Hi team，
The customer is BYD.
Customer has concern about λtotal of the SRAM. They think the λtotal is too high.
Their customer wants λtotal to be lower than 10FIT.
In my opinion, λtotal can be calculated by following formula:
I refer the vaule of λMPF from FMEDA_MPC5775E_Summary_INTERNAL_rev_1_1:
Is the λMPF(λMPF_det + λMPF_undet) a constant comes from factory hardware test?
What the reson of the high basic total failure rate λtotal of SRAM, and is there any way to reduce the λtotal to lower than 10FIT?
Any comments are welcome. Thanks!
Hi Baiguo,
Please find below response to your question:
Question: What the reason of the high basic total failure rate λtotal of SRAM, and is there any way to reduce the λtotal to lower than 10FIT?
Answer: The λtotal (raw failure rate) of the SRAM is fixed and can’t be change because λtotal (SRAM) is calculated based on area occupied by SRAM on die. SRAM failure rate are high because they primarily driven by transient faults.
λtotal= λSR (Safety related) + λNSR (Non Safety related)
The λ which is shown below is λSR
λSR (safety related) can be reduced by changing the percentage of SRAM used as safety related in select safety measure sheet in FMEDA. Thereafter, you can apply techniques such as ECC to ensure that dangerous failure rate (λSPF+ λRF) is reduced to well below your target. Please note that the failure rates are computed using TR62380 standard (for permanents) and based on testing techniques specified in JESD-89 (for transients). You should use these failure rates only if you are performing safety analysis. For reliability purposes, NXP quality team can provide failure rates based on reliability testing.
Question: Is the λMPF(λMPF_det + λMPF_undet) a constant comes from factory hardware test?
Answer: λMPF doesn’t come from factory hardware test instead calculated internally inside FMEDA. These are standard terms used in ISO26262 standard.
Best regards,
Bhavik Vakhariya
