[Coco] DW4 problems

[Aaron Wolfe]

It seems to me that a problem related to 50 vs 60 would not go away or
change when we make an adjustment in the server code or in the
server's serial driver, but those are the variables we've adjusted and
we have seen remarkable if only temporary changes.  You may want to
unplug and replug the serial adapter and see if your FIFO change
suddenly is undone.  I had a wonky adapter that would lose it's
settings sometimes, never figured out why but the un/replug trick
always seemed to sync up what was shown on the screen with what the
adapter was doing.

Do you have another serial adapter available even if only for a quick
test?  Also have you tried removing all other devices on the USB
controller the serial port is using?  Some motherboards only have a
single USB controller so that might mean removing everything (HUD
devices like KB and mouse should not be an issue).  I forget who
reported it, but some storage devices can routinely lock up a usb for
a few hundred ms.

Another sort of similar problem was recently solved simply by moving a
serial cable away from other cables and/or crossing them at right
angles rather than running them parallel to other cables.  EMI sort of
thing, caused data corruption during long transfers.

There are extensive diagnostic logging options that might reveal
something, but I'd really like to do everything we can to eliminate
cabling and the serial adapter first, as those seem to be a more
common source of trouble than I realized.

On Mon, Oct 22, 2012 at 8:59 PM, Bob Devries <devries.bob at gmail.com> wrote:
> Continuing the DW4 saga....
>
> Moving to version 4.3.0 seemed at first to have at least lessened the
> problem; I did a DSAVE last night, with very few ERROR 245's.
> On Aaron's suggestion, I removed the FIFO buffer in the driver for my ATEN
> (Prolific) USB dongle. That seemed to work wonders last night, but this
> morning, it's back to square one.
> I can do a MEGAREAD </X2, and get no errors. The speed result was 2:02,
> where my H0 (Quantum SCSI using SCSISYS V2) gets 0:16 !!
> The IDENT for the DW modules is as follows:
>
> 4 $E1 $02AF3D . rbdw
> 22 $21 $8435E1 . dwio
> 2 $E1 $F66B73 . scdwv
> 1 $E1 $41B964 . scdwp
>
> I'm unsure which CLOCK2 module this is:
>
> 2 $21 $D173CB . Clock2
>
> Can anyone tell me? I'm concerned that timing differences between the NTSC
> and PAL models may be causing this or at least worsening it, and I wonder if
> the CLOCK2 module could be part of the problem.
>
> The PAL Coco3's main crystal is 28.475MHz. I don't know what the E and Q
> clock lines run at; but the technical manual *may* tell me that.
>
> @Aaron:
> Is there some sort of test programme we can cobble together to try to chase
> this problem down?
>
> I have not yet shortened my coco3 DW cable, but based on Gene's comments, it
> may not help.
>
> Comments?
>
>
> Regards, Bob Devries
> Dalby, QLD, Australia
>
>
>
> --
> Coco mailing list
> Coco at maltedmedia.com
> http://five.pairlist.net/mailman/listinfo/coco