That's pretty awesome! What's amazing is these controls were built, basically unchanged, from the mid 1970s until around 2012. They even brought it back after the MP32 disaster. I'm not sure how they did it since the 8086/8088 chips were discontinued in the late 90s. It's amazing how much all of these controls were the same. Mitsubishi, Yasnac, Fanuc, Fadal, Haas, Dynapath, etc all used the 8086 family processors with that card cage design. They all ran some version of DOS. I really think they would still be running that system if they could still get the chips and there was support for USB and ethernet on whatever ancient version of DOS they all ran.
Yeah, that is part of the allure of Fadal, the same basic design was used for their entire run. They upgraded to a 286 with the 1400-3 and a 386sx with the 1400-5, and 486DX4-100 with the 1400-6. They even built the "PCB-0346" which was an Nvidia M6117 embedded SoC PC104 computer on a card, replacing all 3 of the main logic cards. As I understand it, everything that came after the 1400-5 was not very reliable and basically hot garbage. AFAICT, they only would have used DOS perhaps with the M6117 control, since it was a PC architecture. The rest is all bespoke custom x86 assembler. I've done some reverse engineering of the 1400-1 software, but the addressing scheme of the 1400-2 is a little more challenging.
Hi P, this brings memories of the old days designing 6800 stuff in the 70's. You must be having fun since your attention span exceeds the minimum for success. I estimate gcode volumes of about half of your most complex dream parts will fit in the new memory. My 386 based controller holds double and I think I filled it once. I have a pair of twin test xrf devices that use a TI 8 bit with the same design simplicity, but the twin test devices are not as useful as your cnc so they will drift to obscurity, but not without thoughts similar. What up with the v10 truck?
Yeah, it was fun. This board was actually born out of illness; unfortunately I got sick a few times and so I started playing with KiCAD during one of these. I estimate that I probably have around 100 hours into the reverse engineering, design, and debugging of this memory expansion. If you consider that in context for back in the 80's, that's like 2 sprints worth of work in today's measurements. The V10 is a bit more involved and is just one in a bunch of projects I need to complete.
With your knoledge will be interesting if you ca make a card to convert encoder signal to resolver signal. To be able to replace that very old resolver technology.
You don't need to do that, the later cards have both resolver and encoder support. You have to make a choice, stick with the old tech or adopt the new tech. Encoders, fast CPU, and memory all come with a price of about $3800. Converting an encoder signal to a resolver signal would take some neat tricks and it wouldn't be cheap. I personally like resolvers, they are technically higher resolution than encoders. The major problem is that they wear out faster than encoders. For me, I had 3 rebuilt motors from 20 years ago, so my motors got a new lease on life.
No, it's not compatible with a -4 system. There is a company out there selling 384K boards for -4 systems for $250, you'll find them if you search Google for Fadal memory boards.
That's pretty awesome! What's amazing is these controls were built, basically unchanged, from the mid 1970s until around 2012. They even brought it back after the MP32 disaster. I'm not sure how they did it since the 8086/8088 chips were discontinued in the late 90s. It's amazing how much all of these controls were the same. Mitsubishi, Yasnac, Fanuc, Fadal, Haas, Dynapath, etc all used the 8086 family processors with that card cage design. They all ran some version of DOS. I really think they would still be running that system if they could still get the chips and there was support for USB and ethernet on whatever ancient version of DOS they all ran.
Yeah, that is part of the allure of Fadal, the same basic design was used for their entire run. They upgraded to a 286 with the 1400-3 and a 386sx with the 1400-5, and 486DX4-100 with the 1400-6. They even built the "PCB-0346" which was an Nvidia M6117 embedded SoC PC104 computer on a card, replacing all 3 of the main logic cards. As I understand it, everything that came after the 1400-5 was not very reliable and basically hot garbage. AFAICT, they only would have used DOS perhaps with the M6117 control, since it was a PC architecture. The rest is all bespoke custom x86 assembler. I've done some reverse engineering of the 1400-1 software, but the addressing scheme of the 1400-2 is a little more challenging.
Wow, fantastic effort! Thanks!
Very cool Perry. Thx for sharing!
Impressive work!
Hi P, this brings memories of the old days designing 6800 stuff in the 70's. You must be having fun since your attention span exceeds the minimum for success. I estimate gcode volumes of about half of your most complex dream parts will fit in the new memory. My 386 based controller holds double and I think I filled it once. I have a pair of twin test xrf devices that use a TI 8 bit with the same design simplicity, but the twin test devices are not as useful as your cnc so they will drift to obscurity, but not without thoughts similar. What up with the v10 truck?
Yeah, it was fun. This board was actually born out of illness; unfortunately I got sick a few times and so I started playing with KiCAD during one of these. I estimate that I probably have around 100 hours into the reverse engineering, design, and debugging of this memory expansion. If you consider that in context for back in the 80's, that's like 2 sprints worth of work in today's measurements. The V10 is a bit more involved and is just one in a bunch of projects I need to complete.
With your knoledge will be interesting if you ca make a card to convert encoder signal to resolver signal. To be able to replace that very old resolver technology.
You don't need to do that, the later cards have both resolver and encoder support. You have to make a choice, stick with the old tech or adopt the new tech. Encoders, fast CPU, and memory all come with a price of about $3800. Converting an encoder signal to a resolver signal would take some neat tricks and it wouldn't be cheap. I personally like resolvers, they are technically higher resolution than encoders. The major problem is that they wear out faster than encoders. For me, I had 3 rebuilt motors from 20 years ago, so my motors got a new lease on life.
Hi, you do a very nice work there. The expansion board is not compatible with -4c system?
No, it's not compatible with a -4 system. There is a company out there selling 384K boards for -4 systems for $250, you'll find them if you search Google for Fadal memory boards.