ClausB

PM sent.

No, RAMbo and Wizztronics copied my public domain design. RAMbo added some jumpers for the 1200XL. Newell disallowed banks 0-3 but their circuit board was large. (Their solution wasn't perfect - they just remapped 0-3 to other extended banks. You still had 12 banks and you had to know which ones to use.) Every computer project is a mix of hardware and software. As designer you get to choose where to draw the line - what gets handled by hardware and what by software. I followed the Woz way and put the minimum on hardware. I'm done defending my design. Instead of criticizing it, why don't you criticize the software that doesn't follow the clearly written and well distributed rules.

Fun! I used telephone wire myself. Good and stiff, easy to push into vias and solder.

You make good points. Thank you for adding that support. Of course, in the day, we didn't have a whole replacement OS, so one way to patch it was to copy it into underlying RAM, switch in the RAM, and load the patch. I could do all that without having to disable it. If a 16K cart is installed, then the underlying RAM can't be used by the system, so it could be used as extended RAM. In all my documentation I was careful to call it 192K extended RAM. I wrote that bank numbers range from 4 to 15. I made no mention of the side effects of setting banks 0 to 3. I left that for the more astute users to discover, although my driver does use it. It's really a side effect of incomplete decoding. I simplified the hardware as much as I could and left it to software and documentation to use it properly (I had studied Woz's work and followed his strategy). It's quite analogous to undocumented opcodes, which usually resulted from incomplete decoding in the CPU done in order to save gates. If you understand all their side effects you can use them to good purpose.

Another virtue is being able to copy the OS ROM to the underlying RAM without switching out the ROM. I think my driver did that.

Well, since mine was the first XL RAM expansion, even before RAMbo sold it, it should have been supported by all drivers since then. It's really not that complicated.

I don't see what's wrong with bumping an old thread. Often it's interesting to read old posts - lots of good info and opinions from past members. Also I prefer to have everything related to one topic in one thread, rather than scattered over many threads, some of which are just rehashes.

Wow, that was hostile! In case you didn't notice, that same post has the RAMdisk driver which has no trouble using the 192K extended RAM to emulate a 180K drive or two 90K drives without stomping on the system RAM. Also, a rat would be quite unhappy with a nest made of just 6 wires, 5 of which go to the same place.

Were there any ROM carts in the left or right cart slots work you tested? If not, did it go to memo pad when it worked? I would try BASIC in the left cart slot and then test each RAM board/cartridge by itself in the slot behind the 10K ROM. If it comes up READY, then type ?FRE(0) and Enter. If it reports 13 or 14 thousand-ish then that RAM is likely OK.

Just measured mine. 60 Ohms DC. Not sure what the AC impedance is.

Damn you're quick. Before I could edit my mistake! Guess I need a beer.

Try it again, BEFORE drinking the beer.

Some of it is available for use. CC00-CFFF holds the international character set and can be overwritten without crashing the OS.

Teensy boards are great! Easy to interface, easy to program, fast, cheap, and lots of memory. 5V tolerance makes them ideal for retro computers. I've done two projects with them, have one in the works, and more planned. https://www.pjrc.com/teensy/techspecs.html

Not exactly. I wrote a fast multiplier but did not alter the add, subtract, and divide routines.

Where did those fast FP routines come from?

Here's a schematic of Atari's 48K board: http://atariage.com/forums/topic/185500-using-the-atari-400-48k-board-in-the-800/?do=findComment&comment=2335665 It needs 4 wires soldered to the motherboard (S5, RD5, S4, RD4) and I can't think of a way to do it with only 2.

My local salesman knew what a deal the 400 was and let me in on it, even though it was earlier (12/80) and more expensive ($600) and had less RAM (8K). It was the Educator package, though they agreed to swap out the educational software for a Star Raiders cart. Were those extra few months worth $200? Yep.

I enjoyed the demo at VCFMW. I had no idea this existed back in the day even though I worked with CDCs and Cybers at MSU from 1977 to 1982, and I followed Artwick's FS work. Cool that it's being preserved. Keep up the good work.

Because I enjoy the challenge of repair (and I'm cheap), I have tended toward the low price untested lots, and have lucked into working units more often than not. But there seem to be fewer of those these days on eBay and Craigslist, or else I'm looking less often.

This thread explains how the 800 uses 8K and 16K RAM modules. http://atariage.com/forums/topic/177885-atari-800-ram-selection/page-1?do=findComment&comment=2220617

I would guess they are rare, as many were converted to 16K in the day.

Rybags is right. RD5 is ignored outside the 8K left cart space. Yes, it would need to be buffered, similarly to this: http://atariage.com/forums/topic/122750-64k-atari-800/page-1?do=findComment&comment=1483565

The logo on the white edge of that 32K board looks like a stylized BMS. Anyone heard of them?

I have to laugh at the box art. "Intec, used in the Space Shuttle!" Or "mpc for your Atari (which happens to look like an Apple II)!"