Hey everybody,

my firm is constructing machines which comprise a single-board laptop with an Atom x5-Z8350 processor. Since we began utilizing this SBC we regularily have points with the SBC not booting up within the discipline. Clients complain that the display screen stays black and nothing occurs, even after power-cycling the machine a number of occasions. Generally this occurs after the shopper has used the machine for a while, however we additionally had instances the place fully new machines are dead-on-arrival on the buyer website, despite the fact that the machines labored completely after they left our facility.

After analyzing the issue, evidently the difficulty is brought on by a corrupted CMOS. If we clear the CMOS utilizing the button on the SBC, the SBC begins again up once more and works as inteded. We additionally discovered that this concern solely seems on programs which use a coin cell battery for the RTC. Techniques which had been shipped with no coin cell battery will not be affected. If the coin cell battery has a voltage of round 2.9 – 3.0 V, the difficulty occurs probably the most. If the battery is both beneath or above this voltage vary, the difficulty solely happens not often.

My questions:

• Is it doable that inadequate or unstable energy from the RTC battery is inflicting a CMOS corruption?
• Within the Intel documentation AP-728, I learn that there are two reset alerts for the RTC circuitry (RTCRST# and SRTCRST#). SRTCRST# ought to solely be pulled low whereas additionally pulling RTCRST# low to keep away from the system coming into an indeterminate state. May an undesirable assertion of SRTCRST# (e.g. EMI) with out additionally asserting RTCRST# trigger the difficulty we’re seeing?
• I mapped out the circuitry on the SBC which provides energy to the RTC area. Please see the image beneath (hand-drawn schematic) for particulars. If it isn’t readable, please let me know and I’ll create a greater schematic diagram. The second image exhibits the beneficial circuitry for the RTC area in Intel AP-728.
  Mapped out RTC energy circuitry on the SBCBeneficial RTC energy circuitry in Intel AP-728

As could be seen, the producer of the SBC determined so as to add a further MOSFET to disconnect the battery as quickly because the VccDSW3_3 energy rail turns into obtainable. May this trigger the difficulty we’re seeing? I measured the voltage at C1 with an oscilloscope and even when I randomly disconnected the facility provide from the SBC, I couldn’t document a drop within the voltage. The coin cell battery took over as quickly because the system energy rail was gone.

• Is there a approach to completely lock the entry to the Person RAM within the RTC area? Can I exploit a software program instrument and so forth. to perform that? The BIOS of the SBC gives an choice, however it solely locks part of the consumer RAM (tackle vary 0x38 – 0x3F).

I now that this put up isn’t that particular, however we actually want to resolve the issue and the producer of the SBC isn’t cooperative in any respect. We went forwards and backwards with them and all of the options they proposed didn’t assist, so I’m now reaching out to you and digging deeper by myself to seek out the basis reason behind the difficulty.

For those who want any extra information or info please let me know and I’ll attempt to carry them to you ASAP. Any assist could be very appreciated!

Finest Regards