ET-3400 ROM - Feature Creep

I got  most of the PCB artwork done in accordance with my original plan, and then I thought perhaps I could add an extra feature that I hadn't planned originally.

At the moment I have an 8K EEPROM and using one of two 1K banks selectable by a jumper. I am wasting 6K out of the 8K.

I had a look at the memory map for the ET-3400 and it seems that there is no memory use between $C200 and $FBFF below the monitor ROM. Also the add-on ETA-3400 Memory I/O Accessory unit doesn't make use of any of that space.

The other project I am working on (that big mess of breadboards and wires in my earlier pictures) is a Memory I/O Accessory unit that does use all of that space, but not everyone will have that configuration and may only have the stock ET-3400 unit or the stock Memory I/O unit.

There is sufficient space in this EEPROM to have 8K of ROM from $E000..$FFFF if you don't need to have two ROM versions. Wouldn't it be nice to be able to configure the board to make use of that space, or have it as originally planned with two ROM versions. Then it would be up to the user to determine what configuration they wanted.

This is what is known as Feature Creep, where extra features are added part way through a project. Sometimes this can get in the way of finishing a project or getting away from the original intent, but I think I have plenty of time to at least consider it. If it doesn't look like it will work I can fall back to my original plan.

So now my project has two goals:

    1. make a small PCB ROM replacement board with two selectable 1K ROM versions, and

    2. allow for an alternate configuration where the top 8K of memory can be used as ROM.

First thing to consider is whether the /CS1 line from the motherboard will allow me to use the upper 8K because that's a show stopper otherwise.

Page 87 of the ET-3400 manual shows that IC12 is decoded by A15, A14, A13, A12, A11 and A10 (see table below). That places the ROM in 1K of space between $FC00 and $FFFF. The first three address lines A15..A13 are actually done on the motherboard by IC2 to produce the /CS1 signal which means any address from $E000..$FFFF (8K) is done on the motherboard. The extra address lines A12, A11 and A10 are the three signals decoded by the original mask ROM itself to bring the range down to $FC00..$FFFF (1K).

That's good news because it means if we ignore the 3 address lines from A12..A10 then the /CS1 signal will be active and can select any of the upper 8K memory space.

Show stopper avoided !

The tricky part will be trying to fit any changes in the remaining space and I haven't got a lot of real estate anyway.

Using my other ROM Replacement design I found that I needed to make the vias smaller on this design so I checked the PCB manufacturers web site (JLCPCB) to find out how small I could make them and I can go down to half the size of my previous design. This gives me more wriggle room.

Will there be enough room for the extra features, only trial and error will tell...


Comments

Post a Comment

Popular posts from this blog

ET-3400 ROM - Message Program Assembly

I intended to place this programming test in the new 8K ROM section of the ROM replacement board, however the program I was copying from (out of the ET-3400 Manual) was self-modifying and needed a rewrite so that it would run in ROM and use the RAM in the lower 512 bytes for variables. After I got into it I thought it would be relatively easy, but I was assuming that the 6800 CPU programming was similar to 6502 CPU programming since the 6502 was modelled on the 6800. However it turns out that was a false assumption, and my programming experience on the 6800 wasn't up to speed. The 6800 doesn't have indirect addressing mode and the program was self modifying so that it could do a sort-of indirect addressing. Also there were no opcodes to transfer the X register to the ACCA or ACCB register or the stack as far as I could tell. It was going to take me a lot longer to learn enough basic 6800 code to get this done in time, so I reverted to editing and extending the program in it'
Read more

ET-3400 ROM - Schematic Draft

Image
I've done the schematic quickly in Design Explorer 99SE running under virtual Windows XP. This is an old program but free and it works on my old MacBook. The CS/CE logic needs to take the four CSx inputs into the original mask ROM and create a single active low /CE output to the EEPROM. In logic terms we need a logic 0 out only when (CS0 = 1) AND (CS1 = 0) AND (CS2 = 1) AND (CS3 = 1). For any other combination of inputs we need a logic 1 out so that the EEPROM is NOT selected. Since 3 of the 4 signals (CS0, CS2 and CS3) are normally active high we could push them through a 3-input NAND gate (74LS10) that will only go low out when all inputs are high. but we still have to deal with the CS1 input which is active low, so we'd have to mix the output of the 3-input NAND gate by pushing that through a 2-input NOR gate with the CS1 signal to create an active high out of that, then invert that again to create an active low. That would involve at least 2 TTL ICs and will make the PCB l
Read more

ET-3400 ROM - Programming gains and losses

To write the program I wanted I was going to use ideas from the examples in the ET-3400 manual to help me create my little demo program. I also needed an assembler for 6800 code.  Turns out both of those tasks had already been mostly done previously and I had forgotten about them. I couldn't find a  suitable free assembler for Windows. I had used the Task Assembler (aka TASC) in DOS in the distant past for some Z80 code, and I think it could also assemble for different processors. Anyway that was so long ago and I couldn't spend the time looking for it. I also couldn't find a version of TASC on the www that I could access. All my Google searches pointed to CRASM which was for Linux and that rang a bell... I rebooted into my Ubuntu partition and it turns out that I had already installed CRASM probably last year and had forgotten about it. Good, one down, next... I was looking through the ET-3400 example programs and it turns out that Example 4 on page 61-62 was virtually w
Read more