ADC32RF45EVM: Unable to program JESD204B Settings

Hi,

When you attempt to write to the paged registers such as the registers on the JESD tab, and they revert immediately back to zero, that is a strong indication that there is either the device clock to the ADC missing or the SYSREF to the ADC missing.   One other possibility that has bitten me once before was using a bench supply with the current limit set too low.  Once the ADC programming took effect there was an inrush current that caused a supply brownout and I'd lose the register contents I had just written a millisecond earlier, but after the brief brownout everything looked normal again.  Just listing that last one for completeness.

I downloaded the LMK config file you posted and loaded it into the GUI in simulation mode, no hardware, and am looking at what you are setting.   You are using VCO0 of the LMK, so the clock would be 'near' 2.5GHz, and your sysref divider is 512.  The ADC clock has a divider of 1, which is good, but output type LVDS.  we usually set that for 2000mV LVPECL format for larger signal swing.   plus I see a few other settings you have, such as turning on the LMK clocks CLKOUT10 and CLKOUT12.   That's fine.  Are you using the EVM into something other than our TSW14J56, such as an FPGA development platform?

The quickstart tab 'PROGRAM EVM' button simply loads 3 or 4 config files that come with the GUI.  You can do the same thing as the PROGRAM EVM button by going to the lowlevel tab and loading first the proper LMK config file, then the proper LMX config file *if* the LMX is being used, then the proper ADC config file, and finally a file called NL_Config based on the chosen Nyquist zone to trim in the non-linearity correction logic in the ADC for best performance.   That is all the PROGRAM EVM button does, but using the quick start tab to choose a mode of operation takes away the possibility of user error in choosing which config files go together, such as picking an ADC config for decimation modes while picking the LMK clocking for a bypass mode.     But that is why the PROGRAM EVM button overwrites your LMK configuration - the first thing it does is configure the LMK.   The other reason to use the PROGRAM EVM button is to load the configs in the proper order.  Clocking first, then ADC.  You *could* use the quickstart tab to program everything to the mode closest to what you want, and then go to the LMK tab to touch all the things that you want different such as the SYSRE divider and enabling the other clock outputs. 

it is possible that the ADC is not 'seeing' the LVDS clock output from the LMK if it is too small a signal.  Also - what *is* the mode of clocking that you desire?  Do you want to use the LMK to clock the ADC, or the LMX, or external clocking?  The jumper JP3 must be set for internal or external, accordingly.  And if you want the LNK to clock the ADC then there is a bit of soldering involved for the clock AC coupling caps C431/C432.   If you are using the LMX for clocking or external clocking then it doesn't matter that you have the LMK chosen to output LVDS format for the ADC clock.  it isn't going anywhere.

Also, what revision of EVM are you using?  There were some early revisions of EVM that would not have the SYSREF levels be compatible with the config files in the GUI you have, but you would have to have an early pre-release EVM for that to be an issue. 

That SPI GUI for the RF45 with the quickstart tab has only been on the web for a couple of days, and the updated User Guide for it will be on the web shortly.  Did you get the SPI GUI in the last few days?  Just want to be sure it isn't a beta copy from one of our FAEs or something. 

Regards,

Richard P.

Hi Richard,

I downloaded the latest copy of the GUI two days ago. :)

I've installed JP3 and C431/C432, and am using J7 to input a 614.4MHz clock to the LMK, which then outputs 2.4576GHz to the ADC.

Changing the output type to LVPECL 2000mV seems to have made the difference. Thanks a million!

Is there any way to monitor whether the ADC is getting the clock/sysref that it requires to function correctly? I'd sure love to be able to detect this type of situation when I inevitably screw up clock settings in the future. Thanks!


-Brian

Hi,

I am glad the clock amplitude was the issue, and easy to identify and fix.   I am not aware of a way of just seeing that the ADC has an adequate clock.  I just get in the habit after loading my config files of picking a control on a GUI tab to see if the control will 'stick' or not.  I usually flip over to the bit for setting Scrambling and check the checkbox and then clear the checkbox.   if the checkbox won't stay checked, then I look to see if I maybe didn't turn on the power, or my bandpass filter for the clock doesn't match the signal generator frequency, or the sig gen is off.   (We have a lab person who likes to go around turning off the outputs of unattended equipment.   I come back to my bench and find I can't make the GUI control 'stick' then I look to see if he's been by.  :-)   )

Regards,

Richard P.