LMX2572EVM: Auto ramping without cal

Yes, I've been referring to the EVM user's guide, but have not been able to successfully control calibration.

Have you read my questions?

Ed

Hi Ed,

Calibration free automatic ramping and full assist do not work together.

Full assist allows you to manually switch VCO frequency with min. amount of time. 

Automatic ramping will sweep VCO frequency automatically once RAMP_EN is set to 1. 

EVM user's guide section 3.2.2 is a calibration free automatic ramping. Here calibration free means during frequency ramping, no VCO calibration will be executed. As a result, the possible frequency ramp range is limited to below 100MHz. This range is VCO core dependent and is not guarantee. Our recommendation is 50MHz. In your case, you only need 36MHz range, it should work. Were you able to duplicate the result of section 3.2.2?

when you do register read back, were you able to see signal out from the MUXout_TP?

Noel,

Thanks for your reply.

Yes, I have been able to get close to the result of section 3.2.2 of the EVM's user guide.  There is some non-linearity in the frequency vs time plot that needs to be cleared up, but for now would like to focus on the basic ramping.

After implementing the 10MHz ramp of section 3.2.2, I was able to implement a 36MHz/144usec up-ramp, followed by a 36MHz/144usec down-ramp.

I also implemented a 36MHz/100usec up-ramp / 36MHz/100usec down-ramp signal to verify that could be done.

I then tried a combination of the above 2 signals: a 36MHz/144usec up-ramp with a 36MHz/100usec down-ramp.  I have not been able to achieve this without calibration occurring.  Oddly enough, I can get a signal with a 36MHz/100usec up-ramp with a 36MHz/144usec down-ramp to work.  In fact, I have shortened the up-ramp time to 60usec, so that I have a signal with 36MHz/60usec up-ramp and 36MHz/144usec down-ramp that works without a calibration interrupt.

I am still using the original loop filter that came on the EVM, so I expect it is limiting how short of ramp time I can implement, but wanted to see basic ramping occurring before modifying the filter.

My end goal is the have a 36MHz/144usec up-ramp followed by a 36MHz down-ramp with the shortest possible time.  My questions at this point are:

1.  Do you know why I'm able to shorten the up-ramp time (leaving the down-ramp time at 144usec) successfully (no cal interruption), but am not able to shorten the down-ramp time (leaving the up-ramp time at 144usec) without a calibration interruption?

2.  Do you know why I'm seeing the non-linearity in the frequency vs time plot?  Is this likely due to loop filter bandwidth limitations?

An example frequency vs time plot is below (time scale is 25nsec per tick).(36MHz/60usec up-ramp, 36MHz/144usec down-ramp)

Ed

Hi Ed,

Non-linearity is likely due to not sufficiently high loop filter bandwidth. We recently found out that some EVMs were built with wrong loop filter components, as a result, the loop bandwidth is smaller than what we have shown in the user's guide. This problem is fixed in newly built EVMs. 

Not knowing if you have the latest EVM, to play safe, please change the following components:

I don't have idea why the ramp up/down time cannot be reversed. A smaller slope should always be welcomed. Can you share your TICS Pro register configuration?

Noel,

We replaced the loop filter components and have seen an improvement in linearity.  In addition, I have now been able to create a 'triangle' with a 36MHz/25usec up-ramp and a 36MHz/144usec down-ramp, without a calibration interruption.

I am still seeing the issue with calibration interruptions occurring when I try to make the down-ramp shorter.  I have included an example below.

The 1st plot shows a good result for a 36MHz/50usec up-ramp, with a 36MHz/144usec down-ramp.  The TICS Pro file follows that.

The 2nd plot shows a result with calibration interruptions that show up when trying to create a 36MHz/144usec up-ramp, with a 36MHz/50usec down-ramp.  The TICS Pro file for that is also included.

4532MHz_36MHz_50usup_144usdwn_A.tcs

4532MHz_36MHz_144usup_50usdwn_A.tcs

Thanks for your help!

Ed

Hi Ed,

Please remove the RST bit. This bit is not needed here because the end frequency of current ramp is equal to the start frequency of next ramp.