SPRUJ28 User guide

SPRUJ28E November 2021 – September 2024 AM68 , AM68A , TDA4AL-Q1 , TDA4VE-Q1 , TDA4VL-Q1

12.6.3.12.2 Internal Diagnostic Signature Generator Using MISR

A 32-bit multiple input signature register (MISR) with 32-bit Galois LFSR polynomial, P(x)=X32+X22+X2+x+1, is used to capture a signature of active video pixel data (the 10 MSB bits of each color component) for each internal diagnostic region, as shown in Figure 12-374 and the algorithm below it.

Figure 12-374 DISPC Internal Diagnostic 32-bit MISR Implementation

The MISR (32-bit Galois LFSR based) signature generation algorithm is as follows:

Tap _polynomial = 32’hE000_0200

lfsr_q = seed // initial LFSR value - any none zero value should work

For each (pixel data in the internal diagnostic region) {

data_in[31:0] = { 2’b0, r,g,b } // 10 MSB bits of R,G,B components

lfsr_d = (data_in^{(lfsr_q >> 1)})

If (lfsr_q[0]) lfsr_d = lfsr_d^{Tap_polynomial}

lsfr_q = lfsr_d

}

signature = lfsr_q

All MISRs are initialized at the beginning of each frame with a non-zero "seed" value either with the default constant value (0xFFFF_FFFF) or with the programmable seed configuration value in SAFETY_LFSR_SEED register for each group of regions (refer to Table 12-353, DISPC Internal Diagnostic Check Regions Parameters). Note that a same seed must be used to compare signatures between two frames. If different than the default seed value is desired, then a SEEDSELECT parameter must be set before the corresponding SAFETY_LFSR_SEED register is configured with the customer seed value.

All captured signatures from the internal diagnostic regions are memory mapped to read-only SAFETY_CAPT_SIGNATURE registers for each video port (refer to Table 12-353, DISPC Internal Diagnostic Check Regions Parameters). A SAFETY_CAPT_SIGNATURE register is updated with the signature from each sub-region at the end of every frame. This register returns the signature of the last frame data when the MISR is enabled. When MISR is disabled, this register is cleared.

The MISR aliasing (faulty signature matches fault-free signature) probability is:

(2^(L-n)-1)/(2^L-1) = ~ 2^(L-n)/2^L = 2^-n for large L, where

n = length of signature register

L = length of input sequence

Using the above approximation, the result is:

n=4, aliasing probability = 6.25%

n=16, aliasing probability = 0.0015%

...

n=32 (as in DISPC MISR), aliasing probability is ~2^-32 (negligible)