Where "\->" represents the right arrow character (141 decimal) and
n represents the time in ticks (1 tick=1/8192 second), KEYTIME\->
returns the current keytime value, and n \->KEYTIME sets a new
keytime value. For example, 512 \->KEYTIME sets the value to
512/8192=1/16 second.
It works in both the 49G and the 49g+. For the 49G, the default is
1365 (which seems a bit high for many users), and for the 49g+,
the default is 0. Apparently, the 49g+ developers didn't think it
was necessary for this model, but perhaps kept it for
compatibility, or maybe as a policy of not fixing what isn't
broken.
I wouldn't call it a "scanning" rate, but rather a means of
ignoring key "bounce" occurrences. If the same key is pressed
again in less than n ticks, then the second keystroke is ignored.
It has no effect at all on how soon you can press a different key
and have it register.
Personally, although I often have missed keystrokes on the 49g+,
and occasionally on the 49G, I've yet to experience any key bounce
problem (a single keystroke registering more than once) with
either, even with the keytime value set to 0, but it wouldn't hurt
to set it to a fairly low value to prevent key bounce problems.
Setting the keytime to too high a value will make it difficult to
type in the same key repeatedly.
Setting the keytime value has no effect on most missed keystrokes,
except that setting it too high may well cause repeated presses of
the same key to be missed.
For the missed keystrokes problem, the best I can suggest is the
"just press firmly" method; that is, press the key until it
definitely bottoms out, not just until you feel the "click".
Something that may help you to detect missed keystrokes is to
check the "Key Click" option in the "MODE" input form, which will
cause a short beep with each (registered) keystroke as it's
removed from the key buffer. But this might not be any help in a
noisy environment, and may well annoy others in a quiet
environment.
For what it's worth, the information that I have on the 48SX is
that it scans the keyboard once every millisecond and generates an
interrupt when a keypress is detected. Once the interrupt has been
generated, the keyboard handler checks which keys are pressed.
Whether this is accurate, and whether it's still accurate for the
newer calculators, I don't know. Some contend that at least part
of the problem with missed keystrokes on the 49g+ has to do with
switching CPU speeds when a keystroke is detected, or perhaps that
the Saturn microprocessor is emulated by software running on the
ARM microprocessor.
Note that with all of the RPL models, up to 15 keystrokes are
stored in a FIFO buffer, so you can press a few keys while the
"busy" annunciator is on, and they'll be processed when the system
is ready for keyboard input again. If the "Key Click" option (49
series only) is enabled, then the beeps won't occur until the
keystrokes are removed from the buffer.
Regards,
James
Edited: 31 Jan 2005, 5:07 a.m.