We are using AD9517-3 on-chip VCO working at 2048MHz with a 10MHz reference.
Following the ADIsimCLK ("Frecuency Tune Dialog") the VCO tuning voltage
should be around 1.375. But we get 1.68 in some boards and 1.29 in others
In this condition, after a new recalibration the VCO voltage remains the same.
Could you tell us which should be the tuning VCO voltage?
Has the loop filter influence on the calibration process?
Ok, the voltage should be around vcc/2 (in our case vcc=3.3).
But, we are still getting 1.3 volts in some boards.
-Should we suspect that those chips have a failure?
-Or maybe some component is not correctly mounted on the loop filter?
The VCO calibration should center the voltage at half of the Vcp power supply. Vcp is usually 3.3V, so we'd expect about 1.65V +/- ~100mV.
1.68V sounds about right. 1.3V sounds low.
In order to recalibate, you should clear (R0x018=0), issue an IO_UPDATE (R0x232=1), and then recalibrate by writing R0x018=1, and then issue an IO_UPDATE (R0x232=1)
I would clean and DI wash the area around the loop filter. Overall, however, it may not be a big deal as there is a lot of headroom designed into the part wrt the ability to operate over temperature extremes.
I have a correction for you. VCO tuning voltages in the range of 1.2V to 1.75V post-calibration are normal. This can be seen in Figre 15 of the datasheet.
There are different VCO bands that are selected during VCO calibration, and it sounds like the 1.29 V case is on the edge of one of the bands. Therefore, what you're seeing is normal, and the chip will function fine over temp and voltage extremes.