32,768000K-TC 206B-20-E-12,5-TR
Load capacitance: 12.5pF
Series resistance: max. 40k Ohm
Additional capacitors to ground: 22pF +/-5% (Dielectric: C0G / NP0)

The microcontroller and the external circuit was applied with an acrylic coating.

The marking on the IC:

NXP
LPC1768FBD100
SM0442.1   01
ZSD1125A

"Second device revision".

Since we are applying acrylic coating, it is not longer possible to "blow out the RTC".
But that's not a proper solution.

[lpcware]

Content originally posted in LPCWare by Valmir Rodrigues on Thu Dec 20 12:33:57 MST 2012
Hi!
Here in Brazil we are using LPC1769 and the problem is the same! The RTC clock stops or doesn´t run when the main power is on.
I tried 5pF, 18pF and 100K and 10K on Rs and no good results.
The crystal from Shoulder has capacitance = 12.5pF and Rmáx <50K.
The layout is not so far, with good GND
Some help??

Thanks

[lpcware]

Content originally posted in LPCWare by oahmad on Mon Oct 01 14:13:12 MST 2012
We had intermittent failures on our RTCs even with the supposed fix from NXP, I stopped relying on the uC RTC and went with an external one.

[lpcware]

Content originally posted in LPCWare by Daniel Widyanto on Thu Sep 27 01:47:49 MST 2012
Hi,

Just curious, the initial LPC1768 is quite famous for its RTC problem (See LPC1768 errata at http://www.nxp.com/documents/errata_sheet/ES_LPC176X.pdf). But it's suppose to be fixed in Rev A parts.

Which one is your LPC1768's rev ?

Regards,
-daniel

[lpcware]

Content originally posted in LPCWare by Karlsberg on Wed Sep 19 04:54:35 MST 2012
It was just a coincidence. :-(

[lpcware]

Content originally posted in LPCWare by Karlsberg on Tue Sep 18 08:04:35 MST 2012
OK.
The following combination works stable (for now):

Watch Crystal: AB15T (ABRACON)
Load capacitance: 12.5pF
Series resistance: max. 40k Ohm

Additional capacitors to ground: 22pF +/-5%
(Dielectric: C0G / NP0)

Maybe this is just coincidence.

[lpcware]

Content originally posted in LPCWare by Karlsberg on Tue Sep 18 00:58:12 MST 2012
No, the problem is not so easy to solve. ;-)
The crystal is connected to the correct ports: RTCX1 and RTCX2. Sorry.

To take over the parts of the LPCXpresso has not helped.

According to information of a manufacturer of quartz crystals, the resistor suppresses the overtones of the crystals. The cutoff frequency f is calculated: f = 1 / (2 pi Rv C2).
With C2 = 5pF and Rv = 100k the cutoff frequency is approximately 318,310kHz.
Ten times the fundamental frequency.
There may also be another reason for resistance.

In the further studies, we will use a crystal which requires 6pF load capacitance.
But it seems a bit more to consider with the external circuit.

[lpcware]

Content originally posted in LPCWare by Dave on Mon Sep 17 09:21:43 MST 2012
I hate to sound like Captain Obvious, but in your original post, you mention that you have the RTC crystal connected to pins XTAL1 and XTAL2.

Did you mean RTCX1 and RTCX2?  These would be pins 16&18 on the LQFP100 part, and 13&15 on the LQFP80 part...

[lpcware]

Content originally posted in LPCWare by jdurand on Mon Sep 17 09:01:38 MST 2012
I first found out about series resistors long ago (decades) from crystal oscillators. We used to have a lot of trouble with crystals starting up in third overtone and this resistor is chosen to be a low-pass filter over the fundamental but well below harmonics.

I haven't seen the problem on any modern processors.

The "watch crystals" used for 32kHz systems have VERY low power oscillators and some may be so low they're sensitive to board contamination.  One brand of processor I evaluated and rejected for some projects was extremely sensitive and required conformal coating the part of the board with the crystal, capacitors, and processor pins.

[lpcware]

Content originally posted in LPCWare by Karlsberg on Mon Sep 17 04:22:04 MST 2012
We have tried the to add a parallel 10M resistor without success. :-(

Very interesting, in the schematic from date 2010-10-19 the resistor is missing and the capacitors have a value of 12pF.
In the linked schematic from date 2011-02-11, the resistor is shown (100k) and the capacitors have value of 5pF.
About the crystal there is no further information.

We now try to put the components from the LPCXpresso to our own circuit, to see what will happen.

[lpcware]

Content originally posted in LPCWare by Ex-Zero on Mon Sep 17 03:45:14 MST 2012
Schematic of LPCXpresso1769 shows this resistor: http://www.embeddedartists.com/sites/default/files/docs/schematics/LPCXpressoLPC1769revB.pdf

Did you add this resistor already?

Did you try to add a parallel 10M resistor as mentioned in linked thread above?

[lpcware]

Content originally posted in LPCWare by Karlsberg on Mon Sep 17 01:35:14 MST 2012
Unfortunately, the changing to a crystal with 6pF load capacitance and adapting the capacitors to 10pF (also 12pF) has not helped.
The layout seems to be OK, there are only short distances and no signals crossing the area below the components.

On the LPCXpresso LPC1769 (Rev.B) the circuit is working well. However, there is a resistor (R91 = 100k) connected in series to the crystal, which is not present in the schematic.

Are there any specific recommendations for circuit?

In manual of the LPC17xx there are very detailed informations to the main oscillator, but not much to the RTC oscillator.

[lpcware]

Content originally posted in LPCWare by Ex-Zero on Thu Sep 13 09:17:10 MST 2012
There was a RTC discussion here:

http://www.lpcware.com/showthread.php?t=3497&langid=2

[lpcware]

Content originally posted in LPCWare by Dave on Thu Sep 13 08:34:27 MST 2012
Your post indicates the load capacitance target is 12.5pf.  This means that the parallel capacitors should be about 25pf each - if your crystal is located very near the processor, you should have no more than 5pf additional capacitance, so your 22pf caps should be just fine.

Make sure that your caps are as close as possible to the crystal.

[lpcware]

Content originally posted in LPCWare by Karlsberg on Thu Sep 13 07:54:02 MST 2012
That could be the problem!
We will try your suggestion and post the result.

[lpcware]

Content originally posted in LPCWare by wmues on Thu Sep 13 07:42:31 MST 2012
We are using tuning fork crystals with 6pF load capacitance, and 10pF capacitors to GND.
Most modern RTC oscillators do not work any more with load capacitance >= 10pF.