Ls1046 abnormal crash

[411243223]

   Hi, I meet a problem: I desigend a board with ls1046(1.2GHz), reffer to ls1046rdb.  A simple boot which print i++, runs and abnormally crash with no time regularity, maybe crash in a few seconds or in a few minutes or one hour. 

   I have measured the power supply & clk , nothing abnormal found. 

   I found that, the signal "hreset" may drop down a short time, "hreset" I use as input signal,4.7K res pull up, that means I do not drive on the signal . I also found "ifc_clk" change from 50MHz to 12.5MHz,  when Ls1046  crash., If not crash ,it will be always 50MHz. So, it seems LS1046'PLL crash.

    One mistake is that, I used 1.0v to power the vdd, when I check the datasheet, it should be 0.9v(1.2Ghz), will this destroy the IC? I have correct the vdd to 0.9v, but I cannot find anything abnormal in these days.  may somebody help me ? 

  Thanks a lot. 

[411243223]

I have thinking about why "poreset_b" be interfaced these days, And found the vdd power ltc3866 may have some problem . I designed the ltc3866 reffer to the demo, but much high overshoot (p2p 22v max)found on the "sw" pin, that will worsen　the board power supply dc-12v, and the the dc-12v interfaces the "poreset_b".

I check the datasheet of ltc3866, and  want to ask why "sgnd"&"gnd" are not one net, but connect them by 0Ω-res(on the bom ,there is no  this ０Ω unexpectedly )?   Dose anybody have the same problem?

Thanks.

[411243223 · Accepted solution]

Problem is solved．

First, 1.0V vdd will not damage ls1046 of 1.2GHz version.

Second, on my board, the "poreset_B" signal is interfered by the board main power, which is DC_12V.  I found that, the higher Voltage　of  power(like 13V),  more possibly LS1046 crash.  And  when I  measured the "poreset_B" signal , Ls1046 can run well.  

So, I suggest  "poreset_B" have ａresistance pullup and 22pF cap to ground,  that will be beneficial.