[Coco] An 8-Slot Multi-Pak Design for my Color Computer Friends

[jdaggett at gate.net]

On 5 Jul 2010 at 9:05, Little John (GIMEchip.com) wrote:

> Had to have a fourth of July Holiday, so now I'm back. Let's have a look 
> see:
> 
> > First
> >
> > A huge mistake in not buffering the data bus on the main board. If you are 
> > running a stock
> > Coco3, you do run the risk of damaging the MC6809E and the HD6309E if the 
> > board
> > becomes detached while powered up.
> >
> That's easy enough to fix. However, even with the databuffer on the main 
> board, detaching while powered up could cause damage, especially if it were 
> knocked sideways - zap.
> 

Yes that is true. The MC6809 does not like to drive large capacitive loads. The additional 
runner lenghts and connectors add capacitance that degrades noise immunity and potentially 
damaging current transients during witching. Frankly I am not surprised that the original MPI 
has more problems. The duffers should have been on the inside of the Coco and not on the 
MPI side. 


> > Second
> >
> > why the dip switches on the Control and clock lines. A wrong setting on 
> > the dip switches and
> > you have a nonfunctioning board that can lead to frustration.
> >
> If I build it, I'll know what the DIP swicthes do. But again easily fixed - 
> jumper wires instead of the DIP switches if you don't want the DIP switches.
> 

True. Still the E clock line should feed directly to the slot connectors from the buffer. No 
jumpers or any switches. 

> > Third
> >
> > What is the function of Q1?
> >
> It is supposed to tell the ATX power supply to switch on when the CoCo 
> Powers on - this was stated in the text. However, this is pretty lame - A 
> reed relay might have been better or just a switch to turn the darned thing 
> on. Yeah - the switch probably would be best.
> 

Okay that makes sense. Did not realize that you are powering this off an ATX supply.

> > Fourth
> >
> > Your ground plane on the top and bottom are basically floating planes that 
> > are tied at but just
> > a few spots. You have all your locic tied to GND1 and then this large 
> > plane on both sides is
> > GND. You also have fingers that are floating that is bad for noise 
> > immunity. Either use one
> > ground or tie them in a star pattern and eliminate the floating fingers.
> >
> > just a few things I see wrong or bothersome.
> 
> I did use only one ground - the GND/1 is the Library Name for the Ground 
> Symbol that I have chosen - ALL Grounds in the schematic are electrically 
> tied to the GND net name. This can be seen if you load it up in EAGLE and 
> right click to get the properties of any GND connection - they are all tied 
> to GND.
> 

They are? Why then highlighting the plane states GND and the runner from runner CN2 is 
GND1. That means you have two grounds and they are tied together at some other point on 
the board. 

At least I would not do it that way. If they are supposed to be one then make them one. Not 
two and then use some etch someplace on the board to tie them togeter unless you intend to 
do a star connection. If tha twas your intent the star point should be where the board make 
ground connections to the outide world. Not internal to the board.

> If you find any more problems - fire away - it'll help me fix this right up. 
> I'm just a hobbyist - no formal training, so this design MUST be SCRUTINIZED 
> thoroughly in order to work out all of the bugs. -John
> 

If i find time to do a more in depth review I may. Right now I have more important things to 
do. 

james