Subject: Re: Floppy drive?/fixing rd53s
To: Michael Sokolov <sokolov@alpha.CES.CWRU.Edu>
From: Johnny Billquist <bqt@Update.UU.SE>
List: port-vax
Date: 03/30/1998 17:44:33
Okay, I'll jump in this once and give some "nice" answers...

>    This brings me to the following question: are the classical DEC disk
> packs and other media like this too? For example, according to the 4.3BSD
> documentation, RX211 (and RXV21 I assume) cannot format a disk if "the
> header information on sector 1, track 0" has been damaged. However, this
> apparently doesn't apply to the rest of the disk, so at least the
> drive/controller has the ability to write the sector headers itself. The
> documentation also says that the density (single or double, FM vs. MFM
> really) must be specified for formatting, suggesting that you can change
> from one to the other. Also since according to the same doc the track
> format is industry standard (IBM 3740 for FM and System 34 for MFM), so if
> you have a completely degaussed disk, you should be able to format it on
> something less restrictive.

The RX02 formant is very unstandard. In fact, I've never heard of anything
but an RX02 being able to read that format. The thing is, the floppy is
really in IBM 3740 format both for RX01 and RX02 format. But for RX02, the
drive switches to double density for the data part of the sector. So,
changing between RX01 and RX02 format on the disk really is just a
question of noting somewhere which it actually is, and this is done on
track 1, sector 0.

The RX01/02 cannot really format at all. It's all cheating. If the floppy
has been degaussed, it will fail miserably on that spot.

>    It is also interesting what does the interface between RX drives and
> controllers look like. Normally FDDs have the rawest possible interface,
> with raw data on the read data and write data lines (each pulse
> corresponding to a flux transition). If this case the controller has total
> control over the encoding and decides for itself how to go about
> formatting. If RXs have such an interface, it should only take a special
> controller to make the drive format completely degaussed disks. However,
> one thing about DEC RXs makes me doubt that they have such an interface.
> Since the distinction between single and double density is nothing more
> than FM vs. MFM and the flux transition density is the same, neither the
> media nor the drives should care about it, should they? This should be
> entirely up to the controller, shouldn't it? Why are RX01 and RX02
> different drives then? Also I have heard that both are intrinsically dual,
> aren't they? Is it just that two drives are put in the same rackmount box
> and connected to the same ribbon cable like FDDs in IBM PCs, or are the two
> drives really integrated (share the controller interface logic)?

I assume you don't have one close by, or a simple inspection would till
you the answer. The drive has more intelligence in it, and both drives
connect to that stuff. Given the way the RX02 works, I assume you also can
figure out why they are different drives.
Actually, the controller side is pretty simple.

>    RMs and RPs seem to be freely formattable, since the 4.3BSD
> documentation gives the formatting instructions for all of them and clearly
> says that the format procedure actually writes the sector headers. It seems
> to imply that the disks are hard-sectored. Is this correct? It would still
> be interesting, though, whether there are any servo surfaces that cannot be
> written and whether the formatting procedure requires something to be
> already on the disk like it does on RX211/RXV21. BTW, what's the difference
> between RMs and RPs? Which are removable and which are fixed?

The servo surface cannot be written. You cannot even access it. It's
entirely internal to the drive.

I very much doubt that you can rewrite the servo surface, which means that
if it breaks, you loose. However, the format of the data area isn't that
defined by this. It gives the drive an idea where the track is, and an
index at the start of a revoloution.

RP07 has some even more nifty and advanced stuff available at format time.

RM02/03/05, RP02/03/04/05/06 are all removable. RM80 and RP07 are fixed.
(I'm not sure if there ever was an RP01, but that was before my time in
that case...)
The difference between them? Hmmm. Good question. All RM-drives I've seen
has been physically smaller than RP-drives.

>    How about RKs? The handy-for-everything 4.3BSD documentation gives
> formatting directions for these, so I assume that they are freely
> formattable too. Are these hard-sectored or soft-sectored? Do they have
> servo? Does the format procedure have any prerequisites? Someone has once
> posted which RK disks are removable and which are fixed, but I don't have
> that post handy. Do you remember which are which? Just out of curiosity,
> what does their controller interface look like? I have been told that RK07s
> use the same controller as RK06s, so the interface is the same between
> these two, but apparently it is different between RK06 and the earlier
> ones.

RK05 is hard sectored, and you can really format one. RK06/07 has a servo
surface. You loose if the servo surface breaks.

>    How about RLs? This time the 4.3BSD documentation has no answer. Are
> these formattable? Are they hard-sectored or soft-sectored? You later hint
> that they have servo. Is this correct? And what is the difference between
> RL01 and RL02? Is RL01 5 MB and RL02 10 MB? I have heard that RL02 can read
> RL01 packs, suggesting that the two are mechanically the same. I have also
> heard that RL02 packs have two tracks per cylinder. Does this mean that the
> pack is actually a single disk? I remember you and others hint that the
> controllers are the same for the two, so the interface is the same, right?
> Just out of curiosity, what does it (the interface) look like?

RL drives have embedded servo information, so it's not a surface by
itself. You cannot truly format an RL, even if you can rewrite the
headers. You need a special machine to format them, and rumours has it
that it weighted a *lot*, to avoid vibrations during the process.

Mechanically, and RL01 and RL02 are the same, but the RL02 has twice the
number of tracks. An Rl02 drive cannot normally read or write an RL01, but
you could physically patch the logic board on an RL02 to make it able to
read an RL01. But if you installed this patch, it couldn't read RL02
packs, so usually you put in a switch. It's still not possible to write
RL01 packs in an RL02.

Of course the RL02 has two tracks per cylinder, so does the RL01. The RP06
has 21 (or was it 23?) tracks per cylinder. One per surface, pretty
obvious I'd say.

>    How about RAs? I know that RA8x have servo surfaces, but many ST-506/412
> and ESDI Winchesters have servo surfaces too, but the data surfaces can be
> freely formatted even if they are completely degaussed. Can one format the
> data surfaces of an RA8x? How about smaller RAs? How about RA60 packs? Can
> they be formatted? Do they have servo?

As long as the servo surface is okay, you can "format" the drives, yes.
That goes for all the above drives. RA60 works about the same way as RL
drives.

I don't know about RA7x drives, they are too modern. :-)

> > I've been using RL02s for over 15 years and the newest pack I have is
> > about 10 and the reliability is hard to beat.  I've only lost one pack
> > and that was to external field munging the servo track.
>    
>    What's the servo track? I have only heard of servo surfaces, which are
> used by the drive to tell the position of the heads at any time. All drives
> that don't have a stepper motor to fix the track positions (i.e., all
> drives with voice coil head actuators) need servo surfaces, unless they
> have embedded servo. But they have to be surfaces, a single track obviously
> won't help much with determining the head position at any time.

A surface has a number of tracks, the number of tracks the disk is
supposed to have. In the case of the RL drives, these also happen to be
the data tracks, but that's a small distinction.

SInce you know about the construction of servo surface drives, you should
also realize that one cannot format a disk unless the servo information is
ok.

> > I know of RK05Fs and RK05s that are old enough to vote.
>    
>    What's the difference between RK05F and RK05?

RK05F was a fixed disk version of the RK05 with twice the number of
tracks.

> > I know of a
> > RS08 (circa 1968) that is operable, the computer needs work on the core.
>    
>    What's that?

RS08? A hard disk, of course.

	Johnny

Johnny Billquist                  || "I'm on a bus
                                  ||  on a psychedelic trip
email: bqt@update.uu.se           ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol