Interview with Bob Dobkin and Jim Williams
April 19, 2006
Milpitas, CA

RW: Bob Dobkin and Jim Williams are world famous analog IC designers. Bob has over thirty years experience in analog design and developed many advanced products at National Semiconductor and Linear Technology, including the first three terminal adjustable voltage regulator. In 1981, he co-founded Linear Technology and at the time of this 2006 interview, was VP of Engineering and Chief Technology Officer. Jim Williams was recruited from MIT to National by Dobkin. He moved to Linear and at the time of this interview is Staff Scientist there. Jim is well known for his lectures and writing on analog circuits and instrumentation.

RW: Bob, tell us about your family and growing up.

BD: I grew up back in the East Coast in Philadelphia and I started playing with electronics when I was about six. So it goes back well before learning anything in school. And I think that most of the people who have good, intuitive feel for analog circuits started well before they got into school as well, building things, doing projects, and then for some of us, it ended up being a profession.

RW: Well what were your parents like or your siblings?

BD: Nobody else in my family had any inkling to do any electronics. My parents were business people and my mother was a housewife. My brother, one brother's a doctor. The other brother's in advertising and they don't know which end of a screwdriver to pick up, much less a soldering iron. So I picked it up all on my own.

RW: So where'd you go to college?

BD: I went to MIT and never graduated. Went out and did business after being in MIT and I think I learned most of what I know on my own.

RW: Well Jim, how about you?

JW: I grew up in the Midwest in Detroit . I had three sisters. My father was a banker. My mother was a housewife. When I was five years old, my father brought home a shortwave radio. And that was, for me, a seminal event because for the life of me, I could not understand how all those voices could come from all those distant places out of that little box. I still can't. I don't know what gave my father the wisdom to let me take apart his new shortwave radio and it would make a good story to say that I got it back together but I didn't. It was a mess and it stayed a mess. But from that moment on, there was no question in my life what I was going to do with my time.

RW: So did you go to college then?

JW: I went to college for a year and a half, dropped out, ended up at MIT, got a teaching appointment, was there for ten years teaching and doing research on electronic circuits, particularly analog circuits applied to biomedical and biochemical systems. And I was a lifer. I was going to stay at MIT forever. But some people, Dobkin among them, convinced me to come out and try California for a year or so. And it's been a very long year since 1978.

RW: How could you teach at MIT if you weren't degreed?

JW: That's a long story but basically when I got there, I talked to numbers of people over an extended period of time, over six months. And finally one of them, a very well known physicist, got me an appointment.

RW: That's incredible. I wonder if that's ever happened before.

JW: It only happened to me once and that was all that was necessary.

BD: But very frequently now, we lecture at different colleges as guest lecturers. I have a story about taking things apart. I was nine. It was 1952. We just got our new nineteen inch black and white TV set which costs as much as a car. And it's in the living room. And I got my screwdriver and I took it apart. And my mom walks into the living room and I didn't just open the back. I had all the chasses out on the floor. I had all the tubes out of it and I'm looking at every different piece. And she just looked in and left. Didn't say anything. And I made diagrams when I took it apart. I put it back together again and it worked. And I didn't think anything about it till fifteen years later when I heard her telling the story to some friends--- “And I walked into the living room and it was all apart on the floor. What was I going to tell my husband?”

JW: That's why he's my boss. He put his back together and it worked. Mine never worked again.

RW: So what were your first jobs?

JW: My first job was in a television shop. I was too young to work under Michigan law but the arrangement that Mr. X and I had was if I found any money on the floor after I was done working on Saturday and Sunday, I could keep it. And I started working for him when I was twelve or thirteen.

BD: I was about the same age. And I had a little kit. I used to go around fix people's TV sets and radios. And then…

RW: Well so Bob, you somehow got to Silicon Valley , how did that happen?

BD: I was working at GTE Defense Systems at the time. And Widlar had started at National and was doing the best analog circuits at the time. I had written him some letters and he had asked if I wanted to come out and talk to him. And at the time I had also written some letters to a company up in Boston called Philbrick-Nexus. They made op amps as well. And these op amps were little modules. And they really hadn't gotten into IC ' s. And I'd written them some letters as well. And I got a job offer to move up there and make some IC op amps, be their chief integrated circuit design engineer. And that's where I met Bob Pease. At the time, I decided I'd probably be better off going to Philbrick. And started off there. When I got to Philbrick, they were part of Teledyne. There were all kinds of corporate problems between one section and another making the products. I found that you couldn't get anything done and I called up Bob Widlar and I said, “You ready for me to come out?” And he said, “Come out.” So I got packed up and I moved west to work for Bob Widlar at National. And that's how I got out west.

RW: It's amazing isn't it, how these East Coast companies had screwed up semiconductors? None of them had been successful.

BD: They started, the East Coast companies started. There were East Coast companies like Transitron. They they made some IC's. They made transistors. They made diodes and they went down the tubes. Tyco Labs, there's a whole bunch of them.

RW: Philco.

BD: Yeah Philco. Philco had a very nice technology at the time. I went on a tour through there. And they all didn't know how to be in the semiconductor business.

RW: Now how did you meet Jim or decide to recruit Jim?

BD: It was real easy to decide to recruit him. All I had to do was meet him. But I don't remember how I met him.

JW: I sent you a thermometer in the mail.

BD: Yeah.

JW: You did the LM135 temperature sensor and I liked it so I sent you a long thermometer in the mail which broke on the way.

BD: And then Jim came out to visit us at National.

JW: Yeah, I came out to look it over. And I liked what I saw.

RW: So you left MIT for National?

JW: Yeah. I didn't think I'd be out any more than a couple of years. But one thing led to another. And then Linear got started. And for a geek like myself, joining a startup like Linear is akin to joining the Marines. It's gung-ho. I figured that's at least five years. So I cut ties back East.

RW: Well going back to Bob Widlar. Now he's quite a character, right?

BD: Yes.

RW: Tell me about him.

BD: Well Bob was one of the few people I considered to be a genius. He was also paranoid, very hard to get along with and drank incessantly. My interview with him was before dinner drinks, two bottles of wine with dinner and after dinner drinks. And I was still standing so I was hired. Course, I went home and threw up a lot! And he was very egotistical. He didn't think anybody could invent things, at least back in the old days. I think he changed his mind after a while. But he was very good at what he did, made sure that things worked very, very properly. He promoted them properly. I learned a lot in terms of how to be in the analog semiconductor business from Bob.

RW: Well he wrote his own app notes, is that right?

BD: We all did our own app notes. I wrote my own app notes. And he wrote his own app notes. He wrote his own data sheets. He set up the format for them. He gave his own lectures out to the customers. And, at times, he answered the customer phone calls. And that's where we learned about a technique which we call design for minimum phone calls because you make a million IC's; you get half a million phone calls if they don't work.

RW: Well, Jim did—did you know Widlar?

JW: I worked with him here at Linear. And he was very much a character. One night he and I were working late in the lab and he noticed some funny stuff going on on his breadboard and we finally tracked it to RF emissions from the San Jose Airport . And I came up with some idea of screening the circuit which would take some number of hours to really do well. But Widlar was already on the phone trying to get a phone number for the tower at San Jose Airport . And he finally got some number that got close to the tower but not quite there and very coolly requested that they shut down the transmitter for a half an hour so he could finish his breadboarding which they said they couldn't do and he hung up and I thought the FBI would be on us in fifteen minutes but it didn't happen. But that was Widlar.

RW: But in a way he—he set the stage then to what you people do today. You know, more involvement with the customer, trying to do a lot of it yourself.

BD: He set the stage for probably the group at National at the time and the group at Linear because we followed in those areas. But other companies have other techniques where there's a barrier between the engineer and the customer. And the marketing people see the customer and then any information gets thrown over a solid brick wall and the engineer catches it and makes what he will out of it. And the communication goes from the engineer to the applications engineer who sometimes talks to the customer. But it's not the same as we're doing here or even as we did at National. The chains of command are longer and I don't think it works as well.

RW: Well now you guys had good jobs at National. What inspired you to break out of that and start Linear?

BD: Too comfortable. National was really getting big. National had the idea that they wanted to be one of the biggest, if not the biggest, IC maker. And the fabs—a lot of pieces that don't necessarily work well when they're homogenized. And contributor type of development. It takes smart people to do it. It takes processes that are tuned for analog circuits and it takes just handling the analog circuits the right way to get out a product that's a good analog circuit. It was getting harder and harder to do that at National. The chains of command were getting longer. The bureaucracy was getting bigger and we ended up talking over the partitions—wouldn't it be nice if we could do this, if we didn't have to use all the profits to fund microprocessors. And finally we went out and did something.

RW: And so you left good jobs…

BD: Left good jobs with no promise of a company, with our reputations behind us and we got funded pretty quickly. And I think it was within six or eight weeks, we had Linear Technology out of it.

RW: So you had to go pick a place right?

BD: We had to pick a place.

RW: Build it?

BD: Well we had temporary quarters where we started. We picked the lot. We thought right from the beginning, we have to have our own fab. And so we had to build our own fab. So we went out, we picked the lot, built a building, put a fab in it. In a year, we had the fab built and our first wafers out.

RW: Now you—you talked about processes. You used both bipolar and MOS, right?

BD: Yes.

RW: How do you make the distinction of which to use?

BD: Whatever does the job best. We have a function we want to do, we have something that we need to do, we make the distinction, we want to do it the best way possible. It would be nice if we could make a product and we don't have to make another one for ten years. So we pick the best way to do it and that's the way we go do it. And we do have products that last for ten years.

RW: Do you ever go outside today like for MOS or CMOS?

BD: Yeah, we do—right now, the very tiny line width products are too expensive to set up here. Unless you're doing a lot of digital, high density logic, it's not worth it to set up very small submicron fabs. So anything smaller than a half a micron, we go outside to a foundry. And that limits us of course to the circuits that we can do in that process but it's adequate for what we're doing right now.

RW: Well and the equipment that you use is not state-of-the-art now—the equipment you use in your own fabs.

BD: Our fabs are not state-of-the-art, not the most expensive equipment because the leading edge for analog doesn't necessarily mean tiny line widths. Digital functionality increases with the number of transistors you put on the chip. You put more and more transistors on it; it's a more functional circuit. So if you make those transistors smaller, you can get more transistors onto the chip. That's why they're going to smaller and smaller line widths. Analog circuits deal with real world values. Digital is handling information. It doesn't matter whether it's a one or a zero at one volt or ten volts, the digital circuits always just deal with information. Analog circuits deal with volts, currents, amperes, temperature, light, and since they deal with real world values, the quality that they handle those signals is more important. And that's not necessarily defined by the line width. It's defined by precision, by noise, by how accurate the circuit is. So state-of-the-art analog is different than state-of-the-art digital.

RW: And also much less expensive is the fab?

BD: It's less expensive with the fab.

RW: Yeah. By probably an order of magnitude?

BD: Yeah, that's probably true.

RW: Well Jim, how are the engineers treated here Linear?

JW: I know at least one is happy. In general, I think people are happy and they're happy for a couple of reasons. One is because it's an individual contributor thing which is one part or one project or one app note or one whatever it is per engineer, it's your thing from beginning to end, to integrating it into the customer's board ultimately. And I think anybody who's attracted to making things and becoming an engineer will find that a very attractive environment. It's a lot of responsibility but, at the same time, it's a lot of fun. You have ownership. There's much to be said for teamwork and many of the things around here do require teamwork but there's also much to be said for something that has ownership associated with it. That's your part. You conceive it. You develop it. You design it. You finish it and you work with the customer to get it to work in their system. That's rare in today's world.

RW: I was thinking you guys are really, in a way, in the ASIC business, the ASIC analog business.

BD: No, no we're not. I just want to mention that we have more people that retire out of Linear than leave Linear to go to other companies. And I think, you know, what Jim said, engineers get a big kick out of seeing some customer buy a half a million of their widget. And we don't make ASICs. We try to make the majority of our products are general purpose products that'll have a very broad base. And by having products that have a broad base, they have a long lifetime, they're relatively stable in terms of income to the company. They don't have big highs and lows because they're relatively well spread out. And that's opposed to an ASIC which has very specific applications. And if that application goes away, you stop selling it.

JW: Design time is too rare and too valuable to roll the dice on one product for one customer because if you strike out, you've got nothing. I mean, I've seen ASIC requests come through here that look like great things to do technically, wonderful things to do technically, but there's no way we would touch them because the risk factor is just too high. You want to build something that's hard for your competition to make but that appeals to a broad spectrum of customers over a protracted period of time. That's the goal.

RW: Well, what's your test strategy with—you've got a lot of products, right, and they're being—they're being developed. Do you have some automation to generate the test programs?

BD: No, it's all done by smart people. Test engineers are another scarce commodity. And knowing what to test and how to test it to get the kinds of outgoing quality levels that we have is part of making analog circuits and it doesn't happen easily. It doesn't happen with a lot of thinking about it and a lot of custom test equipment.

RW: So do you build a tester per product or a jig or…

BD: We have standard testers that are available that we purchase. And we have a custom interface board that customizes the tester to each of the products. And, of course, each product has its own test program.

RW: Well how about computer aided design? What sorts of computer aids do you use?

BD: The biggest one is SPICE which is a circuit simulator. So SPICE we started with approximately twenty years ago. Prior to that SPICE was run on bigger mainframes. There weren't any small computers that could run SPICE. And now almost all the design is done in SPICE because the simulator does a better job of figuring out what the circuit's doing than trying to make it with parts or breadboard it. And there's just too many parts to do a good breadboard of, a lot of IC's.

RW: Okay. Well you've talked about the importance of engineers. I understand you have them distributed around the world.

BD: We didn't distribute them.

JW: They distributed them.

BD: We need grow the company, we need good engineers. And not all of them can afford to live in California and not all of them want to move away from their families. So we have to set up design centers where the engineers are and attract them into the design centers. And we've been doing that for fifteen years at least. And we've put in many design centers and each one grows. We buy a piece of land. We build our own building there and the design center continues to grow. And the design center has IC design as well as test and some applications work.

RW: And many are in New England ?

BD: Many are; New England, all around the U.S. We have one in Singapore and we've just opened one in Germany now.

RW: Well you talked about retirement, that you have more folks retire than quit. Now how come you guys haven't retired? I mean, you must be well off by now.

BD: This is fun.

JW: When I get up in the morning, I want to get to work. That's what I want to do. There's two fears I have in life. One is sickness and the other is retirement in that order. It's still a real good time.

BD: Everybody I see—most of them who retire, they get old really fast. And doing analog IC's is like a hobby to a lot of people. It's a creative outlet. It's like painting on the canvas but the canvas is silicon. I'm going to quote Jim about when he met one of his wives. He was talking about their first date. Said we were out for a couple hours and I hadn't thought of one circuit.

JW: That's something.

RW: Well okay, we've talked about the positives. What are the negatives of Silicon Valley ?

JW: The biggest negative is, in terms in terms of recruiting, is the expense of living here. That's what necessitated—two things necessitated design centers. One was that and the other was good analog people are where you find them and if they don't want to move, you set them up there. But the biggest negative for me is I often run into people I'd like to see come to Linear and I can't get them to do it because the cost of living here. Other than that, you can't get a good corned beef sandwich in Silicon Valley . I mean, I would like to see Carnegie Deli open here. For a bunch of East Coast ex-pats like Dobby and myself, we'd pay a pretty penny for good deli.

BD: Wasn't so long ago, I finally found a place that makes good Philly Cheese Steaks.

RW: As we're winding up this interview, do you guys have anything else you want to put on the record here to go to Stanford?

BD: We're just getting started. I'm not ready to wind it up yet, are you?

RW: Okay. Why not take over the meeting now.

JW: I think if you want to understand the analog business from either a technology or a business standpoint, you have to realize that for various reasons, there's a dearth of analog design capability out there today, not because people can't do it or won't do it, but because it hasn't been taught. Beginning in the seventies with the rise of the microprocessor, analog was declared dead out of a number of quarters. And as a result, university level analog courses dropped dramatically, not to zero but to a number very close to zero. That provides a tremendous business opportunity for a company like Linear. It's not enough to design the best part in a given category. That won't get you the sale. You're going to have to show your customer how to use that part in the general sense and also in a very specific sense in their particular application. One of the ways we justify our price multiples is we don't just drop off a box full of parts at the customer and send them a bill. Because there's a dearth of analog design capability, whether it's at the monolithic level or at the board level, you're going to have to design your customers' circuit for them more often than not. They're either busy doing other things or they don't have the analog capability to design your product in. For whatever reason, you're going to have to finish the job. You're going to have to take your part which ostensibly you know how it works and design it into their circuit to work the way they want it to work in their system. That's a big part of this company's capability. That's the area I've chosen to concentrate in—helping the customer to help yourself. That's a large part of what this company has to do to sell its product, to counter this problem that there just is not a lot of analog design expertise. And as I've said, it's not because people aren't capable of doing it. There's no patent on intellectual capability. It's just it hasn't been taught for a long time.

BD: It takes a long time for somebody to become an analog engineer. I kind of think about the process of becoming an analog engineer like learning a language. There's all these pieces of analog circuits that you get familiar with and you know how they work and that's like the words. And when you first see a circuit if you don't have any experience, you can compute what each of the transistors is doing and what each of the currents are doing and finally figure out how it's working. And that's kind of like translating a page by having a dictionary and looking up each word in a dictionary until you finally figure out what that page is says. But once you have familiarity with the circuit pieces and how they interact, you get a intuitive feel for where a circuit goes. And it's only until you get that intuitive feel you'll be able to start writing in that language or designing. You can take a big, complicated schematic and you can have an analog designer's experience and he can look at it for minute, two minutes and know what it does because he knows all the pieces and puts it together in his head. Agree with that Jim?

JW: Yeah. An experienced designer can very quickly ascertain not only what something does but if it does it well. It's very, very easy to spot clumsy design.

RW: Well why isn't analog taught in colleges?

JW: It's not—go ahead.

BD: Go ahead. I think it's taught but I think most of the students think that digital is more whizzy. You can do something on a computer and you have your answer right away. You do an analog circuit, you got to wait six to eight weeks till you see whether it worked or not. Couple years ago, I gave a lecture at MIT. And the kids—very few were interested in analog. They all wanted to do an internet company. About four years later, they all wanted to be in analog because it was nice and stable.

JW: It's also expensive to teach. I mean, to verify that an analog circuit works requires laboratory capability. I mean, one of the reasons that digital courses are so popular with universities is they're cheap. Computer based courses are cheap. Laboratory based courses are messy. They require lots of equipment. They require space. They require support personnel and they're expensive. And I think one of the reasons the university system has gravitated to things digital is they can almost entirely be handled with computers. And that's cheap. To teach analog design at the bench based level is slow and very expensive and best done in very small groups. That's not a good recipe for a university that's minding its budget. That's tough. I mean, we take university level graduates here at all levels, Bachelor's, Master's and PhD and we still have two to four years of mentored like training to go into them before they can be genuinely productive. That's just the way it has to be. We do some mercenary work. We go out and we lecture at universities and try and get kids turned on to analog circuitry but, in general, that's a very hard thing to do. The yields are very low. I mean, if you lecture five or six times a year at various universities, you're going to pick up—and those lectures encompass five hundred people total, say, if you're lucky—you're going to pick up one or two kids. That's it.

RW: Well how about off-shoring? How about India ? How about China ? Taiwan ? That's where so much of the digital work is being done now.

BD: I have not seen any really good analog expertise in the Far East . I've seen some in Europe and in Russia but not in India , Taiwan or China .

RW: Or Japan?

BD: Japan has some but it ends up being pretty specialized for the consumer market. They have some good people. And they're making circuits for the consumer area and, as such, they're really not well distributed. So it's a little hard to see what's going on but there is analog circuit design there.

JW: I think the Far East is, like I said, there's no patent on intellectual capability. I think the Far East is preoccupied with digital stuff and Internet stuff and software because that's where the major market is. That's where eighty percent of the action is.

RW: Yeah, but not—not eighty percent of the profits. This is profitless; prosperity in digital.

JW: Well you don't see us starting an Internet division here. We're aware of that.

BD: But the profits don't come without everything being done properly. The circuits have to work when they get plugged in. You have to have deliveries that are on time. You have to have quality that is the world class in terms of parts per million and failure rate. You have to have reliability which is world class. Actually our failure rate is about ten times better than most of the major semiconductor manufacturers. You have to have application support. And you have to have the ability to go in and fix the customer's problem if he has it. And that's part of what we offer.

JW: It's a big deal. You may spend a lot of time fixing a customer problem that has nothing to do with your part. It's elsewhere in the circuit. But it's not a blame game. If you want your product designed in, you got to make the customer's board work.

BD: And the analog designers at the board level, these guys are in short supply as well. And they have long memories. And if they drive your part and it didn't work, they don't try it again.

JW: Independent of whether the problem is with your part.

BD: So we work very hard to keep the customers happy. And they're very important to us. And there's no time that we just ignore customer problems. I mean, we run through the line. If a customer's line is shut down, we've run hand carry runs through the line to get products out in a very short time period because our customer had a problem or we had a problem in the wafers that came out and couldn't ship the proper product in time. Doesn't happen often but it's part of our commitment. Back when semiconductor supplies were over-bought and all the semiconductor companies were booking everything they could, we stopped booking because we wanted to make sure that what we did book, we delivered when we said we would.

RW: Well Intel has dropped, "Intel Delivers.” So maybe you guys could pick that up. Linear delivers. Well can we—could we go for a tour of your lab?

BD: Sure.

RW: All right.

BD Okay, well I can show you how we can kind of verify an application.

So here's a circuit. We have the IC here and these are the external components set up around it. So what we could do is do a simulation here and pick out what we want to look at - and it's simulating right now in real time. So the computer's load step's going to happen right there. So this is the output voltage. You can see that when it gets hit with a load, pulls the output down a bit and circuit responds. And then the other edge

The components are so small that putting any kind of real measurement device on them would likely affect the measurement that we're trying to read. So we get more accurate results actually simulating IC's than we do by building them and trying them out.

RW: Well a scope probe is going to add capacity.

BD And slow things down and make it so it doesn't work right. So SPICE which runs on the PC just like that is how we do the designs. Actually the designs are done on paper and we verify them on the PC because we do it on paper, then after the design is done, then we put it into the PC to see if it runs right and do the questions as necessary.

And that way you can run it over temperature for example. So then it's real easy to do on the PC. And so you make sure that it runs over the full process spread and temperature spread.

RW: Yeah, and digital circuits, the standard load, all speeds were specified with a standard load of fifteen picofarads. And that's because that was what the scope probe was in those days.

BD: After the circuit is designed, it actually is laid out into the chip where we make the layers and the transistors. We convert them from a schematic into a real device. And let's go annoy somebody by looking over their shoulder. Okay. Allison.

Allison: Yeah. Oh. Hello. What are we doing?

BD: We're being invaded. We want to show how we turn a circuit into a layout. And that again is done on computers. And Allison will start with a schematic. Could you just hold up a schematic? And that got the transistors in symbol form. And Allison will take that and she'll turn it into real transistors that'll be the photographic mask for the IC.

Allison: Which is something like this? Is that okay?

BD: She'll actually draw them on there and maybe you could just blow up and show one transistor.

Allison: Wow that's a transistor. Here we go.

BD: Just find something small.- And each of those lines are on different layers or different mask sets. And we will take that. It'll get drawn. It'll get turned into glass plates and the glass plates will be used to make the actual silicon. And on a circuit of this complexity, what's this, about two months to lay out?

Allison: Maybe a little bit less but yeah about that.

BD: Okay. And it gets checked by the engineer very frequently. Okay. Then we'll go into the lab because after we've got the mask made, we make silicon wafers. And that's where we go from there. Take this lab. It's probably better. Go down through here. This is what a finished layout looks like. The outside edge is the pads. This is the circuitry in the center. This is an MOS circuit and it looks like a DAC to me, Digital to Analog Converter. After we get the wafers made, we have to check out to see if the IC works on the wafer. And these stations here are fixtures with very small probes where we can go down onto the wafer and we can actually measure points inside the—the die with wires as small as a few microns. Smaller than a hair. And we're able to go in, measure the points in the circuit and verify that different parts are working. I don't think we can get a picture of what the actual device looks like inside.

RW: But you can go into the interior without using a bonding pad?

BD: We can go into the interior. Most of our circuits only have one layer of metal so we can actually go into the interior at most two layers. And we go through that bottom layer and actually measure that. Why don't we come over here and you can take a look at the—here's a test die and a test fixture. And these are the probes that can go into the center of the wafer. And after we figure out whether it's working or not and when we get the silicon finally fixed, it ends up with Jim who does the applications work and finds out where the real problems are.

JW: I can show you the applications area. There are several. The most basic activity is to take new product and put it through its paces on a breadboard at the bench, to play like you're the customer and beat the product up every way you can think of that it'll get beaten up in the real world to find holes in the product. Another level of activity is the development of circuits using products that you think might be useful to customers in the field. And that's kind of a roll of the dice. You give your best guess at what you think people are going to do with this thing.

A third level of activity is taking products that are already released in the field, in the catalogue, and coming up with circuits that you think will be useful to customers or tangentially useful to customers, spark some idea. Finally we spent a lot of time developing methods and techniques to verify that our circuits are working. And these methods and techniques should be reproducible by customers so that we can both be on the same page in terms of agreeing that a circuit's doing something that it's supposed to be doing or not doing, something that it isn't supposed to be doing. And I archive breadboards for about five to six years before I clean off the whole bench and start all over again.

BD: And I've seen Jim fire up a five year old breadboard, get it working again.

JW: It's easier—it's easier to take one of these things that's bent up a little bit and put it back in shape and get it running than build it all over again when somebody calls up with a question.

RW: And so Charlie Sporck has not seen this desk?

JW: No, Charlie hasn't seen this desk.

RW: We could call…

JW: Saved me the expense of a smoke detector.

RW: I notice you have a Maxim data sheet. That's one of your competitors.

JW: We keep any eye on the competition. And I assume they keep their eye on us.

RW: Now you own stock in Tektronix, is that right?

JW: No.

RW: Huh? You must own the stock in Tektronix.

JW: Not that I know of.

RW: But you've got millions of dollars worth of Tektronix equipment. You're not… on…

JW: It was millions of dollars once upon a time but the way we buy it, second hand, it's cheap.

RW: All right. Jim, great pleasure. Bob, wonderful.

BD: A lot of this equipment is actually owned by the engineers. They buy it on EBay.

JW: Upwards of half the equipment in this laboratory is privately held. We encourage that.

RW: That's fabulous.

JW: Not just for economic benefit to the company either but because the engineer has to find it, fix it and maintain it.

RW: They take care of it.

JW: They take care of it.