bbs00068@uafcseg.uucp (Joel Kolstad) (07/26/90)
I recently got a 16MHz 68000 and the did ever popular trick of taking the 28MHz system clock, dividing it by two using a flip flop to get 14MHz, and feeding it to the 68000. Well, this works real well, except for the the fact the clock gains about 10 seconds every minute! Note that I'm talking about the internal clock of the Amiga -- a setclock opt load restores everything to normal... for awhile! Does anybody know how to fix this slight problem? It's more of an annoyance than anything else, but I'd really like to have my clock read the correct time! Thanks in advance... Joel Kolstad kolstad@cae.wisc.edu...
rsbx@cbmvax.commodore.com (Raymond S. Brand) (07/26/90)
In article <4975@uafhp.uark.edu>, bbs00068@uafcseg.uucp (Joel Kolstad) writes: > > I recently got a 16MHz 68000 and the did ever popular trick of taking the > 28MHz system clock, dividing it by two using a flip flop to get 14MHz, and > feeding it to the 68000. Well, this works real well, except for the the fact > the clock gains about 10 seconds every minute! Note that I'm talking about the > internal clock of the Amiga -- a setclock opt load restores everything to > normal... for awhile! > > Does anybody know how to fix this slight problem? It's more of an annoyance > than anything else, but I'd really like to have my clock read the correct > time! A1000s A500s and [AB]2000s use the processor to generate a signal called the "E-clock". This clocks at 1/10 the cpu clock and is used as the phi(??) clock to the 8520s. The 8520s use this to clock the countdown counters, and the timer.device uses the countdown counters to keep track of time. Normally, the E-clock doesn't change too much, there is a value for NTSC systems and a different one for PAL system, and they're within 1% of each other.What you've done is changed the system E-clock frequency to twice what it should be and confused the timer.device. Fixes: Undo your hack (uh, I didn't think so..... :-) Build a divide-by-10 circuit to generate the E-Clock from the OLD processor clock. Build a divide-by-2 circuit to generate the E-clock from the E pin of you speedy processor. > > Thanks in advance... > > Joel Kolstad > kolstad@cae.wisc.edu... rsbx ------------------------------------------------------------------------ Raymond S. Brand rsbx@cbmvax.commodore.com Commodore-Amiga Engineering ...!uunet!cbmvax!rsbx 1200 Wilson Drive (215)-431-9100 West Chester PA 19380 "Looking" ------------------------------------------------------------------------
<LEEK@QUCDN.QueensU.CA> (07/27/90)
In article <13438@cbmvax.commodore.com>, rsbx@cbmvax.commodore.com (Raymond S. Brand) says: > >Normally, the E-clock doesn't change too much, there is a value for NTSC >systems and a different one for PAL system, and they're within 1% of each >other.What you've done is changed the system E-clock frequ >should be and confused the timer.device. > >Fixes: > Undo your hack (uh, I didn't think so..... :-) > > Build a divide-by-10 circuit to generate the E-Clock from the OLD > processor clock. > > Build a divide-by-2 circuit to generate the E-clock from the E pin > of you speedy processor. > That's should get the right frequency for the E clock, however the E clock is also used to do 6800 style perpherial chip access in the 8520s. The E clock has somehow be sync to the 6800 bus cycle for proper operations, doesn't it ?? Looks like the job for some PAL chips to generate all the timing info + bus signal for a 6800 perpherial access. That's the way the Lucas 68020 accelerator works - a 68020 do not support 6800 perpherials, so Brad Fowles make one out of a regular memory access. >------------------------------------------------------------------------ > Raymond S. Brand rsbx@cbmvax.commodore.com > Commodore-Amiga Engineering ...!uunet!cbmvax!rsbx > 1200 Wilson Drive (215)-431-9100 > West Chester PA 19380 "Looking" >------------------------------------------------------------------------ K. C. Lee P. S. For those interested, the Lucas article is in the now nonexistance Amiga Transactor Volume 1, issue 3 P. 36-43. Pal equations are included.
ammrk@swpyr2.sbc.com (Mike R. Kraml) (07/28/90)
>In article <4975@uafhp.uark.edu>, bbs00068@uafcseg.uucp (Joel Kolstad) writes: > > I recently got a 16MHz 68000 and the did ever popular trick of taking the > 28MHz system clock, dividing it by two using a flip flop to get 14MHz, and > feeding it to the 68000. Well, this works real well, except for the the fact >the clock gains about 10 seconds every minute! Note that I'm talking about the > internal clock of the Amiga -- a setclock opt load restores everything to > normal... for awhile! First I heard of this trick. Has any other net type folks pulled this off? How (well) does it work? Does the other hardware (custom chips and such) get affected by this hack? From what I can tell, he is only feeding the 68000 the new clock rate, is this the case? If so, how the the rest of the system handle this? Oh well, enough questions for now!!! > Joel Kolstad > kolstad@cae.wisc.edu... Any input would be great. Sounds like a nice cheap means of doubling your processor speed. By the way, how much faster does the thing "actually" go now (i.e. compiles and such)?? See ya all later, Mike...