[Coco] 6809 to 6309 switcher

[RETRO Innovations]

On 11/21/2020 9:23 AM, Joe Schutts via Coco wrote:
> Hi Everyone,
> The problem was that it would NOT WORK with the 2 Meg memory upgrade that is available for the CoCo 3. I contacted the person and asked him IF the circuit board was STILL AVAILABLE and he told me that it was available (for a cheap price). I also asked him about the problems with the 2 Meg memory upgrades and he told me that (as far as he knew) there were no problems. He then asked me WHO it was who told me that there was a problem with the 2 Meg upgrade and I told him. He than told me that he would check into it (I guess contact whomever had told me that there was a problem) to see what the problem was.
> So far he has NOT gotten back in touch with me and I DO NOT know what (or even if) the problem still exists...

Hmm, I thought I responded, but in case it went to spam:

Yes, I did figure out the issue, and there is a problem (well, calling 
it a problem is overdoing it.  I'd probably call it it a lack of 
functionality).  The initial design was engineered to support 2 CPU ICs, 
any combination of 6809E and 6309E DIP ICs.

But,

The way it does that is to take advantage of 2 things about the CoCo and 
it's use of the 6809E:

1) The 6X09E can be effectively stopped by bringing the TSC line high 
(Tri State Control).  Bringing the line high causes the CPU to 
essentially go into a holding pattern.  As well, many of the signals 
lines on the 6X09E at (as the name implies) tristated, which means they 
are effectively hidden.

2) There are, though, some lines are are not hidden, various control 
lines called BA, BS, etc. (Bus This and Bus That, so to speak).  
*HOWEVER*, the CoCo does not sue any of those lines, so they can be 
effectively disconnected from the CoCo without any ill effect.

It's this second item that is the cause of the issue for some memory 
expansions.  While the CoCo does not need those bus control lines, the 
memory expansion does.

So, where's the problem?  Well, there is no problem if you consider the 
MMU/DAT portion of these memory expansions as part of the CPU assembly.  
If the CPU is attached directly to the DAT/MMU portion of the memory 
expansion, and then that combination is then installed in the Dual 6X09 
adapter I sell, there is no issue.  The CPU provides the control signals 
to the DAT/MMU board, and then DAT/MMU board acts on them, and all is well.

*EXCEPT*

Using things this way would require the owner buy *2* DAT/MMU boards, 
one for each CPU, which is neither financially advantageous nor easy to 
technically implement (You'd have 2 DAT/MMU boards trying to control the 
memory, which I doubt will work).

SO,

The way people want to use the switcher is to:

Install the DAT/MMU board into the CoCo CPU socket

Install the switcher board in the DAT/MMU CPU socket

Install the 2 CPUs in the switcher CPU sockets.

But, as noted above, that means that the CPU's control signals will be 
sent to the switcher board, which.... does not connect them to the 
combined CPU header.

And thus the failure.

The good news is, the solution is technically simple: tristate those 
signals manually and pass a set of them onto the CPU header, depending 
on which CPU is turned "on".

The bad news is, that required a redo of the PCB and some testing, which 
has not completed yet.

Now everyone knows more about this than they ever wanted to :-)

> Sorry that I cannot be more helpful about this.
> P.S. The gentleman did tell me that (as far as he knew) the circuit board DID WORK on the 512 K upgrade. Also, the circuit board comes as a plain board that you solder the sockets (or the chips) to the board...
I can confirm the "passive" switcher works in this situation.
> Hope this helps...
> Joe...
>
Jim

-- 
RETRO Innovations, Contemporary Gear for Classic Systems
www.go4retro.com
store.go4retro.com