> On 2017-06-22, at 20:01, Gerrit Heitsch <gerrit@laosinh.s.bawue.de> wrote: > >>> At the end of 1986, Commodore changed the process from NMOS to HMOS-II. For a short while, some CIAs were labeled '8521R0' but then they went back to the '6526' label. Usually they were also fast enough for get the 'A' for 2 MHz operation. The HMOS-II-CIA also changed a bit timingwise, that's where the legend originated that the 6526A behaves differently than the 6526. >> They DO behave differently. For example my old test cart shows CIA related failure when using the "6526A". > > Does it still show that when you use a 6526A-1 made in 1985? I expect not. So the definig characteristic is not the 'A' but the datecode. I understand. I was referring to those you mentioned being relabeled back to 6526A from 8521. > If the CIA was made at the end of 1986 or later, it's HMOS-II and internally a 8521. Since HMOS-II is faster, they all tested good for 2 MHz and go the 'A' stamped on them. Some of them were even good for 3 MHz and got a 'B'. Interesting that the tiny incompatibility gets detected by the cart. > Do we know how exactly they differ timing-wise? I guess this is related to what Marko wrote yesterday? > > It's something about the timer IRQ happening one cycle earlier or later than the NMOS version. Need to check the cart's code to see what exactly fails there. It's even more interesting if it's only a one cycle difference. -- SD! - http://e4aws.silverdr.com/ Message was sent through the cbm-hackers mailing listReceived on 2017-06-22 19:00:02
Archive generated by hypermail 2.2.0.