6502.org

Where are the QFP real estate savers?

Perhaps they'd be cool-looking boat anchors.

WDC seems to have discontinued them several years ago. I wish they would bring them back.

The PLCC packages have been around for a long time. WDC will sell onesies and twosies but now that Jameco and Mouser are stocking most of the WDC product line, there are easy-to-use sources for these items.

Incidentally, my POC unit is powered by the '816 in the PLCC package.

The PLCC is new for the W65C51, but as you say, have been around for the other chips for a long while.

As much as I'd like the most compact arrangement, even old NMOS needs to be allowed for. Plus, speed and ease are the essence of these boards. Not everyone has surface mount soldering skills, nor the time to learn them.

You might not need a new board for each new CPU. Years ago I dreamed up a layout that had two 40-pin CPU footprints overlapping. Slightly staggered, they shared the same area on the board. I think it was for 6502 and Z80 -- I found their pinouts are similar enough that it wasn't too difficult to have the address and data busses bridged from one CPU to the other, then out to some 28-pin (or 32-pin) memory sites. Of course you'd only install one CPU; the idea was have an "ambidextrous" board which could be manufactured in greater volume.

The idea's no good for PLCC, of course. And there'd still be a fair amount of point-to-point wiring to be done. But at least the drudgery of connecting the 16 address and 8 data lines (and power & ground) would be taken care of.

Hmmm... Since 6502 and 65c816 pinouts are so similar, maybe those two could coexist with Z-80 in a three-way board design! (Or maybe 6809 instead of Z-80; I haven't looked into it.)

-- Jeff

Or you can pull the trick that Jeri does of a small plug in module for each cpu.


by the way, speaking of the 02/816 pinout, finally revamped the schematic to acommodate both CPUS. my usual stop, www.nightmarepark.com/6502.php click on the pinball mind folder and check out the pdf of it. Let me know what you think there...

I see where you use some 74HC type logic. May I suggest 74F or 74ABT? The latter are far faster and the 74ABT has very strong drive, along with single nanosecond prop times. Only things is most of the ABT stuff is SO packaging, which is manageable.

Is there DIP versions of ABT? One goal is ease of repair in the future so all packages are DIP based. the 245 logic is simply buffers to the outside world, in an attempt to protect the CPLD. The F series I always had issues in arcade games of running hotter and being the first to go bye bye. Speed is not so important as I don't think I will go past 4 MHz in this design in general.

There was at one time. You'd have to shop around. If you are certain you'll never ramp up the clock then consider using 74AC as an alternative. That stuff is still available in DIP.

Thanks, will go to AC if needed for higher clock speeds. All chips will be socketed, so it won't be a major issue. I will try and see about designing the board traces for up to 20 MHz if needed but my programming seems fairly happy at 4 MHz. Would be hard pressed to justify then AC at slower clock speeds since the parts are twice as much cost wise...


EDIT: ABT lower cost and available in PDIP. thankee!

I've not let the fact that ABT is mostly in SO packaging stop me. 50 mil centers can be soldered by hand, even by an old fart like me. You just need a lighted magnifier (5 diopter) or machinist's magnifying goggles, good small-tipped slobbering iron and reasonably steady hands (that last is where I run into trouble).

As for SO chip removal, I've found that a carefully directed hot air blast from my small heat gun (has a 1 inch diameter nozzle) will desolder SO and SOJ packages. These packages aren't as hobby-friendly as DIPs, but they do take up quite a bit less PCB real estate, and thus make it easier to make tight trace routing that enhances high frequency performance. The industry in general is moving away from DIP, so the change is inevitable.

I have no problems in soldering SO packages. I can regularly do 144 TQFP packages by hand at my old job. the secret is in the flux

The thing is a design goal I have in this pinball computer, I am making it for other people to be able to repair it 20 years down the road. Every part is fairly common with multiple sources, except for the Xilinx CPLD, and I use a through hole socket plus test points all around it. Worst comes to worst, someone can make a sub board with a new kind of CPLD replacement. It is still a pretty decent power system at a good price point, so it should work out well.

Why not just include an extra pre-programmed CPLD or two with each machine?