We did strip things as needed , my memory is that they used the real-estate on the 7501/8501 for an unused register bit to implement more with back-bias generation. I remember walking down the hall and overhearing a conversation that the test guy (named was Terry) that was writing the genrad test code for the 7501 test fixture was having trouble with the Port 0 test. I remember asking him if he knew that one of the ports was gone and he lit up saying that that was the problem. He was trying to rotate a bit all the way through the register and of course it wasn’t making it. So not only did we strip and change things on the fly, we dint have mechanisms and documentation that kept up, our eye was on the clock and foremost did whatever we needed to move the ball a few inches; if we made it we would worry later about cleanliness. Quick story about how we figured stuff out on the fly: We were getting black dots on the display of the early TED/264, seemed to have shown up when we married a PCB with keyboard early on. I threw a scope on the master clock and it was "freezing". I licked my finger and went around the pins of the ted chip and found a pin that would cause a beat pattern of black dots on the display (clock was freezing at a 60 hz rate) when I touched it with my other hand on the monitor itself. I strapped the pin to VCC with a Germanium diode to keep the pin from pulling higher than VCC and the symptoms stopped. I walked in on a croup of chip guys talking and literally said " let me guess, you are using xxx pin to freeze the clocks for test initiation purposes (tests need to start from a known place) by pulling it above Vcc" The answer was "ummmmm yeah, howd you know?" I showed them the problem and how I had to shunt it, turns out there was two pins that they had done this to.You can see two Ge's soldered to early Ted chips in pictures and old parts. Basically they learned what the hw already types knew which was that a high impedance pin acts as a "amplifier" , I.E. they didn’t think there was any way a pin could go above VCC when the highest voltage was VCC. Bil Herd -----Original Message----- From: owner-cbm-hackers@musoftware.de [mailto:owner-cbm-hackers@musoftware.de] On Behalf Of Nate Lawson Sent: Friday, August 12, 2011 2:16 AM To: cbm-hackers@musoftware.de Subject: Re: 264 Series and their chips On Aug 11, 2011, at 3:10 PM, Hársfalvi Levente wrote: > On 2011-08-11 18:32, Gerrit Heitsch wrote: >> >> So I noticed when I tried it yesterday evening... Oh well, another >> pretty theory shot to hell... >> >> The reason I came up with this idea was that I read a posting on this >> list where someone stated that the 6510 and the 8500 have the full 8 >> bit port on die and that you could use some clever coding to >> distinguish between a 6510 and an 8500. Go look for the thread called >> '6510 and 8500 differences'. > > Ah, Nicolas Welte's findings about the unimplemented portbits of the > 6510/8500 (...one kind of stuff I like cbm-hackers for :-) ). > > Thinking it over, even that one seem to contradict with the idea of a > full 8-bit port that was supposed to be implemented on the 6510/8500's > die. Bits of the output register don't fade out by the corresponding > I/O pins absence. That might rather suggest that MOS already had a > 8-bit port equipped CPU design, maybe more than one CPU design with > several features incorporated into the 6510 at the time they actually > needed the 6510, and pasted/stripped parts as needed, with no care for > leftovers (ie. those that didn't hurt them). > > It's interesting though, that they didn't change the design once they > switched to HMOS-II. I thought that move needed a more or less > complete redesign of the chip. Still, the artifact has been kept (it's > only the manufacturing process that results in different time until > the fadeout happens), which suggest they didn't change the layout. > ...I'm really looking forward for this thing to be demistified. In "On the Edge", one of the C64 engineers said they made the 6502 into a core before creating lots of variants, such as the 6510. (I believe it was Winterble, but not sure.) Anyway, since the 6502 was hand-drawn, I assume what they did was reconstruct the gate design in more modern CAD tools to allow it to be modified more easily. This also would allow them to change processes, do design shrinks, etc. HMOS is still an NMOS process, just with higher density. Most of the early Amiga chips were HMOS too. I think it wasn't until Commodore started outsourcing some of the chip production that they moved to CMOS. -Nate Message was sent through the cbm-hackers mailing list Message was sent through the cbm-hackers mailing listReceived on 2011-08-14 19:00:12
Archive generated by hypermail 2.2.0.