My Professional Life

I have enjoyed a number of careers in my working life. I started as an electronics design engineer in the defense industry, moving to commercial electronics and broadening my skills to include mechanical and optical design. 

I moved from there to semiconductor sales where I was focused on data communications and defense designs.  That focus on communications eventually transitioned me into the IT world, starting as a network administrator then to senior network engineer, and eventually to where I am now.  Currently I manage a cybersecurity team for a global enterprise. 

Professional Achievements

My technical achievements are listed in this section of the web site and include electronic systems for planes, tanks, cookers, industrial controllers, digital road signs and a whole host of other things.  I have designed my own silicon chip from concept to production.  I have also designed data centers and remote site networks for a global enterprise.

Professional Skills & Certifications

From this varied background I have developed skills in coding, teaching, mentoring and managing projects. I have presented to engineer, executives, and have taught VHDL to other engineers for creating programmable logic devices.  I have often operated and communicated in multi language, multi cultural environments and currently manage an international team across multiple time zones.  I currently hold the following professional certifications:
  • Six Sigma Yellow Belt 
  • Six Sigma White Belt
  • CompTIA Net+
  • Brivo Technical Certification

Contract work can be very satisfying

One of my contract jobs was to design a local office network for about 150 people and assist in the implementation of the network. My role in the design was to plan, design and configure the switching and some of the firewall rules.  In addition I assisted with the patching of the racks, and I have to say I was quite pleased with the results.

ASAs, servers and other odd and ends

Data network switching in blue, telecoms in red.

This project took several days to complete but at the end of it the office had a data and telecoms network that has resilience and security. This was one of my favorite contract jobs and clients.

Cisco 1841 Memory Upgrade

Note: this was written in 2012 and the technical details reflect that time.


As a contractor, you need to keep your skills up, so I kept a lab at home.  The clients I dealt with were mostly small and medium sized businesses. SMBs don't generally have large networks and I would often use the lab to try out ideas. 


I also liked to browse eBay for cheap gear. If you have looked at the Cisco 1841 router, you would see that many are advertised as having IOS 15, but beware; they often don't have enough memory to run. IOS12 will run on 128MB but IOS15 requires the full 384MB in order to run correctly.


This is what happens when you don't have enough memory:

This has happened to me twice now; twice I have purchased 1841's with IOS 15, and twice they have not had the required amount of memory to run IOS 15.


My first 1841 came with IOS12 and 15 so I just dropped back to 12. This is currently my home router. My latest purchase needs to run IOS15 for my lab work, so I decided to try a memory upgrade.

Memory prior to upgrade


This is the Cisco page showing how to add the additional memory. It is very straight forward requiring only the removal of one screw to get the box open.


Of course you can't just throw any old memory in there. The 1841 comes with either 128MB or 256MB of DRAM, with a single slot for upgrading to 384MB. In my case I have 128MB so I needed a 256MB ram module, and after a little research, This is the one to get.


Installation took maybe all of 5 minutes, and the router is up and running again:


Memory post upgrade

And that is all there is to it. I've been working with the router most of the day, and so far run into no issues with functionality or with the extra memory.


Programmable System on a Chip

In my time working for Ambar components, I had spent a lot of time working and training with the folks at Cypress Semiconductor, so when an FAE position opened up in Texas, they offered me the chance to apply which I did.  I was lucky enough to secure that position, and found myself relocating to the Dallas area. 

One of my first successes with Cypress semiconductor was the work I did with a digital sign maker in Texas.  The types of signs they designed and built are the roadside dot matrix signs. 

The PSoC module

Cypress had a product called PSoC ( Programmable System on a Chip). The PSoC was a small micro controller with configurable I/O which in this case was used for ultra bright LED driving and a serial interface for module to controller communication.  The module drove a small matrix of LEDs.  These modules would be connected together to form one large display panel.

The sign comprising PSoC modules

This is the back view of a completed display module.  The PSoC modules are connected via serial interfaces to a PC card, that handles the graphics and communications requirements of the sign.  The signs could be permanently mounted or portable mounted on a trailer (as seen below)
The completed Item

The completed item shown is one of the first trailer mounted signs.  

Optical Networking Components

This was a white paper I wrote while working for FQD, so this is dated around 2000, and was partly updated in 2008.  Many years ago I used to sell optical components to companies like Alcatel, Marconi and Cisco. I put this handy guide together because at the time, I was unfamiliar with these types of components, and needed to get up to speed on them quickly.

______________________________________

Optical Network Components

This guide covers the basic optical transmitter and receiver types that were current at around 1999 - 2001, so the products referenced will be obsolete by now. However, the physics involved will still be relevant today. I have also included a little about fiber types, just to complete the work.


The posts are split into the following:

  1. Laser Diodes
  2. Wavelength
  3. Laser Receivers
  4. Fiber Types


1.  Laser Diodes

In this post, we will discuss laser diodes, and how we use them for data communications.  So what is a laser diode? I'm not going to go too deep with this, as there are some good articles already written and posted that go into a great deal of technical detail about laser diodes and how they work. For the purposes of this article, all we need to know is that it is a semiconductor device, much like an ordinary LED, that generates light at a tightly controlled wavelength.


The other curious fact about laser diodes is that the light they generate, comes out of both ends of the die. Die is a term used to describe the chip itself. In the case of lasers it is not a silicon chip, but rather a Gallium chip of some sort (Often AlGaAs).

Other pertinent facts are that the wavelength of the light is effected by temperature, so to control the wavelength, we must control the laser temperature.


The other fact to consider is that to generate a coherent takes a little time. This means that for the most part, (at the data rates we are talking about) we cannot simply translate our digital ones and zeros into an on/off function for the laser. Rather we must modulate the light output of the laser to form our data stream.


Quality and stability of the light output are very important factors in lasers for communications. The distance we can transmit data, and the rate at which we can transmit it, are directly connected to the stability and quality of the source.

So how do we control the laser?

Generally speaking we control two things - the laser current and the laser temperature, and both are related. As current flows through the laser, it will generate heat, changing the wavelength (and potentially, the quality) of the light output.


To compensate, the laser die is mounted on a Peltier cooler. We can control the current through the peltier cooler to control the die temperature.  Remember the die also creates light out of both ends? Well the light out of one end goes to our fiber for transmission, but the light at the other end is measured with a sensor to detect the light intensity. This way we can tightly control our laser.


What we have coming out of the laser at this point, is a carrier wave. This is a controlled wavelength of light at a known and constant intensity. What we need is data, so how do we do that?   In order to create ones and zeros, we need to modulate the light from the laser. This can be done in three ways:

  • Integrated modulator
  • External modulator
  • Direct modulation

The integrated modulator uses another die (usually AlGaAs again) and that die is cemented to the same substrate as the laser. The 'transparency' of the modulator die is also controlled by a current and it is this current that modulate with our data signal. In many ways this acts like a shutter, turning the light on and off, although in reality it isn't that black and white - it tends to limit the light output to 10% for a zero and 90% for a one.


The external modulator works in exactly the same way, except that the modulator is a separate device and has to be connected to the laser.   


We mentioned direct modulation of the laser itself above as being a bad idea, and it is for very high data rates, but; for short distance low bandwidth data streams it works just fine and as the technology has progressed, we can see 850nm lasers being directly modulated at 1Gb/s for running over 100-200M.


Moving on from our discussion of laser types, lets take a look at wavelength and where these lasers are used.


2.  Wavelength

There is a whole spectrum of wavelengths used for a multitude of purposes out there, but the wavelengths most often used in networking are:

  • 850nm
  • 1310nm
  • 1550nm

There is also the 1480nm wavelength and this is used a great deal in data communications, but generally as a part of an EDFL pump amplifier. That is really outside of the scope of this document but does bear mentioning.


850nm

850nm lasers are cheap to make and use cheap POF (Plastic Optical Fiber) or multimode fiber. This makes them ideal for directly modulated applications where cost is an issue. The drawback for this type of laser is the short distance it is able to transmit over. These types of links tend to be short distance fiber Ethernet and similar.


1310nm

The 1310nm wavelength sits between two absorption bands making it really useful for single signals, rather than multiplexed signals, but it can work well with both single and multi mode fiber.


1550nm

With 1550 nm we have a larger spectrum to work with allowing us to create a large number of channels (1470nm - 1610nm). Matched with single mode fiber, DWDM (Dense Wave Division Multiplexing) systems are capable of carrying over 160 channels of very high data rates over long distances. This is made possible by the ability to tune lasers as discussed in the previous post.


To Compare

    Wavelength  Fiber Type   Data Rate Distance 

    850nM       POF            1Gb/s      100m

    850nm       multi mode     1Gb/s      500m 

    1310nm      multi mode    10Gb/s     1km 

    1310nm      single mode   40Gb/s    10km+

    1550NM      single mode  100Gb/s   100Km+


It should be noted that these are very approximate and not absolute. As technology marches onward, these numbers will change, and are most likely out of date now. For instance, if we look at the current Ethernet

standards:


    Standard     Fiber Distance  wavelength  

    1000Base-SX   mmf 550m      770-860nm 

    1000Base-LX*  smf   5km       1310nm 

    1000Base-ZX   smf   100km     1550nm

*There is also a mmf version that has a range of 550m.


... and then there is SDH, ATM, Metro Ethernet, CWDM and DWDM long haul systems to name just a few fiber systems out there and they are all specified differently, but the above information is good 'rule of thumb' stuff.


3.  Receivers.

There are generally two types of receiver used in optical communication systems:

  • PIN Diodes (Positive insulator negative)
  • ADPs (Avalanche photo Diode

PIN diodes are simple to use, require little circuitry, and are cheap, but have no 'gain' so one photon received = one electron in the circuit. APDs on the other hand have gain (~100) due to their structure and materials used, but they are expensive and require a high voltage circuit to bias them. That is the short version.


For the long version of this answer, there is google and wikipedia and a whole slew of other white papers from folks like Fujitsu and JDSU who manufacture these devices and know a great deal about them.


In terms of their construction, they are very similar:

The 'secret sauce' in most cases for these devices is the lens.


So where are they used?

When I was working with these devices, the rule of thumb was this:


  PIN   1310nm   short distance   2-3Gb/s    mmf

  APD   1310nm   long distance    10Gb/s+    smf

  APD   1550nm   any              10Gb/s+    smf


I suspect this has changed now due to advances in manufacturing and technology in general.

And so that brings us to fiber which I will cover in the next post.


4.  Fiber Types

In simple terms, fiber is the pipe we push the light through; it is the optical equivalent of the electrical wire and just like wire there are certain things about fiber that we need to know.


First of all, fiber is not loss-less. As our light signal travels down our fiber, it becomes attenuated and distorted, just like and electrical signal on a wire. The numbers involved are different of course, but still, we have to design around these issues.


Fiber is also made of glass, and generally about the thickness of a hair. This makes it very hard to join or splice with other fibers. The glass fiber is clad in another material (the cladding) that keeps the light from escaping the fiber, by a process called total internal reflection.


The fiber and cladding are then protected by another outer layer, the buffer, and then the plastic jacket forming the outer layer.


There are two (main) types of fiber for networking - single mode fiber and multi mode fiber. I was going to write a whole lot about this, but two things come to mind; like wiring, we want to keep the fiber cost low, and out of the two, multi mode fiber is less costly than single mode fiber. So we use mmf for short haul lower data rates (850nm, 1310nm) and we use smf for high data rates and long distances (1550nm)


I said there was two things didn't I? The second thing is this - Wikipedia has a great page on fiber and I encourage you to read it.


And that wraps up my quick discussion on optical networking components. I hope that was useful information.


De Dietrich LED Control Panel

Another display module design that I put together was for the French company De Dietrich, who  manufacture high end cooking equipment, and this module was intended for an oven.  The module required an LED display with custom icons, two switches and a rotary switch encoder.

The completed module

This module comprised two designs, the LED display module and the main board it was mounted to.  The LED module consists a printed circuit board and custom plastic molding.  The custom molding includes a flame icon and a celsius symbol.

The icon is top right of the module

As with other Three-Five designs, the LED controller die was bonded directly to the custom LED PCB.  Three-Five Systems had extensive experience with wire bonding, as all LED displays require wire bonding.

Sadly I cannot find a picture of the display in the final unit, so if anyone has one, please feel free to send me a copy. 

Lenze Industrial Inverter Display Panel

This was one of my first LCD designs.  Again I was responsible for the electrical, PCB, LCD and mechanical designs. This was a custom LCD display that was to fit into a plastic molding.  The LCD was a separate custom design that I put together including the metal bezel.

Custom LCD module on the main assembly

The electrical design for this was a little more complex.  Back in the 90s the LCD was driven but both X and Y drivers, mounted in die form under the LCD PCB.  The LCD itself was a custom design completed on AutoCad and manufactured by Three-Five Systems.

Completed Assembly

The remaining electrical design was conventional surface mount tech, as can be seen in the photos of the completed unit.  Some push buttons and LEDs completed the design.

The Complete Unit in situ

And this is what it looks like in the finished unit.  The module was designed for a number of different units, this is just one of the 

Custom LED Dot Matrix Display

This project was a stackable LED dot matrix display.  I don't have the drawings for this any more but this would have been about 25mm x 100mm (or 1"x4" for my American friends).  For this design I was again responsible for the electrical design, PCB layout and plastic molding design.

The Display Module

In order to make this stack correctly, the plastic molding had indents and protrusions to ensure alignment.  The photo below shows this detail.

This shows the location lugs

The electrical and PCB design featured a bank switched LED array, driven by a National 5450 LED driver, with the die directly mounted to the PCB (the black blob shown below) so part of the PCB design included the die layout and bonding diagram for the die placement. 
 
LED driver on the board

I believe this was in red but as for the purpose of the display, that is lost in the mists of time.

The structure of an LED

During my time at Three-Five Systems I worked as a design engineer.  My role was electrical, optical and mechanical design.  As such I learned a lot about the structure of LEDs, and how to design an LED display.  Here is a little about that.

LED Structure

The first thing most people don't realize about LEDs is that they emit the majority of light from sides not from the top. In fact the top of an LED is where the bond wire connects.

LED Assembly

This means that to maximize the light output of an LED segment, the plastic molding needs to have a polished internal surface, and be angled to reflect the light towards the top of the molding where the diffusion layer is.

The diffuser is the layer of thin sheet plastic on the top of the plastic molding.  It diffuses the reflected light from the LED to create a uniformly bright segment.  If there is not enough diffusion, the LED can appear as a bright spot in the segment and thats poor quality.

LED, PCB and plastic assembly

LED Brightness

A DC current through the LED will provide a constant brightness.  The brightness being in direct proportion to the current. More current = more brightness.  However there are limits, namely thermal destruction!  An LED has some resistance to current. That resistance will produce heat and as the current goes up so the heat dissipated by the LED goes up until eventually a limit is reached and LED breaks and stops working.

By using Pulse Width Modulation (PWM) techniques, we can reduce the current through the LED, while appearing to increase the brightness.  This is because of something the eye does called integration.

If we have a high enough frequency for the PWM, the eye will not see the LED flashing, but sees it as a constant source of light.  If we take 2 LEDs, and one we provide a DC current of 10mA. The other we provision with 20mA with a 50% duty cycle, both are using 10mA average but the PWM LED appears to be brighter. 

Getting it all Even

So we have our LED bonded to our printed circuit board. We have a plastic molding that is shaped to reflect the light from the LED upwards to the top of the LED well.  To even out the light we add a diffusing film over the top of the well.  This does two things; it evens out the light distribution reducing bright spots in the LED segment, and it protects the LED assembly from damage and ingress of dirt and other foreign particles that may prevent operation of the LED.

There is of course a lot more that can be written about the structure of an LED segment, but the purpose of this post was to cover the basics and I think we achieved that.

ABB Industrial Controller Display Panel

I worked at Three Five Systems for five years.  I was the only engineer in the UK office and in my time there I did the electronic design, PCB layout and the plastic molding design.

This assembly is the front panel for an industrial controller.  It consisted of 6 dual digit displays (14 segments), and a bar graph module. Both of these modules are custom assemblies, so I had three PCBs and two plastic moldings to design.

the complete assembly

The electronics to drive all of this consist of discrete transistors for bank switching, National 5450s for the LED driving, (remember this was the early 90s) and a couple of voltage regulators.

The PCB back side

I used PCAD to perform the PCB layout, and AutoCad for the plastic moldings.  However, to create these modules, Three-Five would use the silicon die and wire bond directly to the board. In the picture above, the black blobs are where the die is located.

The completed Module

I can't find a decent picture of the completed module, but here is a watermarked picture of the module.  

Log Amp Control ASIC

Quantel was known for digital video in the late 80s and early 90s.  What a lot of people don't realize is that they also had a small team that designed electronics for niche military applications.  Quantel has had a fairly turbulent history and the military part of the company appears to have been spun off into Dynamic Signal Processing LTD.  The company does not appear to be in operation but references to its products can be found on its web site which at the time of writing is still up.

In my time at Quantel, my role was to create a mixed signal ASIC for stabilizing logarithmic amplifiers.  These log amps were used in video and radar applications.  The issue I had to solve was that log amps are inherently very unstable and given that these amplifiers are used in military applications, they had to work over a wide temperature range.  This compounds the issue with stability, making it far worse.

DVLA (Digital Video Logarithmic Amplifier)

The ASIC was defined to provide bias voltages for the amplifiers to stabilize them, based on a temperature measurement from a sensor on the amplifier.  Each amplifier was characterized in test and the appropriate bias voltage values were stored in an NVRAM.

Under normal operation, the ASIC ran a process that measured the temperature, compared the temperature value with the NVRAM to get the bias voltage, and output the bias voltage.  Inside the ASIC was a small processor (state machine), an I2C interface, some logic, DACs and ADCs.

The size of the ASIC was approx equivalent to about 4500 gates, and was to be rated to military specs; -55 to + 125C.  The ASIC was manufactured by Harris Semiconductor (in their fab in Melbourne FL I believe), designed on an Apollo computer design system, using Harris semiconductor FPGA text based design software. (no real graphical interfaces like there is today, or high level languages for definitions)

The only reference I can find to the design of the chip are on the google patents page and this page: Application EP90305332A.  These descriptions clearly show the slow speed mixed signal ASIC used for the control and compensation of the log amps.

MBB105 Helmet Display

This project was a helmet mounted display for the MBB105 helicopter.  The display had some HUD function but was also slaved to the under mounted gun.  As the pilot's head moves, so the gun under the helicopter moves with it, and the optical system identifies hard edges of vehicles and buildings for targeting. 

MBB Bo 105

We take this stuff for granted today but back then (the late 80s) this was cutting edge military technology.  Now, your cell phone camera identifies faces, even as your subjects move around. It's basically the same technology, in a phone!

My part in all of this was again, the analog interfaces for the display itself. In this case the analogue card was shrunk down into a module, that would be mounted to the board. This meant we used all new devices, a lot of surface mount which was very new technology at the time. It presented a whole set of new challenges and that's what made this project fun.

V22 Helmet Mount Display

Given that I worked in the airborne display division and my specialty was analog electronics, I was added to the project team for the V22 Osprey helmet mount display.  My responsibility was the implementation of the analog output card.
V22 Osprey

The analog output card took digital data from the display system and turned that into analog signals for driving a CRT.  Yup back then these things still used CRTs. 

I had just completed the C17 design so it will come as no surprise that the V22 card was a very similar design to the C17 card.  The interface to the digital system was via programmable logic devices (CPLDS) and these interfaces to some DACs.  The output of the DACs was a current and this was turned into a voltage and then buffered off the card using amplifiers.

It's a pretty simple set up but if course the design had to be resilient and military spec to survive the extremes in temperature, vibration and other environmental concerns.

Tornado E-Scope

 The Panavia Tornado was the primary strike aircraft for the royal air force in the 80s.  GEC Marconi was tasked with providing an avionics upgrade package that included a partial redesign of the e-scope

Tornado 

The e-scope is the display for the terrain following radar.  This radar enables the tornado to follow the contour of the land at low level. 

My role in this as a design engineer was to put together part of the display interface. This was a much smaller project than the C17 project for instance. 

F5 Interface Card

A number of the Northrop F5's were used for training in the USA and were due an avionics refit. As a part of that refit, an interface between a radar system and HUD needed to be created.

This was a simple interface card to connect the radar system to the HUD.  Essentially the radar system had an XY analog output, but the signals were of the wrong polarity.  So the interface card buffered the signals and then inverted them, and buffered them out to the HUD.

F5E Tiger II

The design was a quick and easy one, compared to some of the others.  

C17 Globemaster HUD

Much of my time at GEC Marconi Avionics was spent on the C17 Globemaster project.  GEC were putting together an avionics package including the Head Up Display (HUD).  HUD systems back then were usually comprised of two parts; the optics assembly, mounted in the pilot's field of view, and the computer system that drives the optics.

C17 Globemaster

I don't know what HUD technology looks like now but back then, HUDs were based on CRTs with vector scanning. This required analog connections to CRT/Optical module, usually X, Y & Z where X and Y are the left/Right and up/down signal controls and Z is on/off.  

My role in this project was to take the digital data from the frame buffers, and turn them into analog signals. The challenge here was this had to be 14 bit accurate over the MIL-STD temperate range of -55C to +125C.  That is a tremendous amount of thermal problems.

Everyone gets a HUD
Back in those days, surface mount technology was just beginning to be a main stream thing, and the issue when dealing with extended temperature ranges was the surface mount components popping off the boards when subject to temperature extremes.

The temperature and accuracy challenges were eventually overcome and as we can see, the Globemaster is  extensively used, 25 years after I worked on it.

Abrams M1A1 Laser Rangefinder

Way back in the 80s I worked for GEC Marconi Avionics.  I have no idea why we were asked to do this but we had to design the laser rangefinder receiver for the Abrams tank. Of course in a company that does military work you never get to see the big picture.

Abrams A1M1 (pic from Popular Mechanics)

The design challenge was that the return signal from the laser was so faint that thermal noise in the circuit would swamp the signal.  The design solution was a discrete low noise transistor amplifier, mounted on a peltier cooler to reduce the thermal noise. The output signal was then pushed through a logarithmic amplifier to provide a meaningful signal for the targeting systems.

It was an interesting project and as a new engineer, I found it fascinating and exciting.