[Coco] calling all MM/1 owners

[Gene Heskett]

On Thursday 24 January 2008, Joel Ewy wrote:
>Bob Devries wrote:
>> I just removed the I/O board from the MM/1 to more closely examine the
>> RTC.
>>
>> I found that one of the termination resistors for the SCSI bus is
>> oriented the opposite way to the other two. RN5, the middle one is
>> different. Now I'm wondering what effect this will have on the working
>> of the SCSI bus? Could this be my problem?
>>
>> Joel, the termination resistor's 'notch' or 'dot' end should be in the
>> solder pad that's square, am I correct?
>
>Yeah, that's right.  I think that's the common lead for the resistors.
>Resistors themselves aren't directional.  Does it make a difference
>because of the way they're assembled in an array?
>
Yes, there are two common points in a resistor pack as used for scsi 
terminations.  There is a supply pin that is intended to be connected to a 5 
volt supply, and a ground pin.  In a 10 pin pack (they come on various 
widths) you will have a power pin, a ground pin and 8 pins intended to be the 
terminators.  Each of the 8 pins has a 220 ohm internal resistor connected to 
the supply pin, and a 330 ohm connected to the ground pin.

The scsi buss is a wired or buss, and these terminators furnish the pullup 
resistance to hold a line that is not being driven low from anyplace on the 
cable, at a guaranteed logic one plus about 600 mv for noise margin, or 3 
volts.  The active pins have a 220 ohm resistor connected to the supply pin, 
with a 330 ohm connected to ground, which will set this 3 volts if the supply 
is a full 5 volts.  It is often not because of the use of an isolation diode 
to keep the 5 volt busses isolated when one of them is tuned off.  This 
diode, if Si, will drop abut .65 volts and the voltages at the active pins 
will drop accordingly.

The commonly used cable for scsi hookups, the flat ribbon cable, has if every 
other wire is grounded, a characteristic impedance of about (its not a very 
precisely made cable) 120 ohms.  You need that resistance on each END of the 
cable or a signal will bounce from end to end, echoing many times.  Exactly 
the same as when you are measuring the SWR of your antennas for your ham & cb 
radio use.

The parallel combination of a 220+330 ohm is about 132 ohms (and these are 
often +- 20% parts people!) and generally does an adequate enough job of 
absorbing these echo's that it will work if the supply is correct.  With the 
isolation diode, it gets borderline low and you have lost the noise margin 
for the logic one state, which is about 600 mv if the lines are at 3 volts 
idle.  At 2.6 volts idle, the noise margin is down to 200mv, and it will be 
much more sensitive to these echos, and at 2.4 volts the noise margin is gone 
and its not going to work, period, full stop.  This is where you start 
sacrificing goats or whatever trying to make it work reliably but it won't.  
The much lower forward drop of a power schotkey diode will help a lot, but if 
the psu is also sagging with age, and is only putting out 4.82 volts or so, 
then even the schotkey diode's lower voltage loss of about .15 volts might 
not be enough.

A properly setup scsi bus is a work of art that you can run 40 meters at full 
speed for scsi-2!  Build it with 20% tolerance parts & a saggy psu 5 volt 
line and you have a nightmare with only a 20" cable.

So called 'active terms' are in fact a voltage regulator set at 3 volts, with 
a 120 ohm resistor between the output of the regulator and the data lines.  
These are much less sensitive to the supply variations, and since the 330 ohm 
to ground is done away with, draw considerably less power, about 5 watts 
less.  If one can find those in a suitable package, remove the resistors and 
substitute the active chips output into the pcb holes, dependability will go 
way, way up.

-- 
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)
Many people write memos to tell you they have nothing to say.