>MOS did make a 8501R4 in 1986 and there are even 8501 with a 1990 datecode. Those are rumored to last a lot longer. Part of what the process engineers did at MOS was make a chip better by studying temperatures and "schmoo plots", and either fixing things in the recipe themselves (or feeding back to the chip layout guys to change source and not just paint the masks). The process engineers mentioned one time how they laid a temperature sensitive LCD sheet on a die (think "mood ring" from the 70's) and found a 200 ohm "resistor" (nmos gate in conduction mode I assumed) between VCC and ground. >I still have a working 7501R1 and, surprisingly, it needs about 20mA >_less_ power than the 8501R1 and R4. Sometimes things like coming out of Hi-Z faster (newer process) can have an effect like this. They also did some things with a back bias generator that was an onboard charge pump that was supposed to increase speed and reduce leakage by increasing gate potential, but who knows if the charge pump increased power usage. >You could have put a RAM_enable signal on the expansion port without >needing extra gates though. Both C16/116 and Plus/4 have some spare >gates that could have been used to gate _CAS so that a module on the >expansion port could have disabled the internal RAM. But someone in >Management might have noticed... The gates came later but my memory is that the signals might actually have been blessed by management before hardware engineers got involved, some management like Shiraz were very sharp being engineers themselves. Same timeline for decisions like the joystick connector and expansion connector themselves. Plus honestly I wouldn't have thought myself about adding that for a $49 computer. I seem to remember doing a DRAM expansion for the series though now that we are the subject, I remember getting it laid out at Fourth Generation before we did our own PCB layout. With that said, the Japan office did the release, anything West Chester put in would only be considered input to the final design at that point. I saw the power resistor only after they were in production for example. Thinks got <ahem> tense after Jack Tramiel left since the production release resided in an office still run by his son. ->Replace 7805 with switching regulator (RECOM 5V/1.5A type) Didn't exist at the time as affordable or the Tokyo office would probably have put it in. What was there was large and expensive, no single chip, an inductor alone would have cost as much as the 7805 I think. Commodore didn't have a single switcher that I am aware of until the C128 power supply which we had to source from Mitsumi. > Remove R10 (the large ceramic resistor) When I saw this I laughed, I cried, I had a moment of anger and then acceptance. I realized that the CBM Japan office which ultimately took what we sent and made the C116, had assumed that there would always be a minimum current draw of a couple of hundred MA to keep the regulator in regulation. I had never ran across that kind of thinking before, my assimilation into the CBM collective began that day. They literally found a $.08 answer to a different kind of heat problem. For $49 they were only meant to last a year or two. - Add heatsinks to CPU, TED and PLA Again product life wasn't never a consideration that I found, passing FCC WAS a consideration, especially after we had been fined for the products going out that were non-compliant. So the hardware guys didn't have a say, the mechanical guys barely did, the Japanese office that was also responsible for passing FCC made the call on the heat sink/emi shield. Again, the Japanese office was run by Sam Tramiel and was the office that touched the design last before production when I first got there, and there was only one power greater then Sam. - Replace the ROMs with properly programmed 27C128 or 27C256. (saves about 50mA per ROM replaced) Lol... I don't believe that they were cost effective and MOS didn't make them which would have been the deciding factor. The ONLY problem that we ever needed to solve was making it cheaper. Saving 50ma didn't figure into unless it kept the product from shipping. :) Saying to use someone else's ROMS would have been a reason to get fired, this was a time when we made Atari's ROMs. >That makes the system run a lot cooler in general, also meaning cooler >dies for TED and CPU and hopefully some more years. I am serious when I say that it's hard to believe that ANY system are running 30 years later, our typical accumulated MTBF for a consumer system was around 2-3 years... that is the ones that weren't already broken before we shipped them. We would gladly sell you a second system if your firsts system died. Some things changed as we got into the 264 version where we started going to the Japan office and making it less of a total re-engineer for production release. But then the 264 turned into the Plus4 so nothing to crow about. When we did the 128 we brought the engineers from Tokyo to West Chester and that was the first board that was laid out for American and Asian auto insertion machines from the start, so the first to not be re-engineered after preprod release. With Sam moving to Atari the production release power was pulled more formally into West Chester. CBM did realize they needed to get into CMOS and had broken ground on the new line when I left. I was told they never finished . Another example of how the production release was its own animal: There was a last minute chance to fix the serial bus, Bob Russell had come up with the software fix for the problematic shift register hardware and had had the PCB films changed to wire the serial lines to a working hardware pin in addition to the software driven pin, the idea was they could at least cut over at some point very soon. Tokyo looked at the film and painted out the second connection assuming it was an error and the software serial bus became permanent. I am just trying to describe a culture here, not make excuses, but with that said most firmly believed in cheapness at any cost and saving a few pennies was a good day. Hard to believe it was 30 years ago! Bil Message was sent through the cbm-hackers mailing list Message was sent through the cbm-hackers mailing listReceived on 2013-05-02 02:00:04
Archive generated by hypermail 2.2.0.