Writing Quality - Page 8

CD Decoding system

The following pictures show the CD decoding system in detail.

The RF signal is sent to the EFM decoder where the EFM channel bit stream is decoded and passes to the first stage of the CIRC decoder. After the C1 decoder, the data interleave is removed and passed to the final stage of the CIRC.

If the C2 decoder data can be decoded, the data is then de-interleavered and driven to the Error concealment and after the D/A to speakers. In case of CD-ROMs the data is de-scrambled (additional interleave stage) before driven to the Error concealment circuit.

If the C2 decoder fails, it flags the errors for the error concealment circuitry where the remaining errors are concealed before being passed to the digital to analog converter and finally to your speakers. Only rarely (about once every 125 hours), is the CIRC decoder in unable to identify and flag an error for the concealment circuitry. When this happens, an audible click is often heard.

The digital audio stream is error-free unless the CIRC determines that a C2 codeword is contaminated by an error pattern that it cannot correct. If the 24 bytes were flagged wrong, then it would turn 6 samples wrong at once, on both channels. If all the audio data that lie adjacent to the audio samples in the uncorrectable C2 codeword are reliably recovered from the disc, then the values of the un-reliable audio samples are interpolated.

In case of Data discs, CD-ROM chipsets use the C2 error flag to mark data going into ECC layer 3 so that they can use erasure correction for improved performance, similar to the C1 flag being used to mark erasures for C2.

A more detailed explanation of the various strategies used in the C1/C2 decoders follows (source: Digital-Inn):

"…Assuming there are no input flags from the EFM decoder (the normal situation), C1 can detect and correct up to 2 symbols (bytes). However the exact performance is determined by the strategy employed. The two basic strategies are (a) correct 1 symbol or (b) correct two symbols.

In (a), the decoder with detect and correct a frame with one errored symbol. It will accurately detect frames with 2 or 3 errored symbols. It will mostly but not always detect frames with 4 or more errored symbols (see later). For frames with one errored symbol the flags will all be set to definitely correct. For frames with 2 or 3 errored symbols, there are two possible strategies; either marks the detected errors as possibly incorrect or mark the whole frame all possibly incorrect. The second possibility is usually employed mainly due to accuracy; basically it would be as inaccurate as the 2 symbol scheme without the increased correction capability. For frames with 4 or more errors, all symbols in the frame will be flagged as possibly incorrect as there is no way of differentiating the correct from the incorrect symbols.

In (b), the decoder will reliably detect and correct frames with one or two errored symbols. Frames with 3 errored symbols will either be correctly detected (generally resulting in the whole frame being marked as unreliable) or will be misdetected as having 2 errors and will be incorrectly corrected resulting in 5 unflagged errors being passed to C2. For 4 errors, there is a slightly possibility (2^-19) that C1 will misinterpret this as 1 error again resulting in 5 undetected errors being passed to C2 (the same as the 1 correction strategy), otherwise 4 or more errors will be detected and the whole frame will be marked as unreliable.

C2 uses the original information and the flags to try to detect all errors undetected or miscorrected by C1; here the strategies become more divergent. Unfortunately the manufacturers are not giving this information away.

There are other strategies (including multiple pass ones) and it is also possible to apply many strategies at once and then decide which to use based on the results…"