Bruce Carsten: Lifetime Achievement Award Winner

Sep 1, 2006 12:00 PM
By David Morrison, Editor, Power Electronics Technology

Applying Hot Swap to Power Modules

In 1978, Carsten joined Research Industries in Burnaby, British Columbia, where he exploited an idea that he had thought about in the past: hot swap. Although hot swapping had already been used on logic cards, it was not generally being applied in power applications. It was in a project for Alaska Telecom that Carsten was able to exploit this concept to its full advantage.

In 1980, Alaska Telecom was installing microwave relay sites on mountaintops, which required dc power at 24 V, 48 V, ±130 V and +250 V. Power levels ranged from a few hundred watts to multiple kilowatts and were significantly different for each site. Battery backup was to be at one voltage (initially 48 V), with the other voltages produced by redundant converters. Because the range of power levels required varied as much as 3-to-1 on different voltages, the customer essentially asked for 20 different designs.

To satisfy these requirements, Carsten proposed a system with only two modules, a 48-V to 24-V switching regulator and a 48-V to dual 130-V dc-dc converter, where the outputs were connected in series or parallel and either end grounded (at the “card cage” output) to produce the other three voltages. Enough modules were paralleled to provide the power at each site, with two extras for redundancy.

This modular approach would decrease power-supply development time while increasing production volumes. Although the primary reason for modularity was to permit scaling of power levels, it had the added benefit of permitting the hot-swap capability. So, customers could avoid shutting off power when replacing units. Ironically, the power module’s reliability was so high that the hot-swap capability was almost never used.

Making Telecom Rectifiers Switchmode

Prior to 1980, the conventional wisdom was that switchmode converters were too noisy and unreliable for the demands of the telecommunications industry. But Carsten believed he had learned enough to overcome those difficulties. Sensing he soon would be leaving the world of production design, he decided to give this design challenge his “best shot,” as he believed this project might be his “swan song.”

The result was what Carsten describes as “the first successful switchmode telecom rectifier.” This 48-Vdc (nominal) 200-A power converter used bipolar transistors. (The 1000-V FETs available at the time produced unacceptable turn-on losses no matter how many were paralleled.)

The switchmode rectifier provided a great reduction in size and weight from that of the previous rectifier, which used SCR, saturable reactor phase control or controlled ferroresonant transformers. Because that design operated at the 60-Hz line frequency, it required large, heavy filters to achieve the low-noise performance required in the application. At the time, a typical SCR-based rectifier would produce 48-V dc at currents in the 400-A to 1200-A range and might weigh 1000 pounds to 2000 pounds. Carsten’s switchmode design reduced the weight required for 200 A to less than 200 pounds.

With the tremendous reduction in size and weight, the switchmode rectifier gave telecom operators the ability to expand their power levels, particularly in their “neighborhood collector sites.” Previously, the power equipment was so large and heavy it had to be installed during the construction of the building. The prospects for adding power supplies after the fact were severely limited.

Carsten’s switchmode rectifier, model SMR 48/200, was about 93% efficient at one-half to full load. That’s a leap forward from the 80% to 85% efficiency of the 60-Hz power supplies being replaced. However, better efficiency was not an incentive for the customer, because operating cost was not factored into the rectifier’s purchase cost.

One of the innovations employed in this design was polyphase switching.[1] Carsten opted for four phases because they could be easily digitally controlled using two bits with the basic clock plus the shift register giving you turn-on commands for phases.

Carsten’s work on the telecom rectifier also drove him to investigate the impact of high-frequency effects on magnetics design (see the “Advances in Magnetics Design” subhead below to learn more about Carsten’s work in high-frequency magnetics). At that time, the only reference for calculating winding losses as a function of frequency applied to sinusoidal drives. Carsten discovered that when a nonsinusoidal drive signal is applied, winding losses have to be calculated at every frequency component of the drive signal. This required quite lengthy calculations and led Carsten to develop an intuitive method for analyzing winding losses, including a normalization factor (Kr) that made the results easier to apply. Other researchers took Carsten’s work and extended it, creating a formal way of calculating the loss for any wave shape.

In the mid-1980s, Carsten gave a minitutorial on high-frequency magnetics design within a paper session at a power conference in Germany. The material presented in his tutorial was based on his design experience and lessons learned “the hard way.” Over time, this tutorial evolved into a full-day seminar and then a two-day seminar.

Over the years, Carsten has presented more than 120 design seminars in North America, Europe and Hong Kong on a range of power-design topics, including high-frequency magnetics, designing switched-mode power supplies for low EMI, topology selection, and control and feedback methods.

Carsten left Research Industries, which had become Telecom Power, in 1984 when the Burnaby operation was shut down. “I fully expected to look for another job,” says Carsten, “but I had been doing consulting on the side and had enough work to keep me busy for a few months. And it did not make sense to spend time looking for other employment while I still had consulting projects to finish. The projects kept coming in, however, and I have remained self-employed for the last 22 years.”

Reference

Want to use this article? Click here for options!
© 2007 Penton Media, Inc.