Feature Archives | Feedback

The Meeting

The meeting was held in a conference room in Houston. I was representing one of a number of power supply companies that had been asked to attend. All the companies were providing equipment for the new NASA space station. The main purpose of the meeting was to request that each vendor provide simulations of their power supplies. This would allow NASA to simulate the entire power system and check it for stability.

Instability in today's power systems is common. It stems from the fact that switching power supplies are constant power devices. This means that they have a negative input impedance. Negative impedance, when combined with a reactive element, can oscillate.

Negative Resistances Oscillate

Hewlett-Packard's first product was an audio oscillator. It used a Wein Bridge oscillator, a design that uses a light bulb to provide a negative resistance. Tunnel diodes possess a negative resistance and are used in high frequency oscillators such as "Grid Dip Meters."

When you have a reactive EMI filter providing power to a switching power supply, the possibility of an input oscillation exists. When you are putting together a space station, you don't want the power system oscillating at 22,000 miles altitude. This is why the NASA people called the meeting.

The Middlebrook Stability Criteria

The Navy first discovered this problem aboard submarines, which were the first vessels to demand the lighter and smaller switching supplies. Oscillations on the power bus occurred. The Navy commissioned studies to determine the cause of this. A study by Dr. R. D. Middlebrook (Caltech) http://www.ardem.com/ generated a canonical model combining the EMI filter characteristics with the switching regulator's. The study verified that both the EMI filter and the power supply had resonance characteristics, due to the EMI filter's reactive components and the switching regulators output ripple filter. Middlebrook plotted the two transfer characteristics, and showed that if they were separated in frequency and properly damped, oscillation would not occur. The findings of this paper became known as the "Middlebrook Stability Criteria."

Space Station Power

Information provided at the meeting was very interesting. We were told that the space station's batteries and solar cells could produce about two-hundred kilowatts of power, but the total load the station would carry totaled close to a megawatt! This would require having 80% of the load disconnected at any given time. Different loads, for different tests and applications, would be switched in and out as required, using care not to exceed the stations maximum power.

As each load will have its own bank of switching power supplies, their input transfer function may or may not match the transfer function of the previous bank being removed. NASA wanted to be able to study this, and they were asking for the vendor's help. They had no budget for this study, and were planning to keep the cost down by depending upon vendor-supplied models, supplied free of charge of course. My company had done extensive modeling of the supplies we were building and agreed to supply the Pspice model. Other company's models used other software. Some companies didn't have any models.

I thought to myself, "This is crazy. Do they really expect to be able to predict the load curves for all the permutations of load and check it with Middlebrook? I don't think so."

I raised my hand and asked a question: "When you assemble the station, will you be able to verify your models and predictions?" Never, I was told. There will not be an assembled station on the ground. We are only building one station, and it is being assembled in space with other countries participating. "But, but," I said, "What about problem analysis. I remember in the movie Apollo 13 how NASA was able to solve problems by simulating them on the ground with the actual hardware." Not this time. The budget does not support a ground station.

As I flew back to New York, I though about this and decided that it probably wasn't going to be a problem. At the same time, I would not risk a multi-billion dollar space program on my guess. That was a few years ago.

Cheaper and Better?

A number of years ago, NASA reacted to the government funding cuts by declaring a "Cheaper and Better" campaign. Now anyone in this industry knows that Cheaper and Better is an oxymoron. What I had witnessed represented a slight risk that an unstable condition might occur at some load conditions. What frightened me was that there would be no hardware for hands-on trouble shooting if a problem occurred in space.

Possibly NASA and the government should rethink this strategy. It seems that our space triumphs are behind us and our failures are recurring. This week, the Star Wars people fired an interceptor rocket into space. It missed the rocket it was supposed to destroy by miles. The previous rocket fired in October homed in on a balloon decoy and by luck found the target. Two Mars probes have recently disappeared, as well as a spy satellite, all due to some error or failure. The problem is that in most cases there is no smoking gun. These billion dollar toys go off into space and all the evidence of failure goes with them. No black box to find. The one Mars failure which was explained was caused by a dimensioning mix up, where yards and meters were interchanged. How reckless it is to send up a multi-million dollar Mars probe with unchecked dimensions.

What to do?

Should NASA ask for more money to improve reliability? Should NASA cut back on the number of programs and spend the money improving the reliability of the remaining ones? Is manned flight really necessary? I really don't know. Do you?

Frank Greenhalgh
January 27, 2000

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.