Clock Talk
Bill van Dijk’s September 2016
article, "Build the Numitron, A Six-Digit
Clock," is just plain fabulous! I really
enjoyed it, and what's more, I have
learned something new, and that is
important to me. I now have several
questions for Bill.
For soldering surface-mount
components, you mention that your fine
point tip for your soldering iron is almost
needle sharp. It looks like our soldering
stations might be the same. I have a
digital Weller WESD51 powering a
PES51 iron. I bought the model "ETH"
tip, but have always thought I needed a
finer point for my SMD exploits. Have
you found a finer tip to fit onto your iron
other than the Weller ETH?
When I built a PIC16F877 based
clock a few years ago, I used a single 4
MHz crystal. The interrupt routine and
Timer1 perform timekeeping using that
frequency. In my climate controlled
house, I routinely see accuracy better
than 10 seconds per year. In your
design, you added a second crystal with
capacitors to do the timekeeping at a
slower frequency. Why? Is there an
advantage in using the additional parts?
I really admire your use of a
hardware solution for debouncing your
pushbutton switches. The link you
included for the debouncing discussion
( www.eng.utah.edu/~cs5780/debouncing
.pdf) is fantastic. I have never seen a
more thorough handling of the topic.
However, I believe that author does
make one mistake: As voltage climbs,
the high-going transition point for a
Schmitt trigger is not 0.9 volts, so I think
some of his results are in error. It seems
that he has mixed up the two transition
point voltages.
After reading that in-depth
debouncing discussion and then looking
again at your schematic, I became totally
befuddled. You do not seem to follow
that author's logic in your circuit. Your
15K discharge resistors (R19 and R20)
are very similar in value to those
employed in the discussion (R2 is 18K),
so at first glance it appeared you were
on the same track. However, your
debounce capacitors are specified as 0.1
µF instead of 1.0 µF, which would
certainly be a big change in the math
equations. Plus, your charge resistors
(R17 and R18) are only 1K compared to
the discussion's R1 of 101K. Coupled
with the diodes D3 and D4 in your
debounce circuit, that is a very fast
charge time indeed! Could you tell me
how you decided to use the 1K resistor
here?
I used the formulas in the discussion
to do the math for your debouncing
circuit. The port C pins you use on your
PIC are Schmitt triggers, so the threshold
voltages are one volt low-going and four
volts high-going. With your components,
that will take care of bouncing contacts
when a button is pressed for 2. 4
milliseconds. In the discussion, it was
decided that a good target would be 10
to 20 milliseconds. I bet you have good
quality pushbutton switches in your
clock, and with reduced bouncing in
quality contacts I imagine your
functionality does not suffer. If another
builder puts this circuit together with old
or inferior switches, unfortunately, there
could be contact bounce problems.
I am more interested in the button
release, however. I did the math, keeping
in mind the bypass diode. When you
release your pushbutton, that 1K resistor
with the 0.1 µF cap will only suppress
bouncing for 0.15 milliseconds. I
understand that released contacts do
have less bounce than the press-down,
but some bounce hash is present in
READER FEEDBACK
DOWNLOAD our free CAD software
DESIGN your 2 or 4-layer PCB
CHECK your design with xCHECK DRC
SEND us your design with just a click
RECEIVE top quality boards in just days
1
2
3
4
5
MADE IN THE USA | www.expresspcb.com
$61 3 PCBs
2-layer
MiniBoardPro
(Shipped in 2 business days)
$41 3 PCBs
2-layer
MiniBoard
(Shipped in 1 business day)
+ shipping + shipping + shipping $81 3 PCBs
4-layer
MiniBoardPro
(Shipped in 3 business days)
Now Offering
SMT
STENCILS
– Nano Coating available–
December 2016 7
Continued on page 58