Hi Xavier, At 2.5GHz sampling frequency you might need to calibrate each IO lane individually, rather than just sweeping the global clock phase. I don't know where you're getting your calibration code from, but if you have a calibrate_all_delays method (like at https://github.com/jack-h/adc_tests/blob/disentangle/adc5g/src/tools.py) you might want to try using that and see if you have any luck.
Cheers Jack PS. I think harmonizing (and robustifying) adc link training code is going to be pretty high on the CASPER development priorities, and I'm sure will be discussed at the workshop in a couple of weeks. On Fri, 28 Jul 2017 at 05:06 Xavier Bosch <[email protected]> wrote: > Hi Jonathan, > Thank you for your response. > Unfortunately, I only have one ADC, so I cannot discern between hardware > and software error testing each ADC separately. > Yes, it is an ROACH2. At the Simulink level, I deleted the ZDOK_1 ADC and > I removed some Simulink block that was related to simulation (a block that > was reporting data in the workspace) because Matlab was crashing when > compiling. These are all the modifications that I did at the CASPER design > with respect the one provided in > https://github.com/sma-wideband/adc_tests > > thank you, > XB > > > > On Thu, Jul 27, 2017 at 6:02 PM Jonathan Weintroub < > [email protected]> wrote: > >> Hi Xavier, >> >> One obvious question is do you really have just one ADC board? Or have >> you seen the same results on more than one hardware set? >> >> Also, where you say: "I adapted the CASPER design to match my hardware . >> . .” could you give a little more detail on what hardware you are using? >> Is this based on a ROACH2? >> >> Thanks, >> >> Jonathan >> >> >> On Jul 27, 2017, at 8:02 PM, Xavier Bosch <[email protected]> >> wrote: >> >> Hi all, >> >> I got a ADC 5G V2.0 DMUX1:1 and I decided to evaluate its performance. As >> expected the results are poor and its phase needs to be calibrated. I read >> the document in here >> http://www.worldscientific.com/doi/pdf/10.1142/S2251171714500019 and I >> got code from here https://github.com/sma-wideband/adc_tests I adapted >> the CASPER design to match my hardware and I also adapted the calibration >> routines. >> >> The result is always the same: while setting the ADC in test mode, >> channels *a, b, c* report 0 glitches while the channel *d* reports about >> ~2000 out of the 16384 samples each time. I sweep the phase using the c >> *alibrate_mmcm_phase()* function but I never got to reduce the number of >> glitches.So the optimal phase does not exist, hence it cannot be >> calibrated. >> >> I am attaching two histograms for ADC_fs=2,5GHz and a signal of input >> power fo -10 dBm and 1 MHz and 800 MHz respectively. As you can see,* >> a,b,c* channels are withing a reasonable margin while channel *d *gets >> out of control. >> >> Has anyone find a problem like that before? How have you solved? Is there >> more available documentation for that part ? >> >> Thank you, >> >> XB >> >> >> -- >> You received this message because you are subscribed to the Google Groups >> "[email protected]" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to [email protected]. >> To post to this group, send email to [email protected]. >> <HistogramTest_800MHz_m10dBm_sampling2p5GHz.png> >> <HistogramTest_1MHz_m10dBm_sampling2p5GHz.png> >> >> -- > You received this message because you are subscribed to the Google Groups " > [email protected]" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To post to this group, send email to [email protected]. > -- You received this message because you are subscribed to the Google Groups "[email protected]" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected].

