Hello-
We're using an LMX2531LQ1742 to synthesize 904MHz. The reference input frequency (OSCin) is 49.156MHz. According to the LM2531 datasheet (see paragraph 2.7.8), XTLSEL should be set to 1. But when writing 1 to XTLSEL, the Lock Detect output (Ftest/LD) is low and intermittently high, even though the VCO is locked. When writing a 2 or 3 to XTLSEL, the Ftest/LD pin is a solid and steady high, and the VCO stays locked.
What is the recommended setting? Why does the status bit toggle when set per the data sheet? Thanks and Best Regards,
-Tim Starr on behalf of HD@CS
Tim,
What I think is happening is that the calibration is finding a VCO band that works, but is not as optimally centered as it could be. Perhaps you are using a higher phase detector frequency and modulator order, which makes it harder for the digital lock detect. I would recommend using the manual mode for a more optimal calibration that better centers the charge pump voltage as follows:
XTLMAN2 = 1
XTLDIV = 3
XTLMAN = 16*49.152/2.5 = 315
XTLSEL = 4
In addition to this, if you are running a higher order modulator or higher phase detector frequency, this makes it harder for the digital lock detect. For diagnostic purposes, you might try reducing the modulator order to see the impact. Also, for this device, if you are not using the highest charge pump gain, consider increasing this and decreasing the phase detector freuquency. The higher charge pump gain has a large impact on the phase noise. The National Clock Design Tool at http://www.national.com/en/clock_timing/codeloader.html simulates phase noise and fractional spurs.
Regards,
Dean