source: t29-www/en/computer/univac9200.php @ 556

Last change on this file since 556 was 299, checked in by heribert, 12 years ago

Zahlreiche Bugfixes, vor allem an der englischen Seite und im Deutschen. Letzte Änderungen, bevor Homepage offiziell ihren Dienst antritt.

  • Property svn:keywords set to Id
File size: 9.5 KB
Line 
1<?php
2        $seiten_id = 'univac9200';
3        $version = '$Id: univac9200.php 299 2012-09-19 09:38:55Z heribert $';
4        $title = 'UNIVAC 9200, UNIVAC 9300';
5       
6        require "../../lib/technikum29.php";
7?>
8    <h2>UNIVAC 9200, UNIVAC 9300</h2>
9       
10        <p>The Univac 9200 (Univac 9300) is a punch card computing data center from 1966. It is
11           uncommon that such old devices are completely preserved, even all the
12           manuals are available. This system was stored for over four years in the museum
13           archives until we moved it with a freight company into the museum, next
14           to the <a href="univac9400.php">UNIVAC 9400</a>.</p>
15           
16        <p>After moving the devices with a trucking company to the museum building,
17           the restoration started with cleaning all the devices, initially removing the typical
18           old, rotten noise-absorbing mats at the very first. They have been replaced by
19           brand new cellular rubber.</p>
20
21    <p>We expect most of the problems to be with the mechanical parts in the cabinets.
22       Already we have removed the transportation locks and replaced some bearings,
23           drive rollers and belts. The card puncher features 15 timing belts, for instance.
24           The card reader (500 cards/minute) works again and the puncher mechanics are
25           now executable, too. The processor link to these auxillary devices is another
26           issue and will be checked in the near future.</p>
27
28                <div class="box center auto-bildbreite">
29                        <img src="/shared/photos/rechnertechnik/univac9300.jpg" alt="UNIVAC 9300" width="700" height="375" />
30                        <p class="bildtext">
31                                <b>UNIVAC 9300</b> electronic data processing system, with its components (left
32                                to right): bar printer, cpu, "electronic cabinet" (power supply and plated wire
33                                memory), card reader, card puncher.
34                        </p>
35                </div>
36               
37        <p>For aesthetic reasons, we also repainted the cabinets.</p>
38        <p>We will address the bar printer at a later time. In contrast to the already
39       mentioned devices, the printer cannot be controlled manually, so we will
40           have to start up the processor, too. This will be a buggy job.
41           But all the work is not for nothing, <!-- sic! gute uebersetzung -->
42           since the system is really unique, featuring punch card assembler programming.</p>
43
44        <p>We will continously update this page about the progress of the reparation
45           until the device is fully functional again.</p>
46
47        <div class="box center auto-bildbreite">
48        <img src="/shared/photos/rechnertechnik/univac9200.jpg" alt="UNIVAC 9300" width="700" height="368" />
49                <p class="bildtext"><b>UNIVAC 9300</b> electronic data processing system, uncovered while being restored</p>
50        </div>
51               
52               
53        <b id="cards">UNIVAC 9200 Software</b><br>
54<p>Often it is sheer luck that helps saving unique artifacts from scrap. In this
55case a curious student at the Goethe University discovered strange objects in a
56building and informed us. It turned out that these devices were a <a href="/en/computer/punchcard.php#u1710">UNIVAC 1710</a>
57
58card puncher and a cabinet full of punched cards containings programs for our
59UNIVAC 9200. The punch cards, about 65000 pieces, contain programs which were
60developed between 1967 and 1975 at the institute of mathematics and applied
61computer science. We will surely revive some of these old programs on our
62UNIVAC system. <br>
63 The rescue action was actively supported by the University of Frankfurt and
64the "FITG" (Frankfurt) whon we would like to thank for their efforts.</p>
65        <div class="box center auto-bildbreite">
66<img src="/shared/photos/rechnertechnik/univac/lochkarten.jpg" alt="65.000 punch cards" width="700" height="174" />
67                <p class="bildtext">24 boxes containing more than 65000 punch cards extend our software library</div>
68               
69<p>NB: One punch card can hold up to 80 characters - that makes about 80 bytes
70per card. Thus 65000 cards correspond to about 5 MB which is roughly the same
71amount of data that a modern digital camera produces for a single picture.
72Stored on punch cards 5 MB weigh about 160 kg while the cabinet housing the
73cards has a volume of about 0.5 m³ (about 500 liters of volume).</p>
74
75
76
77        <h3 id="blog">Restoration Blog</h3>     
78        <p><b>17.01.2010:</b> There were some mice in the power supply and printer; they
79           bit thorugh some small cables.
80        <p><b>16.02.2010:</b> After locating two defective resistors the power supply is
81           running again. Now we can use punch card devices from the CPU. Actually
82           we cannot read nor punch data.
83        <br>Now there is another malfunctioning device: The <a href="storage-media.php#plated-wire-storage">
84           plated wire storage</a> does not work. We are trying to get it working at least partially,
85           replacing it with a new self-made solid state memory. As you can read on our
86           <a href="/en/devices/plated-wire-storage.php">detailed description of the
87           plated wire storage</a>, this type of memory has always been very error-prone.
88        <div class="desc-right">
89                <img src="/shared/photos/rechnertechnik/9300pannel.jpg" alt="UNIVAC 9300 Front palen" width="400" height="296" />
90                <p class="bildtext" style="width:400px;"><b>UNIVAC 9300 front panel:</b> 160
91                states of processor and periphery can be indicated with light bulbs and
92                selected via switches.</p>
93        </div>
94        <p><b>15.04.2010:</b> The boot process of the device is still crashing. There is
95            an error message from the printer without any reason. We are trying to
96                locate this error.
97        <p><b>02.05.2010:</b> A broken thyristor (hammer driver) raised the
98            "printer error" message. Data integrity was the most important consideration at the
99                time. If only one of the 140 printer columns is not working correctly, the whole
100                printer goes offline to avoid any wrong output.
101        <p><b>10.05.2010:</b> We managed to get the plated wire storage online. We
102            can even start some small test programs via the input switches, but there are
103                bugs while running. This is perhaps the only device that still uses the old
104                plated wire storage. Anyway, we are planing a replacement.
105        <p><b>12.06.2010:</b> After calibrating the optical card reader, we could read in
106           and execute small programs in the data memory. We will report about the high
107           security level of the card reading process later. Surprisingly the plated wire
108           storage still works.</p>
109        <p><b>16.06.2010:</b> The card puncher doesn't work any more. 30 years of inactivity
110           are a long time for computers, too.
111           <br>On the other side, we could execute a printer loop program. The huge printer
112           starts up, but doesn't print yet. After two minutes, a thermal fuse triggers.</p>
113        <p><b>25.06.2010:</b> The fuse is triggerd by a broken centrifugal switch from the
114           printer engine start-up windings. Therefore the winding was always on and
115           constantly dissipated current. Now the engine is running, but print commands
116           are not yet executed.
117           <br>We also found a bug in the memory. Now all 8kB seem to run completely
118           error-free.</p>
119        <p><b>30.06.2010:</b> We located another bug in the printer logic (faulty
120           transistor). For the first time in 30 years, the bar printer works and is capable of
121           printing files from various punch cards. The type face looks good.
122           <br>Now we turn to the damaged printer.</p>
123        <p><b>05.08.2010:</b> After replacing a broken transistor and injecting some oil,
124           the puncher is up and running! Now we are able to dublicate punch cards.
125           Unfortunately we had to disable the error checking functions since the device
126           detected a non-existing error when punching. Locating this error is the next
127           problem.</p>
128        <p><b>26.08.2010:</b> Locating the bug in the device's internal error checking
129           of punched data was hard work. The computer compares the data which should be
130           punched with the position of the hammers in the punch station in a very
131           complex way. One of the 24 inductive sensing elements was broken,
132           furthermore a transistor which amplifies the particular induced voltage was
133           out of order and there was a cold solder joint. Finally the complete sensing
134           station had to be recalibrated to deliver all information simultaneously to
135           the comparing element. The adjustment has only a margin of 5 microseconds.
136           After this repair, all duplicated punch cards are checked too. If there
137           is a wrong punch hole, the computer stops immediately (an event that occurs
138           extremely rarely now).</p>
139           
140           <div class="desc-right">
141                <img src="/shared/photos/rechnertechnik/univac/messen-am-memory.jpg"  width="400" height="313" />
142                <p class="bildtext" style="width: 400px;"><b>Backside of the UNIVAC 9300:</b> With a storage oscilloscope and a logic analyzer logic states are measured</p></div>
143
144        <p><b>19.10.2010:</b> After all test programs were run successfully, we can now declare
145the machine as being fully operational. The next step will be constructing a
146new memory system which is necessary since the original plated wire memory is
147fragile and it is doubtful that it would survive the years to come without
148errors.</p>
149
150
151<p><b>Feb. 2012:</b> Redesigning the memory using modern semiconductor circuits is more complicated
152than expected initially. Despite a lot of effort concerning the timing of
153all signals involved in the RAM's logic the new RAM card is still not working.
154Further tests and modifications will be necessary. Fortunately the original
155wire memory is still working perfectly.</p>
156
157        <p><i>This blog will be irregulary continued.</i></p>
158       
159<p class="small">We would like to thank Dr. Frank Berger and Dr. Juergen Steen (both from the
160"Historischen Museum Frankfurt) for their suuport and many spare parts they
161donated for this machine. Repairing such complex circuitry would be next to
162impossible without known good boards for swapping etc.</small>
163
Note: See TracBrowser for help on using the repository browser.
© 2008 - 2013 technikum29 • Sven Köppel • Some rights reserved
Powered by Trac
Expect where otherwise noted, content on this site is licensed under a Creative Commons 3.0 License