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

[TrickyNekro]

For testing you can also use electrolytic caps, just use at least 10x bigger values.

When you can do it, it will be nice, in order to see the results.

Best Regards, Lefteris
Greece

[hgg]

I will update when I receive the tantalums.
(I don't want to desolder and resolder the caps twice on the small smd contacts)

Thanks!

[hgg]

Hi,

This is an old thread about the Vantage Pro2 wireless weather console, but I wanted to give it
a second try.  I was getting an error of "PLL not lock", and the console could not receive any data
from the transmitter.  The sensors transmitter works fine because I have another console.

I've replaced some capacitors and the crystal but the problem remained.  I did not get the error so often,
but again no data.  I measured the 5V & 3V voltage rails and they are spot on and without any ripple.

The console uses the CC1021 single chip RF transceiver.
http://www.ti.com/lit/ds/swrs045d/swrs045d.pdf

In the datasheet (page 12) it is mentioned than the typical value of the external clock is 300mVpp.
The thing is that I measure 132mVpp. 

Do you think that this might be the actual problem?

Thank you.
George.

[codeboy2k]

In the datasheet (page 12) it is mentioned than the typical value of the external clock is 300mVpp.
The thing is that I measure 132mVpp. 

Do you think that this might be the actual problem?

Page 12 mentions that the external clock should be 300mV , but they are not using an external clock, they are using an external crystal and internal oscillator.

If you probe the pins with your oscilloscope probe you can load it down and make it either not oscillate, oscillate intermittently or oscillate at a lower amplitude than it would without the probe.  The point is, you cannot easily probe a crystal and believe what you see.

Do you have a function generator?  If you think there is a problem with the internal oscillator then you can try to test it with an external oscillator from your function generator.  First, de-solder the crystal and 2 caps. Then using a DMM set on continuity check, find out which pad of the crystal is connected to pin 10 of the IC.  Set the function generator at sine out, 14.7456 Mhz , 300 mV peak to peak.  Connect the function generator signal to that pad through a 10 nf cap as stated on pg 12 of the datasheet. Now you have an external oscillator at 300 mV p-p

If it locks better with the external oscillator, than you know there is a problem with the internal oscillator.

So, if the internal oscillator is the problem, then the easy solution might be to make an external crystal oscillator on a piece of copper clad board to stuff into the device's case.  NOTE: you cannot use a full voltage digital pierce oscillator here, i.e. the common 4049 or 74HC04 hex inverter oscillator that you can find all over the net.  Using that will require re-programming a register internal to the device, and you don't have access to the software to do that. So you will need to keep the external oscillator within the limits that the software has programmed the radio, and that will be a sine wave oscillator between 300-600 mV, according to the datasheet.  The software will not be programming the device to expect a digital square wave at full Vdd voltage.

Google around for an oscillator circuit that is a 2 transistor crystal oscillator that can work up to 20 Mhz.  That will get you a sine wave with buffering and enough voltage drive at the output.  A single transistor oscillator will not have enough drive, and will be loaded down by the IC.  You need a second transistor to provide buffering and voltage gain.  You will just need to trim the output to about 300 mV p-p.

You can also search for and use an opamp crystal oscillator circuit if you want to go that route, but you will still need to trim the output voltage to 300-600mV p-p.

Here's a 3 transistor oscillator:
http://www.seekic.com/circuit_diagram/Signal_Processing/Oscillator_Circuit/1O_2O_MHz_CRYSTAL.html

See also figure 1e and figure 4b in Linear Technologies AN12 - Circuit Techniques for Clock Sources. 

[hgg]

codeboy2k   thank you very much for your detailed instructions and explanation !!
I will first try with the signal generator and see what happens.
Thanks!

(by the way when I probe the 1.8Mhz crystal of the ATMEGA128 I get 324mVpp)

[hgg]

ATMEGA128  crystal @   1.84Mhz has an amplitude of 332mVpp
Transceiver  crystal @  14.74Mhz has an amplitude of 132mVpp.

But when I probe them again with the oscilloscope & probe set to x10  I get

ATMEGA128 @ 580mVpp and
Transceiver @ 600mVpp

So the oscilloscope probes at x1 were loading the crystals but when set to x10
the oscillators look ok 

What do you think?  Are the oscillators working according to the datasheet?
or shall I give the console to my girlfriend?

     

[codeboy2k]

You should always use x10 on your scope probe.  I don't even use x1 anymore.

Again, it's hard to trust when probing a crystal directly, but it seems like there is enough voltage drive there that it's probably ok in this case, as you seem to be getting reasonable values at 500mV - 600mV

However, if you do have a function generator, I would still try it out.  Your error is a PLL lock issue and that is definitely clock related, so it's still a possibility that the internal crystal oscillator on the radio IC is having stability issues. This might go away if you test it with an external clock source and then you'll know for sure.   So if you have the equipment available, I'd still test that out.

[hgg]

x10 made a difference indeed.

Although the signal looks better, I am still not sure why the transceiver cannot lock the frequency.
If you have a look at the attached picture, the transceiver sends on pin 10 a series of short and
then a series of long pulses. Immediately after that, it looks like that the oscillation begins.
You can see that on the right top line.  Sometimes the oscillation lasts for quite long and some
times it stops immediately after the pulses.  I removed the crystal and the loading capacitors
just to see what happens and pin 10 showed exactly the same except the right thick line which
is the actual oscillation.

The problem is that my function generator cannot generate 14Mhz... and I will try to make one
from scratch.  I have some of these 14.74Mhz crystals.  I've tried connecting to pin 10 the output
of a crude crystal tester circuit I had with an output of 800mVpp and although PLL did not lock,
it did not stopped trying even after an hour.  Normally after 10 minutes if it cannot lock it will
display the error "PPL NOT LOCK" and will halt.  I need to test it again with an accurate frequency
and 300mVpp amplitude.  How accurate does the frequency needs to be?

[PA0PBZ]

To me it looks like the radio chip is started, the CPU waits for the PLL lock and then when it does not see it in time it resets the radio chip and it starts again. Since you already replaced the crystal the problem is most likely in the radio chip. What are the chances you can get one and successful replace it?

[hgg]

It looks like that its programmable. from the microcontroller(?)

5.3.1 Configuration Software
TI provides users of CC1021 with a software program, SmartRF Studio (Windows interface) that
generates all necessary CC1021 configuration data based on the user's selections of various parameters.
These hexadecimal numbers will then be the necessary input to the microcontroller for the configuration of
CC1021. In addition, the program will provide the user with the component values needed for the
input/output matching circuit, the PLL loop filter and the LC filter.

Not easy to find it but even if I do, soldering it successfully is another story...   
If it turns out that the problem is in the transceiver chip, then bye bye console.
How often these kind of chips fail though?

[PA0PBZ]

you don't have to configure the chip, the data is already in the Atmel controller.

How often these kind of chips fail though?

Once too many I'd say 

[Deathwish]

Google around for an oscillator circuit that is a 2 transistor crystal oscillator that can work up to 20 Mhz.  That will get you a sine wave with buffering and enough voltage drive at the output.  A single transistor oscillator will not have enough drive, and will be loaded down by the IC.  You need a second transistor to provide buffering and voltage gain.  You will just need to trim the output to about 300 mV p-p.

:schematic in first few minutes. I used it the other day and it runs with over a 20Mhz crystal

[hgg]

Once too many I'd say 

Bad news...   

I will build the oscillator tomorrow.  Thanks for the video link.  I will try the w2aew one which is
buffered and uses only two transistors.  The amplitude though has to be 300mVpp.  Can this
circuit fine tune the frequency?

How critical if the frequency accuracy.  In the crystal tester circuit I have, the 14.745Mhz crystal
oscillated at 14.742Mhz...

[danfo098]

Have you tried to do anything with the PLL loop filter (C55, C67, C57, R53 and R54)? According to the datasheet they use loop filter number 2. Check out page 48 in the datasheet for different loop filters with different bandwidths and corresponding lock times.

Just probing the filter somewhere will change it for better or worse. Lowering the bandwidth should produce less jitter which is good but it will also increase the time to lock.

[hgg]

Have you tried to do anything with the PLL loop filter (C55, C67, C57, R53 and R54)?

I've read that part but I haven't done anything with that part of the circuit.

I've managed to build the 14.745Mhz oscillator (at 14.742Mhz..) with a nice sinewave, connected it to
pin 10 using a capacitor, but nothing.  It still does not want to lock.  Maybe the transceiver chip has failed
after all.   Who knows.        I might give it a last try with a proper signal generator.