I have modified the concept layout. Here it is:



The standoffs are threaded and can be stacked. There are many different length's available to support different height requirements. They also come in brass, aluminum, or nylon.

Here is a picture of a 5/8" nylon one:



Using the ribbon cable allows for variable-height boards and is cheaper than using a backplane.

Each daughter board has a 32 byte window and a dedicated IRQ. The main board will decode the address space and provide an IRQ flag register for the 4 daughter boards along with its own IRQ for the RS-232 channels.

There will be two main boards to start with. One will have a simple 65C02, 32k RAM, 32k EEPROM, and RS-232. The other will have a 65816 supporting up to 4.5MB of RAM, and using a microcontroller to preload the RAM from either a serial flash/EEPROM or purhaps an SD card. The serial memory will be accessible via a secondary RS-232 channel.

User IO for both modules will be through the primary RS-232 channel. This will have a Pololu RS-232-to-USB bridge with drivers for Windows and Linux.

Daughter modules will include dual VIA's, dual 65SPI's, Video and keyboard IO, and sound.

I plan to get the proper spec's for these out within the month. I'm still trying to get my website updated.... wish my work didn't interfere with my play so much!

Daryl

The bus that the ribbon cable segments is going to extend to the daughter boards: is that the MPU address and data lines, or a separately driven bus a la PCI?

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

It will be driven by the CPU. By limiting the number of boards to 4, I hope to keep the bus length under control enough to use an 8 MHz clock.

I will test it fully loaded to see what the max clock rate is.

Daryl

I have a feeling you're going to run into trouble with it.

Years ago, I was working with an SBC design in which the address and data lines were extended to a receptacle on the board, into which an IDC plug could be inserted. I was playing around with a crude SCSI host adapter that connected to this "expansion" port and quickly discovered that the system would go awry with more than about 6 inches of ribbon cable. Ribbon cable has quite a bit of distributed capacitance, which may sabotage timing and prevent your system from working.

If you decide to go the ribbon cable route it might be worthwhile to alternate between signals and ground so no two signals are adjacent in the cable. This was how it was possible to achieve high speeds on single-ended SCSI buses with flat ribbon cable (before the development of LVD). Even then, you may not have enough drive to achieve a stable bus. It'll take experimentation.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

Add in bus driver chips like 245s I think and also resistance blocks at each end chain to prevent reflections....

_________________
"My biggest dream in life? Building black plywood Habitrails"