Learning from a mentor

When Raouf Ismail, president of the company I was working for, Cambridge Aeroflo, asked me to join him in his new startup, I readily agreed. Aeroflo was his 8-year-old startup. He was an avid glider and engineer with an MBA from HBS. His initial designs of altimeters and other airflow and pressure-based instrumentation did not do well. That was when he had a call from Digital Equipment Corporation asking if they could design an airflow switch that toggled its output when the flow fell below a threshold. The airflow switch was a big success, and every DEC machine has several to monitor airflow levels to detect fan failure or filter blockage.

I joined them after my MS in Manufacturing Engineering from UMass. With my electronics engineering and product design background, I was asked to help with product design for better production. Surface-mount technology was growing, and that was what I started with. One of the first projects I got given was one from AT&T. Those days AT&T was both a carrier and a telecom equipment manufacturer.

Engineers often undervalue their creations

The problem that AT&T was facing with their 4ESS circuit switcher was intermittent contact failures in board connectors and IC sockets caused by vibration from the fans. Those days the largest DC fan vendors were Comair-Rotron (US), EBM-Papst (German) (they were separate companies), and Nidec (Japanese). Their high-end 48VDC fans had a third wire other than power and ground. This tachometer output generated two pulses per revolution. Even fans from the same manufacturer and production batch ran at slightly different speeds when powered with the same voltage. When there are several fans in a box, the difference in their rotation speeds caused strong Beat vibration. This is similar to what you may hear in a propeller plane when the rotational speeds of the propeller are slightly different. For AT&T, the beat vibration of 0.5 Hz made the whole rack shake every 2 seconds in some cases leading to contact failures. AT&T wanted Aeroflo to find a solution to this. My colleague and a brilliant engineer, Adam Cohen, and I were on this project. We realized that the fans could not be controlled unless the power to each fan is separately changed. Another option was to make the fan that can receive a control signal. We broke open fans, traced the circuits, and created a change such that an external PWM signal could modulate the fan speed without changing the supply voltage to the fan. We tested such modified fans and found that we could control them precisely to the same RPM even if the fans are from different manufacturers. We wrote up the specifications for the new fan and trained fan manufacturers. AT&T was so pleased that the problem was solved. This new fan design became the 4-wire fan that is commonly available in every laptop to large electronic systems. I had specified the wire colors as Blue (control), Yellow (tach), Red (power), and Black (Ground), which is still in use. Now it may be a multi-billion dollar product line. We did not patent the idea. As engineers, we thought if we could think of such a fix, anyone could do it. We underestimated the potential of our creation. Lesson learned!

Even entrepreneurs sometimes undervalue their company

AccuSense was started because Aeroflo faced a problem due to their success. Aeroflo had borrowed some $250k from one of its customers, McLean Engineering, after DEC and other computer companies failed in the early 90s. As a condition, McLean demanded and got the right to buy Aeroflo for some fixed amount (I believe it was $2M) at any time. At that time, Aeroflo had about a month to survive without the injection of new funds. So Raouf had no options but to agree to this term. A few years later, Aeroflo was flying high with some $1.5M cash in the bank. McLean exercised their option and bought Aeroflo for $2M cash, almost a free deal. The company may have been worth three times that. Before McLean could exercise their right, Raouf launched his new company Cambridge AccuSense with sensor products that I had designed after the loan deal with McLean. He argued that they did not know of these products, and at that time, he had the sole right to make any decision. So overnight, we rented a 1500 square feet space next door and relocated our SMD assembly machines, all sensor product inventory, computers, and component stock. I moved with him to start this company with him.

AccuSense released several successful sensor and instrumentation products that I created as the chief designer. Some of them became industry standards that are used by leaders in automotive, military, IT, industrial controls, and biosciences.

In December of 1996, I left AccuSense to start Degree Controls. Seven years later, DegreeC would acquire AccuSense and make that its sensor division.

 

Learnings

Hot Body Anemometry: One core way to measure airflow is to heat a body to a known temperature and measure the heat loss. The heat loss could be affected by the temperature and the velocity of the air. For lower the temperatures the the heat loss would be higher. Also for higher the flow, there are more air molecules hitting the body, hence higher transfer. We can measure the air temperature easily, but how do you measure the power loss? We used a Wheatstone’s bridge to heat up a thermistor (NTC) to a constant resistance and temperature. When the heat loss is higher it takes a higher voltage to maintain the temperature so the power loss is V2/R. This core concept was used in several applications.

 

  1. Have a set of tools, methods, algorithms, and design. Think in terms of the application first and identify what tools you would use from your kit.

  2. If you have several 10mm nuts to deal with you would prefer to use a 10mm spanner, not an adjustable one or a plier. If you understand the problem well, you can design the most optimum solution. The customer will always pay premium price for an exact solution.