Interview with James Downey
May 20, 1995
Los Altos Hills, California

RW: Let's go do an interview.

JD: All right. Thank you.

RW: Well Jim, tell me about your early days in college and before Fairchild?

JD: Well, before Fairchild there was two colleges and a job. I graduated from Santa Clara University in 1962 with a Bachelors of Science in Electrical Engineering. And went on down to the University of Arizona for a Masters Degree. Thought about going on for a PhD for awhile but decided that would take too long. And so in 1964 came back to the to the Valley. First job was at General Electric. In those days GE was still in the computer business and they had an advanced peripheral equipment laboratory they called it in Sunnyvale. And I was put onto an optical character recognition project early on. That's where I met Bob Shriner and Bill David Alan and some other guys. So that was good experience. But in 1966 or maybe late, late '65 GE decided that they couldn't compete in the computer business, even though they had been the first to develop an all solid state commercial computer, the Erma System for the Bank of America. But they, GE had this saying in those days that a GE Manager could manage anything and they brought these guys in from things like the small DC motor division and things like that and they just couldn't, couldn't manage the computer business for sure. And so they finally sold out to Honeywell. They tried to get all of us in the lab to go down to either Phoenix or Oklahoma City as part of Honeywell and nobody went. Harold Bytel went for a few months and came back. And at that time Fairchild was doing very well and I was impressed with Fairchild 'cause I had used their products in some of the discrete amplifiers and that sort of thing that I'd built up. And so they were hiring at, in the, at the R&D Facility in Palo Alto. And so some of us went up that way and that's how I got into Fairchild.

RW: Now who hired you?

JD: Actually I was hired by Bob Seeds, but not to work directly for Seeds. In parallel Shriner went over there and also Harold Bytel when he came back went, went up to R&D. Later on Bob Zyhackian joined and, but that was when Fairchild had the Systems Technology Division. But the three of us were at R&D and initially I worked on some of the circuitry the Uniface Circuitry for the, for what was going to be the 8008 tester. And then I got into the integrated circuit activity and I worked I, I was never sure I was, it was a combination between Shriner and Vedez, Les Vedez.

RW: Well, you also took the Grove's semiconductor course.

JD: Oh yeah right.

RW: I mean that was more advanced than anything that was being taught at universities.

JD: At that time the, what ultimately came to be known as the "Green Book "Grove's treatise on Semiconductor Physics was in loose leaf form. And he was it was a syllabus that he was continuing to change. And I think he was also at that time beginning to use it or he had been using it up at UC Berkeley. But that was sort of like required, you know, I mean you had to go through the Grove course in order to earn your stripes at R&D. And it was good because it was, it was a fresh comprehensive look at the basics of the technology.

RW: Well, then how did you get into ASIC?

JD: Well, I started out, I can remember after doing the interface circuits for, for the 8008, well, the tester that became the 8008. There were there was a, a fixed effusion pattern approach that had been defined by Vedez and Seeds maybe a year or so earlier called FMGA, later on called micromatrix. And there was a bipolar version of the same type of thing where there was, in either case it was a fixed effusion pattern that was customized by two layers of metal toward the end of the wafer fab process. And that's how the custom function would be made.

RW: And that's called the gate array now?

JD: It called the gate array now it was called a, whatever FMGA stood for. Fairchild


JD: MOS... Fairchild MOS Gate Array I guess.

RW: Gate Array, yeah.

JD: And so I can remember Shriner asking me "Do you want to go on the MOS side or the bipolar side?" And I said "Well, what's MOS?" Well, I knew a little bit about it from the devise Physics course, but I said "I'll try that, you know, it sounds like fun". That was kind of a watershed decision. And that's how I got into it.

RW: Well, the Gate Array version never really worked.

JD: No it didn't. The in the MOS Case it was a it was a pre-diffused set of eighty gates and it was customized by two layers of metal. And all of that could be done. What wasn't accomplished at that time was to make stable MOS devices. And so with the two layer metal process they were all hopelessly unstable and there were so many ions going through people hadn't learned how to getter yet. And to make other things to make them stable and so they, basically it never worked. And ultimately I got frustrated at that and said, you know, to Shriner and Vedez "Isn't there something else that I can do?" And we started playing around with this standard cell concept.

RW: And that was the first standard cell.

JD: I think, well there was, there was an effort or there was a description of an effort at NSA I think at that time. They were talking about doing standard cells, but I, I don't know if they had implemented anything. Certainly, you know, ours was the first implementation out here.

RW: And it worked.

JD: And it worked.

RW: It worked well. Can you show your book from 1966.

JD: Sure. This book from 1966 is the Handbook for the micromatrix, which was the pre-diffused pattern and this shows the two layers of interconnect and the bonding pads and you can see the date on here is 1966. I joined in February of '66 and this is in November, October of 1966, sorry. One of the things that we did here, I've got some extra papers in here. One of the things that we did in this technology was to not only describe how we did the circuits and I'll show some of the functions here, which are similar to the building blocks that are made today, if I can find one; like so. Different cells structures and then the metal interconnect pattern that would accomplish that function and so on, different cell structures. But besides having to design, the design approach we also had in those days a bread boarding approach, no longer do people do this sort of thing. But we would actually print out the two layer interconnect system that we had for creating the cells and interconnecting them and then populate a PC Board like this with eighty metal cans of transistors, I'm sorry, of gates and then yellow wire it to plug it into an exerciser to test it.

RW: This was really pre-simulation, logic simulation.

JD: Pre-logic simulation. What we could do though at that time was to prove out the test sequence there was a test...really can't see much here, but this is a flow diagram of how we go about creating a test pattern a sequence of test patterns to exercise that bit of logic. And this would be ready by virtue of the PC Board mock-up or bread board prior to the integrated product coming out. These worked, but unfortunately the two layer metal devices didn't work or at least they didn't work for very long.

RW: But then the standard cell ready did.

JD: The standard cell did work.

RW: It started small and grew to gee, I don't know, a thousand gate complexity and we had hundreds of designs.

JD: Yes we started out with a thin in those days an MOS that was a thick oxide process which had a very thick field oxide to prevent field inversion, but there were lots of processing problems associated with thick oxide MOS and there was another opportunity or another process called a thin oxide which had antimony field stops and that process was running in prototype format in the, in one of the lines in the R&D Fab. In fact, right across from the office that you and I shared Rob. And so I decided to put together some cells, some functions in that process and we made, you remember the term kit parts? We made a bunch of kit parts, which were individual little functions with the antimony field stop process. And that worked and then we, we put together a concept which is shown here for a standard cell gate array really standard cell a ray where each of the cells are different individual functions interconnected in the original technology two layers one diffusion and one metal. This is a paper that I gave I think, it's at the IEEE Session in St. Louis in 1968 if I'm not mistaken. Jim Cofert gave a paper at that same conference.

RW: And today in 1999 it's exactly the same thing except that it's ten thousand times more complex.

JD: Yes.

RW: But it's the same organization, same concept, same ol', same ol. '66 to '99.

JD: Right. Well we quickly out grew the you got tired of the high impedance associated with the diffused interconnect. And I can remember reading an article or paper written by Ken Moyal who was at National at the time extolling the virtues of the, the then improved thick oxide metal gate process and detracting the silicon gate process, which had been invented at Fairchild and which National didn't have. So I decided it would be a good idea if we reproduced this gate array family or this cell family in the silicon gate technology and got the go ahead and do that. And of course the, that was a much better family because we really said we had like two and a half layers of interconnect. We could still use the diffusion a little bit at least inside the cell and then we use the poly silicon and the metal to complete it to an XY coordinate interconnect system. And that worked very well and that's the family that we really took off on and built up not only the circuit group, but also the CAD group that you described.

RW: Well, I remember we did the first credit card verifier for TRW.

JD: I remember that. Yeah.

RW: And I, I remember I went down and I told you "Hey it works, you know, the first time."

JD: Yep.

RW: And you said "You mean it works the first time?"

JD: You mean it really works?

RW: And I said "Well that's the idea isn't it"? And we went into production on that.

JD: Yep. You know, yeah and that was fun.

RW: Now, at some point in time you moved to the out of R&D into operations.

JD: Yeah, well, we moved out of, actually there were two of us one from R&D into the Ellis Street building, the Rusty Bucket, and that was in the summer of 1968 I believe. And then sometime after that, actually I use to complain about not being able to get the stuff built in a timely fashion or with good enough yields and somebody said well in that case then you do it. And so I went from the engineering function to the wafer fab operations functions.

RW: And along the line we had some interesting characters that were running this operation, there was Jack Gates.

JD: Oh, yeah.

RW: Tell us about Jack Gates.

JD: Wild Jack. Jack Gates was had been an operations guy at AMD and he went off to Stewart Warner. He was the GM at Stewart Warner. And I remember him telling me the only thing he made money on at Stewart was harvesting the apricots in the back 40. They were in Sunnyvale at the time. And then he was hired back into Fairchild by, I think Blanchet, Gene Blanchet, to replace John Sentus. And Jack had the responsibility for, well our activity the MOS, the custom MOS, the custom bipolar and I think the, I think he had the responsibility also for the standard bipolar memory activity. At any rate Jack was an extremely outgoing guy who is just typical of the, of Fairchild middle management at that time. He fought everybody. Everybody was the enemy up and down the line.

RW: I remember one time walking down the hall and he had Ray Zoniger, he'd picked him up about a foot off the floor and he had him, he was bang him against the wall.

JD: Yeah, I remember that. Yeah, Zoniger had done something and that, that was Jack's physical approach to things.

RW: Well then he finally, he finally got kicked out.

JD: Yeah.

RW: And then they brought in...

JD: Uncle Roy.

RW: Roy Polick.

JD: Yep.

RW: Another heavy hitter.

JD: Uncle Roy came in from RCA and he was he was a piece of work. He was a he mostly he would chase after the secretaries and play politics in the front office. I don't think he really understood what we were doing. He was an expert at ticking off customers and he did a good job of that. I'm not sure what he brought to the party or why but I guess they...

RW: Chaos.

JD: There was some chaos.

RW: Well you remember the meeting at, at Ricky's? And I was going to be the engineering guy for all of engineering not just custom and I was going to do memories and I was going to have three engineers to do memories. That seemed a little inadequate to me.

JD: Yes.

RW: And I wasn't going to have control of layout. I've never heard of an engineering group where you didn't control your own layout. That was going to be done by the CAD people. And so I turned it down and caused a certain amount of hard feelings there amongst a lot of people.

JD: Well, I ultimately left Polick's organization and was assigned to work with George Scalise on a joint venture activity with TDK. Fairchild and TDK were contemplating a joint venture activity that Leo Dwart somehow had a hand in. And I, I look back, I can remember looking back at that time, there had been twenty one organizations in something like sixteen months since Roy Polick had come to the company...

RW: In his...

JD: his area of responsibility. It was just you, you said it a minute ago, it was chaos.

RW: Yeah. Well he went on back to RCA to destroy, to destroy RCA, which is now no longer, no longer in the business.

JD: No longer in the business. That's right.

RW: And with his laser disk and they took a billion dollar write off on his laser disk business. He's quite a, quite an experienced character. So then at some time you left to go to AMD and become the...

JD: Well, yeah.

RW: MOS guy.

JD: In, in late 1972 it was in the in the fall of '72 that, that I was considering, well I was asked to join the Fairchild TDK joint venture effort. And I spent a few weeks looking at that including spending some time in Japan to consider it. And I, my assessment was it was not a wise thing to, to be with Fairchild particularly if your 6000 miles away. And the, the company had unfortunately gotten so political at that time that, you know, the power bases were changing from day to day. In the meantime, AMD had gotten going. Jerry Sanders and John Kerry and, and Stan Semonson and others had started the company in '69 after Les Hogan had a fired Jerry and some of the other Motorola people had fired Kerry and Semonson and so on. I mean, it was, that take over was a blood, or the post take over was a blood bath. So those guys started AMD, which is bipolar company and mostly a linear bipolar company. But Jerry knew that MOS was the, the technology of the future on, on a cost per function basis. And had tried to bootleg or boot strap an MOS effort within AMD which was going no place. So he and Sven started talking to me about the same time as I was making the change, in fact, Jerry would from time to time, my wife's mother lives in Southern California and Jerry would go down there a lot and occasionally he would meet Kathy on the plane and say what's it going to take to get Jim to come over to AMD? And eventually we got together and I joined in December of 1972.

RW: And what was your duty?

JD: My responsibility was to get them into the MOS business. And so what I did was I started hiring some of the people that I knew from Fairchild and a few others and we had the fortunate situation of having no baggage, we could start over again from on the process and, which we did. We created a new silicon gate P-channel silicon gate process which had, which was clean from the start. We had the advantage, of course, of all the learning process that, that we had gone through and in parallel hired by Sarkision and we started to designing circuits. And we produced something like 25 new circuits in the first year or fifteen months, something like that. And that was in the '73 early '74 time frame that we did that.

RW: And today MOS are actually CMOS is what huge must be a huge percent of AMD Business?

JD: Oh, yeah. MOS grew from that starting point that I described to it, it just grew steadily in percentage of sales and in absolute numbers through the 70's. AMD like the rest of the industry went through a terrible time in 1975 late, actually late '74 early '75 and then it took off. And we had, you know, just really good growth and towards the end of the decade phenomenal growth. We had three years in the late 70's early 80's where we had more than 50 percent per year growth compounded. And MOS during those days was half the company, a little more than half the company and, and then when the CMOS clicked in it, it just, it became the dominant percentage. I couldn't tell you what it is today, but it's, it's a big percentage.

RW: Well, what's Jerry Sanders like to work for?

JD: Well Jerry certainly is interesting to work for. He's very smart. He, he always has an opinion as to where he's going. He does listen, some people think he doesn't listen, he will listen to ideas, but if he's made up his mind he doesn't want to hear, he doesn't want to hear any objections or any counter positions any longer. And he is, he's demanding. He demands that you, that you perform and that you, that you even anticipate where things are going so that you don't get caught by dumb mistakes. And of course everybody knows the flamboyant reputation and he certainly is that and you, to work successfully with Jerry over a long period of time you have to cater to that, which some people find difficult.

RW: The thing that impressed me about him was, in the early days anyway, that he would nurture engineers that were politically unacceptable at Fairchild. Guys that, that dressed funny or, you know, spoke up or whatever and they were very unpopular at Fairchild and the Peter Alfkies...

JD: Yeah.

RW: ...and many many and he understood that engineers were the basis of the company and he nurtured them instead of treating them like shit.
JD: More than that. I can remember in the, in the Thompson Place Building, which is where the whole company was located when I joined, there were, there were signs, you remember Elliot Sopkin. He had, Elliot was the PR guy both inside and out and there were signs throughout the building "The Engineer is King" and little caricatures showing this. And so there was a culture that was created there that made people feel that they were the center of the show and that was important. That's, I had forgotten that, but you're, you're absolutely right.

RW: As opposed to Fairchild where, you know, the front office was Mahogany Row.

JD: The front office was the, was the center of the universe.

RW: I'll never forget, remember Bob Frankavitch?

JD: Yeah I do.

RW: The beard?

JD: Yep.

RW: And he was our most productive mass designer and he was on swing shift. And swing shift was a lot more productive than day shift on doing layout. And one day one of those ubiquitous guys with the three piece suits that you'd see come and go at Fairchild and caught him in the hallway and said, "What are you doing here?" And gave him a bad time and Bob almost left over that. Then there were rumors that he was smoking marijuana on swing shift since he was the most productive designer. I, like General Grant - Lincoln said about General Grant, I said "Find, find out what kind he's smokin and give, let's give it to the rest of the mask designers." Going back to, going back to Fairchild, it's interesting that we just didn't just come up with a, with an organization for laying these out. That we also came out with a full computer aided design package, the first DLSI Tester, the eight thousand that later became the century. A complete methodology that is essentially the same as, as used today. So it was, it was more than Silicon, it was really the whole package.

JD: Yeah it was. We, you know, as part of these same documents where we describe the organization, the conceptual organization of the, of the array we also developed flow charts as to the sequence to be followed to generate the circuit, to, to generate the art work, to generate the test programs, etc. And we really systematized the approach to design not taking away the synthesis that has in creativity that has to take place at a certain point. We also, do you remember we generated check lists so that, and every time you made a new mistake you got a reward. But if you made the old mistake over again you got kicked.

RW: That's right.

JD: And so we built up a body of knowledge that and when we got better and better, got and better and better at making them right the first time.

RW: Well at LSI Logic we extended that and the check list got to be literally thirty, forty pages long. All of whom had been generated by actually making the error, because we were doing ten designs a day at our peak rate. But then we, we actually put it in to software and so the software would be doing a lot of the checking, so, but same concept.

JD: Yeah. Yeah.

RW: We were...

JD: Remember my friend Jack Gates? I, going back to Jack he would say "You make a new mistake you get a gold star in your collar, but if you make an old mistake I'll kick your ass". Good ol' Jack.

RW: Well anyway back to AMD what, when did you leave AMD? And then, and what were you doing then?

JD: Well I, I left AMD in 1987 and at that time I was the Senior VP of Operations and I had been for oh, I think five years or so, five or six years. Senior VP of Operations at AMD at that time meant that I was running all the wafer fabs and all the back end international assembly test operations of which there were four, and all the business units. What we called managing directorates. AMD went along with the rest of the industry, it went flat in 1985 and actually went backwards in sales. We had reached a billion dollar run rate in 19... our fiscal year '94, uh, '84 and then went backwards to about, I don't know, seven hundred fifty million, something like that. And it was a frustrating time. Jerry Sanders had, had previously announced a no layoff policy and if there ever were a layoff he would be the first to go and that sort of thing and so we, we were running R&D at thirty percent of sales and losing money and there was a lot of pressure. The Japanese were, had already decimated the DRAM business and then in '85 did the same thing to the EPROM Business. So we went from having twenty percent of our business, some two hundred million dollars in nonvolatile memory with a twenty five percent or so pretax margin in 1984 become a roughly a hundred million dollar business with a twenty percent pretax loss in 1985 and that was not a lot of fun. I can remember the average, the average backlog price for the 256K E-Prong at that time was, went from something in the range of sixty five dollars in 1984 to something in the range of six to seven dollars in 1985, tough to, tough to keep up with.

RW: Well, the Japanese had specifically targeted those products.

JD: That's right.

RW: And they were losing money as well.

JD: Oh, yeah. But they, it was a war of attrition.

RW: Yeah.

JD: So anyway for a number of reasons in and then, there was a, the PC down turn in 1985, very large downturn. And all that coupled together our business and that of many others went, went flat to down in those years. In 1987 AMD bought MMI, really Jerry did, to, to get the sales back up to the billion dollar rate. And I didn't think that was a very good idea and eventually left the company.

RW: And then you started in a, in a whole new direction with magnetics.

JD: Yeah, I was going to take some time out after I left AMD and Jack Pilletto got a hold of me. He, Jack had been at Hambrick and Quist after his time at VLSI Technology. And Jack suggested that I talk to some people who were struggling with a magnetic component company. And I didn't know anything about magnetics and I wasn't very interested, but they called me and persisted and so I went and took a look. And the name of the company was Read/Write. They were located in Milpitas. A small company losing money, small sales, with, and what they were trying to make, the product was thin film read/write heads, recording heads. And Max Thor was heavily depended upon them and then a few other people to a lesser degree. And there was no consistency in the production. So I talked to a number of people, one led to another, to another, I probably talked to thirty people before making the decision to, to give it a try. And what it seemed to me was that the thin film head process is a wafer fab process followed by a machine, which is similar to semiconductor, followed by a machining process, which is not similar, followed then by an assembly and test process, which of course has its own characteristics, but conceptually is similar. And the wafer fab was just in disarray and I figured that if we brought some of the disciplines that we had developed at AMD into that technology we might have a chance of making it, so that's what I did. I hired some, some of the guys from Austin, you know, Frank Digiswaldo came and joined and we brought in a fellow named Bob Anderson and Paul Schroeder and some other people and pretty quickly got it turned around. And to, to the point where the company went profitable in the fourth quarter of 1988, some three quarters after we got there.

RW: So now you're a magnetics guy?

JD: Yeah, one could say I, you know, I've got a bit of an understanding there.

RW: Well you went off and did some financial kinds of things didn't you? Or...

JD: Yeah, for a little while. After I was at Read/Write for two years and we had this, we the fortune, good fortune at Read/Write of, by the time we were getting our act turned around I also took a gamble and switched from two inch wafers to four inch wafers, which may not seem like a lot, but it's a factor of four and it was in the right direction. We also went from what was called then a standard head to what the magnetics industry called a seventy percent head but what we call, what we call if, if the standard was standard then the next one must be a micro and behind that was coming a nano and behind that was the pico. And so we infused some semiconductor technology into that industry, semiconductor terminology I should say and manufacturing technology. But the, we changed to the seventy percent device, which means that it shrunk it to a .7 of the dimension of the standard in each direction. Well, that means that we got twice as many candidates on a wafer per unit area and we increased the unit area by a factor of four. So we had a lot going for us in terms of increasing the output and we stabilized the process so that one wafer looked like the next coming out the end of the line. I can remember in late '88 we had four or five development programs going on with Connor Peripherals who was then the hottest disk drive guy around. And they told us that we were doing a good job and that we probably would get one, maybe two programs in production in 1989. It turns out we had all four sole source so that was what really propelled the growth. We grew the company to forty million dollars in sales in fiscal '89 ended September and continued to be profitable throughout that period. So it was, it was a lot of fun. That was the turn around was very satisfying.

RW: What does it take to do a turn around?

JD: Well, I think to do a turn around you got to figure out what, why it's not working and then obviously react to, to that problem. And in the case of Read/Write there was good technology there and there were good technologists there. The problem was there was virtually no discipline in the manufacturing area and so, you know, we figured that out. I guess the other thing that it takes to do a turn around is to manage the cash flow. We had, you know, the venture capitalist had put up so much money and there was no more to be had, make it work with that or forget it, and so you, you learn to manage by cash flow as opposed to manage by any other performance parameter.

RW: Well, did you have to lay people off?

JD: We did. Not long after I got there we you know it was clear that the, the company was staffed far beyond its ability to generate revenue and given its cash flow consideration there was really no choice but to do that. So we did that and focused on the one area that could turn the company around, which was to get some consistency coming out of fab. The machining processes were pretty, were reasonably predictable. And so we, we felt that if we could get a consistent flow of material coming out of wafer fab we'd be in good shape and that's the way it turned out.

RW: Anyway then you went on to do financial?

JD: Well yeah. I started to say a few minutes ago we, we've had some good fortune. Not only did we get going with Connor in early '89 and we're ramping up our new four inch wafers with the shrunk devices, but also CDC decided, or the PMI, the subsidiary of CDC, was sold to Sea Gate and Sea Gate announced that they would honor existing contracts but then they were going to take the output of this new subsidiary company to Sea Gate they were going to use it internally. Which meant that our major merchant market competitor was being taken off the field. And what I wanted to do was to, to capitalize on that since it was not, you know, the prospects of getting more money from the current venture group weren't great, lets go to Japan and get a partner and do the typical geographical licensing thing...

RW: Sure.

JD: And so on. And here again my path cross with Jack Pilletto who was out kind of freelancing at the time and Jack worked as a intermediarian and we had Sumitomo, Sumitomo Metal actually set up to, to do a deal and the, the people at H&Q didn't want to do that. We had a parting of the ways and so in early 1990 I was looking back at Read/Write. And that's when I got into the financial activity. I some people looked me up and these people had some interests in various companies out here on the West Coast, one was a health care company, another was a semiconductor equipment company it was a fledgling semiconductor equipment R&D company and some real estate activities. And so I joined up with this ventured group, privately held venture group, and managed some of these activities for them for a few years and that went on for two years I think. It was a good experience, but I got tired of it because I really prefer to have a direct involvement as opposed to an investor involvement.

RW: So then Billetto recruited you to Cartesian?

JD: Not directly I in late 1991 CoMag, who was the leading supplier of thin film media, bought a small thin film head company in South San Jose called DasTech. DasTech had been around even during the days that we were at Read/Write, but DasTech was not a very, they were under financed and they weren't, you know, a major force in the market. When Comag bought DasTech Steve Johnson whose the, the CEO at Comag contacted me in early '92 to join, to explore my joining and taking over the, the operating responsibility down there. And I couldn't do it because I had some commitments to these, this financial activity that I was working with. But later on in the year when those were done and you know it, it appear that there wasn't going to be very much more investment forth coming from these, from these people and furthermore I was getting tired of the role that I was in. Steve was still looking for a solution to his management problem and so he and I got together and in the, let's see I guess it was October time frame of 1992 I joined DasTech as President and CEO. DasTech was a wholly owned, privately held company sixty percent Comag and forty percent Asahi Glass who is a partner of Comags in, in thin film media enterprises both here and in the, in Japan. And, and so we wrote a business plan and it was another turn around, but this time was inside a more of a corporate structure. And that was going well, in 1993 there was a slow down in, in the disk drive industry and, and we were stretched out as were all the head manufacturers. And toward the end of '93 the Comag board decided that they really should focus on, on thin film media and not be involved in heads and so they made a decision to get out of the head business and that happened in 1994 early - first half of '94, so I, by that time Jack and, and some of the others at, what was called linear peripherals at the time...

RW: Right.

JD: ...were contacting me and I started to get interested in that and eventually joined up with them in the August time frame of '94.

RW: Now this product has tremendous potential. I invested in it first and second round and I liked it because it didn't require really any analysis of markets or anything of that sort. If the thing works you make a lot of money. If it doesn't work you loose everything.

JD: That's it.

RW: It doesn't I mean...

JD: It's a binary product.

RW: That's the beautiful part about it, and a tremendous potential. Now you did, you've done two rounds, you're just in the process of the third round...

JD: Right.

RW: And all of a sudden it doesn't work.

JD: Well it's not, it's not quite that bad. What we, what we had planned to have going in the April time frame was the first prototype. And what we found was that the motors and this is a linear actuator with an X and Y, an XY actuation that creates the sufficient, the positioning and the sufficient velocity for the heads, we use magnetic heads with the same, built with the same technology as flying heads except that they're in a different form factor. And we use magnetic media also built with the same technology as the rotating media, but in a different form factor. And we create the motion between the heads and the media by a spring mechanism in an XY coordinate system as opposed to a rotating circular coordinate system. And what we found in this prototype was that the motors were undersized, underpowered and that the forces that they had to drive were not properly taken into account. And in the small form factor of our first product that translates into a reduction in the amount of capacity that we can ultimately deliver. But we've spent several weeks now analyzing that problem and we're, we think we're back on track. We, we have two major corporate investors who want to participate in this current round. Both of them have had technical teams come and take a look and they're satisfied that this is a glitch but not a, not a killer defect and both are interested in pursuing the investment and in fact we're off to the Far East the week after next to see if we can secure that.

RW: But isn't this the, the classic kind of thing that, that we do? I mean otherwise anybody would, could do this sort of thing.

JD: Yeah. It you know, you think you know, you try to figure out where you're going to have problems and as you know you, the places where, that you think about sure enough you don't have problems, it's the one you don't think about that gets you and that's what happened here.

RW: It's, when you, when you go into these things you know there's a chance that, that you're going to fail, don't you?

JD: Sure, there's a, I mean in all of our conversations about this technology and the products that can be built with it, they're all along the lines of what you said earlier, if this works it's going to be home run, maybe a grand slam home run. But there's no guarantee that it will work. There are some, like any new technology this is not a revolutionary step, this is a, sorry, an evolutionary step, it's a revolutionary step and when you do that you, you are plowing new ground and it's tough to predict how fast that your going to be able to get that done and with what degree of success you're going to be able to get it done. We are encountering as expected, problems that people have not had to solve before. And the measure of success will be how what time rate of success we have in solving these problems.

RW: But there's a lot of people aren't, that are unwilling to take the chance on failure. In fact, I think most people don't want to risk that. And...

JD: Yeah I'd say that's true.

RW: It seems to me that we have culture here in the Valley that it's okay. I mean if you give it your best effort and your not crooked it's okay to do it.

JD: Yeah. And interestingly the way we have, as you know, the way we have financed this company so far is with all individual, industry individual investors and they all understand the risk that's involved, they all understand that in developing a new technology you may find some problem that is not solvable in a realistic time frame. And, and that would constitute the enterprise going belly up. And I think both the insiders, the employees and the investors understand that. And, and maybe it is a, it's a cultural thing that people around here accept and probably don't accept in other parts of society.

RW: Well, Japan, if you fail in Japan you got one shot and that's it. You're really a failure.

JD: Yeah. Yeah.

RW: Now it is I, it's exciting having seen this whole structure build up here in the Valley a venture capitalist and entrepreneurs and you just sort of think that everybody's that way and then you travel to another part of the country or something and people, a job is a job...

JD: Yeah.

RW: You know it's eight to five and its, it's a different, it's markedly different culture and, and we, when I was doing my book "Silicon Density" all the people that I interviewed were quite successful and I've only known them as being quite successful. And when I got their stories they had all failed, they'd been fired, laid off, whatever, I was really surprised. And that was consistent except for the Japanese people who hadn't. So it's exciting.

RW: Well anyway thanks, Jim.

JD: Okay. Well, thank you Rob I enjoyed the opportunity to, it's been kind of fun, you know, thinking about the, the micromosaic and even back to micromatrix the very early days spanning back nearly thirty years.

RW: Well you made me a lot of money Jim, I can say that.

JD: Well, I'm glad to hear it.