Part Number: LMX2572LP
Other Parts Discussed in Thread: LMX2572
Hi there,
We are currently using the LMX2572 as a PLL in a direct conversion transceiver. We are trying to use this in a "frequency hopping mode". To do this we are using the fully assisted calibration mode to get the fastest lock time on the LMX. We do this by going to all the frequencies, saving thier values and loading these over SPI to the PLL when we want to go to a particular frequency. This all works as expected and the output is changing frequency.
The problem is we see a large spike on the received audio. This happens as the SPI transaction occurs and each regster is written, causing a sharp phase discontinuity on the LMX2572 output. This sharp phase discontinuity causes a large spike on the recieved audio as it is a FM reciever. This large spike gets stretched due to filtering etc ruining some of the audio for a few hundred micro-seconds.
We have looked at the LMX2572's double buffering method as well but because the channel divider can not be double buffered, this will not work for all frequency bands. This is because we may need to 'hop' over a channel divider threshold. Changing the CHN_DIV will then result in the same phase discontinuity for the same reason the fully assisted does.
We have also tried putting the charge pump in a tri-state during the fully assisted register transfer. This helped a little bit to reduce the spikes. We have also tried tuning off the output using OUTA_PD and OUTB_PD but this did not help much.
Is there a way to change frequency fast (600 times a second) and maintain phase continuity to an acceptable level so that there are minimal audio spikes on a FM receiver?
Below is a logic capture of the recieved audio on the FM reciever. The spikes are visable where this audio should be flat.
If you need any more data or explanation, I can provide that. Thanks for any help.
-Evan
