Inspiration from the ZX81 ------------------------- Jan 25, 2005 Subject: Re: [cosmacelf] 1802 ZX81 leealanhart Richard wrote: > I just had to throw and Idea in here. I was thinking it wouldn't > take a lot of work to create an 1802 based version of this machine. > The hard part would be to finding someone that could make the > keyboard layout for less than an arm. The rest could be done with > software. an we wouldn't have to have RF modulator because almost > all VCR's have video connections. It could have a CF port for storage. The ZX-81 was a classic Clive Sinclair "mad genius" product. Very *VERY* tricky and clever. The original ZX-80 had about 11 generic chips; the ZX-81 got this down to 4 chips with one being a simple custom part. The keyboard was just the PC board itself with a metalized mylar overlay; when you pressed, it flexed enough to short traces on the PC board. The video circuit was even simpler than the 1802/1861. It had a Z80, static bytewide RAM, and ROM. These parts worked normally to hold the program and data, just like any microcomputer. But, at the start of the top video scan line: - The Z80 did a JMP to *execute* the ASCII data to be displayed in the RAM. - But, the hardware opened the Z80's data bus, and pullup resistors jammed in a NOP instruction instead. - The Z80 executed 32 NOPs in the time it took for one scan line. - Therefore, the Z80's address bus incremented 32 times. - This read 32 consecutive bytes from the RAM (the characters to display on the first line). - The data from the RAM was patched to the *address* inputs of the ROM. They selected the bit pattern lookup table for that character. - The data from the ROM was latched into an 8-bit video shift register, and was shifted out to become video. The top row of dots for the characters is displayed. - The Z80 finishes its 32 NOPs at the end of the line, and continues with another 32 NOPs during the next scan line. Its lower address lines A0-A4 still go to the RAM, and scan the same 32 charcters again. But the Z80's A5-A7 lines keep climbing; they are the row counter and go to the ROM to select which row of dots gets output. - This scheme continued to generate 24 lines of 32 characters, each in an 8x8 dot matrix. - When the Z80's address finally counted to FFFF (end of RAM), it wrapped around to 0000 (the ROM), and normal program execution resumed again. So, like the 1802/1861, about half the Z80's execution time is taken up in supporting the video. But the Z80 is running at 3.7 MHz (about twice as fast), and it only needs 4 clock pulses per bus cycle (compared to 8 for the 1802). So, the ZX-81 gets 32 bytes per scan line compared to 8 for the 1802/1861. Nowdays, for an 1802-based "ZX-81", I think I'd lean toward an LCD display instead of a TV set. $10 will get you a small character-based LCD, and even a 240x64 pixel full graphics display is under $50 new. The interface circuit to drive it is trivial (basically just an 8-bit output latch). You'd need a ROM with software to operate it as a usable display. Jan 25, 2005 9:01 pm Subject: Re: [cosmacelf] Re: 1802 ZX81 leealanhart What always amazed me about Clive Sinclair is how accurately he could target a market, and the design a product ABSOLUTELY suited to hit it exactly. The Sinclair ZX-80/81 went from non-existent to the #1 selling computer in the world in just 4 months! He sold at the peak (to Timex), and then happily watched as the market changed (the target moved), and the ZX-81 was no longer on-target. mc71de wrote: > He never sold to Timex... after the ZX80 was mainly sold in kit form > (only a lower percentage was built by home workers, it is said) and he > could not fill the demand for a prebuilt unit, he contracted with > Timex to manufacture the ZX80 in their british plant (Scotland, i > think). Part of the deal was that Timex got the exclusive rights for > the US market (renaming the ZX81 to the Timex 1000). Timex was clever > enough to include 2K ofRAM, and by the time the Spectrum came out, > they backed out and sold their own design (Timex 2068) instead. Thinking a bit on this... There's no point in building yet another laptop, no matter what the CPU. It's been done to death. But, there *may* be a point to designing a little programmable thing (like the ELF). It isn't a "computer" (that word's been taken :-) but you could call it a "controller". Instead of a normal CPU, perhaps it contains an FPGA that implements some "CPU" that is very easy to program and use for a beginner -- kinda like the 1802. This would be in stark contrast to the PICs, ARMs, and all the other modern chips, which you essentially have to be a professional engineer to use. Maybe this FPGA directly executes Pittman's IL code, so it runs Tiny BASIC as its native language. This would make it easier for kids to program, yet 100 times faster. Since the IL code is extensible, you could add keywords or functions as needed. Like the BASIC stamps, you could add keywords to support LCD displays, hobby servo motors, etc. Jan 26, 2005 6:28 pm Subject: Re: [cosmacelf] Re: 1802 ZX81 leealanhart Remember I said "mad genius"? [Clive Sinclair] really is a genius. The ideas for the products he built were all great. There really *were* large markets for all the products he had designed: digital watch, pocket TV, digital multimeter, oscilloscope, scientific calculator, several computers, personal electric vehicle, etc. If he could produce a product at the price/feature point, he'd be rich! But he's also mad. He built these products *before* conventional wisdom said they were possible. To do it, he had to "pull a rabbit out of his hat" -- find some fiendishly clever invention or design or construction shortcut to do the impossible. "Uncle Clive" was very good at finding such tricks. However, normal design, manufacturing, and sales people often had a lot of trouble making his tricks work in the real world. The ideas weren't bad; in fact they were often downright brilliant. But the implementations were often horrible.