"PLL NOT LOCK PLEASE REBOOT"... Davis Vantage Pro2 Weatherstation Repair

[hgg]

Hello,

I was wondering if someone can help me identify a problem I have with a wireless weather station board
that stopped working.  Its the Davis Vantage Pro2, a very nice wireless station with a lot of functionallity,
until now...

http://www.davisnet.com/weather/products/weather_product.asp?pnum=06162

The PCB board is from the console that receives a data packet every 3 seconds from the wireless sensors
transmitter that is on the roof of the house.  It stopped receiving data and gives the error message:
PLL NOT LOCK PLEASE REBOOT.

I have contacted Davis and they told me that : "There is a radio calibration process that occurs during
startup.  If the calibration process cannot complete itself it triggers the PLL error.  Sometimes rebooting
will allow it to recalibrate properly and all is ok.  Sometimes rebooting does not help and the console's
PCBA must be replaced.  It appears we need to replace the board."

Although they need to replace the board they will not do it because I live outside USA.  They told me to
contact the local dealer in Greece, which I did and they could not help me as well even after contacting
Davis themselves.  That's what I call excellent customer support...

Anyway, the wireless sensors transmitter works fine because I've tried another console and it works.
The faulty console receives some data now and then.  Maybe after two or three hours or so.  So it
works intermittently so to say.   I have included a hires photo of the board.

Can you think what might be wrong with the board?  Logic or passive component fault?  I would say
that one of the CC1012 RF transceiver components might be faulty.  Can you identify the blue chip
on the bottom right corner of the board?  What kind of chip is that  
Finally can it be from a faulty crystal?

So, where do we start the diagnosis of such an error?

Any help is appreciated!
Thanks.
George.

TOP of PCB Board:   http://postimg.org/image/5wpng2qar/ 
(click on the image again for a high resolution version.)

Underside of PCB:   http://postimg.org/image/onrkq8kv7/

[hgg]

Hi,

Just installed the board back, powered it on and while it initially got some data packets, then
it stopped again.  I probed the three crystals and while two of them oscillate according to their
specifications the third crystal next to the transceiver chip does some strange things.  It stops
and starts continuously.   Generates a frequency for a while, then nothing and repeats itself.

The other two look ok.

The big one is 1.84Mhz next to the ATMEGA microcontroller and the smaller 32.768Khz next to it.


1.84Mhz


32.7Khz

Below is a video that shows the erratic behaviour of the Q-14.7-539 crystal and then it shows the
other two as well.

 

Is this normal for a crystal to behave like that?  Doesn't it have a constant oscillation?

[AndyC_772]

Possibly not, when the extra capacitance of a scope probe is added to the oscillator circuit.

[hgg]

Hi Andy,

Did you reply to question "Is this normal for a crystal to behave like that?"
or to the question "Doesn't it have a constant oscillation"

I am probing ground and the pin of each crystal.  The other two are perfectly stable.
Do you think that the Q-14.7-539 crystal is the problem?  The thing is I cannot find
that part number anywhere...


Crystal

[hgg]

There is definitely something going on with that crystal.

I read in the datasheet of the CC1021 transceiver chip that through programmable configuration
it has a crystal oscillator power up / power down capability.  Does this mean that it can power
up or down the crystal itself or its internal frequency oscillation?

The thing is that while all the other crystals have a 400-500mV, that one has zero Volts.

CC1021 Datasheet:  http://www.ti.com/lit/ds/swrs045c/swrs045c.pdf

[AndyC_772]

I was replying to both questions - the answer is the same!

The crystal is just one component of the complete oscillator circuit, and getting such circuits to oscillate at all can require some careful calculation, component choice and layout.

Adding the extra capacitance of a scope probe to that system can easily destabilise it. It may stop oscillating, or shift slightly in frequency, or start and stop in an unpredictable way just like you've seen.

That's not to say that the scope probe "is" the cause of the symptoms you're seeing, just that it could be. It could equally well be that the circuit is working exactly as intended, and that it gets powered on and off at various times by software, maybe as a power saving feature.

[GK]

Crank the timebase sweep speed right down to the "roll" mode to see if there is a distinctive pattern to the on/off switching of the oscillation. If it's a faulty oscillator it will be unpredictable and random.

[Anks]

You could also take the crystal out and make a circuit to test it. Bit brute force but it will remove the variables.

[hgg]

Adding the extra capacitance of a scope probe to that system can easily destabilise it. It may stop oscillating, or shift slightly in frequency, or start and stop in an unpredictable way just like you've seen.

Ok, I see.  Not so easy then...

GK, I've just measured it in 'roll' mode and here is what I get:


Wave1


Wave2


Wave3


The thickess of the signal is because of some 50hz noise.  I don't know how to remove that...
I get mostly the first waveform, but sometimes the third and once the second.  It has a mind of its
own... 

Generally speaking is it a good practice to switch on and off the crystal? 

Pippy, it is logical to assume that the sensor transmitter does not transmit packets continuously in order
to save power, especially because it works from solar power charging a supercapacitor.  So if it does
not send packets continuously and the base crystal switches on every 2 seconds for example, it will be
very difficult to get synchronized.  Isn't that logical to assume?

Anks, I might do that as well but its an smd component and I am not very familiar with soldering
and desoldering them...  If I do, whats an easy way to test the crystal?  Can you just supply it
with a low voltage/current source?

[hgg]

Ok, I tapped lightly the console on the table and got the following afterwards.  (not during)


Wave4

I tapped it again and I got a smaller signal...
Whats going on here?...

[Anks]

There are loads of oscillator circuits on the internet just do a quick Google you could even buffer the signal so you probe doesn't interact with the crystal. Most older SBC would have such circuit with a watchdog attached that you got get a idea from.

[hgg]

Anks I might use the Pierce oscillator if I finally remove the crystal.  Thank you.

Does anybody know what kind of chip is the following?
(its on the right bottom side of the board)


greenshot download

It goes through the board and below is its underside:

Underside

I am no expert... but to me it looks some kind of EEPROM.  Its too far from the micro controller
though.  I don't know.  Could it be some type of diagnostic chip since its located near a JTAG
like connector?  (You can also upgrade the firmware in that console.  Could it be that it is
stored in that chip?)

If the calibration that the Davis technician mentioned is some kind of algorithm and the data have
been corrupted,  it could be beyond repair and I am wasting my time.

I am still trying to think if its a component fault or a digital logic error.
Any ideas?

[Anks]

I'm no expert but looks like one of those temp and humidity sensors but I could be wrong.

[G7PSK]

There is what looks like a red dipped tantalum capacitor its always worth checking those.

[Anks]

On second look I'm willing to bet that some sort of UV detector.

[SeanB]

Red dipped unit is a NTC thermistor. The ceramic board is a light sensor, probably used to measure daylight level and or cloud cover.

[electronupdate]

Since the FW is reporting a PLL lock error that would appear to be the logical starting point.

The wireless communication's IC is from TI (CC1021).  It states that the PLL lock status may be found at pin 9.

As noted the wireless comm's is probably not active all the time to save power so you will need two channels on the scope.. one monitoring an enable , the other the status of the lock. 

Presuming that the lock signal does not occur, or is intermittent, it will be likely that the circuits associated with cc1021 should be looked at.  As the #1 failure in most electronics is mechanical it would be good to scope the input antenna portion to make sure that's in good shape  and then work inwards

[hgg]

Indeed, I forgot that the console also measures inside temperature and humidity...

SeanB the board is from the console that displays all the weather data and is inside the
house, so the little blue chip doesn't look like to be a light/cloud sensor but even if it is a
humidity sensor why does it need to have 12 soldered pins!?!

The little red part could be the temperature sensor. 
I applied some heat with a hair dryer... and its resistance went down from 10Kohm to 4Kohm
and it was raising again as the unit cooled down.  When the temperature reached again the
initial value, its resistance was exactly 10Kohms again.

electronupdate I agree.  The problem should be somewhere near the CC1021 chip.
The problem is that when you power up the console the error message is the first think you find.
Then you go into a menu that lets you choose which transmitting station to use and several
other parameters and when you go out of this mode, then you get the on/off signal from the
crystal.  It also displays the letter R which in the manual it is stated that console waits to
re-establish a connection.

The Davis technician said that "there is a radio calibration process that occurs during
startup.  If the calibration process cannot complete itself it triggers the PLL error",
so the error happens before any RF synchronization attempt is made!

I've probed pin 9 on the chip and it has a 2.99 voltage.  What does this mean? 
Locked or Unlocked?  The chip works with 3V.  Does this mean that the pin is high,
and therefore actually PLL Locked?

There is another small detail.. in the datasheet of CC1021.
"There is a finite possibility that the PLL self calibration will fail. The calibration routine in the
source code should include a loop so that the PLL is recalibrated until PLL lock is achieved if
the PLL does not lock the first time. Refer to CC1021 Errata Note 002."

Below is the Errata Note they are talking about:  (Note 002)
http://www.ti.com/lit/er/swrz007/swrz007.pdf

Maybe someone can interpret the above information and suggest a testing method.
It looks like that Davis did not do their homework, because more people are having the same
problem.

I still don't know where this "source code" mentioned above, is stored and how you can access it.   

[Alana]

Have you tried to visually check for bad solder joints and re-heat some that look suspicious?

[hgg]

Yes, that's the first thing I did.  I looked at all the joints with a loupe and although they
don't have a lot of solder they look ok.  I will check again just to be sure.

From the datasheet:

"An external clock signal or the internal crystal oscillator can be used as main frequency reference.
An external clock signal should be connected be connected to XOSC_Q1 while XOSC_Q2 should be
left open.
...
Using the internal crystal oscillator, the crystal must be connected between the XOSC_Q1 and XOSC_Q2
pins.  The oscillator is designed for parallel mode of operation of the crystal.
...
The crystal oscillator is amplitude modulated.  This means that a high current is required to initiate
the oscillations. When the amplitude builds up, the current is reduced to what is necessary  to maintain
approximately 600 mVpp amplitude."

Crystal Oscillator Circuit.

                                 XTAL
XOSC_Q2 ____.____|[]|____.____XOSC_Q1
                        |                     |
                       =  C1         C2 =
                        |                     |
                      GND               GND


There is continuity between pin 1 of the crystal and XOSC_Q1 and between pin 2 and XOSC_Q1,
so they have used the external crystal with the internal chip oscillator.  In that way they can
turn it on and off I think.   They also mention that in order to initialize the oscillation high current
is required.

Could it be that some part of the power supply rail has failed and cannot start the oscillation of
the crystal connected to the CC1021 chip?

[gman4925]

Does anybody know what kind of chip is the following?
(its on the right bottom side of the board)


greenshot download

It goes through the board and below is its underside:

Underside

I am no expert... but to me it looks some kind of EEPROM.  Its too far from the micro controller
though.  I don't know.  Could it be some type of diagnostic chip since its located near a JTAG
like connector?  (You can also upgrade the firmware in that console.  Could it be that it is
stored in that chip?)

That is the Barometric Pressure Sensor.
Search Google for HP03S and you'll see similar ones.

[hgg]

You are right indeed!  Its the barometric pressure sensor.  Nice.
I bet that it is the humidity sensor as well.

[kayvee]

It's probably one of these  from HOPERF.

http://www.hoperf.com/sensor/barometer_16/

[dr_p]

Is the power rail OK? stable, no noise, etc.? Check at the bottom of the PCB, U11 looks like a linear regulator. Also, you might want to probe the big via between the DAVIS logo and 14.7 crystal. Shorted tantalum capacitors maybe?

Any device added recently that might interfere with the radio communication? Have you tried using the indoor station closer to the outdoor one? Check the antenna for defects?

[hgg]

dr-p hello,

Is the power rail OK? stable, no noise, etc.?

Not sure about that.  When I probe the circuit I always get a lot of 50hz noise.  You can see that
on the oscilloscope signals I've posted.  Only once or twice I had clean signals with no 50hz noise.

Indeed U11 was the thing I checked first and it has 4.9V at the middle pin and 3.0V at the right one.
The circuit draws around 5mA and 140mA with the backlight on.  The voltage on C11 is stable when I
switch on the backlight.  Just a few uA drop.

Also, you might want to probe the big via between the DAVIS logo and 14.7 crystal. Shorted tantalum capacitors maybe?

I don't understand.  How do I do that?

Any device added recently that might interfere with the radio communication? Have you tried using the indoor station closer to the outdoor one? Check the antenna for defects?

There is no problem with RF interference.  I've checked with a different console and it works ok.