As some of you know who have read this thread before,
I have used these Dallas/Maxim DS32KHZN/DIP TCXOs exclusively
in my clock experiments since they became available back
around year 2000 or so.

They have always proven to be very stable, but not always
tuned to the exactly correct frequency.  A typical prototype
clock I have running with one now gains about 1.2 seconds
every 14 days.

Well, I finally decided to do something about that.
I calculated how many extra pulses it is making over
those 14 days, and divided that number by 14 and then again
by 24 to arrive at the extra pulses per hour.

That number turns out to be 117 pulses to somehow remove
every hour.
I made a pulse remover.  It's a very simple affair that
uses a 4040 ripple counter, a 4073 triple 3-input AND gate
package, and 1/2 of a 4013 D flip-flop.

It is set up to decode count 117 (75 hex) (0111 0101 binary).
Upon decoding that count, the flip-flop is reset.
When the flip-flop is reset, it holds the 4040 counter
at reset as well.
During the time the counter is running, the outgoing
pulses are gated off to the clock's counters via an AND gate.

It works because I tested it triggering the flip-flop
every second instead of every hour.  During that test
the clock lost exactly one second in 4 minutes, 40 seconds
just as the numbers say it should, if the pulse remover
is working.   So after having tested it this way I moved
the lead over to the input to the hours counters, so that
the pulse remover gets to run once per hour.

It's still too early to know whether or not this pulse
remover will only be triggered when it's supposed to be
and at no other time due to noise.  But if it can be made 
to operate correctly, I think it can hold the possiblility of
drastically (as in orders of magnitude) improving
the accuracy of this TCXO.

A slightly more complex version can be made that has
the integer that represents the number of pulses to
be removed to be set in on two tiny hexadecimal rotary
dip switches.   That would provide a resolution of one
pulse, and be selectable from 0 to 255 pulses removed
per hour.   It appears that finally there is an accurate
and fairly simple way to adjust the speed of these clocks.

The first version of this circuit uses only 3 dip
packages.

If all works well, this can be retrofitted into all of my
clocks by making a little add-on board and installing it.

This thing will even provide a way to compensate for
crystal aging.

Time will tell.    Just thought I'd run this past all
of you.  I am sure most of you already have heard of such
pulse removing ideas before.   If a slow one was ever 
encountered, a "pulse stuffer" could also be made to
correct that.

Chuck


$4.95/mo. National Dialup, Anti-Spam, Anti-Virus, 5mb personal web space. 5x 
faster dialup for only $9.95/mo. No contracts, No fees, No Kidding! See 
http://www.All2Easy.net for more details!

-- 
You received this message because you are subscribed to the Google Groups 
"neonixie-l" group.
To post to this group, send an email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/neonixie-l?hl=en-GB.

Reply via email to