Have you separately confirmed that the IO space works? It seems to me that your 55/AA /FE test with ram doesn't exclude the possibility that IO could be flaky (too). I mean it's technically possible just from that test that IO is a problem.

I guess I would try this ram chip in the previous project, or in some different memory map that doesn't use the 688, and see if that works. But you know, it may just be a bad breadboard connection.

I'll do a before and after re-wire photoshoot! But I'll have to post it in a day or two, it's getting pretty late here.


That is correct, and when the LEDs show $00 it stays that way. ($00 $00 $FE)


I hope that I didn't seem like I was shutting down anyone else's ideas. That wasn't my intent at all. It was just the fact that it works at all seems to imply that sort of issue, and it's repeatable. Perhaps I'll make a short video of the problem.


I checked out DigiKey - they have the same free shipping when order > $100 but I found that Mouser's prices on the items in my cart were generally $0.50+ cheaper. It's all right though, I placed the order. I've got some AC '139's on the way as well along with a bunch of other goodies.

Steve

(facepalm) I forgot to grayscale the schematic. Argh..

I read somewhere that both HCT and ACT are more susceptible to noise, and that's why I am going to snag a few HC/AC replacements.

Steve

I did a bunch of tests with the 6522 and some LEDs that didn't use RAM. I also changed the IO address in the program to see if the 6522 was somehow enabled anyway and it wasn't. I wrote a small program that would do a little Knight Rider effect and let it run for like 10 minutes because my son thought it was cool. It seems to work.


I should put up a picture of sbc v2.0, that would require a rebuild. The 61512 has four more pins and I didn't leave enough room between the 6502 and the RAM for that.

Steve

It's alive! There must've been either a bad connection or a crossed connection. It works now. I'm running my LED loop test except I changed it to store the $55 value to $C000 and $AA to $7000 so as to test each side of the 32k threshold. It's been running for about 5 minutes without issue now. I've uploaded some pictures. I have a shipment from Mouser that should come in this week. It has some 74AC10s in it which should allow me to do away with the inverter.. but I don't know if I want to have more than one 6522 which would necessitate keeping the 74AC11 so I can share the IRQ line with everything.

I also found that I left one gate on the inverter floating.. perhaps that was the issue.

Thanks!
Steve

Way to go, Steve. Congrats'...

They are interesting parts... I've often wondered if they could be used to insert ROM and I/O into 64K Bank 0 on a 65816 system.

Cheerful regards, Mike

Hello again all!

I've made some changes that someone may find interesting. Slightly. Well, not really.

This new logic will allow you to place the IO page anywhere on the map. I've tested it at $FE00 and $0000 and it works. I wonder if there's an application where having IO on zero page would be useful. Perhaps if you were controlling something where every clock counted.. but I suppose the 6502 may not be the best solution in that circumstance.

I did get some 74AC10's and replacing the '11 and removing the inverter worked fine, but you couldn't place IO in ROM space with that setup.

Anyway, I am running into another odd problem. The computer works well for around 10 minutes. After that it crashes. At first I thought it was some dodgy wiring (it may still be) but I can cycle the power without touching the wires and it restarts without issue, and then crashes. My main thoughts are poor wiring (wires too long, perhaps some interference, dodgy connections perhaps) or flaky RAM chips (I got them from Ebay.) Can anyone think of other reasons I may be able to test for or check out?

Cheers!
Steve

Hm, the time delay reminds me of the other trouble you reported earlier. Are you using the same power supply as before? Perhaps that could be the culprit.

Zero page is a valuable resource, we all agree. But how one chooses to use it is a decision that'll vary according to one's priorities and circumstances.

In my own case, speedy IO is a high priority, and I had no hesitation in devoting a portion of zero page to IO. The speedup is non-trivial -- partly because all loads and stores happen one cycle quicker, but also because of being able to exploit the very efficient 'C02 bit manipulation instructions: SMB / RMB (set/reset memory bit) and BBS / BBR (branch on bit set/reset).

However, many 65xx fans are obliged to support legacy BIOSes that make zero page largely unavailable. In those circumstances you wouldn't put IO in zero page unless you were absolutely desperate for the fastest possible IO.

BTW, the reason my own computers have oodles of zero page space available is this. Except for IO, almost all my use of zero page (including BIOS-like routines) takes advantage of a stack located there. This makes allocation and deallocation of the precious space automatic, and I never experience crowding.

-- Jeff

_________________
In 1988 my 65C02 got six new registers and 44 new full-speed instructions!
https://laughtonelectronics.com/Arcana/ ... mmary.html

I suppose if a fridge or freezer starts up somewhere that could give a power supply a bit of a wobble...

I had a weird PSU issue some time back - I was USB powering my system via a USB serial adaptor and had moved the cable from a rear USB socket on my PC to a front one. The issue came when I hooked up an LED board to it for some tests, and after a lot of experiments it seemd that when I lit up 3 or more LEDs on the panel it was enough to cause the voltage to drop. Moved the cable back to the rear USB sockets on my PC and it was fine....



It was good in the Atari 2600 which only had 128 bytes of RAM and the TIA device (Television Interface Adapter) was also mapped into zero page too, so you saved a cycle when accessing it - was that cycle worth it? In the 2600 YES as the CPU itself was racing the beam to generate the video signal and every single cycle counted...

In todays applications? Personally, I'd suggest we're doing it wrong, but back in the 2600 days it was all they had - for the money people were prepared to pay for it...

-Gordon

_________________
--
Gordon Henderson.
See my Ruby 6502 and 65816 SBC projects here: https://projects.drogon.net/ruby/

Check with your finger see if any part is too hot.

Also press/wiggle/flex your board for possible intermittent connections.
Bill

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

I've been experimenting trying to find the source of the crashing. There were some interesting developments.

I was running the SBC at 6MHz, and it would crash around the 10 minute mark. I switched to a 1MHz can and it crashed after an hour. Of course I thought this meant perhaps there was some bug in my program that would only manifest after however many clock cycles that would be, but I put in an 8MHz can and it ran for about 4 hours before it crashed. Must be a co-incidence.

I redid the layout for the board. Every chip moved, I also made my 8 LED IO device out of a 74HC574 latch and some glue logic (using the third gate of my 3-input AND gate - no wastage here! It used to be just manipulating pins on the 6522) and modified the code to suit, and the problem still happens the same way, which although not completely ruling out bad connections does seem to make it less likely. Also just placing my hand down on all of the wiring while it runs does nothing.

I noticed that after a crash the current usage would go up by about 100mA. The LCD also would act funny. I also noticed that the LCD would act funny on some resets in general. I've used this LCD for over a year without issue, I'm going to test it in one of my other boards.

I wanted to see if the RAM chips I'm using may be the culprit so I re-wrote my LED code to eliminate the LCD code and all JSRs. It's now all inline with no RAM usage. Crash still happens.

On a lark I just pulled out the LCD. Board ran for 6 hours with no issues. Remember, there is no code driving it now, it's just in the circuit connected to the 6522 but I don't make any usage of the 6522 at all. I put in a 10MHz can, and it's been running now for a couple of hours. 10MHz has never worked before.

So it looks like the LCD is causing some sort of issue. It's a 1604 display, and like I said I've been using it for over a year. I have a 2004 display in my second SBC which I'm going to swap in probably tomorrow because of Christmas engagements but I'm kind of wondering why it would cause issues now.

The things you learn when you just mess around with stuff because the code is so simple it has to work. Oh, I also modified the code to add an stz $A000 just to add a RAM access and it doesn't seem to be a problem.

What a ride.

Have a merry whatever you celebrate!
Steve