Recovering the data from Series80 tapes
Message #1 Posted by Vassilis Prevelakis on 5 Nov 2005, 5:40 a.m.
The following is a summary of a thread on the classic computer
mailing list.
The problem with the old HP 9162-0061 data cartridges is that the
magnetic coating peels off from the (transparent) tape causing the
machine to stop accessing the tape with an EOT warning. This is
actually very good news because it means that the damage is localized
(a few mm of tape with scrapped off magnetic coating).
BTW here is a picture of a damaged tape:
Tony Duell suggested that tapes stuck onto themselves and I took apart
two old tapes and found the following:
Assuming that the tapes were rewound before storage, there are two
stretches of weakened magnetic coating on both tapes I looked at; the first
was short while the second longer. These correspond to the length of tape
in contact with the drive belt
In
you can see that on the take up reel (left) there is less tape in contact
with the drive tensioning belt, while on the right hand reel (which has most of
the tape and hence a longer radius) more tape comes in contact with the
belt.
Looking closely at the belt, I can see remnants of tape oxide stuck on it
(in fact its stuck so well, I could not remove it).
So how come we get an EOT error when using these tapes?
The tape has tiny holes near the beginning and end signifying BOT and EOT.
(BOT) (EOT)
+-------------------------- .... --------------------------------------+
| |
| o o o o o |
| |
+-------------------------- .... --------------------------------------+
I assume that the first hole signifies the end of the data region and then
if the drive sees one hole its EOT, while if it sees two holes (prob. more
than one) its BOT.
So when it sees the blank region, it interprets it as one hole and
reports end of tape.
Using a marker to paint (on the back side of the tape) over the region
does not work, since the markers I tried allow some light to go through.
I stuck a piece of splicing tape and that took care of the hole, but I
got a DATA error. I guess the 85 needs some of the info that is missing
from the tape.
I have also noticed that the tape does indeed stick onto itself on the
right hand reel, but if you unwind it slowly you can get it unstuck without
damaging it. I assume that conditioning the tape by warming it up before use
may reduce this sticking problem, but I still would not want to fast wind
a tape in that condition.
Given the above, I am even more confident about data recovery, as we may
be able to use the existing tape mechanism of the 85 to read the data
by skipping the regions that appear blank until the tape un-spools and
the drive stalls.
Proposal 1
To do this I would need to manually rewind the tape till the BOT sensor,
mount it on the tape drive and advance it past the read/write head
till I run out of tape (tape unspools from the supply reel).
While this is going on I am recording the output of both heads via the
audio in port of my PC sound card (maybe use two separate sound cards to
avoid cross talk?).
Data extraction rig:
1) audio connection: what kind of set-up would be required to connect the
heads to the audio card. From the schematics I see that there are essentially
three connections to each head:
Head 0: H0, CT0, and HC
Head 1: H1, CT1, and HC (HC is common to both heads)
But I am not sure what these are used for.
2) running the tape at a constant 11ips. There is a circuit that uses
the tachometer wheel to control the tape speed, but again I am not sure
how to program (?) the tape controller IC (U1) to move the tape.
(I may need to disable the BOT/EOT sensors, but that should not be a
problem).
I am lucky to have a 9915A as my test bed which allows far easier access to
the tape drive mechanism than the HP-85 (and does not have these evil
ribbon cables).
Unfortunately I do not have a scope, but I can borrow one for a short
period if necessary. I have a logic analyzer.
Proposal 2
Another plan is to get the tape moving at 11 ips and then tap the output of
the read/write IC (1820-2418) in the HP-85 tape drive to get the bits from
the tape.
This allows the use of existing hardware. But still requires programming
the tape controller.
I wish I could understand this infernal assembly language of the 85
so that I could look at the internal routines that access the tape
for ideas.
Tony suggested to look at the diagrams of the 9815 and 9825 for
clues on the tape drive mechanism and the pre-amp. I have looked at them
but I do not have access to a 9815 or 9825.
Do note that the speeds of the various tape drive differ:
HP 9825 22ips
HP 9835 22ips
HP 9845 22ips
HP 85 11ips
HP 9815 10ips
Any comments, words of advice are always welcome.
**vp
| 
Go back to the main exhibit hall