Mickey Mouse logic, address decoding, power management, etc.

[No True Scotsman]

Thank you, guys. Nothing is set in stone design-wise yet, and I'll take these recommendations under advisement.

[No True Scotsman]

Mouser doesn't have the 74FCT521 in DIP, Jameco doesn't list it period, and it's insanely expensive coming from the UK on Ebay, at ~23 USD for one chip.

I think I'd best see if the 688 will manifest actual problems in my build before making substitutions. I'll go ahead and buy the 138 and 139 in HCT as suggested though, since I haven't bought them yet.

[AndrewP]

If you haven't placed your order yet Mouser do stock the SN74F521N which is a dip from the F family and is only a nanosecond or two slower than the FCT variant. If memory serves it pulls all the way up to about 4.5V so it should play nicely with HC chips.

[edit] apologies - memory did not serve me well. I forgot I swapped out my wide SOIC F521 for a narrow SOIC FCT521. The F variant will only output up to about 3.5V just scraping into the input range of an HC chip. The FCT variant output all the way beyond 4.5V. Unfortunately the reason I swapped them was because I used the wrong (narrow) footprint not because I needed CMOS level outputs.

[No True Scotsman]

AndrewP wrote:

If you haven't placed your order yet Mouser do stock the SN74F521N which is a dip from the F family and is only a nanosecond or two slower than the FCT variant. If memory serves it pulls all the way up to about 4.5V so it should play nicely with HC chips.

Would it still be better to get the HCT decoders?

[AndrewP]

I'm probably the wrong person to ask as I went all in on LVC - I wanted minimum propagation delays out of discreet ICs.

I like speed. But as a hobbyist with no real background in electronics the lessons were pretty rough.

I think mouser stock the AHCT138 and AHCT139; and also the AHC138 and AHC139 decoders in DIP. I would go with either of those families first before choosing F. They don't draw the huge amount of current that the F series does and the slower, calmer transitions will make your life easier. In terms of using something other than HCT I'd say experiment. Buy a couple of each if you can - they're relatively inexpensive - but start with a slow clock and HCT first before moving onto AHCT or something.

[GARTHWILSON]

No True Scotsman wrote:

Mouser doesn't have the 74FCT521 in DIP

Although SOICs can't be socketed like DIPs can, they're easy to solder by hand; so there's no need to be afraid of them.  I tell how at https://web.archive.org/web/20240116060 ... =20&t=1278, with pictures of the results.  I don't even use a thin soldering-iron tip.

[No True Scotsman]

I would need through-hole breakout boards to solder them to Veroboard. And I have very poor depth perception due to being blind in one eye. I'm lucky if I can pour coffee into my cup instead of the floor.

Messing around with the ESP32 board today, I found that it probably doesn't have enough flash memory to do everything I want it to do. I'm going to go ahead and order a genuine Arduino Nano ESP32. It can earn its keep by burning bootloaders onto the Non-duinos I already have, as well as replace the antiquated DEVKIT V1.

I'm also going to order a genuine WDC 65C02, since I don't know for sure what these two chips I have really are.

[barnacle]

If I might bring your attention to this: viewtopic.php?f=4&t=7910&hilit=breadboard

It takes either through-hole dip, or SOIC up to 20 pins, or a mixture, with convenient decoupling capacitor mounting places. There are a couple of places to fit oscillator cans, too. Let me know if the Gerbers are of use to you.

Neil

[No True Scotsman]

Listen. I haven't won the lottery. I can't just pile everybody's dream BOM into a shopping cart and click the button. There are some things I need to buy because the parts I have aren't up to snuff, or are of dubious quality. But I still have to use at least some of the stuff I already have. This is already getting into "may as well buy a kit" territory as far as pricing goes.

I mean the 688 has been fine for over a week, and now it's suddenly too slow for a 1 to 4 MHz computer? It's absolutely astounding that anybody manages to get a 6502 computer working at all, if they're that finicky.

[BigEd]

> I mean the 688 has been fine for over a week, and now it's suddenly too slow for a 1 to 4 MHz computer?

That's an interesting observation! Any idea what changed? If this is a breadboard construction, then it may well be a bit fragile: some connections not quite good enough, especially the power supply - both the resistance and the decoupling local to each chip.

Could you share a photo?

[No True Scotsman]

I haven't built the computer yet. We've been discussing 688 decoders here for the past 10 days. Nobody said a word about the propagation delay being unacceptable until now.

As far as me soldering any SMT device with one eye, it probably ain't gonna happen. That's just the way it is.

[BigEd]

Oh, I see - sorry for misunderstanding.

What you tend to get here is lots of opinions - hopefully most of them right. You often also get ideas which are in effect micro-optimisations - not necessary, but moving towards some ideal world.

It's not always easy to see which bits of advice are just nice-to-have, as opposed to that'll-never-work.

Edit: of course all flavours of advice come from a place of good will and enthusiasm. But they may not fit a person's current constraints or preferences.

[AndrewP]

Just as BigEd has said, there's a lot of enthusiasm on this forum. And a scary amount of experience. The advice you get is gonna be good but ... not always suitable. A 1Mhz 6502 build is not a finicky build; you'll absolutely be able to do the one you've sketched out on a bread board with HC/HCT parts.

Where things start becoming trickier is with faster clock speeds. You mentioned you might want to try an 8Mhz clock. That's only 125ns per cycle; 62.5ns for PHI2 low and again 62.5ns for PHI2 high. Propagation times for HC/HCT chips at 5V is generally around 15ns so chaining even two together and throwing in the relatively slow '688 (at 20ns'ish) has already consumed almost all the time in a half-cycle. That's not including the time taken for the 6502 to setup its address lines nor the time it might take for RAM or ROM to respond to a read.

I think a lot of the discussion on faster parts might have come from the above.

My advice comes from a South African context - ordering from Mouser for me is big deal. The cost of postage is usually (a lot) more than the cost of parts so I tend to bulk everything I could possibly want into a single order and hope I haven't missed anything.

Mostly I'd say start with what you've got - for example the 1MHz clock you have is a nice speed. If something in the build doesn't work I know someone here will be able to point to the problem. But more likely it will work and then you can start increasing clock speeds and swapping in faster components.

Finally I'd add that ordering the WDC65C02 in your basket is a good idea. That makes it a known quantity and easier to debug if things do go south.

[No True Scotsman]

AndrewP wrote:

The advice you get is gonna be good but ... not always suitable.

I've no doubt the advice is solid, and I do appreciate it. Thanks!

AndrewP wrote:

You mentioned you might want to try an 8Mhz clock.

I'll try it for long enough to see if it works, which sounds like it'll take no more than a few minutes. I'm not going to jump through hoops trying to make it work at that clock speed. Farming the keyboard and video out to an ESP32 ought to reduce CPU overhead enough that 1 to 4 MHz won't seem overly sluggish. Nobody ever complained about the C-64 being too slow anyway.

AndrewP wrote:

... ordering from Mouser for me is big deal. The cost of postage is usually (a lot) more than the cost of parts so I tend to bulk everything....

That's what I'm doing. Postage isn't egregious in the US, but I still try to get the most mileage for the money. The tariffs on crystal oscillators from Mouser can be substantial though. Two-legged crystals strangely do not have a tariff.

AndrewP wrote:

Finally I'd add that ordering the WDC65C02 in your basket is a good idea. That makes it a known quantity and easier to debug if things do go south.

That's what I'm thinking. And I can already see that the outdated ESP32s I have can't handle more than basic example programs.

All things considered, the SN74F521N is reasonably priced, and it can't hurt to have a few on hand, so I'll go ahead and order them.

[barnacle]

My apologies; I did not intend to cause such a ruckus. It was merely a concern that the part was relatively slow compared to other options.

Neil