Notes from the vault 0x06 – The lives of a Wang 700 calculator

Blog note new 20Sep2023. Updated 28Nov2023 (Fig.09 with inset). This note is in group NOTES FROM THE VAULT – and will be included there when it’s ready to be published. This note is meant for some slow reads, some now and some tomorrow. Then finally, some next week.

Intro

Fig.01 – The Wang 700 is still around, 1970 to 2023

This calculator or computer contains no microprocessor, the “program” is a state machine in a diode matrix, it has core for code and data, can do scientific calculations and is programmable. And it has nixie tubes. It’s from 1970 and it’s made in the US. It weighs around 18 kg. It is just beautiful engineering and computer science!

I have saved it from the trash container two times since it was taken out of use. First in 1982 when I brought it home. Then I gave it back to the newly started “Autronicasamlingen” (Autronica collection or museum), where it had a sordid position since it was no Autronica product. This was in the nineties. Then I saved it from the container again, when that collection was dismantled in 2016 (see Fig.03b). Some stuff went to Autronica Fire & Safety and some went to Kongsberg, since the original Autronica company had been split, just before the turn of the millennium. And some was ready for the container. But not this guy. It was ready to be brought home, again. And now, in 2023, it has been properly named “mine”, so that I in fact can give it away.

Since I picked it from my cold loft and photographed it, I never powered it up. That’s why Fig.01 doesn’t show the nice nixie tube display.

Over there is back home

Fig.02

Before Christmas 2023 it will be On 15Nov2023 it became the property of an American computer museum! Status on the photo is as of 30Oct2023, where it’s awaiting the next level of packing. Previous colleague Morten S. at Autronica certainly helped me with that next level. Post in shop didn’t have a scale that matched, so they approved my digital handheld luggage hanging baggage scale’s value of “about 32 kg.”, also probably rounded down by my memory.  The package details, as seen on the net, was 33.8 kg, only 1.2 kg short of not being handled by Posten here in Norway. After it arrived I got this response: “All unpacked, excellent packing job, thank you! There is no damage. And what a beauty!”. Yes, it is back home now!

The last rescuing

Fig.03b – The WANG 700 stowed away and not on display in the AUTRONICA COLLECTION when it was dismantled in Oct2016 (see Fig.03 here)

On the outside, this wasn’t my product. Well, the nice nixie tubes, the clicks and feeling of keyboard, cassette tape recorder and weight certainly attracted me. I mean, I never used it at job. I think it was mostly used to calculate transformers. Coil type, number of windings, type of iron and what do I know? These were formulae where a slide rule, seemingly, had lost its mission. The first microprocessor based calculators hadn’t arrived yet. And Autronica needed to calculate transformers since we used power supplies internally in our products, as well as developing and selling battery chargers. Even transductor motor driven regulators, for incandescent lighting at large cinemas, like the early Colosseum kino (movie theatre) in Oslo (here). I just assume that this calculator helped the engineers with all of this.

Maybe the new owners, who were also sent a pile of cassette tapes with programs on, perhaps some day may be able to tell more precisely.

Fig.04

For me, I had taken care of this computer for other purposes. I didn’t feel that something as nice as this machine should be thrown away. I must have felt this for more of the type. See photo of my cold (but dry) loft.

In 1982 (the year of my first salvation of the Wang) I hadn’t read the Turing book (Andrew Hodges: Alan Turing the Enigma) which I bought and read when it arrived in 1983. I hadn’t been to Bletchley Park and written about it in a Norwegian computer magazine (095:[3]). And there seen a Colossus machine being reborn. I wasn’t allowed to photograph it, but the man in charge said he’d take a walk. I hadn’t even been to Science Museum in London to absorb computer technology, from a Chinese Abacus via Babbage‘s Difference Engine to Ada Lovelace to the new then Personal Computer (PC) and on. But I did have a basic admiration of the stuff. After all, I had been in my rather cosy basement cave since I was a youth, or maybe even child; and I don’t know how many radio gems people presented to me – and I tore almost all apart to forever gone, and I arranged all the parts in numbered drawers and boxes. I still have some very old stuff on the same loft. Some about this in a note in Norwegian here: Radiobyggboken (Google translate there).

Disclaimer

Some of this stuff are assumptions on my behalf, or – between qualified and dubious guessing.

Some would be quoted from the 700A REFERENCE MANUAL, WANG LABORATORIES INC. 700-0251B 4-70 2.5M. Autronica had two, I kept the worn copy myself. It’s also at [1] – but a newer version is downloadable at [2].

If you find anything plain wrong, or even slightly wrong, or just have something you’d want to add, please comment below or mail me here.

Read-only memory (ROM)

Fig.05

At the bottom of the Wang 700 is a large diode matrix which implements a read-only memory (ROM) which functions as the control store of this machine. This store is accessed via address decoders. We can spot the wired connecting top side of it beneath the protective plastic in the above Fig.05 (lower part). (The trigonometric functions  “are executed  from the core memory” [1] page 1-1, which might indicate that constants like ℇ (≈2.71828) and π (≈3.14159) are not in the diode matrix. I assume this also in general goes for all transcendental functions. They may even be nowhere, since they can be series expanded in subroutines).

Fig.06

The above picture (as Fig.05) shows the wire wrap section. It connects all the plug-in circuit boards. We also see keyboard, plus (bottom) the top of the plastic protected diode matrix.

Fig.07

Here we see the diode matrix. There are sights of it through the machine’s steel bottom ventilating holes. I have inserted rulers in this photo, so that I didn’t have to count them all. (There also are numbers on the wired side of the board, see below Fig.09). The number of diodes would be (30 * 2) for the left and right edges plus + (60 * 31) for the central dense square, which would add up to 60 + 1860 = 1920 diodes. I don’t know the bits width of the hardware logic state machine, which would be taken up by the control bits needed to have the machine tick through its microcode (?) to realise each instruction. I assume that the instruction number would have its index from something like (N-1) * sizeof each control-store element.

Anyway I think it’s much smaller than the the internal data representation, which would most probably be binary, which would be floatN, with N somewhere between float32 (32 bits available for 7 decimal digits plus exponent) and float64 (64 bits available for 15-17 decimal digits plus exponent). Including a sign bit for both the mantissa and the exponent. However, the display reflects two of the 122 ferrite core registers [1]. The display itself has 12 digits + 2 digits for the exponent and I guess, one bit for each of those for sign. If per digit representation, one digit is 4 bits and a data word bit width is 4 * 14 = 56 bits plus 2 bits for signs = 58 bits. But there may be a binary-in-register (floatN) to binary-coded decimal (BCD) for the display functionality in one of the boards. Another complexity here is that the Wang 700 has a RECALL RESIDUE function, “a unique Wang feature that makes double-precision arithmetic a simple operation on the 700.” [1] page 1-1. I assume that the 12 digits of the display already shows double precision. We’ll come back to this, because it’s more related to the the display and the ferrite core.

Fig.08

The chip 7008A from Texas Instruments (“TI”) seems to be a line driver for the matrix. I cannot find this on the internet. I only find a dual port RAM, which it cannot be. I will search in my set of old data books.

That being said, look at the beauty of this!

I am quite certain that they did not have automatic placement of components in 1970. Nor that they came on a roll. In other words, they mounted by hand, every other one the same direction, and the others the other direction.

For the machine to work correctly with all instructions today, all these diodes have to work. These are glass encapsulated diodes which would have to have survived for 53 years. They don’t have their type printed on them, but they are obviously small signal diodes.

Fig.09

From the Wikipedia Diode matrix article (below) I quote: “A person would microprogram the control store on such early computers by manually attaching diodes to selected intersections of the word lines and bit lines. In schematic diagrams, the word lines are usually horizontal, and the bit lines are usually vertical.” In the above Fig.09 we see the connecting wires between these intersections.

Hand mounting the diodes is one thing, hand soldering these wires just adds to the admirable work needed – and the artwork thus presented. After 53 years some small and accurate hands deliver. I think about these people. I was 20 at the time, and I have no problem going back there. I can even feel the warm summer air towards my skin. Back then, it was now. These also were people of blood, and I guess, tears.

1970 and before were times when service manuals were half of the job (well..). I have some of those, even with hand drawn 3D perspectives. Like in the service manual of the 1959 Philips 17 TN 250A (17TN250A), as seen in the above Fig.04 (also mentioned in blog note 100). There were no flat component assemblies; everything was soldered in a chassis, mostly between tube sockets. True 3D. But any Wang 700 service manual probably might do with 2D. I wonder if there, any place in the world, might be a service manual for the Wang 700?

I think that the inductances seen would increase the speed, to avoid ringing on the lines, to balance out the reactive components of the transmission lines. I have not had the Wang open and scoped while it was at work, so I have no idea of the speed here. Maybe the new owners might have this secret revealed.

Now, look at the wires. One type of varnished enameled copper wire is from 1916 (I found this ref.: “History and Future Prospects of Magnet Wire Development”, Sugawara e.al. here). But I think they used oiled cotton tape to insulate the wires when my father wound power station generators around 1938-1945 (203:[A private aside]). These wires on the Wang are as jewelish as they look! Not one short is accepted! Not one, even after they all have been bent, twisted, de-insulated and soldered by some tiny hands.

Aside: the above Fig.09 is, as the rest of the Wang 700 photos here, shot with a Panasonic Lumix DMC-TZ100 camera on our living room floor, with a tripod. I did use a circular LED lamp to avoid too much shadows. The others are produced from the best JPG files it produces. However, Fig.09 is made in macOS Pages with two raw files as the source.

About the Wang 700

From the  Wang Laboratories Wikipedia page [4] (18Nov2023) I read that:

(Intro) “Wang Laboratories was a US computer company founded in 1951 by An Wang and G. Y. Chu.[1] The company was successively headquartered in Cambridge, Massachusetts (1954–1963), Tewksbury, Massachusetts (1963–1976), and finally in Lowell, Massachusetts (1976–1997). At its peak in the 1980s, Wang Laboratories had annual revenues of US$3 billion and employed over 33,000 people.”

(Calculators) “From 1965 to about 1971, Wang was a well-regarded calculator company. The dollar price of Wang calculators was in the mid-four-figures. They used Nixie tube readouts, performed transcendental functions, had varying degrees of programmability, and used magnetic core memory. .. In 1971, Wang believed that calculators would become unprofitable low-margin commodities and decided to leave the calculator business within a few years.”

(The Wang 1200) “Harold Koplow, who had written the microcode for the Wang 700 and its derivative the Wang 500 rewrote the microcode to perform word processing functions instead of numerical calculations.”

The WANG MUSEUM by Jan van de Veen in the Netherlands also has a Wang 700, plus lots of pictures and stories [3]. It also has some quite interesting facts about the 700:

In December, 1968, WANG Labs announced the new 700-series calculator to compete directly with HP’s 9100 calculators. As many people who know high-tech industry understand, the announcement of a product doesn’t necessarily mean it exists. Even back in late 1968, the notion of “vaporware” existed, and WANG’s introduction of the 700-series was just that. Even though the product had been announced, shipments weren’t promised to begin until mid-1969, which, as it turned out, was a milestone that WANG missed by a few months. In February of 1969, McGraw-Hill Publishing’s Product Engineering magazine published a cover-story on the new (and not yet available) WANG 700, featuring a proud, cigar-toting Dr. Wang surrounded by a number of 300-series keyboard/display units and his new baby, a prototype WANG 700. Cooling was a problem, but soon a fan was introduced that solved the problem.

Also worth reading is the The Old Calculator Museum‘s pages, see [5].

Not wire wrap

Fig.10 – Not wire wrap, by a closer look

This wiring is not of the sort wire wrap. (The term is not not even relevant, so I haven’t included it in the Wiki-refs list below.) Wang must have needed thicker wires than the 30-28 AWG, which would have been required for wire wrap. And the posts would have needed to have a square profile and be gold plated, so that the silver-plated wire coating could have properly cold-welded to the golden posts, to produce the mega solid connection that wire wrap technology gives.

Wang have used some kind of connector that seems to be clamped onto their silver-plated(?) posts. Probably also mega solid. I have no clue what this is called. I assume that the cables’ insulation is PVC 196:[My materials science notes]. I love the material, but I’m not certain if I should..

What is has in common with wire wrap is that it’s hand made and beautiful!

The keyboard and its printed circuit

Fig.11 – The keyboard’s printed circuit from below (lower of picture is closest to the display)

This board is hand soldered, you can see this from the fact that only the through-holes are soldered.

This printed circuit board would be an example of hand made etched traces technique. However, not hand drawn as such, but as “the sweeping curves in the traces are evidence of freehand design using adhesive tape” (Wikimedia photo here). I remember these rolls of thin, black tape, in some kind of a dispenser, to be applied carefully so that it didn’t break – as that would cause broken wires by conception to appear on all the etched boards. Obviously the copper is etched away for all of the surface, but not the area covered with tape. So a successful wire better be made with a continuous tape. (Aside: when I was a youth I “borrowed” my mother’s nail polish bottle, containing that nice brush, to paint the wires. Then I think I used ferric chloride (Iron(III) chloride) that I bought a the local pharmacy to etch away the copper.

The above is about making one of a board. I don’t know how they produced thousands, but I think the technology of using photosensitive material and UV light was available at the time.

The four keys just below the display (“Run”, “Learn”, “Learn and Print” and “List program”) are mounted on the keyboard’s metal frame. The cables on the short side are the wires to the key (matrix?) I think. The extra, rather thick cable, goes to some output connector. This looks like a true add-on, mounted in the rear of the machine, “in the air” between the TYPEWRITER and the I/O connectors.

About the keyboard from the top. They might be micro switches, because they feel and sound expensive.

Memories from the users

27Nov2023. I am writing on this, to be pasted into here when ready. See Notes from the vault 0x06 – Autronica (password needed).

More to come..

References

Wiki-refs: An Wang, Binary-coded decimal (BCD),Binary number cassette tape recorder control store, Colosseum kino, diode matrix, ℇ (mathematical constant), float32, float64, Harold Koplow, Iron(III) chlorideMagnetic core memory, Microcode, nixie tubeπ (Pi), Programmable calculator, Transcendental function, Wang Laboratories

  1. 700A REFERENCE MANUAL, WANG LABORATORIES INC. (1970). On the back cover: 700-0251B 4-70 2.5M. Newer version:
  2. 700A/B REFERENCE MANUAL, WANG LABORATORIES INC. (1970). I found it at manualslib.com. No back cover, so the internal version is not seen
  3. WANG MUSEUM, by Jan van de Veen, The Netherlands. See https://www.wangmuseum.nl. Especially see Desktop calculators
  4. Wang Laboratories on Wikipedia (too important not to also be in this list)
  5. The Old Calculator Museum, see https://www.oldcalculatormuseum.com (but no physical Wang 700 there. However there is a very interesting (“overstated”) chapter about the 700 at The Indiscrete Calculator by Stuart P. Roberts, General Sales Manager at Wang)

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.