Re: How much strain can a IIGS power supply handle?

[mjmahon@aol.com (Michael J. Mahon)]

Rubywand wrote:

>     The main limitation on the original PS is the pissant leads that connect
to
>the motherboard. These limit the current you can draw to something less than
about
>2.5 Amps on the +5V line. You can dram more current; but, by then, you have
>glitching problems due to noise and low motherboard voltages.

The power leads in question are about 12" long, and are a single 18 AWG
wire for +5 and two 18 AWG wires in parallel for the ground.  Since the +5
carries the most current, and is a lower voltage, it is the most perturbed
by wire resistance, so we can just look at it to simplify matters.

The resistance of a foot of 18 AWG copper wire is 7.5 milliohms.  That
means that the voltage drop in it is 7.5 millivolts per amp (or the delta-V
is 7.5 mV per delta-I of 1 amp).

Since the stock IIgs supply +5 is rated at 3A, and a heavy-duty one is
rated at about 5A, it seems safe to assume that the IIgs load on +5
is less than or equal to 5A (probably less for most, since the vast
majority are getting along fine with a stock supply).

This means that the voltage drop on the +5 wire is about 37.5mV and
the drop on the ground wires (in parallel) is about 18.7mV, for a total
drop of less than 56mV in the +5 supply.  (Of course, the +12 current
returning through the ground wires will contribute another amp or two,
but that will add only another 3.5mV-7mV drop.)

So the total resistive drop from the power supply to +5V loads is
about 60mV under maximum load.  If we assume that there is a
delta-I of 2A (probably high), then there is a max delta-V (due to
the power supply wires) of less than 25mV.

TTL logic is spec'ed for 5V +/-10%, or +/-500mV.

The IIgs board is bypassed with plenty of monolithic and electrolytic
capacitors, and some inductive decoupling as well.

Therefore, any problems with power supply impedance, at either
high or low frequencies, is unlikely to be attributable to the wires
connecting the board to the power supply.

If the relatively small resistance of the power supply leads is sufficient
to cause a problem, then there must be other factors which are out of
spec--for example, bypassing capacitors on the main board may be
defective, the power connector may be somewhat oxidized, the power
supply +5 voltage may be low (near 4.5V), or the output filter capacitors
in the power supply may have diminished capacitance.

Replacing the power supply leads with heavier gauge wire is not an
appropriate solution to any of these problems, though it may appear
to solve or reduce them for a while--until whatever component(s) are
actually degraded are slightly further degraded--whereupon the
symptoms will recur.

I view this as an aspect of the "monster cable" myth.  Electrons only
know about resistance, and a certain amount of resistance is always
allowed for by any electrical design, since it is unavoidable.  (Yes, I
know about inductance and capacitance, and both are negligible issues
for the wires we are discussing--that's why there are bypasses.)

Reducing resistance below the nominal design level is not worth much.
Eliminating failures that increase the resistance substantially above
design tolerances is what is needed.

-michael

 Email:  mjmahon@aol.com
 Home page:  http://members.aol.com/MJMahon/