• Please review our updated Terms and Rules here

Honeywell 200 resurrection

I'm interested that you consider shunt regulators to be inefficient but CVTs efficient. I was under the impression that CVTs were notorious for spewing out heat and EMR. Regarding the shunt regulators I don't know why Honeywell used them but it is probably, as I mentioned on the programming page of my website, that there are no right or wrong solutions, only appropriate and inappropriate ones and I can only guess why Honeywell's solution was appropriate, so here goes. The CVT in the PSU provided the primary stabilisation of the supply voltage and the shunt regulator mounted in the centre of the backplane completed the task. Being in between the logic boards where the power lines enter the backplane the regulator provided the best possible control of the voltage delivered to the boards, but that probably isn't the whole story. Located there it had two tasks, firstly to compensate for voltage variations caused by cables and dodgy connectors between it and the PSU and secondly to compensate for varying demand from the logic boards. If we assume that the first effect was relatively constant then the major dynamic effect came from the second. The shunt regulator solved the problem by sinking the excess current when many logic boards went idle and the effect of this was to maintain constant power consumption and therefore constant temperature within the backplane enclosure as well as putting a constant load on the PSU. For reliability of electronic equipment constant everything is good as one can easily design for it. Contrast this with a modern efficient computer which consumes varying amounts of power from moment to moment and uses variable speed fans to compensate for the temperature fluctations, or a complex switch mode PSU which is very efficient while it works but then commits suicide when it can't decide how to tackle a situation. That's no problem nowadays when there's always another computer close by to replace the one that's misbehaving, but in the commercial computing world of the 1960s reliability was much more important than efficiency. There are other arguments, such as the fact that a shunt regulator doesn't have to pass all the current for a device like a series regulator and it doesn't reduce the voltage at all when it's not needed to. Personally I'd be more concerned about the CVTs and apparently so were Honeywell. The PSU cooling fan is mounted separately from the PSU chassis and connected to it by a plug and lead, which includes a blocking loop which stops the PSU working if the fan is left unplugged, whereas the regulators were just plugged into the backplane along with all the other boards.

Anyway I don't think 1960s American equipment was designed with efficiency at the top of the list of requirements. As an example, the lights in the Honeywell display panels had current running through them all the time, just not enough to make the ones that were supposed to be off glow. Someone mentioned that they remembered that the display on the H200 seemed to react to signal variations much faster than those on other computers, which was probably for this reason. Other benefits were that the transistors controlling the lights only needed to sink part of the current to turn them on and the lights never cooled down so much and therefore lasted longer, another example of constant being a good thing. I reckon that in total the lights on an H200 panel consumed around 7 Amps at 14 Volts, that's almost 100 Watts, when they were all on, which actually only happened during initialisation, but they must have consumed a good proportion of that figure when they were all "off" as well, so a few amps sunk by a 5 Volt shunt regulator won't be a big issue. Rather than worrying about power consumption one should just use the computer in the winter and save on the central heating bill. The branch offices of the company where I worked had Honeywell Level 6 minicomputers which made good room heaters as well as doing the office work but the staff had to keep the windows open in the summer -- and those machines had complex multimode switch mode PSUs.

Regarding storage, I am hoping that my Dataplex machine is still serviceable. It has two magnetic card drives which use big cards the same size as punched cards, each card having 50 tracks with 256 bytes per track, making 12.5k bytes per card, a forerunner to the floppy disk drive. The Dataplex uses all 7400 ICs, the original full power series, not even the LS series, so it is reasonable technology to use with the H200. Its memory uses long serial shift register ICs which cycle continuously, so the machine just has to wait until the right byte comes around in the register cycle each time like a virtual disk drive. At least the machine was designed, patented and built by a British company and it has very few light bulbs and no shunt regulators or CVTs so far as I can remember.
 
...this will not be as efficient as the original constant voltage transformers used, and will probably be heaver! ...

Someone offered me switch mode PSUs to use in the H200 in place of my CVTs but I pointed out that I need their weight inside the cabinet to stop it falling over when a backplane is swung out for maintenance. That's what you might call well balanced design.:) The biggest item of ballast will be the 1 kilowatt autotransformer used to provide the 120 Volt AC supply. With that feeding current into the CVTs I can say goodbye to any thoughts of efficiency.
 
Somebody told me that many surplus machines from the Honeywell 200/1200/2200 range were bought up by EUR Systems in Mechanicsburg PA and cannibalised for spare parts for their own machines, which they were still using in year 2000. I wonder what happened to them after that, but the fossil hunt will have to wait as we have enough parts to play with for the time being.

Someone kindly tracked down the chap who rescued a big pile of components from the machines that used to be at EUR. He has ten times as many logic boards as us and over 35 backplanes, but plans to sell them as scrap. Well, that's his right of course and understandable, but if he doesn't change his mind our project will continue to rely on European resourcefulness rather than American resources. The shipping costs across the pond would be appalling anyway but fingers crossed chaps. Of course, if anyone Stateside is thinking of building an H200 I know a man ...
 
I've only managed to get one email out of the man so far and don't know whether he's willing to correspond any more, so details are sketchy. He scrapped part of his collection around the year 2000 and bearing in mind the escalation of the price of gold since then he believes that sitting on the remainder for another couple of years would be worth his while. My colleague Marcel did some rough calculations and based on 10,000 logic boards each weighing 100gms estimates they might be worth $10,000. However, without an inventory of them we wouldn't be able to determine which were of interest to us and the man has also dumped all the documentation for them, so we can't see much value in them for us except as sources of components, i.e. ironically everything except the gold on the PCBs! The backplanes are more difficult to assess and unfortunately the man hasn't mentioned how much he got for the ones he's already scrapped. He claims that the contacts in them have small dots of gold at the contact points, which must be very small as I've not seen them in the ones in my possession, but the sockets are very deep. However, there are 5760 contacts in each backplane and he has more than 35 of them, so it all adds up. Apart from the contacts a backplane contains about 9 pounds of aluminium alloy. How one values a backplane on that basis I don't know, but I doubt that the value is significant compared with the cost of shipping a 30 to 40 pound package across to England. He hasn't even suggested a price as he's sure that gold prices will continue to rise steeply in the future. As the price of gold is an indicator of the amount of panic in the financial world his view of the future seems pretty depressing to me. I was hoping that we'd seen the last of the financial meltdowns since the dot-com collapse in 2000 decimated my own retirement investments, which have only now barely recovered.

As I suspected, this project is just as much about putting together people as putting together hardware.
 
it is very sad that the guy scrapped part of his collection and dumped all the documentation, maybe with some luck you could save some of the logic boards and backplanes, for your Honeywell 200 replica
 
We only expect to need one more backplane and we're still hoping to do a deal for that. Someone near Harrisburg PA has offered to arrange collection and shipping for us if the deal goes through, so that part is covered. The logic boards probably aren't worth the shipping cost even if we told him to break off the edge connectors for the gold content and ship the rest to us for the components to reduce his asking price. We're going to have to design our own boards for some of the logic anyway, so we could make use of loose components provided that they were similar to the ones we already have, which we don't know at present. If they were then we'd still really need some kind soul stateside to strip the components and send just those to us to reduce the shipping weight. Well, I did say on the website that there was a reason for putting ".org" in the domain name, because I reckon in the end this machine will just be held in trust by us, not owned outright by anyone. So far as authenticity is concerned, we don't have the detailed design for the original machine, which was second generation not third anyway, so I don't have any qualms about using custom made reproduction boards for our own design where necessary, especially as this will make the machine more compact than the original. I am hoping that by and large the electronic components will all be 1960s Honeywell original though, at least in design if not actual date of manufacture.
 
The good news is that we did the deal on the extra backplane and now have it on site in England. That gives us a total capacity of up to 360 logic boards, but I dread to think how much juice that many would drink. With the socket shortage problem solved I can turn to other important matters. I also now have the five regulated voltage supplies needed by the core memory drivers installed on my workbench and can start testing the individual boards, but not yet. My wife and I are about to head for the south of France to spend a week cruising up the Rhone. There are other things in life apart from vintage computers and I need an excuse to get out of my workroom.
 
This is exciting. I used the H-200 at my first job in 1978. Yes, 1978. It was a high school co-op job at my city's Dept. of Data Processing. I guess they didn't have much of a budget for buying new equipment.

Computers were so much more physical then. I absolutely loved the front panel with all the illuminated buttons, I got to load and unload disk packs and magnetic tapes, and punch IBM cards and feed them into the reader.

I was less enamored of COBOL and RPG.

- Bob
 
We now know that machines from the H-200 series, maybe 1200s or 2200s, were in use right up to year 2000 and it was only the millenium bug problem that retired them even then. Apparently up to the last they were very reliable and only needed the air filters cleaned and worn out fans replaced to keep them chugging along forever. I am told that even the chief designer of the H-200 wasn't aware that they were still in use for that long. If this project achieves nothing else it brought together the chief designer's daughter and the man who was possibly the last maintenance engineer for these machines, one of them with the original design proposals stored in her basement and the other with a graveyard heap of components in his garage. That must have been an interesting conversation. I've always claimed that this project is as much about the people as the machine itself.

I am told that the chief designer of the H-200 used the word "elegant" a lot. I think it sums up the H-200 in many respects and reflects the views of others that I've read in this thread.

I didn't like using COBOL on the H-200 much. (or was it a 1200 by then? Probably.) "Elegant" was never really a word to describe COBOL, but it was a practical language for its time. For a long time I wondered why the words used in it were always the longest possible, like "PERFORM" instead of "DO", "VARYING", "ENVIRONMENT", "CONFIGURATION", etc. Then I realised that it was designed to be a self-checking language in that any mistyped or misplaced word would most likely be meaningless and cause a compiler error. Contrast that with a concise language where virtually anything typed would probably compile into code which did something, even if not what was intended. However, it could be annoying with the long turnaround time for batched overnight compilations to discover that one's programme had failed simply because one couldn't spell a particular long word correctly. The very first COBOL programme that I wrote on my training course had just two errors, or rather two occurrences of the same error, in it and I felt at the time that I wanted to take my case to the court of appeal. Grace Hopper maintained that she wanted a programming language close to English and I maintained that in English "HIGH VALUES" is two separate unhyphenated words! We don't mind Americans adopting a modified version of our mother tongue up to a point, but there is a limit. So much for my view of COBOL.

If anyone is wondering how the construction work is going, there's a lot of groundwork to do before anything interesting happens. At present I'm wiring up hundreds of power connections for the second generation memory driver boards, some of which need five separate supply voltages to work. My Dutch colleague Marcel is heading for the States on a business trip in a couple of weeks time and plans to visit that chap in Pennsylvania with the mountain of components in his garage, but even for Marcel there's a limit to how much luggage he can carry home. That said, last year he did turn up at my home with the entire memory system for the H200 in his rucksack, so who knows. At this rate, by the time this computer is finished its parts will have clocked up so many air miles that it will be able to pay its own visit to the USA.
 
To be fair to COBOL, most if not all, dialects had some shortened word forms, such as COMP for COMPUTATIONAL.

But the length of the words was never a serious issue back in the day. Coding was a small part of the program development process. Users often imposed more Draconian standards on themselves--I recall a government COBOL project where variable names were highly codified--and all about 30 characters in length.

Terse languages can be very hard to read. Glancing, for example, at a LISP program usually does not immediately reveal its function. APL can be very difficult to parse mentally.

I recall that my first PL/I program run on an S/360 produced more pages of diagnostics than there were statements in the program. When the I/O clerk handed me the thick listing, I immediately wondered if he'd forgotten to burst someone else's output from mine.
 
I have the DVD of Billion Dollar Brain by my desk even now. It has been a valuable source of images of the H200 and contributed to the accuracy of the pictures on my website. My comments about Karl Malden's efforts as an H200 operator in the film are already on the StarringTheComputer website.

My globe-trotting Dutch colleague Marcel has just taken time out from a business trip to Boston and Montreal to visit Harrisburg PA and pick up some more circuit boards there. We don't need them as such but need the original transistors on them to build other boards. Other commitments have taken me away from building HoneyPi recently, but I am currently constructing a relational database to record all the backplane connections as I install them so that the design is fully documented. We are following the outline organisation of the memory unit as in the Honeywell documentation but have to plan out the detailed construction ourselves, so keeping our layout documented is important.
 
Hi Rob,

Ironically this is my first post on this web forum. I was actually looking around the web for info on the Honeywell and stumbled upon this thread and your website. I'm a little too young to have known any operational H200's, however I have a special place in my heart for them. When I was a teenager in the 80's, I was given the control / operator panel of the H200 as a gift from a friend's dad who worked on decommissioning computers. It's a neat thing with big plasic buttons and lights and stuff. Anyway, I noticed earlier in the thread that you were going to attempt to recreate one. I still own the control panel, although it has sat in a garden shed for many years now. Last time I had it out it was still in pretty good condition. When I was a teenager, I even could light up it's lights (with a 9volt battery), so I know most of the bulbs work. Anyway, I'd like to see it go to help your project, if possible. Please contact me off the thread.

David
 
What a wonderful and generous offer! Did you accept it, Rob? Will you post pictures when you get it?
 
I'm waiting for David to lay his hands on it and send photos, but yes it is a wonderful discovery and having it would save a lot of time on the project. There is just the small matter of shipping it from California to England though. I was fortunate that someone called in a favour to get that backplane shipped from Pennsylvania cheaply earlier this year but this operation could be a bit more costly. Anyway, it will be some time before we need the control panel, so there's no hurry.

I've been working on the project for a year now and I intend to expand my website to document progress early in the new year, but first I need to make some progress worth reporting. So far the work has been on providing bench power supplies and power buses on the backplane and there isn't a single bit of logic installed yet, but now I have all the necessary supply voltages present and have tested and repaired the timing logic, so come the new year with power and timing signals in place I will be able to start on installing and testing the memory unit components.

Being pedantic about the project infrastructure I am currently working on the relational database which documents the design of the backplane wiring for future reference. I was working at this on my laptop while simultaneously casting one eye at the TV and flicking through channels for something to watch when I coincidentally ended up watching Billion Dollar Brain, the film which features the H200, so I was actually viewing the real H200 control panel in operation while designing the wiring layout for the machine. I noticed that there weren't that many lights flashing on the panel in one scene and concluded that a very small programme was running at the bottom of memory. Personally I would have written something that used the entire memory to provide a better visual effect, but there we are. If the project is a success then I may yet get the chance one day.

Happy new year to all those taking an interest in my activities.
 
Yes, it is interesting. Assuming that it is accurate in the detail then the H120 had only three logic drawers compared to four in the H200. Their colours suggest that one was used for the CPU and the other two for memory. I understand that the H120 used IC logic, probably similar to the ICs that I plan to use, which would account for the CPU taking up less space than in the transistorised H200. My intention is to install the CPU in a single drawer, so this picture is a good omen. Despite physical similarities between my design and the H120 my machine will be an H200 as I am working to the functional specification for that machine. In particular the H200 had a faster memory cycle time. This extract from a 1965 magazine advertisement mentions the differences between the two.
H120-H200.jpg
 
Back
Top