• Please review our updated Terms and Rules here

Honeywell 200 resurrection

Interesting. I remember two identical CPUs on one system. I know that it required special in-house developed code in the operating system to use the second CPU.

It seems likely that it was a custom creation then. If they were both complete CPUs then all they could have shared would have been the main memory and presumably somehow the peripherals, but the original design didn't make any provision for such a configuration and the chief designer, Dr Gordon's personal description of it never suggested the idea as a possibility. The H200 was planned as a cheap easily understood basic business computer with a modular design that allowed for expansion but not in that way to my knowledge. In fact Dr Gordon wrote that he didn't expect the typical H200 user to have much of an understanding of computers. I suspect that the actual people who did use it probably surpassed his expectation though, so customisations could well have occurred.
 
Yeah, never underestimate the power of the "Hey, boss, we really need X"/"You already have Y, I'm not paying $$$ for X"/"Well, okay, if we rejigger it just so, we can get by with Y and Z for only $$" exchange.
 
Yeah, never underestimate the power of the "Hey, boss, we really need X"/"You already have Y, I'm not paying $$$ for X"/"Well, okay, if we rejigger it just so, we can get by with Y and Z for only $$" exchange.

The balance of the comparative costs of people and hardware has swung towards competent people being far more expensive than powerful hardware whereas back in the sixties the converse was true. Therefore in those days we were very inventive to squeeze the maximum performance out of the relatively feeble machines. However, as you say, there always comes a time when the hardware doesn't appear to be up to a task and people have to think a bit to help it.

When I retired I spent ten years working as a volunteer computer technician for a local charity whose IT budget was zero in my early years. They were using PCs in pretty conventional ways so I applied the innovative skills that I had developed during my paid career to push what they already had to the limit, for which they were very grateful, albeit not financially of course, but I'd already feathered my nest well before retirement. That's what I liked about the H200 and still do, the potential of the machine resulting from its well thought out design. No doubt when Dr Gordon wrote what he did he had the attitude that the hardware needed to be smart to compensate for the shortcomings of its users, but the machines actually often ended up in the hands of people who were smart enough to see their full potential. Dr Gordon eventually developed Alzheimer's and sadly by the time he died he didn't even remember working on the H200, which he had always lauded throughout his earlier life. When I told his daughter that some of the largest series 200 machines had been in active business use up until year 2000, when the millenium bug issues would have required them to be reprogrammed, she said that even he would have been surprised by that.

Maybe what we do here that is special is that we appreciate the otherwise invisible artistry that went into the design of the old technology. Personally I can appreciate the skill that went into design of a computer in a way that I could never appreciate the skill that Shakepeare put into his plays or a old master put into his paintings. As for music, I still think the Beatles were rubbish.
 
Yeah, never underestimate the power of the "Hey, boss, we really need X"/"You already have Y, I'm not paying $$$ for X"/"Well, okay, if we rejigger it just so, we can get by with Y and Z for only $$" exchange.

The problem these days is, so often the boss says "we really need to go to cloud xyzzy" when he doesn't have the budget and do not understand why it will more...
... when I worked my bos was desperate to look modern and move from tape backup to cloud backup. The only problem was he didn't have the budget or the internet connectivity and didn't understand the risks of using a service where you have to pay every month or they throw your data away. I am so glad I retired....
 
It seems likely that it was a custom creation then. If they were both complete CPUs then all they could have shared would have been the main memory and presumably somehow the peripherals, but the original design didn't make any provision for such a configuration and the chief designer, Dr Gordon's personal description of it never suggested the idea as a possibility. The H200 was planned as a cheap easily understood basic business computer with a modular design that allowed for expansion but not in that way to my knowledge. In fact Dr Gordon wrote that he didn't expect the typical H200 user to have much of an understanding of computers. I suspect that the actual people who did use it probably surpassed his expectation though, so customisations could well have occurred.

From what little I could glean from the documentation, the memory was "time-sliced" between CPU and peripherals, presumably in an on-demand fashion(akin to modern DMA). Seems plausible that one CPU could have been rigged up like a peripheral, to grab memory cycles. Possibly there was a way to do that more-symmetrically. However, the much larger problem would be how to get two independent OSes to cooperate (not step on eachother). I've worked a lot with modern SMP (Symmetric Multi-Processor) software, and from that experience I'd have to think it would have been quite a challenge to run two H200 CPUs fully-symmetric (although the machine could have been partitioned). I suspect it was more of a master/slave arrangement. Interesting idea though - Rob when do you think you can get two CPUs running? ;-)
 
I added an image of the inside of our 120 console here: https://www.ricomputermuseum.org/co...eresting_computer_items/honeywell-120-console

The datacenter that I worked in during the late 70s had a large dual-processor Honeywell DPS-8. This machine could be SMP with up to four processors.

It is quite possible that my coworkers in Dallas were being a little competitive when they found a way to make their 200 system dual-processor too. I agree that modifying the operating system to be SMP would be difficult task, but making it master/slave might not be so difficult.
 
Rob when do you think you can get two CPUs running? ;-)

I can't promise any symmetric running but as I am currently using an Arduino microprocessor as an interface to the system for my PC during testing it is quite possible that I may continue with that setup while also running the H200's own CPU once I have actually built it. Will that do?

Dave, I fully understand. It is infuriating when one is employed as an expert on a subject and then the boss decides that he knows better just because he is your boss. The really annoying thing that I have experienced is the boss taking advice from an outsider and then my having to maintain the damned equipment bought against my own advice. I had this with an HP all-in-one colour inkjet printer/photocopier/fax machine bought against my advice when it was virtually never used for colour printing and the chance of all three functions working successfully at any time was pretty low. After long telephone calls to an HP support centre in South Africa the machine was returned for replacement ... and then so was the replacement. Even when it was working they discovered that it was mixing the three colours to make black, so they were forever having to replace all the ink cartridges when they hardly ever printed in colour. To solve this problem they bought cheap second source cartridges which clogged up the system and throughout all this fiasco I only occasionally gently remarked "I told you so" because they were a charity and I was working there as an act of charity. I didn't retire until I had trained a replacement whom I could trust to support them at least as well as I had.
 
It is quite possible that my coworkers in Dallas were being a little competitive when they found a way to make their 200 system dual-processor too. I agree that modifying the operating system to be SMP would be difficult task, but making it master/slave might not be so difficult.

Yes, I'll accept that. Provided that both processors had access to main memory via the DMA capability it would be possible to set up semaphores in main memory to synchronise their activities. It is amazing what can be done when the need arises. When we had the H200 with just 4k characters of memory and three punched card peripherals (This actual machine is illustrated on my website) programming conventionally resulted in long run times and late shift working because the peripherals were so slow. To fix this I unwittingly developed my own version of multithreaded programming despite never having heard of the technique. Each peripheral had its own software thread and a system of semaphores ensured that whatever task could currently be done would be with data stored in memory until another peripheral could process it. This technique vastly speeded up the run time but our resident Honeywell engineer, the aforementioned guy with the presenr day broken leg actually, mentioned that we were wearing out the bearings on the card processing equipment far faster than usual. Honeywell must have based their maintenance on typical usage, not continuous flat out operation through manic multithreaded programming. The ridiculous thing is that I did this in 1966 using a machine with a tiny memory and while timesharing between programmes became the norm in machines with larger memories multithreading within programmes was not so commonplace for a long while. No doubt this remark will attract many comments here about early use of multithreading, so bring it on. You have to bear in mind that at one point I was the only person programming that machine without anyone else to advise me on current techniques. It was a one horse site and for some time I was that horse's only trainer and a novice myself at that, but the machine itself just deserved to be programmed well. That is why I have such great respect for its design even now. I didn't have more experienced people to guide me then, just a computer that did that itself by being what it was.
 
This technique vastly speeded up the run time but our resident Honeywell engineer, the aforementioned guy with the presenr day broken leg actually, mentioned that we were wearing out the bearings on the card processing equipment far faster than usual

One of the printers on our DPS-8 was purchased as a 600 LPM. We used to pull a wait state card and it would run at 1200 LPM, but light a fault light on the console. We made a cover for the fault light so the field service people could not see it. The field service people would always complain that our printer needed much more service than at other sites. We always blamed it on the cheap paper and ribbons that we used.
 
M_T, yes it wasn't what hardware you got but what you paid for that seemed to matter and the level of maintenance needed was a factor in that so far as Honeywell were concerned. With one of our later larger machines we went through a short period when it had a particularly heavy workload, so our field engineer changed a connection in it to effect an upgrade to a faster machine. Whether it would perform reliably at that faster speed depended on how well maintained by him it had been, so it was his call. We found the Honeywell field engineers very helpful, as were all the Honeywell staff that we encountered. No doubt they made a big effort to give Honeywell a good foothold in Britain where ICL, formerly ICT, was the most likely choice of computer supplier for businesses. In fact I discovered that the best source of Honeywell 200 series documentation here is held in the ICL archives now at a site run by our national Science Museum. Their archives contain original proprietary documentation from all of their competitors, so is a great source for any vintage computer researcher. Some years ago I spent hours there searching through all the H200 files in it to glean useful information. In fact I have copies of the control panel physical assembly manifest which lists every nut bolt and washer as well as all the major components along with diagrams showing where the parts fit. However, building my own panel from scratch would have been pretty daunting, so being offered an original was great news, even if I have been waiting over seven years now for it to be sent.

Back in 1965 when the deputy manager of our company was choosing our first computer to replace our ageing tabulator equipment he decided to opt for the relatively new Honeywell 200 series of small computers, which was all that we initially needed. That was a radical decision as the machines were designed to operate on US mains supplies at a different voltage and frequency from UK supplies, so we had to install a rotary generator to power the computer room to provide a US environment within it. In fact everything about the computer system was alien to us, from the different standard letter size paper used for documents to the UNC screws used in the equipment. There was also the very coarse language, quite blue in fact, that was apparently used by Honeywell executives at meetings with our far more genteel management. As our work and financial products were so specific to our long established company, which had been formed by an act of parliament in 1835 and could still write cheques directly on the Bank of England itself when almost all other cheque accounts there had been closed a very long time ago, we wrote all our own software from scratch using every feature of the H200 in ways that even Honeywell hadn't envisaged, just to prove to our board of directors that the deputy manager had made the right decision. A few weeks ago I heard that he has just died in his nineties. When he retired from the company many decades ago he came to me and said "We really made that first machine sing, didn't we?" because that early success resulted in the company staying with Honeywell and upgrading to ever larger machines for several decades exactly as Honeywell themselves hoped. Towards the later part of the century new less enlightened management came in and replaced the Honeywell mainframe with an IBM one because they didn't know Honeywell systems, which threw us back into the stone age in comparison to what we already had. At that point I refused to work with the mainframe any more and moved across to the PC network, which ran IBM OS/2, but in fact I had been working on the branch offices' collection of Honeywell DPS6 machines for some time anyway. Although IBM mainframes were apparently stuck with decades of backward compatibility stopping them progressing, OS/2 was far better than WIndows at the time. Unfortunately we eventually changed over to Windows and some of my more advanced software developed for OS/2 simply couldn't be converted because Windows didn't have the ability to provide any equivalent functionality. Back then Windows hadn't yet moved over to the object model based structure used by both OS/2 and Apple computers. In 2000 the company, ailing for many reasons, was taken over and I took early retirement, but the senior manager who had overseen the replacement of Honeywell kit with IBM had been "asked to leave" long before that. Sacked for buying IBM? Really? 165 years in business was a pretty good run for the company though.
 
Last edited:
My contact in California has now sent me photos of both sides of a control panel driver board and the mystery of the capacitors connected to the SCR gates is resolved. The other ends of the capacitors are grounded and together with series resistors from the input pins they form low pass filters so that the lights respond to the average signal levels rather than every single transition of the data lines. That would make sense if the lights were powered from a relatively low frequency AC source as otherwise any single data pulse during an AC cycle would turn an SCR and its light on until the end of that cycle, which would result in most of them being on for most of the time given the high pulse rates on the data lines.

I still have to work out what the other odd components on the boards are for. I'll post the pictures later.
 
My contact in California has now sent me photos of both sides of a control panel driver board and the mystery of the capacitors connected to the SCR gates is resolved. The other ends of the capacitors are grounded and together with series resistors from the input pins they form low pass filters so that the lights respond to the average signal levels rather than every single transition of the data lines. That would make sense if the lights were powered from a relatively low frequency AC source as otherwise any single data pulse during an AC cycle would turn an SCR and its light on until the end of that cycle, which would result in most of them being on for most of the time given the high pulse rates on the data lines.

I still have to work out what the other odd components on the boards are for. I'll post the pictures later.

That's funny, I had to do the same thing for my simulator... so to speak. A typical GUI monitor refreshes at about the same rate as AC cycles... so I had to reduce the rate at which the lights changed in order to get a more-pleasant looking display.
 
That's funny, I had to do the same thing for my simulator... so to speak. A typical GUI monitor refreshes at about the same rate as AC cycles... so I had to reduce the rate at which the lights changed in order to get a more-pleasant looking display.

I am impressed by your efforts to achieve authenticity with your simulator Doug. I have felt obliged to devote my available time solely to constructing the hardware version, so haven't been able to attempt what you have achieved even though my grounding is primarily in software development, not hardware engineering at all. When I originally wrote the introduction to my website I stated that the flashing graphics would get frantic in real time, as you have found. I also suggested that to make the simulation fully functional the buttons ought to be operable on a touch screen. Have you considered trying that?

My friendly English former Honeywell 200 engineer has checked with his old counterparts in the US about the existence of a dual processor H200 and they have no knowledge of a genuinely dual machine either although similar machines were sometimes run in parallel on tasks to speed them up.

Here are the photos from California of both sides of a slightly damaged 3IDA0 board. I have traced the interconnections of the miscellaneous diodes and resistors alongside the SCR driver circuits but it isn't possible to determine what their function was without knowing what they were connected to externally, so that will have to wait until I get the whole panel and can trace the wiring.


Click image for larger version  Name:	3IDA0 Back.jpg Views:	0 Size:	165.5 KB ID:	1224033


Click image for larger version  Name:	3IDA0 Front.jpg Views:	0 Size:	137.7 KB ID:	1224034
 
I also suggested that to make the simulation fully functional the buttons ought to be operable on a touch screen. Have you considered trying that?

Yeah, I've wanted to look into touch screen GUIs for most of my simulations, but don't have the means right now. I wrote it in JAVA so that it was "portable", but (most/all) tablets and smartphones won't run JAVA. I've not had the incentive to add a touch screen monitor to my PC as I sit pretty far back from the 37" TV I use right now. I need to read up on how the JAVA GUI handles touch screen and what needs to be done to support that. It may be fairly transparent, at least in some minimal set of features.
 
Just in case you're wondering I'm still around and still waiting for that control panel to be shipped.
 
Hi Rob.

Funnily enough, I was only just thinking the other day that I hadn't seen a post from you for a while.

Nice to see you are still around!

Dave
 
Has anyone ever located any actual Honeywell software? Like on mag tape? If they were 9-track tapes, we could at least have a chance of archiving them. Depending on how protective Honeywell was of that stuff, they might have guarded against it being retained when a machine was removed. I assume that Honeywell always "owned" the machine, and so insisted on having it all back? Or were they a bit more relaxed about things?
 
Just being SCRs does not always mean that they were driven by AC. They may have been intended as a latching function of something. They could then capture a short pulse and hold it until it was later cleared by either shunting the SCR or opening the current source. It can then work as a holding device instead of a two transistor latch.
Dwight
Good to know, I hadn't thought of it that way! I have a control panel for the H2015 that I started reverse engineering and noticed that the 3IDA0s were using SCRs instead of transistors, so I assumed that the lights were powered by AC. It goes along with some of Honeywell's use of logic circuits in their larger control panels where the CLEAR buttons on the left for CONTENTS & ADDRESS (as seen in m_thompson's photos of the H120) cleared the entire row since the individual bits couldn't be toggled on/off, but only on (the latching function that Dwight mentioned, so pressing CLEAR would have unlatched the SCRs). I believe those larger panels used the same 3IDA0 boards. The output of the SCRs would have not only been used to drive the lamps, but also be the source for the input to the logic circuits, which would have had to have been DC.

This thread is always an interesting read. I've downloaded and played with Rob's emulator (really only a program for clearing main memory...great for sequencing the blinking lights!). The H200 restoration must be a labor of love and I've been following it for several years. I have a bunch of H200/H2000-series stuff that I've either kept since I was a kid or collected over the years since then.

Currently, I am looking to make the lights and switches on the H2015 CP quasi-functional (just for show). As a stopgap (because I couldn't wait), I clipped some power supply leads to the AC and DC ON lights...it feels like I'm walking into a computer room again! :) My plan is to lay out a PCB or two that plugs into the existing edge connectors and drive everything with an Arduino. The easy part is the lights as they are either all or mostly driven by three 3IDA0s. The difficult part is the switches which all seem to go to that monster connector on the bottom. I would like to keep it all original, if possible. I doubt I could find a mating connector for the bottom, so I might have to do some cutting of the wires or just do without the switches. TBD...

Bill
 
They seemed to be a bit more relaxed. My dad had his own business back in the 70s and early 80s, buying/selling/maintaining H200/H2000-series computers. I still have a diagnostics tape from back then (hopefully the pic attaches). I'm 99.9% sure this is a 7-track tape.H200-diagnostics-1MB.jpg
 
Back
Top