This page is a backward trip down memory lane - looking at some of the products I have contributed to - with a touch on the companies they were created for. All pictures are from publicly available documents and web-sites. I can disclose more details from the way-back stuff than from the newer experiences...

The not-so-secret sauce in any innovative product is obviously team-work. There are many people that I want to name as fantastic colleagues. However, I don't want to forget any - and there is also the privacy thing - so none named - none forgotten.

GN

The name "GN" comes from "Great Northern" - which again stems from the direcly translated Danish name: "Store Nord". Over 150 years this company has been behind a lot of very different products - a bit like NKT and Kryolitselskabet - companies we will get to.

I work in the Hearing division, which long before my time has been through some rough weather, but now runs more steady. It recently merged with our sister-company, known as Jabra.

This is my first job in many years in a company making actual consumer products. A typical setup for a "Hearing" user consists of:

• A pair of hearing aids
• A TV-Streamer that broadcasts BLE-Audio - aka - "AuraCast" (or older standards) into the hearing aids in the room - encryptioned if preferred
• A "spouse mic" that similarly broadcasts sound from microphones or other sources to the hearing aids
• A charger that charges a set of hearing aids overnight
• A phone

Modern hearing aids are very complex, with a lot of digital signal processing - as well as connectivity to phones and streamers. Because hearing aids are very small and are often worn for many hours, power-usage is an extremely important factor.

Since hearing aids are medical equipment, safety has been important here always. Medical safety and automotive safety (see Lithium Balance) are not so far from each other. In the latest years, cyber-security has come into focus, and I have worked on assuring general compliance with various standards.

Lithium Balance

Lithium Balance was bought by american Sensata some years ago - when I was there - but seems now to be "alone again". The name comes from the process of "balancing" Lithium batteries - making them equally charged and with the same voltage. Without this help from the BMS - Battery Management System - the total battery cannot hold the as much charge, and will wear down faster. The graph below shows State-of-Charge versus cell-voltage versus temperature.

In all the other companies I have worked in, we discussed milli-amps - if not micro-amps. However, in BMS'es it is about amps - hundreds of amps. Where voltages usually are 12V, 5V - and even lower on the CPU-cores, a BMS handles hundreds of Volts. Thus, it's an entirely different HW-platform with PCBs in fewer layers and larger distances between anything. Due to the high currents you need to consider how you turn on and off...

Safety - related to the automotive world, as well as trains, buses etc is extremely important. Still, the embedded software is not so different from other places.

As I was CTO when I left - not so long ago - I do not have so many stories that I can tell, but I do remember the "startup-spirit". Once, we moved to bigger offices next door and got new furniture. In other companies, you would hire moving men to set it all up. At LB, we bought a number of height-adjustable tables that came as DIY kits. I clearly remember all the developers on their knees at the floor, wawing power-tools some had brought from home - mounting tables. It was a fun and cheap teambuilding exercise.

Bruel & Kjær - later HBK

Like some of the other companies named on this page, Brüel & Kjær has been around for more than 75 years. I actually did my masters on "Time-Delayed Spectrometry" for this company long ago. However, since then, B&K had been split in three - "Sound and Vibration" (S&V), "Medical" and "Condition Monitoring Systems" (CMS).

I started as a consultant on Windows-software in the CMS company on a big project. This project was, however, cancelled due to a merger with a German company.

A bit tired of consulting, I walked across the street in Nærum and got myself hired "for real" in the "Sound & Vibration" company, which you may say, was the "core" B&K. Years later the history sort of repeated, as B&K, S&V merged with their German sister company - HBM - and became HBK.

I spent many years at B&K (and a few at HBK), and was involved in many products - both COTS - Commercial-off-the-shelf - and custom-made solutions. The latter was often for aerospace customers. We will only look at a few standard products here.

The 2245 has built-in Wi-Fi, BLE and GPS - an interesting device! The GPS naturally supplies coordinates, but actually also delivers a very good timestamp. 2245 also became B&K's first Linux-based product - something I had a big role in.

Before the 2245, I was heavily involved in the development and management of "LAN-XI". This was a very innovative product, where many modules could be connected in a star with an Ethernet switch in the center - and still sample in synch. Over the Ethernet we transferred data (TCP/IP), Precision Timing (PTP) and Power (PoE).

The same modules could also run standalone with a battery-module (same formfactor), or rackmounted with an ethernet switch basically being the backplane (and optionally, the same battery module). I learned a lot teaching TCP/IP and Cyber-Security at DTU (Danish Technical University) that I used in both LAN-XI and the later 2245 - and vice-versa.

We wanted a single module to deliver a steady stream of data over Wi-Fi to a tablet - and we also needed multiple racks of modules to do the same - in relative synch - to a powerful PC - via Ethernet and good switches. This meant many hours with WireShark for me.

Another very popular LAN-XI feature was the switchable "faceplates". These allowed users to adapt to whatever connectors they met. On the figure below you also see the LED-rings - using colors to show status.

In my earliest days at B&K I worked with the flagship PC program - "PULSE Labshop". This was probably the first program in its class to drop industrial PCs with built-in DSP-boards, and instead use the "native" intel CPU alone - eventually running from a Laptop (we sort of repeated this trick when 2245 became our first SLM without a dedicated DSP). Labshop was based on C++ and MFC - and later extended with ATL.

IPBlaze

At one time, I actually left B&K to work as CTO and developer in a small startup called IPBlaze. The only product (to be) here was a TCP-offload Engine - aka TOE.

I developed a lot of Linux-driver code for this TOE, while others worked on the FPGA. The FPGA handled the central "connected" state in TCP - where all the datacommunication takes place. The rest was done in software and socket layers were needed to make it usable from standard applications. I also created a Trac-based web-environment for documentation, bugtracking and CVS version-control (you can get this now with git instead). Without customers, however, the pay stopped flowing, and I returned to B&K after 15 months - just in time for the LAN-XI adventure, and with Linux knowledge waiting to be applied in an SLM...

At B&K it was common to say to people leaving the company: "Have a great course!". Just like me, many came back for another round.

NKT Elektronik & DSC Communications

NTK - "Nordisk kabel og tråd" - is best known for huge cables, nails and screws. However, in the nineties, the sub-company, NKT Elektronik, was heavily involved in building the digital highways of the world.

While I was there, NKT Elektronik was sold to DSC Communications and moved from Brøndby to new buildings in Ballerup (where GN resides today). I clearly remember the press-release after NKT had sold us - one of it's four pillars: "Hurtigere frem på tre ben" or "Moving faster forwards on three legs." After I left the company, it was sold - first to Alcatel, then Tellabs - and I believe that the remaining part is now in India.

The figure below shows the Synchronous Data Hierarchy - SDH - which I was hired to work on. SDH allows devices - multiplexers - to "dive into" the high-speed datastream in the backbone and extract/inject lower-speed datastreams. Earlier solutions required the whole datastream to be disassembled and reassembled.

At first I was a bit disappointed when I learned that the HW could do it all - in fact it needed to, as no SW was fast enough. The embedded software that we did was the management - setup, alarms, reconfig, statistics etc. Nevertheless, there were a lot of interesting challenges like dual-CPU system, bank-switching, HDLC and much more.

Managing a small test-tools department, I invented a test-script language and tool - "WinTester" - to match our stripped-down ASN.1 API. This gave me the opportunity to learn and use Lex/Yacc.

DataKomm - Production Testing

I knew the owner of DataKomm from Storno (we will get to Storno later), where we worked together and became friends in private life. DataKomm was a small consulting company in Måløv, and I worked there "on and off". For some time I also had a role as "Technology-manager". A very memorable assignment was for production testing.

A factory was making their production tests more advanced - moving to a Motorola 68k based platform, using the brilliant little RTOS called OS-9 (Linux-like before Linux). I was assigned to work with the factorys gifted mechanical engineer, who needed some assistance with the new platform. We were a fantastic team. He knew exactly WHAT he wanted, and I soon found out exactly HOW we could do it. Probably the most harmonic project I have ever experienced.

I learned a lot about 68k and OS-9, and got to setup a toolchain from the bottom, with makefiles, compiler/assembly-switches, PVCS version control - and obviously the coding. There was a Eurobus with 10 PCBs, a screen where we used semi-graphical characters and "hand-held" shadow-memory for flicker-free updates, as well as custom-keyboards, printer, digital and analog I/O.

It was via DataKomm that I later began working with NKT Elektronik as well as B&K CMS.

Dantec Electronics

Dantec was earlier known as Disa. It had been owned by A.P. Møller, and later sold to Kryolitselskabet Øresund. I was hired into the medical division, which made measurement equipment for muscles and nerves as well as urology. There was also a division creating laser measurement systems. After I left the company, the two divisions were sold seperately.

I clearly remember my first day at Dantec Medical. I was to work in the DSP-team, and the other guys said: "There is a DSP-presentation in Lyngby at 10:00 we need to see. We'll take the department's BMW - you drive". To a 25-year-old guy, this beats flowers on the first day.

Since the R&D department was geographically a few kilometers separated from the rest of the firm, we had inherited the CEOs old car, so we could stay in touch. I cannot however, remember the topic of the presentation.

My first project at Dantec was the Evomatic, which I have some stories about in my book Microcontrollers with C - mainly relating to Bit-Slice development. The follow-up to that one was "Counterpoint" - shown below. This used a 80x86 for controls (I think it was 80286 - but it might have been 80386), and a "Sharc" DSP from Analog Devices for signal analysis.

In Counterpoint I did a bit of DSP - like on the Evomatic - but very often there was not enough DSP-work to go around, and I started taking on more classic realtime embedded tasks. I volunteered to do a self-test program that we used in production and service. This gave me great contacts around the entire company. It was also technically interesting, because you start the program assuming that nothing works and gradually test pieces, before you use them in the next step. In normal programming you assume that your RAM, screen, keyboard etc works. This gave me insight into the HW and made me the perfect candidate for a tour to the US - updating HW and FW in prototypes that had been shipped earlier.

It was great fun to be part of the Counterpoint introduction at a medical congress in Sorrento, Italy. We completely surprised the competitors with our advanced system.

The team I was in, worked very close to the guys doing the urology equipment. As they where finishing a new flow-meter, it was mounted above a gent's toilet for general usability-tests. When you came out, a small strip was printed. It had a flow-graph and "voided volume". Soon we had a competetion going. Everybody had strips on their wall with huge voided volumes. We were drinking a lot of water, sitting cross-legged at our desks - more and more uneasy - until we stormed to the toilet.

One of the last projects I did at Dantec was very special to me. It was the kind of project nobody wanted because it was an upgrade, but again I volunteered. And again it was very rewarding. This kid-brother to Counterpoint was called "Cantata" - see picture below. We took the electronics from an older instrument and revamped the mechanics. We also replaced the old 8085 with a newer 80x86 - my demand to take on the project. This allowed us to reuse a custom-made "windows-like", event-based system from Counterpoint and completely trash the old software. DSP multiplications was done in special HW, but everything else was done in assembly (DSP) or Pascal in the 80x86 - much faster than the old 8085. I did all the software - DSP, UI, drivers, OS adaption - the works.

While I was at Dantec I wrote some English end-user documentation - including a manual. Afterwards it was decided to translate it to Danish, and the company hired a translater to let me code instead. When I did the review, I saw that in the chapter on configuration - where I had written "... save the changes" - it had become "spar på ændringerne" - "hold back on the changes".

Storno

My first job after graduating as MSc was at Storno - a company making cell-phones for cars. As you may know, the "Storno" name is a contraction of "Store Nord". Thus the circle is complete - I actually began at GN.

When I started, Storno was owned by GE - General Electric. A couple of years later it was sold to Motorola.

At this time companies needed new developers so much, that I never applied for a job - I was "fished" at the university.

At Storno I worked in "Advanced Engineering" doing all kinds of experiments and try-outs. As I had just graduated with a DSP-project, some work was DSP-related - mainly based on a PDP-11 running RT11. Other work was making experiments on an artificial radio-path as well as in real life - and connecting everything in the lab together with "Kermit".

At this time NMT-radios were the standard, while GSM was in the making (my bosses were always in meetings about GSM). A mobile phone would consist of a handle with a wire to the dashboard, where a big thing was mounted (see picture below) - talking to an even bigger thing in the trunk!

I did not participate in any "real products" - the picture below is from a radio introduced while I was at Storno.