Please let me know your experience level. This will help target the discussions properly.

If you are new to power design, don't be shy about it - we all had to start at the beginning.

I remember my first day on the job in 1981. They handed a schematic to me and told me that's what I would be working on. I spent two days looking at it, trying to piece together the capacitors, diodes, magnetics and switches that made up a switching power supply. At the end of that, I went to my manager and asked for help. He told me it wasn't even a power supply at all, it was a power distribution unit with relays and timers. It was a humbling experience.

What was worse is that was in the days when the engineers drew the schematics in a logical way so that they actually made good sense, though obviously not to me in this case.

My first real develoment job was with an private avionics company to develop the aquistion, storage and display drver portions of an airborne weather radar system in 1976. After finishing that, I went to my boss asking "We cannot fly this with bench supplies." He responded with. "We were just waiting to see who first finished to develop the power supplies." Thus started my career in power supplies.
After spending a year on the bench, before the advent of PWN controllers, I had finalized the design of a cigarett box sized 30 W flyback converter, and a buck converter that worked down to -55 deg C. It was the most educational period of my life in the SMPS area.
Today, the development cycles are much shorter and the luxury of self-learning is much reduced. I chose to remain with SMPS's because of its combination of RF, digital, and just the mystery of the unknown and the undefinable. It still fascinates me today.
I hope the new entrants into this field, with your sophistticated modelling tools remember Murphy's collerlary: "It takes 98 percent of the time to develop the design/prototype, and it also takes 98 percent of the time to perfect it."

Enjoy.

Marty Brown

I like your story. I hadn't heard of that before.

I have an 80/20 rule on development. When the prototype is finally producing full power, it looks to the program managers that you are 80% done. In reality it's 20% finished. Now you have to get it production ready, do full dvts, blow it up a few more times with stress, temperature and surge testing. Then work out all the production bugs.

In early 1979 I was developing a 600W high voltage switching power supply for a spectrometer intended for measuring relaxation time of carriers in highly-compensated semiconductors and energy relaxation in high-temperature superconductors. We needed a reliable controllable power supply for this instrument. As I understood later I designed and used a phase-shifting hard-switching bridge module, long before the late Frank Goodenough published his article in 1981. Just because that was my first approach to the power supplies of that complexity, the idea of phase-shifting was viewed the first one.

In 40+ years of mostly embedded system design (when embedded systems design meant designing a product with a small microcontroller as the control element) I never really functioned specifically as a power supply designer. Rather the power supply was an important part of the design. Sometimes the power supply design was a simple linear supply, sometimes it involved selecting modules from a catalog, and more recently, actually designing a simple switching regulator (as simple as a switcher can be!). Hope this experience doesn't preclude my membership in this group, by necessity I have an interest in power supply design. I continue to do microcontroller based system design and/or analysis.

I started with a DC-DC converter in 1997. It was a flyback wit 30V input and 300V output. Straight away it had to get a UL approval for a medical application. Those UL requirements still annoy me more than the trouble of blowing up MOSFETs and getting the loop stable.
After that I worked on a telcom supply (48Vdc in) that had to have an efficiency of over 80%. I was so proud to reach 85% with my push-pull design with transformer driven synchronous rectifiers.

One thing I learned from my experienced boss very soon was never to tell the sales people that the prototype was working before it was tested to its limits. Otherwise they would run to the phone to tell the customer the first prototype was on its way.

I agree with Marty, after designing power supplies for over 15 years it is still a fascinating job. There's always room for improvement and there are surprises in each new project.
Educating the younger generation also becomes a challenge. Is it just my feeling or have the technical colleges stopped teaching about analog electronics all together?

To Harme-Jelle Zwier, over the last few years it indeed appears that engineering schools don't teach analog design. Everyone seems to be digital designers, and I use the term designer loosely as most are VHDL/Verilog coders. As someone once said, digital systems are made from analog parts. Ignoring the analog behavior leads to lots of problems. My two cents worth.

I started in power in 1960. My first project was to complete an inverter design for 5kva three phase 400 hz inverter. The input was 200 to 300 vdc from submarine batteries. The switching elements were SCRs with active commutation and the whole thing was convection cooled. That was a major learning experience. One of the problems we encountered is that some of the SCRs would lose forward blocking with a combination of temperature and current. Temperature alone was not enough so we couldn't screen them on the curve tracer. What we did was run the inverter until it failed and time it with a stopwatch. Then we would cool one of the SCRs with freeze mist and run it again. If the time to fail extended substantially, we had found the bad part. Another project was a capacitive discharge igniter for the Saturn rocket. It had to function immersed in liquid nitrogen which brought its own set of problems. At that time capacitive discharge igniters were theoretically 50% efficient in terms of the energy stored in the capacitor being discharged in the arc because the trigger gap dissipated as much energy as the spark gap. We found that by adding a diode we raised the efficiency to an actual 70% and extended the life of the trigger gap tenfold. The only problem is that it burned up the spark plug. With over 50 years in the power conversion field I have many stories that there simply isn't room to tell here but I can say I've had my share of ducking flying shrapnel from exploding parts.

Looks like we have a very experienced group here.

I broke the groups of years of experience up this way deliberately.

Year 0-1 You don't really know anything. If you did, there is no way you would take on the project that you've just been assigned because it's probably impossible.

Year 1-5: Reality sets in. You realize what a deep and complicated set of skills you are going to need to survive and maybe are having second thoughts. But you can't get out now even if you wanted to - once your first power supply is done, you will be a power engineer for ever. Sorry.

Years 5-10: acquiring the rest of the skills you need. EMI, PCB layout, testing, thermal, control, magnetics, manufacturing, etc.

Years 10-30: After 10 years, you are ready to call yourself a power supply designer. You actually start to think you know what is going on in this field.

Years 30+: You come to the realization that you are only just scratching the surface of what you still don't know, and you are not afraid to admit it.

How do you recommend apportioning experience to the group of members design integrated circuit controllers and regulators? We may not have strong backgrounds with the final components, but we certainly have to work on the issues with the actual designers of the applications

If it's power/analog related and driving the final components it should probably count. It is always a bit of a surprise to the semiconductor designers at our course how much of an influence the magnetics have on the performance of the final converter - usually much more than the choice of controller.

9 years in the world of power electronics (Motor drives) however not that much experience to the high tech SMPS designs that always look so much interesting from the motor drive side of the fence.....

Nice comments on the 80% or so of work needed at each stage of a project! That's so true.

Started out with linears as a teenager, progressed through SMPS flybacks for radio tracking of crocodiles in the Australian outback, then forward converters for magnetron sputtering of thin films all with bipolar transistors, then on to the present day with FETs and multiple independent 6kV bipolar outputs for driving electrostatic chucks. This last unit has all-digital control in a palm-sized unit and can deliver ac or dc between -6kV and +6kV on each output.

Loads to learn at each inflection point - now our microprocessor is doing floating point calculations at its maximum rate, interleaved with driving 5 FETs with its PWM outputs. So different from the old analog control days.

First real power supply development began around 1980 for a Navy-Submarine Gas Management System, a few KW and about 10 outputs. Started with crafting the spec., then later discovered I had to design it. That was baptism by fire.

After so many years, you still never know enough. The frequencies increase and materials change. We need to constantly accumulate new knowledge on top of our experiences.

Most of the present power technologies require power engineers to be RF experts with a heavy knowledge of physics. Switching in MHz and measuring noise in GHz was not an issue decades ago... today's engineers are not prepared well for this future in power and the old dogs learning the new tricks won't be around for much longer to teach them.