U3S continuing

now using a FOX 27MHz TCXO in the Synthesizer v1.4 PCB
At 27.000.000 almost exact. The CAL setting for correction is disabled.
The 27Xtal  was generating around 27.003.0120
So now the drift is coming close to the K3 stability.
My old IC706  impresses me even more it can run WSPR stable at 144.489 MHz without the Icom optional 30MHz TCXO (obsolet)
Will not order a 30MHz VCXO  with the experience about TCXO's  i have now.

QRP LABS note about some TCXO having discrete frequency steps is unfortunately valid for this  FOX924-27.00
Intolerable a for PLL frequency reference of course.
If i had known that, obviously i would  have chosen a better suited type.
Taitien 27MHz VCXO might be a better choice it is a bit gambling.
Now a Taitien 27MHz is used the Si5351a is very stable stand alone and generates high quality WSPR tones.
The specification does not mention the method ,  temperature compensation or correction?  
My  23cm l.o 1050MHz by DF9NP PLL  uses a TAITIEN 10MHz VCXO .
It does not suffer  frequency steps,  this PLL does not need an external reference.
That also means that the PLL chip is not corrected by  I2C or SPI commands in a timed loop.
Those commands can cause interruptions and or discrete frequency steps in the rf output.

The FOX924-27 TCXO  frequency corrections are rough discrete steps.
The TCXO X-tal is drifting until it reaches it 2.5 ppm limit? and is then corected in a step.
These steps are magnified at the higher frequencies 144 Mhz  and 50 MHz and are disturbing WSPR mode.
Disappointing the 144.489 result picture below.
Possible for other timing functions the average frequency is good enough.
Statistically the frequency average is stable is that what the specs say?
Max 2.5ppm sounds okay ,but  if the frequency varies constant just within 2.5ppm  then it is bad.

statistics are very flexible
Put your head in a hot oven and feet in ice you have statistically spoken a perfect pleasant temperature.
One reason managers and governments  like statistics so much.

It is a very interesting experience these devices and QRP LABS kits
But still the U3S Ultimate is not performing as good as it could and should do.
It is okay for HF WSPR including 28 MHz  above it is not suited for WSPR now.

144.490500 MHz trace
at 50MHz it is a little less dramatic  50% is decoded
Again i payed my lesson this FOX924 is unsuitable as PLL reference.

As comparison:
below shows the 1296.200000 signal from the GPSDO 10MHz locked ADF4351A signal generator ( by F1CJN )
Received by the DF9LN PLL stabilized DB6NT 23cm XVTR  at 146.200000 MHz IC706
that clearly shows my point the FOX924 seems useless for stabilizing radio equipment
On reservation of mistakes  my side of course.

A strange issue with U3S  extended call   XW=1
Using PA0O/U  WSTJ/x and WSPRV2.21 programs hang every time in decoding.
It seems not possible to use extended call coding.

Still have 2 defective Synthesizer PCB's
U3S can not find the Si5351A  the Progrock red led stay's lit   when the board is plugged in.
I tried to repair by replacing the Si chips but that did not help. The I2C debugger shows the address is C0 hex but the Si chip does not or responds with NAK  not yet found the cause.
 endless I2C connection loop  on the defective devices.  beam A SCA   B bottom SCL at Vcc side do not realy know what conclusion to draw.
NAK is no responce at all ie. no I2C bus connection 
Or negative responce what would mean the Si does receive but replies with NAK 
It is a lot effort for this low prize PCB but being a HAM  forever..

WSPR last night only 1 decode of WD2XSH/17  as if my antenna broke after that decode
heard S21ZEE at 1826kHz good but and called but the signal faded with their SR
my call was in the log so they heard well that night.
Rare at 160  but not atn   in log 2013 S21ZBC  160m CW
2 days left before they return home.

 Another project.. The  ADF4350 35 -4400 MHz signal generator locked to Trimble 10MHz GPSDO
Put in a case  finally it had to be done one time.

i came to this circuit to interface the Trimble  10Mhz 30mV signal to the 3V3 TTL level input of the ADF4351A PLL port   be careful there is no ESD protection on the ports are direct connected to the chip.  Adding protection diodes must be done.