[Coco] MPI problem or ...???

[Gene Heskett]

On Saturday 23 June 2007, Brian Blake wrote:
>Gene,
>
>The card edges are all clean, and I've cleaned the switch is good as I could
> (alcohol) for now. Jacksonville is the nearest town to me that has anything
> other than a Radio Shack to buy electronics supply stuff and cleaner from,
> unless I place an order with Jameco or Mouser.
>
>Anyway, I used the 26-3124 manual for a little diagnostice last night,
> PEEKing at the value of 65407, and those all came back with the correct
> values for each slot. The next step was to start POKEing values into 65407
> and then PEEKing them to see if the value was the same. So, I used the ones
> listed in the manual first, 0, 255, 85 & 170. The 0 & 85 came back fine,
> the 255 & 170 did not. So I poked some more values in looking for where the
> screw up begins. It begins with value 168, which returns 160 when PEEK'ed.
> Every value POKE'd into 65407 larger than 167 returns a number 8 less
> thanwhat was POKE'd.

Trouble shooting by remote control is a pita.  First, I need to teach you how 
to visualize those numbers in hexidecimal, eg not 0-255, but $00-$FF.  Or 
maybe even plain binary because thats where you'll visualize the data line 
status best.

You are telling me that when an $A8 is written, an $A0 is read.  Thinking in 
binary, that $A8=10101000, where the right most 0 has a weight of 0 or 1.  
This is data bit d0.  The next 0 to the left has a value of 0 or 2 and is 
data bit d1, the 3rd one from the right has a value of 0 or 4 and is data bit 
d2, and the 4th one has a value of 0 or 8 and is data bit d3.  So it is data 
bit d3 that would appear to be stuck low.

Now, as you are incrementing your decimal number, there should be a pattern 
that will emerge from this that doesn't match what you are saying IF this is 
the case, because it should come back as 8 less for even the second 8 in the 
range of 0-15, the 8 to 15 should come back as 0 to 7, likewise it should be 
8 shy for 24-31, 40-47, 54-61, 72-79 yadda yadda.

There is more than one chip in the data path of the mpi that could fail, but 
I'm leaning toward the 74ls245 (u2) (since I know there is a turnaround data 
clash that draws considerable current in that chip already), however it 
appears that the 74ls367's also could be involved.

This would be a lot simpler if you had an oscilloscope and the training to use 
it.  Someone really adept at using a logic probe could probably find this 
also.

Anyway, how am I doing so far?

>Example;
>POKE 65407,170
>OK
>PRINT PEEK(65407)
>162
>OK
>
>I can't go any further in the process right now due to the MPI not even
> responding to the presence of a PAK being installed.Even a FDC still boots
> the the ECB screen. Any other pak, regardless of the slot it's in, causes
> it to boot only to a green screen.
>
>Brian
>
>----- Original Message ----
>From: Gene Heskett <gene.heskett at verizon.net>
>To: CoCoList for Color Computer Enthusiasts <coco at maltedmedia.com>
>Sent: Thursday, June 21, 2007 11:41:30 PM
>Subject: Re: [Coco] MPI problem or ...???
>
>On Thursday 21 June 2007, Brian Blake wrote:
>>John,
>>
>>Yes I do.
>>
>>Not having a service manual for this paricular unit I am currently thinking
>> one of the IC's is bad, just had not gotten the scope out to look at any
>> of the outputs. Just don't know where to start looking at this moment...
>>
>>George,
>>
>>Again, I haven't actually got the scope out to look, but, the DMM isn't
>> picking up anything greater than .001vac. I'm not thinking that's enough
>> to cause these problems, but, again, I've only recently got back to
>> CoCoing...
>>
>>
>>Thanks guys, keep the good info coming!!!!
>>
>>Brian  (<- not an expert ;) )
>
>A couple more things to check Brian since these things are nearly old enough
>to vote.
>
>1. clean up all the blackened coating on the card edge connector that plugs
>into the coco.  Get brutal, as in a small piece of 0000 steel wool and some
>contact cleaner & make it shine on both top and bottoom, and that includes
>those 'ears' on each side as they are the main ground connection between the
>coco and the mpi.  Just don't power it up till all the steel wool has been
>removed.  That black stuff is just as good an insulator as that same black
> on a battery post of your car battery.  I got so tired of that that I sawed
> it off, found a pc card with a gold plated finger pattern, and transplanted
> it. That very tedious job was 15 years ago, and haven't had any problems
> with that since.
>
>2.  If that mpi has not had the usual jumpering of all the card socket pin
> 8's together, there is a probable problem there because the slot selection
> switch has become oxidized and needs to be given a shot of cleaner and
> exercised. Lots of exercise, then a couple more shots of contact cleaner.
>
>Gene, a C.E.T.



-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Don't kiss an elephant on the lips today.