29 Mar
2014
29 Mar
'14
4:05 a.m.
On 2014 Mar 28, at 10:38 PM, John Ball wrote:
My PDP-11/84 came with an RA82 disk and after three
years of work I
have it
finally coming up to speed and going ready. I had to redesign the
circuit
that generated a tach pulse for monitoring the speed of the
spindle. The
original circuit never wanted to work for me.
http://i11.photobucket.com/albums/a166/ballsandy/Computer%20related/
logic.jp
g
http://i11.photobucket.com/albums/a166/ballsandy/Computer%20related/
IMG_5731
.jpg
The new circuit however is unreliable. At complete random the drive
will
throw both nonfatal and fatal speed errors however at no time can
you hear
that the speed of the drive has changed (and for error DA and error
26 you
would definately hear it), nor can I catch the last moments before
spindown
on an oscilloscope as to see if the transistor is locking up or if
there's
some sort of interference. Again, it occurs randomly. I can be
seeking or
reading or writing or just idling the drive and it will happen and
it makes
the drive impossible to use. Could this just be me using too low
speed a
transistor or could this be something else that I have not yet
considered?
The reason I'm bypassing the original 74LS14 is that even when
replaced it
leaks about a volt onto the input pin and forced the signal high. I
could
not figure out a way to pull that down and still get the signal to
come up
properly so I simply removed it from the equation.
It's a little hard to say without seeing the original (or rest of)
the circuitry, and I'm not familiar with the RA82 internals,
particularly the frequency and periods at which this circuit is
operating, but a couple of comments in general:
While optical sensors are comparitively clean relative to mechanical
contacts, the edges can still be a slow transition by electronic
standards, due to the 'grey' region as the edges of the optical
interruptor take time to pass through the light beam.
The 'LS14 inverter is a Schmitt-trigger device, it would have been
there to clean up the transitions from the opto-sensor. The Schmitt-
trigger function provides hysteresis on the transition thresholds,
but it's also providing voltage gain, which sharpens the edge before
it's fed on to following circuitry. If the following circuitry is a
counter or edge-triggered device, the slow transition directly from
the opto/3904 may be producing erratic triggerring.
In addition to the missing Schmitt inverter, there is no pull down
resistor on the base of the 3904. While the 3904 may still switch as
it is, the on-to-off transition of the 3904 may be slow. It also may
not be going fully off, as even slight illumination of the opto-
transistor will start it conducting and feeding current through the
base of the 3904, starting to turn the 3904 on. You might try adding,
say, a 10K R to ground on the base of the 3904, or from the opto-
transistor emitter (pin-3) to ground.
Re the LS14 'leaking a volt' onto the input: a TTL input will tend to
'pull-high' as the conduction path for a TTL input is indeed through
to +5, it's normal to see some +V on an (open) TTL input.
TTL is current-sinking logic, the 'excited' state of an input is for
electron current to be flowing from ground, through the driving
output device, into the TTL input and through to +5.
The better question might be what was wrong with the original circuit(?)