$800 Atari 800XL upgrades part 2: PokeyMax fault, AKI firmware and Rapidus stability

[@BooktownBoy]

Good spot on the PokeyMax fix 😎👍

[@flashjazzcat]

Thanks!

[@bitdigital8052]

I just bought a 3.1 and a 3.2 is on the horizon 🤦🏽 Loving these hour long+ videos too. Funny how time flies when you mentioned how long you been modding the xm301's. I remember looking at the threads on AA admiring your work.

[@flashjazzcat]

Isn't it amazing where the time goes. Such long videos aren't my preferred format (it's just the way things turned out since there was such a lot to cover) since they require an enormous amount of editing work (I started editing this one on Tuesday evening and worked on it for the rest of the week, intermittently shooting extra footage and recording voice-overs to try and clarify the already bewildering narrative). I think everyone's had enough of Rapidus 800XLs for now, anyway. :)

[@gingertom64]

What a lot of work you put into that 800XL. And as always, excellent explanation of what was cause and effect of all the issue's.

[@flashjazzcat]

Thank you!

[@donpalmera]

And here I was thinking Amiga people were crazy with their franken-machines.

[@flashjazzcat]

Yeah. I have a couple of A500s and I fitted one with an IDE adapter some years ago. That was enough.

[@richardkelsch3640]

I nearly shot soda pop out of my nose with the Quantum Leap clip. Classic FJC and it's awesome.

[@cathrynm]

The upgrade videos are always worth watching. Rapidus adds to the excitement, will it work or not? Need to watch to find out.

[@flashjazzcat]

Thank you. And yes: Rapidus can always be relied upon to turn a one-day upgrade job into a four-week psychodrama. :)

[@TheWoj76]

Still watching, but Ha! I am not the only one that tripped over the SRAM board :O

[@skipbreakfast]

I, for one, enjoyed your guitar skills.

[@flashjazzcat]

I have been considering spawning a second channel focusing on the guitar, but I'm in the unfortunate position of the mainline computing channel bearing the name which would be more logically applied to said second channel. :)

[@waltciii3]

Now I realize I have 0.47 Side 3 loader. I will look into reflashing my Side3 now.

[@flashjazzcat]

Just remember that if everything goes to Hell as a result, it means your system wasn't stable to begin with. ;)

[@w1katari]

35:05 - demonic laught :)

[@flashjazzcat]

'Insanity is doing the same thing over and over again and expecting different results.'

[@MoparStephen]

Wow - I bet you will never have been so happy to see a machine leave.

[@scrameta]

I see the intermittent key repeat with the pokeymax and the TKII as well. Though the timing of when TKII pulls KR1 and KR2 has a wide dispersion so I thought that was why. To work properly with pokeymax a keyboard emulator need to respond on KR1/KR2 within a fairly tight timing of the request. I don't have a lot of flexibility on this since I used an i2c io expander, the tiny one you re-soldered! i2c isn't so fast.

[@flashjazzcat]

Interesting - thanks!

[@rhayadercomputers6468]

Just wrote to the pope to ask for you to be the patron saint of Atari as you have patience of a saint :) that thing would have gone out the window if i had it

[@flashjazzcat]

LOL. I appreciate it! The stakes are raised a bit higher when I've spent such a lot of money on the upgrades and I have to sell the finished product - parts and labour - to the expectant client. It's certainly not the way I prefer to do things.

[@richardkelsch3640]

The only way to get all of the bells and whistles to work on an Atari 8 bit, is to completely redesign the circuit board of the computer, chip locations and its clock bus. The 1088 doesn't do it properly. It just reproduces the bad original design. Personally, something along the lines of what a ZX Spectrum Next is for Sinclair folks. A completely modern system with legacy interfaces. Place the entirety of the computer on an FPGA and include things like PokeyMax, UM1B, etc right in the core. Give it a parallel bus, SIO as well as joystick ports. Slapping a V8 in a Model T isn't exactly a stable idea, and that is what is happening here. The Atari circuit design was only stable for the chips on the board and nothing else. Everything else is a crap shoot. The stock configuration is barely stable enough to pass QA. It's full of reflections, over and under-shoots, poor clocking and "it's enough to sell" design. It only needed cartridges and SIO to work, and they did. Also, NTSC has an overall faster buss speed than PAL systems. Everything is derived from the system clock and to get 60Hz etc, the primary clock needs to be faster. This means all of the timing is slightly less forgiving on a NTSC system than a PAL system. Anyone developing for NTSC needs to have an actual NTSC system to develop on and not just guess.

[@flashjazzcat]

I broadly agree with all of this, with the caveat that the designer of Rapidus appear to be aware of measures which could be taken to improve the situation if only he had the time and/or inclination and/or some disagreements regarding intellectual property could be overcome. Every device designed for these machines has to account for all of the obstacles you describe above, but some make a better fist of things than others (the featured system working perfectly when the single upgrade pushing things over the edge is removed). Hard agree regarding NTSC, too, since testing on NTSC machines often seems to be an afterthought where upgrades originating from mainland Europe are concerned. We've just had an NTSC anomaly with VBXE which took fifteen years to become apparent, but which could have been observed during testing with a simple software tool or test harness. Likewise, Rapidus' 'U1MB support' was apparently coded up without ever being tested with a real U1MB. I'm perfectly happy with my U1MB/VBXE/Stereo Pokey/SIDE3 XLs and XEs, and certainly am not inclined to throw my weight into further investigation of a problem I first identified some eight years ago and which has yet to trigger a firmware update.

[@TheWoj76]

Oh, and now also seeing it - it is an NTSC machine with a PAL VBXE palette, isn't it? And second, in the end you did not try the VBXE vertical NTSC alignment test?

[@flashjazzcat]

Yes, but when I update the U1MB plugin, NTSC palette can be silently soft-loaded at power-up. I didn't try the alignment test, no. What is it meant to establish?

[@TheWoj76]

That VBXE in NTSC mode has its all business shifted down by one scanline with respect to what the documentation suggests and to what Avery implemented in Altirra, and what made me implement a key activated workaround in my game in case this is not a universal rule / feature.

[@flashjazzcat]

@@TheWoj76 Ah yes - I am familiar with the NTSC overlay bug, but I assumed it was determinate and thus guaranteed to exist on all NTSC installs. Now you mention it, though, I should try the game on this machine just so I can experience the issue first-hand. Moreover, I was going to run the test the other week before I got side-tracked by massive problems running an XEX on the machine before I put 74F08 in it. :D

[@TheWoj76]

The game should actually run fine, if the bug is consistent, which I assumed it is, and you seem to confirm it.

[@flashjazzcat]

@@TheWoj76 I'll test it later and report back. Thanks for reminding me!

[@ndmmt-wu7kz]

Is there a way to obtain hdmi output and/or useful 80 column graphics output?

[@flashjazzcat]

Yes: VBXE has hardware 80 column text and 640x200 graphics. HDMI output will require a scan converter, however.

[@rafaelmmartinez9170]

$800? On upgrades ? Makes no sense

[@flashjazzcat]

In this case the client requested the most powerful machine possible, and the vendor therefore recommended everything but the kitchen sink. For me, the sweet spot is Ultimate 1MB, VBXE, and maybe a stereo POKEY board. A 20MHz 65C816 is not part of the appeal of the 8-bit Atari as far as I'm concerned, but to each their own.