JB: This past January, we ran a Cover Feature on the MAC series mixers, a line of hermetically sealed mixers based on low-temperature-cofired-ceramic (LTCC) technology. How have they done so far commercially?

HK: The MAC mixers are intended for military applications and anywhere high reliability is important, such as in test equipment. In fact, one of our customers who manufactures test equipment had equipment failures. It turned out the equipment was used in a high-humidity industrial environment. We switched over to a hermetic ceramic package and the failures went away. That is just one example of where hermeticity made a difference in reliability.

Whenever you introduce a new product, the customers go through an evaluation phase. Customers will buy a small quantity for evaluating performance. We have been fortunate to see a good response from many of our customers. They are recognizing that the reliability, performance, and very attractive price of the MAC series provide real value to them. The other big deal about these new mixers is the operating temperature range and size. These are tiny rugged mixers only 0.06 inches high. We have operated them without failures at temperatures up to +125C for 1000 hours.

JB: So, it is normal with a new product to take some time to gain traction in the industry?

HK: It has to take timethis is our experience. But these low-cost hermetic mixers are going to be around for a long time. In my opinion, this is a "right on" product that meets both existing and emerging needs.

JB: Some people have an impression of Mini-Circuits as simply a "components company." But obviously, the company is changing over time. The power sensors, the moves into test productswhat got you started in these different directions?

HK: Mini-Circuits has been in this business for a long time, and we have built up different areas of expertise. One of those areas is in testing, since we have to test our own products. So we developed our own robotics systems, our own machines for automated testing, etc. And we've learned quite a bit about components and how to test them.

We also have a strong software team. The question was: How do we use our resources to further enhance our ability to do business? So you go where it is natural to go, where you have the talent and the resources, and you try to put everything you have together. Portable test equipment seemed like a very logical step for us. And I believe the industry is ready for change, especially in test and measurement.

We looked at available equipment for test and measurement in the laboratory and for production. Laboratory test equipment is well served by many leading manufacturers of equipment. But what about test equipment for production? Many different test functions for equipment used in the laboratory may not be necessary for production testing. If you don't need all this capability for production testing, why not provide a solution that provides value to the customer? We know the business because we are testing our own components. So we understand where there is and isn't value in equipment functions used for production testing.

For example, we developed a power sensor that when used with a personal computer (PC) turns into a smart power meter. Why do we call it a "smart" power meter? If you look at a traditional power meter, it is simply a power meter and you use it to measure power. I consider that the "dumb" part. But a smart power meter should be able to do a lot more.

Sometimes you are in a noisy environment, where the value of a measurement can change from one instant to the next. So we take many measurements and average them, and we get pretty good results. You can take as many as 100 measurements and get an average. So now the accuracy will improve. Secondly, we use power meters for burn-in and life testing. Wouldn't it be great if you could continuously monitor measurements over the entire test period and review the data every 15 minutes automatically to look for variations? Our smart power meter software can easily produce a plot showing variations in power over time.

The third is to allow users to remotely monitor a circuit or system and, if a problem is detected, it can send a warning signal over the Internet. These are new features that make the power meter smart. Sometimes when you are measuring the gain of an amplifier, and you have attenuators in the transmission path, you would like to have the attenuation contribution subtracted from the measurements. The power meter provides an offset function that allows the user to zero out the attenuation and measure gain directly.

JB: Mini-Circuits has always offered tremendous quality for the price, and this is apparently part of the set of guiding principles for this company. Can you describe briefly those guiding principles?

HK: Our guiding principles come from seeing things as our customers see them. When I buy something, I expect it to work for its intended use and for as long as I need it. Some people might say that kind of thinking is like Pollyanna, but I don't believe that. You have to be smart when you take this approach. Every manufacturer will tell you that they have products that have good quality, but what are you really doing to make it happen?

So one of the things we learned very early, starting with the Motorola Six Sigma quality program, is that variation is a key factor. So in everything we do, we design it to reduce variation. For example, we previously used toroidal cores to manufacture transformers. But, because the shape is circular, there is no reference point for placement of the wire windings. We switched over to rectangular or square transformer cores, so we were able to define the wire positions during winding. It is a question of designing to reduce variation.

We talk about Skinny Sigma and Fat Sigma. We want Skinny Sigma. So all of our testingeverything we dowe compare to previous production runs, we look at variation, and we use this as a monitor to see that the present run is within well-controlled limits. But we make a big effort to reduce variation in all our processes. We went to component attachment by means of welding rather than soldering because it results in greater precision, reducing variation from different solder heights. Some may say that striving for this greater precision must be more expensive. The truth is, it is less expensive.

We also pay great attention to materials and components at incoming inspection, with the goal of minimizing variation. Even with our marking methods, we strive for consistency. We used to use engraving or ink stamping. Now we use laser marking because it is very legible, it's fast, and you can't rub it off. We continue to invest in order to reduce variation and to establish process monitors in order to be able to measure our quality. And I have to tell you, it pays off big-time.

If you think of the number of units that pass through our manufacturing, we are talking between 50 and 100 million pieces every year. What kind of rejects can you expect to get? If we have 1% rejects, we are out of business. Even one-tenth of a percentwe couldn't handle it. We work with defects measured in parts per million. Whenever there is a variation issue, we put engineering, quality assurance, and the whole team to find the cause and solution, because we don't want rejects at all.

Consider semiconductor manufacturing and qualification. Everybody accepts 1000 hours for life testing. We perform 5000 hours for life testing on semiconductors. Why? Because we saw that sometimes after 1500 hours or 2000 hours, you can get rejects. So, do you ask the customer, "Would you be happy if the part lasts 1000 hours and then we don't know what's going to happen?" The customer doesn't want to have a problem at all. We try to look at reliability from the perspective that we are not to give a customer a headache. And the only way to make it happen is to do a tremendous amount of testing, provide quality control of our processes, and continuously make improvements to our processes and material control to reduce Sigma by getting smarter and tighter.

Even if there isn't a problem, if we see a way to tighten up a process, we will tighten it up. My philosophy is that if we test 100 units and they all pass, but one is marginal, I would rather throw away that one and sell the 99. Because the question that will always come up is, "Why was it a little bit different than the rest?" And most of the time, you don't know. And if you don't know, there is an uncertainty. And if you eliminate uncertainty, then you get better quality.

Editor's Note: To read MWRF's interview with Harvey Kaylie in its entirety, be sure to check out the online version of this article at www.mwrf.com.