2 Replies Latest reply on Aug 22, 2013 10:59 AM by erikwolfner

    Atom Z520PT - SCH US15WPT - ThermTrip# on startup

    Community Member

      I am posting this question on behalf of our hardware engineer James.  I am the software engineer developing the BIOS for a new Atom embedded design.


      Our problem is that the Atom CPU asserts the ThermTrip# signal shortly after initialization.  This occurs after the chip-set microcode has started execution, but before the Atom CPU fetches the first instruction.


      The board is a custom design based on the Intel Swift Current 2 reference design.  Here are the details of the hardware.


      • CPU: Intel Atom Processor Z520PT (512K Cache, 1.33 GHz, 533 MHz FSB)
      • Chip-set: Intel System Controller Hub US15WPT Chipset (Intel SCH US15WPT)
      • RAM: Micron MT47H64M16HR-3 (8 Meg x 16 x 8 banks, 3.0ns @ CL = 5 DDR2-667)
      • The RAM is operated in a single-channel configuration, with four chips to provide 64-bits of data.
      • The memory is soldered onto the board (reference design uses sockets)
      • SCH microcode is SchMicrocode _VLoD2.023x_HiD2015x.bin


      We have the following SoftStraps settings in microcode.


      • rank  1
      • tcas  4
      • trcd  4        
      • trp  4
      • device density  1024Mbits
      • device Width    X16 devices
      • lpc clock loads  clk1 = 1, clk2 = 0
      • FSB mode cmos
      • FSB lane reversal  not checked
      • split Vtt disable
      • erratum 75 enable
      • erratume 72 enable
      • ddr voltage 1.8V


      Here is a bit more detailed description of the problem from our hardware engineer, James. The System Controller Hub starts loading the microcode  but stops at ucode address D7FF. Then H_CPUPWRGD from the SCH goes true and ThermTrip#  from the processor goes true at the same time. This causes a catastrophic shutdown. I think that the thermtrip# is completely erroneous since the processor is still cold and has been powered on for less than a second. Voltages look good  and proper power sequencing has been observed. What could be causing the ThermTrip# ?


      Any help is greatly appreciated.