This is quite true, but several of the timing + RAMDAC devices I've seen have the RAMDAC addresses externally accessible, IIRC. In reality, I guess, all he has to do is to attach an 18-bit-wide RAM (or three paralleled 256 x 8 SRAMs, or the equivalent, using only 6 of the bits) to his three 6-bit color channels.



Yes, they're VERY old. However, they don't require that much external logic. They require external registers to keep data movement synchronized with the pipeline around which the chip is designed. Since there's typically a delay for the refresh RAM, and a delay for the character generator or RAMDAC, an internal parameter can be set to compensate for that timing skew.

Actually, I believe they're rated for 2.5 MHz or greater in their later incarnations, with the 2 MHz limit mainly impacting the bus interface. However, since the pixel clock is always external to the 6845, his pixel rate won't outrun the MC6845. Since there are several "free" IP's for the 6845, and since he doesn't really need the entire 6845, I would guess he'd simply do that job in programmable logic, though, where his clock rate can go MUCH higher. The 6845 architecture can provide conceptual guidance, though. The bigger part of the 6845 is the microcontroller interface, which can be omitted in this case, since, as you say, it's a sequencer, and doesn't affect the data. IIRC, the datasheet for the 6845 indicates that what it expects to "see" as a clock is the "character clock" meaning that the pixel clock is divided by the number of pixels per character before being fed to the 6845.



A few of those EPSON devices might work, but will be difficult to implement in prototype, not that it's much easier with current-generation programmable devices.



Much depends on whether he wants a working display, as he might if he's doing this on his job, or if he wants to learn how to do this sort of thing. Because the type of device he's got is somewhat dated, I'd guess it's the latter. Only he can clear that up. Unfortunately, programmable logic is not generally sold in convenient-for-prototyping packaging these days. Perhaps he can find an appropriate part in PGA or PLCC-84 package. Prototyping sockets for current-generation parts tend to cost upwards of a week's pay, and the adapters to allow you to work with them cost as much again. Generally, unless you can afford to spend on the order of $15K on your device, you can't use 'em in the raw. The "easy" way is to buy an evaluation board that provides enough I/O's.

Can you suggest a commercial device that actually does this job on its own?

RE