Heath GC-1000 Most Accurate Clock Modifications

542 views
Skip to first unread message

gibb...@iccgcc.decnet.ab.com

unread,
Sep 25, 1991, 8:52:32 AM9/25/91
to

Below is a summary of the modifications I have done to the Heathkit GC-1000
Most Accurate Clock to improve its performance:

1.) I replaced all of the IC sockets with machine tool pin sockets. Having to
re-seat some of the IC's every now and then is unacceptable. I can't believe
that HEATH used such CHEAP IC sockets! (Well, maybe I can...)

2.) Re-designed the clock drive to the microprocessors. The 470pF AC coupled
3.600Mhz clock drive relied on the fact that the input to the F8 uP's had a
diode input clamp (probably the substrate of the uP) and the input voltage would
swing from -0.6V to +3.5V. I replaced the cap with a 74HC14 CMOS driver with
am 82 Ohm series resistor (to match the impedance of the driver to that of the
PC board to reduce ringing). The voltage now swings from ~ 0.2V to ~ 4.8V.
Risetime of the clock remained about the same.

3.) I also replaced the output transistor push-pull circuit that drives the
external 3.6000 Mhz output reference with a 74HC14 driver. I actually tied 3
of the inverter gates on the same chip in parallel (prop delays are almost
identical, so you can usually get away with this) to drive this output. Created
a nicer looking waveform. This was the other half of the chip used for the
clock driver section (I used 1 gate of the other half for the CPU clock driver,
2 gates remain unused (and tied off)).

4.) I added a second +5V regulator (78L05) and separated the D/A Latch, R/2R
resistor ladder and all associated circuitry that runs the Colpits Oscillator
/ Varactor Diode / Clock Driver circuitry. One cut on the top side of the PCB
by the input inductor (and the feed-thru hole) isolated the clock section +5V
very nicely. The problem I noticed is that when you turn on the displays, the
main +5V regulator would droop about 100mv causing the D/A voltage to the
varactor diode to droop. This was enough to shift the 3.6000 Mhz frequency by
~ 15-20 Hz. Still within the spec of the clock for output freq accuracy, but by
adding this second regulator the frequency now shifts less than 0.1 Hz (thats
the resolution that my freq counter can measure to). I also bypassed the clock
section with a few 10uF Tantalum caps to reduce switching noise.

5.) Eliminated the display ghosting (display shut off but the 5/10/10MHZ
indicators still glow a little). This was accomplished by simply grounding
the unused side of the display enable switch (labeled 3 + 6 on the schematic).
You also need to cut the power to the decoder IC U101 (pins 1 and 16) and take
these two pins directly to the power connector pins 1+2 (+5) on the display
board (before the switch). If you don't, when you shut the display off IC U101
will load down the uP lines to it and the D/A latch will always get loaded
with 00H, thus screwing up the ability of the clock to tweek its own oscillator
frequency.

6.) Replace all of the caps associated with the 100Hz and 1Khz tone decoder
circuits with polypropolene or stacked foil caps. This reduced the clocks
sensitivity to temperature drift (and thus lousy performance) during times
when the display is on and the insides of the clock heat up. I also replaced
the 2 - 5K Ohm open face pots with 20 turn adjustable pots - much easier to
adjust accurately. The 2 phase-locked loop adjustments are made much easier
by just tying a high-impedance probe on pin 5 of the 567 PLL chip and adjusting
the frequency to either 100.0 Hz or 1000.0 hz. Much easier to adjust than
Heath's method. (In talking to the techs at Heath, they recommend this method
over that in the manual). Caps and everything were ordered from Digikey for
about $10 total.

7.) I added a MOV and a .001uF 1KV cap on the AC input to help line noise
rejection and spike suppression. I personally run the clock off of a 12V
8AH GEL-CEL (I had sitting around for a while, figured I better use it or
loose it) and then use a float-voltage charger for the battery (not a cycle-
voltage charger). I also added a 0.1 uF ceramic disc cap on the output
of the transformer (input to the full-wave bridge) for added noise suppression.

8.) The transformer that HEATH supplies in the GC-1000 is just as bad as the
Radio-Shack transformers: They skimp on the wire size AND the # of turns on the
primary thus giving LOUSY line regulation and they run HOT! DUMB! DUMB! DUMB!
I only plug my unit in when I move it (in case the 12VDC connector unplugs
and wipes out the previous months of clock oscillator tweeking). If you use
the transformer, replace it with a real one that can handle the 800ma load
without sagging so badly (and getting so blasted HOT!).

9.) I also added a computer interface to directly look at the 5-10-15 Mhz
band indication, HI-Spec LED and the 100Hz and 1Khz tone decoder outputs.
(The interface is nothing more than a 74HC14 inverter tied to the appropriate
lines on the F8 uP). I have a CMOS Z80 system monitoring these lines and when
the clock goes into or out of Hi-Spec, I kick the clock's serial interface and
store the time and band info in an EEPROM. I'm still writing the code, but
have the basic system working now (capturing data). It will basically give
me information on when the bands are 'open' to Ft Collins, CO. (which is
1240 miles west of me).

I plan on using the 100Hz and 1Khz data for a later project - To be able to
decode the WWV data stream myself (probably using a 68HC11 uP) and create a
real serial interface that tells you what time it IS, not what time it WAS
1-2 seconds ago (the HEATH serial 'bit banger' interface STINKS!)

I'm still debating on whether to attack the receiver section of the clock -
It works ok, but it COULD be a WHOLE lot better... hmmmmm...... anyone else
tweeked the receiver yet???

[It's a fun little project that keeps me out of trouble....]

--
John Gibbons N8OBJ Macedonia, Ohio

Internet Address: gibbonsj%iccgcc...@consrt.rok.com

"Welcome My Son, Welcome To The Machine" - Pink Floyd

Reply all
Reply to author
Forward
0 new messages