My Landis&Gyr Adventure

I (Ed Thelen) was there from 1989 to 1996
Table of Contents
- Exit Measurex
- Off to Work We Go
- Two Fun Projects
- - TVA (Tennessee Valley Authority)
- - Churchill_Falls_Generating_Station
- Not so fun
- - MMU Memory Management Unit
- - Misadventures
- Retiring

Exit Measurex
As related at the end of my Measurex adventure:
  • I had refused to work on memory banked systems and was fired as Measurex had chosen to stay with Intel's 8086 and trying to memory map different sections of code and parameters.
  • I would not work on what I regarded as a big mess.
  • Measurex was kind, gave me "make work" projects, such as determining the computer loads of mixes of gauges, slice widths, scan speeds, needed controls, ...
  • After about six months of that, and I still was stubborn, I got called into my manager's office, saw his manager, and I knew the end had come. Fired!!
  • They let me stay in the building for another 5 hours (end of day) to say goodbye to folks. This was really kind !!
  • I knew that I would have to run hard to catch a new job - But I didn't realize that I had forgotten how to run hard -
  • Since I had been fired, I could apply for unemployment, and part of that deal was that I had to show good evidence that I was out looking and interviewing.
  • The pickings were iffy -
    • a company now called Landis&Gyr (had been something like BrownBovary? Engineering) was advertising and looked promising. I carefully tailored an application/resume to meet their advertised requirements - no problem actually But heard nothing - twice -
    • a head hunter said that Lawrence Livermore was looking for real time LSI-11 - right up my alley, (help steer laser beams at a capsule of tridium), but a long commute - I applied , nothing, told to be patient
    • a company called "Cochlea" (yes, part of the ear) wanted to listen to newly manufactured nuts falling on a steel plate to listen for defects - gads
    • a couple of others even less likely
    • I avoided start-ups as 80 hour weeks were no longer attractive to me.
  • While sitting on my hands for a few weeks, I started writing a disassembler for the Intel 8086, for fun -
  • A job fair was advertised. I printed more of my updated resumes, and headed for it.
  • At the second desk were two reps from Landis&Gyr. Hmmm - lets give it a big try :-))
  • The first person I met was Ron Miller. We hit it off just fine. I seemed to be what he was looking for. When I left, I glanced and saw him writing a little plus mark on my resume :-))

Off to Work We Go
Oddly, I don't remember any other interview at Landis&Gyr ...
Their job offer was about what I had been making at Measurex :-)
I accepted promptly, somewhat gratefully, no bickering about salary.

So my first day - the usual thing is to introduce the new person around to lots of folks.
I am worse than many about remembering names of people. So I carried a little pad of paper and pencil, and unobtrusively as possible, I wrote all the names and hints to remember who was who.

The next days I went out of my way to greet them and say their names. My new co-workers were impressed ...
I know this is an old sales trick, but this is the first time this techie was so organized.
I heartily recommend it to all ;-))

I soon learned that Ron Miller had intended to hire only one person but wound up hiring another also. This person was Chetan Kitira, from India. We still exchange holiday greetings.

I think the second or third day, our co-workers were excited.
It turned out that Ron Miller, our manager, liked to give a public quiz to new hires, and all the folks has been through this public exercise, and wanted to see how the new victims fared :-|

The quiz was adding, subtracting and multiplying in hexadecimal. Fortunately for me, my previous experience at Measurex with the 8086 had sharpened me up on these hex operations, and I survived and thrived.

Our processor was the Intel 80386, and much development was on an Intel supplied development system MDS 80. Link with an in-circuit emulator so the emulator could be plugged into the Landis&Gyr system under development, and compile, load, and run the code at almost full speed, with handy things like break-points and register examination. and we used the Intel language PL/M-86.

As at Measurex, Landis&Gyr wrote its own "real time" operating system, which was not general purpose and left out many "bells and whistles" demanded by the general world.

After about 6 months, Landis&Gyr decided to abandon the Intel PL/M compiler in favor of a C compiler for the 8086. Mitch Allies reminded me of several reasons to switch:
  1. Intel had declared that they would stop supporting Intel Fortran. - that didn’t bode well for PL/M.
  2. Our owner, Landis&Gyr Zug (Switzerland), had decided to convert from their own unique structured language (on-of) to C and, as our boss, they thought that the whole world could run off of 2 masters, a small system master and a large system master.
  3. Some pressure came from the other side of our building because it was harder to find PL/Mers.
  4. People with our own group thought C was nice (Scott Datallo).
Kind of a “Perfect Storm”

We purchased a converter that did a pretty good job, except that in C all subroutine calls must have their calling parameters defined before usage. (This saves a pre-pass looking for subroutines.) We got the conversion complete and working in a few weeks, much to my great surprise. (Turns out I'm a pessimist, and usually right.)

To make life interesting, the applications, running on top of our operating system, were written in FORTRAN. The process engineers (mostly from India) very much preferred FORTRAN due to its predictability and safety. (It is very easy to shoot your foot off with the C language.)


I got to do some special projects which were fun.
One was for TVA (Tennessee Valley Authority)
The request was to permit some of the smaller installations to go into a special mode, reasonably automatically. It turns out that sometimes it is good to have the equivalent of a gigantic capacitor in the electrical network. The game plan was to turn off the water to a turbine generator, drain the water from the turbine for frictionless rotation, and run the generator as a motor, but with a lot more field excitation which makes the motor look to the network as a giant capacitor. The formal name is "Synchronous condenser". https://en.wikipedia.org/wiki/Synchronous_condenser

My usual scheme of working is to write a working specification, determine the interfaces, define the user interface, define the acceptance tests, then start coding.
My big frustration with this project was that I couldn't get the Landis&Gyr product manager to ask the customer to sign off on a proposed user interface. Often the demands of a user interface shows flaws or omissions in the specification.
I completed the exercise, but did not get to go the Tennessee to install it - Since I never got any complaints, I worry that they never used it.


Churchill_Falls_Generating_Station
Another project was a special terminal interface to some added control software (from the Control Engineers) for the water flow control in the high flat ground in source water for Churchill Falls Newfoundland, Canada
https://en.wikipedia.org/wiki/Churchill_Falls_Generating_Station

I got to go with the project manager to install that version. Actually I was just along for the ride, knowing nothing about what was going on in the control side of the operation. In spite of the lovely looking control room in the Wikipedia picture, our computer was in a concrete room (echo chamber) with several terribly noisy transformers. Really unpleasant !!! We did get a great tour of the generator and turbine and out-flow areas. Huge, impressive !!!

On the way back, I got to talking with a specialist in transformer oils for the large transformers that convert the 15,000 volts from the generators to about 300,000 volts for transmission to the province of Quebec and on into New York State. Apparently the intense voltage fields slowly breaks down the oil, and if it breaks down too much, you get arcs, explosions, and fires. The staff at these transformers MUST monitor the condition of the oil, or there will be big, expensive excitement !!!

Also on the way back, the project manager tried to find out how old I was. The US entry form has an area for a birth date, and I left it blank, fearing such an attempt ;-)) I have always looked, and maybe acted, very young for my age I had learned not to talk about previous experience, kids, ... and other age give-aways. I was over 60 at that time.

And some not so fun
MMU (Memory Management Unit)
Some time in my experience at Landis&Gyr, the 80386, with a "memory management unit" MMU was available and Landis&Gyr needed the added memory mapping that this processor provided.

Ron Miller asked me to quote how long it would take to get this memory mapping going for our operating system. I figured more functionality than he thought necessary. Among other things, I thought a method of letting the control engineers have access to the MMU for debugging their programs.

Ron disagreed, and thought my estimate way too high. He assigned Mitch Allies to the job - poor Mitch !!! After several months it was clear that Mitch was not keeping up with Ron Miller's very optimistic estimates, so another guy was assigned to help. The total effort for the two guys was about what I estimated. I have always felt grateful to Mitch for taking on that task !!

But the process control engineers needed constant help from a programmer familiar with the MMU while debugging :-(

Misadventures OH GADS - There were some wild misadventures !!!
We tried to keep the software of the various project somewhat related - it is all too easy to add modifications/improvements that are not incorporated into the "standard" code base.

We didn't use some of the tools that try to help, such as checking out software, and checking it back in after modification/improvement.

But we did try to keep the .h files that C likes to use to define things consistent - Ho Ho Ho
Frequently you would re-compile something with "unexpected" results, and had to find who did what with which .h file. It was a maddening environment !!! Lots of lost hours and half days.

Also we had all the project machines on a network, and placed the .h files on one machine - we kept project code on the project machines.
Several times the machine that had the .h files got shipped, and we could not compile anything until the machine arrived at the customer's place and got connected to the INTERNET so we could down load the .h files to some local machine.

Retiring - Leaving Landis&Gyr
  • Ron Miller's group was being broken up, and my next job was to link ORACLE database functions into the next generation of software.
    I had a friend who was an ORACLE database guru, I figured my brain was getting too slow, the integration was going to be a nightmare.
  • I had also recently discovered two great hobbies
        a) the Computer History Museum at NASA Ames, this later moved to Mountain View. ( 1, 2 )
        b) an old Nike site, SF-88, being restored ( 1, 2, 3 )
    and needed more time for these :-)
  • Oh Yes, and Betty did not like our little cracker box house in Cupertino so we had moved to Fremont - without checking the commute time traffic ... gads !!!
  • It all piled up suddenly - I gave two weeks notice, and was out of there, to the hobbies that were taking increasing time. :-))
  • I was a few months from 65, so retired to have a lot more fun :-)) As the Japanese say "Sayonara"!

After I left, another firm purchased Landis&Gyr, and my retirement checks are now paid for by Siemens of Germany.

A reason that I waited 20 years to write this up was the sour feeling I had about software development at Landis&Gyr.


This page started Jan 3, 2018
My e-mail address is ed@ed-thelen.org