Third member of Apple’s IC Technology Group located in the Bandley 3 building along with the Mac 1 team
Designed a full custom infrared mouse chip, which was likely the first COT (Customer Owned Tooling) silicon ever developed by Apple. This chip was developed with Bob Bailey – a fellow member of the IC Technology group.
Designed the Apple Real-time Clock and Parameter RAM chip which continued to operate and retain data using the back-up battery on the motherboard when the Mac was turned off or unplugged. This chip was also developed with Bob Bailey. The RTC chip was a full custom design and shipped in high volume in a variety of Macintosh computers
Developed three generations of the Apple Sound Chip (ASC), including a version that included an integrated audio DAC. The ASC shipped in high volume in a variety of Macintosh computers. The Sound Chip was the first chip developed by Apple to use Standard Cell ASIC methodology.
Assisted Bob Bailey in the development of a glue logic cost and board footprint reduction chip, called the AMU (Address Management Unit), for the Macintosh. This chip shipped in high volume in a variety of Macintosh computers.
Was responsible for chip design CAD tool selection for Apple. Evaluated and “baked off” most of the 3rd party tool options then available.
Developed the design for a Fast AppleTalk chip which used a Lead-Lag Digital Phase Locked Loop to provide a 4x speed improvement over existing wiring.
I was hired into the fledgling Apple IC Technology group in June 1983 (about 7 months before the Mac I was introduced) by Martin Haeberli likely in no small part due to a strong recommendation by Dave Patterson at Berkeley, from where I had just graduated with a masters degree with a focus on processor architecture and implementation. The IC Tech group, now numbering three, was housed in the Macintosh building (Bandley 3) and reported to Bob Belleville, who ran Engineering at Macintosh. I definitely remember interviewing by “walking around” Bandley 3, but I don’t recall being interviewed by Steve Jobs.
Infra-red Mouse Chip for a Tail-less Mouse: The first chip the Mac IC Technology group built after I joined was demonstration project for an infra‐red mouse chip for a battery powered wireless mouse. This development would demonstrate the group’s capabilities, and exercise the CAD tools and our relationship with the foundry. This chip, which was designed by Bob Bailey and myself, encoded mouse x and y movements in a simple 8 bit packet appended with error correction bits. Most of the die was dominated by a giant transistor designed to drive the infra‐red LED. However, the forward on‐resistance of CMOS is relatively high, which means it is an inefficient choice for driving a high current device like an LED, resulting in wasted power and a short battery life. CMOS cannot compete with a standard NPN/PNP (bipolar) transistors for switching efficiency. The chips were fabbed at VLSI Technology and several prototype mice were demonstrated in December ’83 – but the prototype was never productized.
Apple Infra-Red Tail-less Mouse Chip
Macintosh Clock Chip: The Real‐time Clock Chip (RTC) also served as the system parameter RAM, which meant that data had to be retained on the chip SRAM on battery power alone after the Mac was shut down. Hence very conservative low leakage design practices were followed for the SRAM design. This chip was built in a single metal single polysilicon CMOS process and was designed by Bob Bailey and myself. Bob handled the SRAM design and the I/O padring, and I handled the rest. Since there was only one layer of metal, and we had a fixed maximum die width requirement that was very challenging to meet, Bob had to get creative with the SRAM design to minimize the size. To accomplish this – the power rails to the SRAM cells were implemented using the implanted diffusion layer (rather than metal). This was definitely not standard practice, but since access speed was not a concern, this approach was feasible but required more than the usual design and simulation effort. The control logic was laid out using VLSI Technology’s version of the Cal Tech sticks tool, which was a fairly direct implementation of the methodology used in the original Mead and Conway “Introduction to VLSI Systems” text. The chip was fabbed at VLSI Technology’s fab in east San Jose The clock shipped in high volume in a variety of Macintosh computers..
Apple Real Time Clock Chip and Parameter RAM
Apple Sound Chip: The Apple Sound Chip (ASC) was an ambitious project for one engineer. The ASC was fabricated in a double metal single polysilicon process and was the first chip at Apple to use VLSI Technology’s standard cell methodology as well as their new off‐the‐shelf SRAM compiler tool. The chip had two 1K byte SRAMs that could be configured as FIFOs (First In First Out memories) using chasing pointers. FIFO mode enabled the processor to stuff audio values into the FIFO in an opportunistic “dump and run” mode, while the chip fed the FIFO output values to the audio DAC at a steady rate. The chip also support a wavetable mode, where the processor could load wavetable samples into the SRAM and thereafter the chip would continuously cycle through the wavetable values in SRAM and transmit them to the audio DAC. The first version of the ASC employed an on‐chip audio DAC to create a monolithic sound solution. We retained a reputable local 3rd party analog design shop (Sierra Semi) to build the DAC. I had the resulting DAC design reviewed by a professor at Berkeley (Bob Brodersen I think) – who thought there would be problems because the design was not differential, and that a single ended design would be too noisy. But we elected to go with the single ended design anyway – and sure enough – it was too noisy (you could hear system bus activity in the computer’s speakers, etc). Power supply noise was being transmitted through the substrate of the chip. So we removed the DAC, and in a subsequent revision to the chip went to an all digital implementation with an external audio DAC. A takeaway from this experience is one needs a very healthy respect for ground and power supply quality when dealing with analog processing that require a high signal‐to‐noise (SNR) ratio. This applies to chip as well as board level implementations, and this experience was to carry through to the board level work I did later at SuperMac and Chromatic Research.. The Sound Chip shipped in volume in the Mac II, the Mac IIcx and other models. The first revision of the digital only version of the ASC is shown below. The second cost reduced version in a more advanced process, along with the first version that incorporated an on-board DAC is shown below in the slide show.
In August of 1987 my supervisor, Walt Peschke, told me of an opportunity to join a pirate outfit headed by Steve Sakoman that was building a new portable computer – and I jumped at the chance to be one of the first members of the Newton team, see: APPLE-NEWTON
Apple Sound Chip: Digital Only Rev 1
Mac Cost Reduction Chip (the AMU): The original Mac motherboard designed by Burrell Smith included a number of TTL parts and PALs (Programmable Array Logic) to generate display timing and control, memory control (including DMA for display refresh), and other signals. Bob Bailey was asked to design a 68‐pin quad flat pack ASIC that integrated all of this logic to reduce the cost, power, and board area footprint. This chip was called the AMU (Address Management Unit), and was designed in late ’84 and early ‘85. Since this was before the days of the Verilog Hardware Description Language, and there existed no other logical description (other than PAL equations) of the Mac to drive any EDA design or simulation tools, I assisted Bob by writing an event driven cycle accurate simulator in C that took as input the equations that described the behavior of the Mac motherboard signals. These equations (or rules) were parsed by a YACC (Yet Another Compiler Compiler) generated program as part of the simulator. There were “rules” defining internal and output signal behavior. System stimulus was captured using a Tektronix DAS logic analyzer on actual Macs, and then the results of the AMU simulator were compared against captured behavior of real Macs. The AMU shipped in volume in a variety of Macintosh computers. An excerpt from the AMU simulator code is shown in the slide show below.
A RISC based implementation of the Apple II 6502 Processor: In mid ’85 I performed an analysis that showed a simple RISC style implementation of a 16‐bit binary compatible superset of the 8‐bit microprocessor used in the Apple II 6502, along with some judicious use of on‐chip caching, could substantially improve performance – to the point of potentially outperforming the 68000 used in the Mac, and given the simplicity of the 6502 the implementation was “doable” by a small team. This was a more direct approach than emulating 6502 compiled binaries by a different processor as was done some four years later in the Mobius project in the Advanced Technology Group (ATG). I set about completing a feasibility study that went through several revisions (Turbo‐I and Turbo‐II), which included a complete micro‐architecture design of the processor along with resource usage diagrams for every clock phase of every instruction. When the design seemed solid and I was ready to move on to an implementation, I sought the counsel and the support of my mentors in the IC Technology group (to whom I owe a huge debt of gratitude), Bob Bailey and Walt Peschke. As usual, when they felt it was time to impart some wisdom upon me, they said, “Pete, lets go for a walk”. As we walked around the local residential neighborhood in Cupertino they explained to me that marketing/sales/biz dev would have no idea what to do (how to position, etc) with such a thing and I would just end up with a black eye. Of course they were right and I stopped working on it. Their warnings were prescient, as four years later Jean-Louis Gassee was to shut down a similar project called Mobius in the Advanced Technology Group (ATG) where the ARM microprocessor was used to emulate another architecture. .
Fast AppleTalk Chip: After the Sound Chip, I began work on a fast AppleTalk chip that was intended to increase AppleTalk speeds by 4x primarily through the adoption of a Digital Phase Locked loop (a lead/lag sampling phase detector). I had completed simulation of the DPLL and was ready to begin serious design of the chip when I transferred to the Newton team (called “Special Projects” on my transfer form) in early September 2007. This design helped me build a better understanding of signal integrity issues for unshielded serial communications that later served me well at Travertine Systems (a startup I founded developing networking technology using the unshielded twisted telephone wire found in every home).
Remembering Steve Jobs: Steve cared most about those engineers involved in the look and feel and the aesthetics of the product – such as the UI software team and the industrial design team. The guys doing the nuts and bolts hardware engineering – including IC Technology– were more of a “necessary evil”. That suited me just fine as finding yourself in Steve’s spotlight was frequently not a pleasant experience. That being said, the years he spent with Woz chasing down parts and hand building Apple I’s and Apple II’s gave him an appreciation for the engine under the hood – which was lacking in a number of future Apple CEOs.
Steve would regularly hold “all hands” meetings in the central atrium of the Bandley 3 building. This was a large open tiled area that later also housed the Bosendorfer piano, the BMW motorcycle, and some video games. There Steve would, among other things, exhort us to work harder. During one exchange someone said to him, “Steve, at some point we need to go home and do our laundry and pay our credit card bills”. He immediately responded saying he would have a washer/dryer installed by Monday if we wanted it, and he didn’t understand the credit card thing because he just deposited a bunch of money (I think he said something like ~$35K) in his credit card at the start of the year and never had to worry about it. The thing was – he appeared dead serious about his responses. Either that or he could deadpan so well that he fooled all of us. Steve would use the all-hand meetings to laud team members he thought were “great”. This would frequently involve very publicly handing out what we called “the grey envelopes”, containing usually some form of monetary remuneration. But Steve had an innate uncanny ability to sense whatever it was that would most motivate a subordinate – be it cash, recognition, flattery, fear – whatever.
For me the work ethic transition to the Mac group, coming straight out of grad school, was not difficult as I was used to working all hours at Berkeley, and much of the team was single. But an ethic of extreme sacrifice was cultivated – which was captured and reinforced with paraphernalia like the “90 hrs a week and loving it” T‐shirts. Marriages & mental health were too frequently a casualty of the pace and pressure. But make no mistake about it – working with a talented highly motivated team that is hell bent on “changing the world” for the better, with a large $ war chest backing them up, is infectious and exhilarating. One Saturday evening I was working late in my cubicle doing chip layout, and who should quietly appear behind me but Steve and Ted Turner. They probably had dinner together somewhere and Steve brought him to Bandley 3 for a tour. You just never knew what was going to happen or who was going to show up at Bandley 3 under the pirate flag.
Trouble at the Mac Factory – a Priority Interrupt: Walt Peschke, the manager of the IC Technology group after Martin Haeberli moved to marketing, initially insisted on initially calling the sound chip the “Foley Sound Chip” (it fortunately later became known as the Apple Sound Chip, or ASC), and when I asked why he said “so they know who to call when it breaks”. Sure enough, some months later I received a call on a Sunday evening that the Mac factory in Fremont had shut down because of a problem with the sound chip, and that there would be an emergency meeting at 8 AM in Fremont the following morning (with Debi Coleman in attendance). It turned out that the factory engineering team had cost reduced the motherboard, eliminating a couple of layers. This resulted in a significant reduction in the quality of the power planes feeding the sound chip, resulting in faulty behavior. This just goes to show the pitfalls of “designing in isolation” when dealing with larger more complex systems incorporating chips, boards, busses, and power supplies.
Apple II GS Chip Bug Chase: I was called in to help debug a problem with one of the chips in the Apple IIGS. This chip was fabbed at the AMI factory in Pocatello Idaho – so I flew to Pocatello to investigate. At one point I noticed a large plot of the chip (a gate-array) hung on a wall and I took a long close look. I asked one of my hosts “what is this signal” and pointed to a long red trace running the length of the chip and along most of one side – much of it in the padring area. He said it was a signal to implement a recent bug fix and I replied “but the whole thing is in poly”. The challenge with running signals over long distances in polysilicon (as opposed to metal) is that poly has high sheet resistance, so that signals are very slow to transition when transmitted over long wires. Problem solved. I recommend eating at the Sandpiper next time you are in Pocatello (it is still there).
The Importance of Mentors: As a newly minted bright-eyed bushy-tailed graduate student, I was fortunate to have some grizzled veterans by my side to counsel me. My consiglieres were Walt Peschke and Bob Bailey. They were far keener observers of the omnipresent political machinations than I ever was. Whenever I would ask a question about why such and such happened, or why a certain person disappeared, they would sigh and say “Pete my boy, its time for a walk” – and we would walk the nearby Cupertino residential neighborhood when we knew we would likely rendezvous with the ice cream truck and they would explain what was going on. They were full of pithy sayings, dozens of them – I only remember a few, such as “the bureaucracy will always link arms to expel the infidels”. When I excitedly told them about my plans to improve the 6502 using some caches and RISC design principles that would enable the Apple II out-perform the 68000 in the Mac, they simply said “Pete my boy, this turbo 6502 idea of yours, we have to say were behind you all the way on this one . . . way . . . . way behind you”. But probably the most memorable pearl of wisdom is what I have come to call the valley creed. Early on at Apple they took me for a walk and said “Pete my boy, there are three basic rules that apply to a career in the valley, and if you can accept these rules, then you can thrive here. If not, then you should leave” – of course I said, “Ok, what are they?”. Walt said, “#1 – there is no justice”. “#2 – there is no mercy”, and “#3 – this is the most important – are you paying close attention Pete? – #3 is . . . . no one cares”. There you have it – the valley creed.
The Importance of Fighting for the Vision When the original 128KB Mac first shipped it did not sell as well as was hoped at the initial $2500 price point, and would soon thereafter be described as “struggling”. It became clear that the business market was ripe for a WYSIWYG productivity tool, but delays in the LaserWriter laser printer were hampering adoption by the business market. So Jerome Coonen of the Mac software team was sent over to Adobe to assist with PostScript rendering speed problems – which were largely due to a large amount of floating point arithmetic calculations. Jerome, a recent mathematics PhD under Kahan at Berkeley, doubled the speed of printing on the initial LaserWriter “with artful use of custom fixed‐point arithmetic types and supporting trigonometric functions”. So the LaserWriter shipped, and the Mac found the market that would sustain it until Moore’s law enabled sufficient functionality and/or lower price to gain traction in the consumer market. A takeaway for me is that the product vision and first mover advantage are of primary importance. It is likely that the vision that is the “right vision” – once realized in product form – will initially cost more than everyone would like. But Moore’s law is so powerful and so predictable (especially in the 80’s and 90’s) that the increase in functionality and/or decrease in product cost enabled by it would allow the market to “catch up” to a product with an ASP that was initially “a little too high”. That is provided you have the confidence in the product vision to support the product initially, perhaps through lower margins and/or finding alternative initial markets (which were both done for the Mac). This was an important lesson that should have been applied to original “Slate” form factor Newton.
Apple Macintosh Slide Show:
Apple Clock Chip die photo. Single metal – single poly CMOS full custom
First Apple Sound Chip (ASC) with integrated A=audio DAC
Apple Sound Chip, digital only version, Rev 2
Apple Sound Chip corporate ad/marketing collateral
AMU (Address Management Unit) Chip example simulator code