What is the purpose of Translator in HDDs ? Is it not redun

[Matiw]

Gurus,

If I am not wrong, translator is used to change LBA address to CHS address. Then the CHS address will be translated by the PCB circuitry into a specific position (sector) on the platter. Why is it not possible for the PCB circiuiry to translate the LBA address directly to a position on the platter without going through one more "unnecessary" (for my little knowledge ) translation stage ?

Thanks

[lcoughey]

Where are you going to keep track of that information? Bad sectors, sector remapping, etc?

[deftrue]

Hard disk storage is not linear. You still have a number of platters, a number of heads and a number of cylinders. So at the end of the day, you need to map a linear addressing system like LBA or an absolute PBA, which are very comfortable, to the real scenario containing magnetic surfaces, with one or two heads, which in turn contain a number of tracks, with different sectors on each track. Moreover, a hard drive may contain thousands of defective sectors and a number of defective tracks. You want to keep linearity on your LBA system, and you want to avoid gaps. So you need a mechanism to skip bad sectors and defective tracks without scrambling the LBA system, and you must be able to add new defective elements to the tables in case it is necessary, without affecting (or little affecting) your linear LBA system.

Can you think of a better way?

[Matiw]

lcoughey/Deftrue

Implementing an LBA-only addressing system will not affect the linearity of the addressing system. Suppose a drive comes from the factory with 500,000 user accessible LBAs, and another 25,000 spare LBAs for defect management.

The manufacturer can identify the defective LBAs in the user-accessible part of the disk and then map them to heathy LBAs from the spares (LBA to LBA mapping); this will be the G-list. By the same token, the P-list will be constructed (during the life time of the disk) by mapping defective user-accessible LBAs to spare ones. This way the user will always see 500,000 consecutive LBAs.

For example, let us suppose that LBA-5 is defective, the HDD operating system (firmware) can map the defective LBA-5 to LBA-500,005. Any application that wants data from LBA-5 will be directed to LBA-500,005. For the application programs, everything is linear.

[deftrue]

Hum... what about the mapping to the real platters and surfaces?

[Matiw]

LBA to Physical location of Sector translation (head and platter movement) is done by the PCB electronics , as before; only, CHS is now replaced by LBA.

Hum... what about the mapping to the real platters and surfaces?

[Doomer]

lcoughey/Deftrue

Implementing an LBA-only addressing system will not affect the linearity of the addressing system. Suppose a drive comes from the factory with 500,000 user accessible LBAs, and another 25,000 spare LBAs for defect management.

The manufacturer can identify the defective LBAs in the user-accessible part of the disk and then map them to heathy LBAs from the spares (LBA to LBA mapping); this will be the G-list. By the same token, the P-list will be constructed (during the life time of the disk) by mapping defective user-accessible LBAs to spare ones. This way the user will always see 500,000 consecutive LBAs.

For example, let us suppose that LBA-5 is defective, the HDD operating system (firmware) can map the defective LBA-5 to LBA-500,005. Any application that wants data from LBA-5 will be directed to LBA-500,005. For the application programs, everything is linear.

That's the definition of translation
What's your initial question is about then?

[Doomer]

Update:
I think I got your initial question
you think the the purpose of the translator is to change LBA to CHS, that's incorrect
Man purpose of the translator is to address defects - change LBA to PBA (virtual linear space that includes all sectors)
PBA then changed to CHS using zone tables, that's how newer divers work (although you can say that zone tables are part of translation system)