Feature Archives | Feedback

I am not sure why power supply design engineers always end up working long hours into the night, but I have my suspicions. We like to blame The Company but you have to admit it: Engineers always tend to minimize the problems at the beginning of the job.

It starts this way:

WEEK 1

Project Manager: Grindle, I need a kilowatt power supply with 18 regulated outputs, global power input, EMI, PFC, Battery back up, hot swap and air cooled without a fan. How long will it take? What will it cost?

Grindle: No problem, man! Probably take about six weeks to design breadboard and troubleshoot. I would estimate the cost at $300 for the parts.

Project Manager: By the way, it must fit in this 50 cubic inch volume right over the oven.

Grindle: Better make it seven weeks.

Grindle's Diary

WEEK 5

I completed the design Friday of this week. Right on schedule. We have Doug our mechanical engineer working over the weekend laying out the board.

WEEK 6

The PC board arrived Thursday and was populated by Friday night. Charlie, my Technician, and I worked on it over the weekend. It took the entire weekend to locate a missing via on the board and some outright layout mistakes. Charlie is very handy with an Exacto Knife and reworked the board. By Sunday night we had the board up and running.

WEEK 7

The breadboard has a few minor problems. A filter capacitor value was miscalculated by a factor of 10. Easy to fix on the breadboard,but the packaging people might not enjoy this news. The loop is not stable and oscillates at 15 kHz. (the crossover was supposed to be 7 kHz). Other problems include out of spec ripple, and there's also noise pickup at the PWM chip, causing double pulses that have destroyed the efficiency. Maybe that is why the transformer is running so hot.

Project Manager: How are things going? Will you be ready to release the design next week?

Grindle: I think so. I should have it ready as promised.

I really believe I can do it. But deep in my heart I know that only if I have the luck of Bill Clinton can I ship this baby on time.

Week 8

It took all of week 8 to stabilize the sucker. It was quite a job. Unfortunately the real problem was that the PC layout was done by auto-router and the thin leads to the filter capacitors were responsible for the high esl/esr and the instability of the loop. I really should have looked at layout when Doug gave me the disk with it on it. I wonder if it is still in my desk drawer. Well, unfortunately we must redesign the entire PC board. And they think it is my fault! Hopefully the new layout should fix the ripple problems, too.

WEEK 9

Worked with Doug during the days this week making sure the layout is correct this time, and worked nights with Charlie on the EMI filter section. We released the new board layout Thursday and sent it out for express service. At night we discovered that the Conducted EMI was 20 dB out of spec at 200 kHz. (2nd harmonic). It turns out the filter design has a resonance close to this frequency that aggravates the whole thing. We worked all week trying to add capacitors and change inductor values. Each time the EMI was still way out of spec. Charlie (Technician) is mad because he must stay each night to help me. He misses his family. Me, too.

WEEKEND 10

At home I spent the weekend modeling a spice load to simulate the PWM . Then I attached it to the Simulated EMI filter. Bingo! It showed that we would be 20 dB out of spec. Maybe I should have used this technique when I first designed this thing. Nah, everybody would accuse me of "playing with your computer again, Grindle". Well, this will show them. By Sunday night (forgive me, Lord) I had completely redesigned the EMI section using an additional PI network. It would now be 20 dB in spec. All on my computer. How I love this stuff!

Monday of WEEK 11

I come in to work so proud, and proceeded to tell everyone that I had fixed the EMI problem using my computer at home. Of course, I also had to explain that this would mean calling off the rush job we put on the PC board and re-lay it out for the new EMI filter design. Yes, I know it will delay the program by at least another week. For some reason everyone was mad at me. Charlie said I should have done this at the beginning of the job. Sure. Have you ever tried to work at your computer for twelve straight hours in this place? No way! I get asked ten questions and hour and must typically attend two meetings a day, and when you work here nights, you generally do the things that you can't do at home.

Friday of WEEK 11

The new board came in today. We had to pay an extra $1000 for the fast turn around. The parts will be mounted and we start testing tomorrow.

Saturday of WEEK 11

Charlie said he expected it would take him about five hours on Saturday morning to load the new board. I arrive at 1 PM only to find Charlie still loading. He had to take his son to Soccer and decided to stay for the game. By 4:30 the board is up and running. When we apply the loads, we see our efficiency is where it should be, the ripple is in spec. Wow! Pulse loading the main output shows a good phase margin. We are home free. That is, if we meet EMI.

Sunday of WEEK 11

Charlie and I meet at 2 p.m. and run EMI. It passes. It is actually 20 dB in spec exactly as predicted. We are blessed. Wow. I can't wait till tomorrow.

Monday of WEEK 12

Sunday night I came down with the flu. I guess all this work has lowered my immune system. I staggered into work Monday smiling, knowing that I could just relax and read the pile of magazines on my desk. I tell everyone that we are done. I am so happy. Maybe they will appreciate me now. Maybe give me a raise and a better computer. Maybe even a bigger monitor.

At 3 p.m. the big boss's secretary Angie, came to my cubicle to get me. She said Mr. Blackridge wanted to talk to me about the job. I think to myself, Yes, Yes, he has noticed how hard I was working. He wants to personally thank me. Wow, I will ask for a P200. and a 21" monitor!

I was still quite weak when I went to Mr. Blackridge's office. I started the conversation smiling and saying "We did it, Mr. Blackridge!"

"Sit down Grindle" he said. "Yes, you did it, six weeks after we had promised the customer his prototype. Six weeks late !"

"Sir, I worked very hard those six weeks," I said.

He replied, "It is a shame you didn't work smart."

Then he said, "I just had purchasing price the bill of materials. It is $100 more than you estimated when we bid the job. At these prices, we can't make any money. This job is a disaster!"

Of course, I then said, brilliantly, "Sir I think I could cut the cost by about the $100." And, he astutely asked, "Grindle, how could you do that?" I again answered, brilliantly, "By operating at three times the frequency. The costs grew because of the EMI and need for multi-layer boards. At twice the frequency everything will get smaller." He then asks, "Why didn't you do this the first time?"

I couldn't tell him that I thought it would be too hard to design and get operating in the original timeframe could I? So I said, "I guess I should have." He mumbled something under his breath and said "How long is it going to take you to redesign to operate at three times the present frequency?"

"About five weeks," I said, and walked out.

Oh, how I love this work!

Frank Greenhalgh

Comments Frank's column hit a nerve and the responses we are receiving are proof of that. Read the comments as well as Frank's response

About the Author

Frank Greenhalgh has been working in power supplies and systems for 38 years. He has many impressive accomplishments and patents. Over the years he has made significant contributions to Trio Laboratories where he held the position of Chief Design Engineer and was then promoted to Vice President.