[Coco] Coco to PC cable

[Gene Heskett]

On Wednesday 11 March 2009, John W. Linville wrote:
>On Wed, Mar 11, 2009 at 12:18:31AM -0400, Gene Heskett wrote:
>> On Tuesday 10 March 2009, Jim Hickle wrote:
>> >--- On Tue, 3/10/09, Frank Pittel <fwp at deepthought.com> wrote:
>> >>While I believe
>> >>in competition I'm afraid that in this case it will divide the community
>> >>and in
>> >>the long run this will be a bad thing.
>> >
>> >Right.  It's bad enough that we right-thinking people, i.e., those with
>> > SCSI controllers, have to endure the disunity brought about by the
>> > unwashed, IDE-using rabble.
>>
>> Shhh, Jim.  We aren't supposed to let on about that. :)
>
>That is worth a chuckle, but in case it was intended as a serious
>point...
>
>It is one thing to require two drivers to enable using either kind of
>available hardware.  After all, most people aren't building their own
>hard drives, and there are reasons (including general availability)
>to prefer one type over the other on an individual basis.  (FWIW,
>I'd be happy to see a SATA adapter for the CoCo!)  But in that case
>you are adapting to an outside reality to bring the community together
>onto common platforms.

I agree, we really should be using one or the other.

But school is in session for a moment folks.  I get down and dirty here.

The legendarily fickle scsi interface doesn't have to be, if so darned many 
compromises weren't inherent in the average setup.
1) SCSI is a wired OR interface, built on the TTL signal level specs, eg 
anything below .60 volts is guaranteed to be a logic zero, and anything above 
2.4 volts is guaranteed to be a logic one.  In between, its a gray area.

And to remove as much cable ringing as can be done by terminating the ends 
(only) of the cable with something loosely representing the characteristic 
impedance of the cable itself was the general idea, it is in fact a 
transmission line and can have what the hams would call a high VSWR.

2)  That impedance runs, in the common flat ribbon cable with every other 
conductor in the ribbon being a ground, is about, give or take 10%, of 120 
ohms when viewed as a transmission line.  And since the specs allow for a 
nearly 40 meter cable, using logic devices that can respond to a 5ns noise 
pulse, that termination is critical.

3) Unforch the common passive terminator is a 220 ohm r connected to the 5 
volt line, in series with a 330 ohm r connected to ground. The logic line is 
connected to the junction of the 2 r's, and was INTENDED to sit at about 3.0 
volts even, which gives a 600 mv noise margin in the logic 1 state.

4) How ever, as an afterthought, they then added a diode to the top of the 
resistor network to prevent the sharing of the 5 volt logic supplies when the 
power to one or the other was turned off.  But nobody warned them the diode 
should be a very low voltage forward drop type, such as a schotkey.  So all 
the scsi cards built use a common, fairly high currant silicon diode, which 
brings the effective supply voltage down to the 4.3 to 4.4 volt range due to 
its forward drop.

5) Ok, so whats the resting logic 1 voltage now?  Given precise resistors, 
which will never happen as they are usually +-10% or more, the voltage is now 
2.61 volts.  That has reduced the noise margin from 600mv to 210mv, and the 
circuit begins to get flaky.  Add in that most power supplies age low, and are 
only making 4.88 volts when they are a year old, and there goes most of the 
remaining noise margin.

And guess what, then we are sacrificing virgins and hiring rain dancers to 
make it work.

OR, designing with active terms, which are effectively a very high powered op-
amp whose output is a regulated 3.0 volts, it can src or sink large currents 
as long as the average is low so it doesn't get too hot.  And all the term 
resistors are 120 ohms and tied to that regulated 3.0 volts.  This alleviates 
99% of the SCSI problems right there.

6) But IDE and ATAPI have essentially the same problems too, which is why the 
IDE interface specs say the Maximum cable length is 18".  And they have slowed 
the logic enough it mostly ignores the very closely spaced echos even such a 
short cable suffers from.

I think SATA is going to be around for a while, so any new drive interfaces 
should be SATA.  But that is another horse entirely I haven't even petted.  
But in the case of the coco's, that short IDE/ATAPI cable means you either 
repack the coco, or figure out how to park a hard drive very close to the mpi 
the interface is plugged into.

7) I see it as SCSI at least, was intended to be used with cables long enough 
to reach a drive in a coco usage situation. OTOH, with those teeny lappy 
drives at 100GB now, one of those could be attached to the interface card 
itself with a 3" cable and should work like a dream on that length of cable.

I could easily take the 40GB out of a 1 touch case and mount it directly on an 
ide interface, and I'd probably have the coco with the biggest hard drive in 
this group!  As is, its a now very elderly Seagate Hawk of 1GB.  It will do 
for me.

Test time is tomorrow afternoon. :-)

>OTOH, requiring different software (both on the CoCo and on your
>server) for two such similar services when they are both under
>active development within the community is IMHO a waste of talent
>and resources.  In this case you are forcing users to choose between
>platforms or to maintain near-duplicate environments.  It just seems
>unnecessary.
>
>Make no mistake, Roger and Boisy are entitled to the fruits of their
>own labor.  And everyone is entitled to support one or the other
>(or both) as they so choose.  I'm just saying that we would all be
>better off if these two projects could find a way to work together.

That would be the ideal situation.  But that is not our choice, its between 
the fellows with a dog in this 'fight'. :)

On a secondary note, the usb bluetooth keys have arrived, and I'm a bit 
confused as the kernel doesn't seem to recognize the makers code & doesn't 
load the right drivers for it.


-- 
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)
Yow!  We're going to a new disco!