TK1 - BCT calibration EVREF parameter


I have build custom board and the shmoo result for 924MHz if I change EmcXm2DqsPadCtrl3 parameter are very different:

The board successfully pass sanity check with:

SDRAM[0].EmcXm2DqsPadCtrl3 = 0x5d75d700;

The board fail sanity check with:

SDRAM[0].EmcXm2DqsPadCtrl3 = 0x5d75d720;

I have checked on TRM TK1 and I found that this bit enable and disable:


I don’t understand very well the meaning of this parameter, someone could explain in which way this parameter can change the result?

The Jetson-tk1 reference board passes with both of them the sanity test.

Best Regards

Hi marcoitalia,

For 924MHz, EMC2PMACRO_CFG_XM2DQS_E_VREF_DQS is set to “1” to enable VREF_DQS receiver. Since it failed in sanity w/ E_VREF_DQS= 1, you need to run DqsIvref shmoo w/ your 924Mhz_CFG in which SDRAM[0].EmcXm2DqsPadCtrl3 = 0x5d75d720, and pick up optimized value of DllXformDqsOffs, Xm2DqsIvref from it to generate new value of EmcXm2DqsPadCtrl3 and EmcXm2DqsPadCtrl6.

To run DqsIvref shmoo:

  1. Edit script …\extras\scripts\, only keep DqsIvref item in “standard_axes_list”, remove other test items.
  2. Edit .bat file …\runme_instead_TK1_Full.bat, remove all other frequency but 924Mhz.
  3. Run runme_instead_TK1_Full.bat to get the DqsIvref shmoo result.

Hi Trumany,

thank you very much for your clarification. I try to follow your advice to produce a new bct.
Probably differential mode is more strong against disturb than e_vref_dqs mode.

Best Regards