[Coco] deluxe rs232 pak / MPI

[Aaron Wolfe]

On Tue, Nov 24, 2009 at 9:09 PM, Gene Heskett <gene.heskett at verizon.net> wrote:
> On Tuesday 24 November 2009, Aaron Wolfe wrote:
>>On Tue, Nov 24, 2009 at 7:30 PM, Gene Heskett <gene.heskett at verizon.net>
> wrote:
>>> On Tuesday 24 November 2009, Aaron Wolfe wrote:
>>>>Hi,
>>>>
>>>>I just acquired an rs232 pak and MPI.  The MPI seems to work fine with
>>>>my coco 3, at least my drivewire rom pak works fine through it.
>>>>I haven't been able to get the rs232 pak to work in basic or nitros9,
>>>>with or without the MPI.  I have the original documentation from
>>>>tandy, was wondering if things have changed for use in the coco 3.
>>>>Also I have the 6309 instead of 6809, maybe this has some effect.  The
>>>>MPI is 26-3024 "updated for use with the coco3".
>>>>
>>>>the docs say to do "EXEC &HC000" or "EXEC 49152" to access the setup
>>>>routine in the pak's rom.  This results in various lockups on my coco
>>>>3.
>>>
>>> The slot switch must be to the slot the pack is in at powerup unless the
>>> MPI has been jumpered.  If it has been jumpered, then the packs eprom
>>> should be pulled as there are other packs you may have plugged in that
>>> will also have that as an exec address.
>>
>>hmm.  I can't get the EXEC to work even when the rs232 pak is directly
>>connected to the CoCo, no MPI.  Could this mean a bad rs232 pak?
>>
> Good question.  Unforch I don't have a pat answer. I believe someone else
> answered with an exec &HE010 to put it back in the rom only mode first.
>

yes, this allowed me to access the rom's setup program.. somehow I
missed those answers before.

>>I think I have to have the MPI's switch set to the slot my drivewire
>>rompak is in so that the HDB DOS rom loads.
>>Does this mean I cannot use the rs232 pak as the same time, or maybe
>>this just means I must wire the IRQ lines together?
>>
>>>>looking through the nitros9 source I see that FF68 is base address for
>>>>T2, and FF6C for T3.  Tried both with basically the same results.  I
>>>>see there is a compiler flag for MPI usage, it is set.  I am not able
>>>>to try nitros9 without the MPI because currently I must boot from
>>>>drivewire  (really looking forward to that SD card pak, if you're
>>>>listening Roger :).  the rs232 pak does not appear to have been opened
>>>>for any modifications, label covering the screw was intact until I
>>>>opened it just now to look for obvious bad components.
>>>>
>>>>when I "iniz /t2" or /t3, sometimes I get error 246, sometimes it
>>>>works.  when it works, an irq routine shows up in the "irqs" command
>>>>and the device is listed in "devs".  however, all attempts to pass
>>>>data do not seem to work.  sometimes the coco locks up about the same
>>>>time data is sent into the port, so I assume this means something
>>>>somewhere is working, sort of?  might just be coincidence to be
>>>>honest.
>>>>
>>>>should I put the rs232 pak in a specific slot of the MPI?
>>>
>>> I believe that most folks put it in the front slot, aka slot 1, byte
>>> $FF7F=00.
>>
>>is there a setting in NitrOS-9 to tell it to set FF7F to 00?  Would
>>this break the drivewire rompak in another slot?  (this rom is how I
>>am booting nitros9)
>
> Nitros9 normally maintains $FF7F at $33 in order to access the floppy
> controller in the back slot.  The convention for IRQ service routines is to
> read it, save it, and set it for the slot that this IRQ service routine is
> preset to access its hardware at.  And when the IRQ has been serviced, put
> the saved value back into $FF7F.
>

The driver must assume slot 1 for the rs232 pak and manage $FF7F
correctly, as it is working fine there now while running off the
drivewire for disks.  I'll have to take a closer look at the driver to
figure out how/where it's doing this.

>>>>do i need to wire all of the interrupt lines in the MPI together to
>>>>make this work?
>>>
>>> Mine has been so wired for about 22 years, excellent idea if for no other
>>> reason than all IRQ's _do_ get through to the cpu.  In your case, it
>>> sounds like the IRQ from data rx'd may have gotten through, but the slot
>>> switch, or the logic at $FF7F, is preventing the IRQ service routine from
>>> accessing the pack, which in turn clears the IRQ, so you effectively have
>>> an IRQ stuck on.
>
> And I forgot that doing so also parallels all the pullup resistors, one for
> each slots pin 8, presenting the device attempting to pull down the IRQ line
> with perhaps more current to sink than it can do comfortably.  There are 4
> resistors along the front edge of the pcb in the MPI that are all alike, and
> may be as low as 2700 ohms each, I've seen 4700's and 2700's there.  It is
> safe to break out 3 of them, there by reducing the current the IRQ asserting
> transistor has to sink to do its job.
>

sounds tricky for a software guy like me..  my soldering will make you
laugh (or cry).

>>>
>>>>saw some posts about that.
>>>>any tests I can do in basic, with the rs232 pak directly connected?  I
>>>>
>>>>saw a post about peeking values, they said:
>>>>> ?PEEK(65384) usually reports 0 initially
>>>>> ?PEEK(65385) usually reports 16 initially
>>>>> ?PEEK(65386) and ?PEEK(65387) should report 0 initially
>>>>
>>>>the answer in the thread was that the user doing the troubleshooting got:
>>>>
>>>>
>>>>?PEEK(65384) 0
>>>>?PEEK(65385) 112
>>>>?PEEK(65386) 0
>>>>?PEEK(65387) 0
>>>>
>>>>I also get exactly these results, if that means anything.
>>>>
>>>>Basically I am not sure if the rs232 pak works, it was in storage for
>>>>many years.  I am completely new to using an rs232 pak or an MPI, and
>>>>actually new to OS-9 as well, so there are a huge number of unknowns
>>>>here.  any hints or recommended practices are much appreciated. I'm
>>>>not sure what tests to do to start eliminating possibilities.
>>>
>>> There is a possibility that the Aztec inverter supply in the pack has
>>> gone to that great retirement pasture.  Since you have the MPI, you have
>>> the plus and minus 12 volt supplies the chips need available in the mpi.
>>>  You have to remove the Aztec assembly, and wire the cards edge connector
>>> to feed those voltages in where the Aztec was feeding them.  ISTR I had
>>> to superglue a couple of strips of thin copper plate into the blank
>>> spaces to do that. PITA, but it works for me, and has been for at least a
>>> decade.
>>>
>>> That rewired pack will NOT work in a coco3, as the coco3 has no -12 volt
>>> line at all, and the load on the +12 line may be the straw that writes
>>> fini to the coco3's power transformer if its not actively cooled.
>>
>>I guess that option is out, I am using a coco3.   But does that mean I
>>could never use the rs232 pak on a coco3, since the coco3 have no -12v
>>line?  sorry if I'm being dense.
>
> If the aztec supply is pulled, then it can never be used in the coco3, only
> in the mpi.

ok makes sense.  I think I'm lucky to avoid this issue for now.
thanks for all the advice!

>>
>>>>-Aaron
>>>>
>>>>--
>>>>Coco mailing list
>>>>Coco at maltedmedia.com
>>>>http://five.pairlist.net/mailman/listinfo/coco
>>>
>>> --
>>> 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)
>>> The NRA is offering FREE Associate memberships to anyone who wants them.
>>> <https://www.nrahq.org/nrabonus/accept-membership.asp>
>>>
>>> Oh don't the days seem lank and long
>>>        When all goes right and none goes wrong,
>>> And isn't your life extremely flat
>>>        With nothing whatever to grumble at!
>>>
>>> --
>>> Coco mailing list
>>> Coco at maltedmedia.com
>>> http://five.pairlist.net/mailman/listinfo/coco
>>
>>--
>>Coco mailing list
>>Coco at maltedmedia.com
>>http://five.pairlist.net/mailman/listinfo/coco
>>
>
>
> --
> 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)
> The NRA is offering FREE Associate memberships to anyone who wants them.
> <https://www.nrahq.org/nrabonus/accept-membership.asp>
>
> The nation that controls magnetism controls the universe.
>                -- Chester Gould/Dick Tracy
>
> --
> Coco mailing list
> Coco at maltedmedia.com
> http://five.pairlist.net/mailman/listinfo/coco
>