30 May
2016
30 May
'16
6:37 p.m.
Hello,
I'm very interested on TU58, as I have a lot of tapes that I need to dump,
plus I would create some new (console at start) for my vax 730.
What I'm trying to do: use a drive mechanism connected with to modern mixed
signal microcontroller with USB port.
The device will be able to control the tape speed and to sample with ADC
and DAC the analog signals on the read / write heads, transferring the bare
samples with almost no filtering to a PC.
Then all the processing can be done via software.
In this way, a weak data track can be maybe recovered, and an empty tape
can be formatted.
It's impossible to use the tape drive with a DEC machine or via serial
port, but I'm not trying to emulate the standard TU58.
If somebody is interested, I can share my ideas, I already begun with the
hardware.
But I have to fix the capstan too...
Andrea
30 May
30 May
9:26 p.m.
On May 30, 2016, at 15:37, shadoooo <shadoooo at
gmail.com> wrote:
If somebody is interested, I can share my ideas, I already begun with the
hardware.
Yes, please do share!
--
Mark J. Blair, NF6X <nf6x at nf6x.net>
http://www.nf6x.net/
31 May
31 May
3:07 a.m.
tisdag 31 maj 2016 skrev shadoooo <shadoooo at gmail.com>:
Hello,
I'm very interested on TU58, as I have a lot of tapes that I need to dump,
plus I would create some new (console at start) for my vax 730.
Great! That is exactly what I want to do. But I also like to recover other
tapes from for example HP 9825 / HP 2645. I have studied the QIC-117 spec
some more but concluded that the drive is not dumb enough. The forward
physical and rewind physical only work with full speed and it isn't
specified if any data is output during these commands. However if one
sacrifice the entire drive and just use the mechanics it would be really
dumb so that could be a route. Formatting wouldn't work using this type of
drive since the width of the magnetic head is different.
What I'm trying to do: use a drive mechanism
connected with to modern mixed
signal microcontroller with USB port.
The device will be able to control the tape speed and to sample with ADC
and DAC the analog signals on the read / write heads, transferring the bare
samples with almost no filtering to a PC.
The bit period is 41.2 us nominally and it is using 1/3 2/3 encoding which
means that there are two flux transitions per bit one at the start and one
at either 1/3 or 2/3 of the bit. So I guess that to decode it one would
like to have at least 2 samples per each third or 6 samples per bit. That
would mean around 160 kHz sampling rate.
For writing I think that it is more on/off switching of a couple of
transistors to have the flux go either way.
Then all the processing can be done via software.
In this way, a weak data track can be maybe recovered, and an empty tape
can be formatted.
One thing to think of is that a TU58 drive does not have optical BOT / EOT
detectors. They make use of special patterns on the tape for this. So to
format a blank tape can be a tricky operation, possibly manually
positioning the tape and then just write the nominal number of blocks and
hope the tape will be long enough.
It's impossible to use the tape drive with a DEC
machine or via serial
port, but I'm not trying to emulate the standard TU58.
If somebody is interested, I can share my ideas, I already begun with the
hardware.
But I have to fix the capstan too...
Looking forward hearing more about this!
Andrea
/Mattis
4:16 p.m.
tisdag 31 maj 2016 skrev Al Kossow <aek at bitsavers.org>:
On 5/31/16 12:07 AM, Mattis Lind wrote:
Because I want it to be as generic as possible. If it is possible to move
the heads I can read tapes that are written not only by a TU58 but also on
a any DC100 compatible device. I dont know what sort of head HP made use
of for their HP9825 and HP2645 for example.
I here compare what AJ have done with a Wangtek 5099 drive. Being able to
read a tape written by a completely different drive (DEI) by just
positionig the head correctly. The resulting data I then decoded in
software and the original tar archive was recovered. Now the encoding was
MFM in th S8000 not GCR which was used normally by the QIC-02 and SCSI
drives with nine tracks and DC300.
/Mattis
tisdag 31 maj 2016 skrev shadoooo <shadoooo at
gmail.com <javascript:;>>:
I don't get it. Why don't you do this with the TU58 mechanism?
However if one
sacrifice the entire drive
and just use the mechanics it would be really
dumb
11:38 a.m.
On May 31, 2016, at 00:07, Mattis Lind <mattislind at gmail.com> wrote:
One thing to think of is that a TU58 drive does not
have optical BOT / EOT
detectors. They make use of special patterns on the tape for this. So to
format a blank tape can be a tricky operation, possibly manually
positioning the tape and then just write the nominal number of blocks and
hope the tape will be long enough.
The TU58-XA mechanism doesn't have the BOT/EOT sensors, but I believe that the tapes
still have the BOT/EOT tape perforations and the mirror behind the tape to form a 90
degree optical path. So, it should be possible to add physical BOT/EOT sensors to a
TU58-XA mechanism as part of a custom tape imager/formatter project.
The approach I'm hoping to take is different, though, and inspired by my perhaps
irrational level of distrust of the tape cartridge design: Transferring the tape to be
imaged into an audio cassette housing, and then imaging it on a hacked up transport, thus
eliminating that #*&!$*! capstan and belt mechanism from the equation.
I wonder how the TU58 tape's magnetic properties compare to any audio cassette tape
formulations? If I could find a way to create new tape belts, then it would be nice to be
able to overhaul old cartridges.
On May 31, 2016, at 03:06, Tony Duell <tony_duell at yahoo.co.uk> wrote:
Now that I have done. I started off with aluminium
alloy rpd (I think
1/2" diameter, but I can check my notes), turned it down, turned a
groove for an o-ring, drilled the centre hole (oddly for a US machine
this is 3mm, not 1/8" -- and note the motor has different diameter
spindles on the 2 ends!) drilled and tapped for the grub screw. With
a good tape I get a good, stead signal at the right speed at the output
of the read amplifier.
I like your approach for making a substitute drive wheel.
--
Mark J. Blair, NF6X <nf6x at nf6x.net>
http://www.nf6x.net/
12:34 p.m.
On 5/31/16 8:38 AM, Mark J. Blair wrote:
Transferring the tape to be imaged into an audio cassette housing, and then imaging it on
a hacked up transport
You may want to use a data cassette like the MT-2ST
http://www.ebay.com/sch/161622290065 has a decent picture. I've got the manual on
bitsavers
12:38 p.m.
On 5/31/16 9:34 AM, Al Kossow wrote:
On 5/31/16 8:38 AM, Mark J. Blair wrote:
the problem is head width and position. i'd have to check if the DC-100 carts being
talked about (DEC and HP) all
put the heads in the same place with the same width. the problem with using QIC transports
is the head width is too
narrow for the old carts, reducing the contact area and increasing the chance of
dropouts.
Transferring the tape to be imaged into an audio cassette housing, and then imaging it on
a hacked up transport
You may want to use a data cassette like the MT-2ST
http://www.ebay.com/sch/161622290065 has a decent picture. I've got the manual on
bitsavers
4:25 p.m.
tisdag 31 maj 2016 skrev Al Kossow <aek at bitsavers.org>:
On 5/31/16 9:34 AM, Al Kossow wrote:
the problem is head width and position. i'd have to check if the DC-100
carts being talked about (DEC and HP) all
put the heads in the same place with the same width. the problem with
using QIC transports is the head width is too
narrow for the old carts, reducing the contact area and increasing the
chance of dropouts.
On the other hand it is possible to shift the position of the head and
thereby circumvent drop-outs by splicing together reads from many different
tracks. A simple teqnique that I used sucessfully when reading S8000
tapes.
/Mattis
On 5/31/16 8:38 AM, Mark J. Blair wrote:
>
> Transferring the tape to be imaged into an audio cassette housing, and
then
imaging it on a hacked up transport
You may want to use a data cassette like the MT-2ST
http://www.ebay.com/sch/161622290065 has a decent picture. I've got the
manual
on bitsavers
5:01 p.m.
tisdag 31 maj 2016 skrev Al Kossow <aek at bitsavers.org>:
On 5/31/16 1:25 PM, Mattis Lind wrote:
It depend what you mean. It was running fine just 25 years ago. So yes it
was running. But since then I haven't tried. Before I fire it up again I
need to give it a complete overhaul. Check PSUs etc. Just hopefully the old
Finch drive is still doing fine.
The purpose of this tape-reading is to find a method to make a copy of the
tapes without having a running system.
/Mattis
A simple teqnique that I used sucessfully when
reading S8000
tapes.
so did you ever get your S8000 running?
5:04 p.m.
On May 31, 2016, at 09:35 , Al Kossow <aek at
bitsavers.org> wrote:
On 5/31/16 8:38 AM, Mark J. Blair wrote:
I tried using plastibands in a TU58 cartridge, with zero success. Once stretched, the
width of the plastiband was too narrow, and it kept on slipping off the edge of the tape
spool and jamming things up thoroughly. These are the ones I tried using:
http://www.amazon.com/Baumgartens-8-inch-Plastibands-BAUSF5000-Assorted/dp/…
I think they might have worked if they were at least twice as wide, at around the same
circumference. Is there a different type available that might work better?
--
Mark J. Blair, NF6X <nf6x at nf6x.net>
http://www.nf6x.net/
If I could find a way to create new tape belts, then it would be nice to be able to
overhaul old cartridges.
plastibands work ok for DC-100 carts. the bigger size is a bit too narrow for DC-600
though 
6:06 a.m.
New subject: Fw: TU58 yet one more time
Yes, it's still me. My normal ISP seems to have installed spam
filters (without warning me!) which (a) drop the classiccmp messages
and (b) bounce most outgoing mail!
On Monday, 30 May 2016, 23:37, shadoooo <shadoooo
at gmail.com> wrote:
Hello,
[...]
the device will be able to control the tape speed and
to sample with ADC> and DAC the analog signals on the read / write heads, transferring
the bare
samples with almost no filtering to a PC.
When I was fiddling with the tape drives in my 11/730, I pulled the
8155 from the controller board so I could control the tape motion
by pulling signals low with wire links, I then hung a 'scope off the
output of the read amplifier. I found that most (old, defective) tapes
gave a very weak singnal, and that the flux transitions wrre almost
equally spaced. I think you would not be able to recover anything from
that. But you can but try.
As an aside, I assume you've looked at the DEC data recovery circut. It's
clever. It's almost indpendant of tape speed. Basically, ramp up on the
transition at the start of a bit cell, change to ramp down on the second
transition, then sample and reset, and start to ramp up again on the
transition at the start of the next bit cell. Depending on whether the
sample at the end of a bit cell is +ve or -ve it sees if it's a 0 or 1
[...]
But I have to fix the capstan too...
Now that I have done. I started off with aluminium alloy rpd (I think
1/2" diameter, but I can check my notes), turned it down, turned a
groove for an o-ring, drilled the centre hole (oddly for a US machine
this is 3mm, not 1/8" -- and note the motor has different diameter
spindles on the 2 ends!) drilled and tapped for the grub screw. With
a good tape I get a good, stead signal at the right speed at the output
of the read amplifier.
-tony
2974
days inactive
2975
days old
13 comments
5 participants
participants (5)
-
aek＠bitsavers.org -
mattislind＠gmail.com -
nf6x＠nf6x.net -
shadoooo＠gmail.com -
tony_duell＠yahoo.co.uk