Hey, hey; an expanded XR4 for CP/M was my idea! ;-) Yeah, it was great to finally meet Phil in person; really sorry that meeting up with you couldn't work out.
I think EXTRAM (and presumably also XR4) used RAMRST, at least in the M100; where is this Vnicad you speak of? If you mean Vb, I don't think it's available on the bus? As I discussed with Phil I was actually working on an EXTRAM clone (although not specifically for CP/M) and exploring how XR4 did its bank switching; do you have any specifics? m ----- Original Message ----- From: Stephen Adolph To: Model 100 Discussion Sent: Wednesday, October 25, 2017 5:40 PM Subject: [M100] new project I met up with Phil Avery recently in person, which was a real hoot. In the process I got a front row demo of a T102 running CPM. Very cool! Now, this T102 was special as it was equipped with a Remem - which provides very flexible flash/ram storage. Specifically, 4MB of flash and 2MB of SRAM. What's important here, is that that big pile of SRAM makes for 2 very fast RAM disks for CPM. Phil and I discussed the challenge of how to make CPM more obtainable. Remem was cool, but never easy to install and awful to build. Keeping the serial port free is also nice as it allows for the link to the outside world. So, where we landed was that an all-SRAM REX could make CPM more achievable, as it would provide not only ram in the critical 0000-7FFF memory space, but also supply ramdisk via bank switching in/out of the optrom bank. Large SRAM chips (meaning 1MB or bigger) tend to be 3.3V IE they need to be buffered to use them in M100-land. Anyhow, long story short, I am awaiting 3 "REXCPM" board now, which look a lot like REX except with a 2MB SRAM chip, extra buffer chips, and 3 wires that need hookup to the M100. Kinda like (exactly like) a mega-EXTRAM. Back in the day, there was a product called EXTRAM that put RAM in the optrom socket - 32kB. Then there was XR4, which was EXTRAM x 4 or 128kB in the optrom socket. XR4 used the IO/M signal to allow PORT based bank selection. Here, REX is instead listening for an unlock sequence on the address bus to enable bank selection. I think EXTRAM used 2 wires - Vnicad, /WR I think XR4 used 3 wires - Vnicad, IO/M and /WR. REXCPM plan is to use 3 wires - Vnicad, /WR and RAMRST. RAMRST may or may not be needed. That signal is intended to protect memory during power down. Perhaps the developer of XR4 found an alternative way to protect the memory. Anyhow if I can eliminate RAMRST I will. Anyone have any thoughts on the subject? In practice this would install the same way as REX2- In M100 - 3 wires over to the system bus socket in T102 - 3 wires over to an adjacent RAM chip in NEC - 3 wires over to a memory module. Losing power would wipe the card..so that's a big difference from REX. Having said that, SRAM does have advantages, and in principle one could have many of the same REX features working on REXCPM - memory backup, option roms etc. anyhow, that's the update. Steve
