Re: Freakin' Second Sight Board

[rubywand@aol.com (RUBYWAND)]

Paul Schultz <pschultz@med.wayne.edu>  writes:

>>
Could you please explain how adding a thicker wire will decrease noise on
the +5V and +12V lines?  I received and installed my new power supply and
had the same results with my SS....it locks up the computer on boot-up and
puts garbage on the frozen screen.
<<

    Ohm's Law says: E = I x R.  The Voltage drop across a resistance
equals the Resistance (in Ohms)  multiplied by the Current (in Amps).

   The leads coming from the most power supplies have a low resistance--
less than an Ohm-- but, they are still "resistors".  At currents of an Amp
or so the drop through the +5V lead and connector to the motherboard may
be around 0.25V.  If the PS normally puts out a little more than 5V, then
the at-motherboard voltage may be pretty near the desired 5V  (e.g. 5.2V -
0.25V = 4.95V).

   A Volt-meter reading at the motherboard PS plug does not show
instantaneous spike voltages. Each time a circuit switches, there is a
change in current drain. Quite a few circuits are switched with each clock
transition; so, the change in current can be substantial at 1 x  and 2 x
clock frequency.  Other events, like turning ON a disk drive, can also
produce brief up or down shifts in current drain. Either way, you have
brief changes in voltage across the PS lead(s) through which the current
is drawn.

   The brief voltage changes are called "spikes" because they are VERY
brief. The larger the current shift and the greater the resistance of the
PS wire, the higher the spike Voltage. Since these spikes are in series
with the circuits connected to the PS and since they are difficult to
eliminate via bypass capacitors, they propagate throughout the system.
They are a kind of "noise".

   Now, as you plug in Memory Expansion, Accelerator, Hard Disk interface,
etc. cards, the current drain increases. At 3 Amps, the average voltage
drop across the +5V lead may be 0.75V with at-motherboard voltage going to
less than 4.5V. Current shifts also become larger. So, you have a 'double
whammy': the lowered supply voltage reduces "noise immunity" of circuit
components AND noise spikes become larger.

   At some point, noise spikes appear which cause latches, memory IC's,
etc. to switch state. If the latch is on a RAMfast, you may get a disk
read error. If a memory chip is affected, data will be corrupted, program
instructions may change, ....; in short, your computer is likely to
malfunction.

   THICK power supply leads have less resistance. Less resistance means
that the voltage drop or loss across the thick lead is smaller. Noise is
reduced AND noise immunity is increased. Another effect which usually goes
unmentioned is that voltage regulation also improves because the 'sensed
voltage' is closer to the actual at-motherboard voltage.

   Sometimes a "Heavy Duty" power supply will get rid of glitches because
it happens to employ heavier lead wires and may have outputs set a bit
higher (e.g. 5.5V at the +5V output).  Other times, a heftier PS will not
eliminate problems. The leads may be thicker; but, are not thick enough.

   Fatter +5, +12, and Ground leads may eliminate glitchy performance. A
lot depends upon which cards you've installed, which slots they are in,
and whether your IIgs is a ROM-01 or ROM-03. Unfortunately, there is
another weak link in the power delivery 'chain': the motherboard circuit
traces supplying power to the slots are fairly skimpy, especially on
ROM-01 boards. A heavy power user in Slot 7 can produce significant noise
up and down the entier Slot 'backplane'.

   The 'ultimate cure' is to remove the motherboard and tack on leads from
+5V on a couple Slots (e.g. Slot 3 and Slot 7)  to the +5V point at the
power connector and from Ground on two or three slots to the power
connector Ground or the motherboard Ground plane. A +12 lead to the Second
Sight board's Slot and seems to be a good idea. (The same goes for any
Slot holding a souped-up 12-14MHz Accelerator c