Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!wuarchive!udel!rochester!kodak!ispd-newsserver!ism.isc.com!hobbes!gerardka
From: gerardka@hobbes.ism.isc.com (gerard)
Newsgroups: comp.sys.ibm.pc.hardware
Subject: Re: What is difference between Unformatted and Formatted HD capacity?
Message-ID: <1991Apr10.032238.19876@ism.isc.com>
Date: 10 Apr 91 03:22:38 GMT
References: <1991Apr4.203413.7550@news.larc.nasa.gov> <5994@trantor.harris-atd.com>
Sender: usenet@ism.isc.com (Ism Usenet News)
Followup-To: comp.sys.ibm.pc.hardware
Organization: Interactive Systems Corp.
Lines: 82

In article <5994@trantor.harris-atd.com> sonny@trantor.harris-atd.com (Bob Davis) writes:
>
>	When I compute drive capacity by the formula:
>
>	Capacity (in bytes) = Cyls X Hds X Sectors X 512
>
>have I computed "Formatted Capacity" or have I computed "Unformatted
>Capacity"? Seems like on all the drives I have used, this formula
>has given me the actual number of bytes available for my use -- which
>I think must be called "Formatted Capacity".
	Yes, you've calculated the formatted capacity.  This
	is more significant than the unformated capacity,
	since you cannot redefine the track overhead.
>
>	Precisely what additional data is recorded on the media that
>is "overhead" and chews up some of the "Unformatted Capacity"?
	The "overhead" consists of several "things".  The data
	that you read and write is part of the "data record".
	The data record also consists of
	1. a sync byte, to indicate the start of the record,
	2. error correction code (ECC) bytes, to detect and
	   possibly rectify read errors.  The ECC may be 5 to
	   7 bytes depending on the implementation in the disk
	   controller.

	Preceeding each data record is its identification record.
	The length will vary depending on controller requirements,
	but the basic information includes:
	1. a sync byte
	2. the cylinder number
	3. the head number
	4. the relative sector number
	5. perhaps a "bad record" flag
	6. ECC
	When an ID record is clobbered, you may encounter
	"Bad Sector ID" messages when reading that track.
	The ID record (and a blank data record) are written
	during a "low level" format.

	Preceeding the ID record is often an Address Mark, which
	provides sort of an interrupt to the controller to indicate
	that an ID record is coming up (after skipping over a data
	record).
	An Index mark has a similar function, but occurs only once
	on the track.
	BTW, if all data records are of equal size, the ID + data
	is commonly refered to as a "sector".

	The biggest comsumers of overhead are the gaps between these
	entities.  Gaps provide 
	1. time for the controller to process the information just read;
	2. areas where the write head can glitch the surface when
	   write current is turned on or off;
	3. a preamble to ID and data records so that the binary zero
	   state can be established for the incoming data stream.
	   (There are no "hard" 1s and 0s written on disks that use
	   MFM techniques to encode data on magnetic surfaces.  The
	   values are relative, and are defined by flux changes.  A
	   bad preamble could cause all bits in the record to be
	   inverted.)
>
>	I have seen some drives having voice coil head positioning
>mechanisms that have an odd number of heads. But certainly all drives
>have an even number of platter surfaces. Is the missing head's
>platter and its bytes counted as part of the "Unformatted Capacity"?
>
>	Why is there a missing head on such drives? Is that head
>*really* there and it is just that the platter surface is used for 
>some overhead function? If so, what?
>
>Thanks.

	The "missing" surface is the servo surface, which has a
	corresponding, read-only head.  The servo information is
	written only during manufacture.  The servo data, which is
	never included in capacity numbers, is used to provide
	cylinder positioning information for the head assembly.
	(On old, floor-standing 300 MB SMD drives, the heads could
	be positioned fractionally off-cylinder, to compensate
	for drive to drive misalignment.)

	Gerard