After a few months of testing ( April – June 2017 ) the 7490 TTL decade counter chip based clock out with the newly added OCXO it was ready to be mounted in a housing.
Monitoring it against an accurate timebase ( Linux PC with NTP ) while piled up loosely on a small table in the office, it was running good and steady. No flaky counts, no flaky digits that were not lighting or staying lit when they should not be.
After the photo was taken above, the power supplies for the clock were consolidated. Under testing it was powered by 2 wall supplies, 5V for the TTL chips and 12V for the OCXO. For the final version I used a 7805 regulator on the right hand side board to down regulate 12V from a Radio Shack 12V/2.3A power supply to 5V for the TTL chips. I got lucky and spotted the power supply during the last few days that Radio Shack was still in business in April 2017.
Not having a good set of tools on hand to build a metal housing, a good option was a wooden one. Micheal’s craft store had both the base that the boards were mounted to and the shadowbox. Both were unfinished and reasonably priced, I believe $5 for the board and $12 for the shadowbox in 2017.
The base was drilled out to recess fit 10mm standoffs. Holes were drilled for the M3 screws and then countersunk with a larger bit to accommodate the 10mm standoffs. Once again I made out good at the Radio Shack closeout and got a bunch of standoffs.
The clock mounted on the base in a state of readiness to drop into the shadowbox as a subassembly. This trick is a bit that I learned from working in manufacturing in the past. To make things efficient and serviceable it was common to mount boards on sub assemblies and then mount the subassembly into a piece of machinery.
Finally the subassembly gets dropped into the shadowbox and is held in place by four small wood screws that are on the back.
The last post on this clock built out of 7490 TTL decade counter chips gave a bit of back story on it and the initial troubleshooting of it’s timebase drift and noise issues that caused extraneous counts to occur. Now that the noise is figured out and solved, it is time to do something better than a plain crystal with a trimmer 7400 ( NAND Gate ) oscillator for a timebase.
Driving the clock with an OCXO
The goal was to allow the clock to be driven by an oven controlled crystal oscillator (OCXO), which would surpass the performance of the oscillator on the board. The on board oscillator can be easily pulled off it’s frequency and slaved to an external source. By using a 7414 Schmitt Trigger Hex Inverter, it is possible to take the low level sine wave output from the OCXO and convert it into a digital signal capable of entraining the on-board oscillator. By using an external master, I can pull the master OCXO off if I need it to calibrate anything else, meanwhile the clock keeps running off of it’s board oscillator. Most of the time the OCXO is not being used and can just remain with the clock.
The hookup is pretty simple, I breadboarded it and found out the biasing one of the inputs of the 7414 mid range using a pair of 33K Ohm resistors in series from +5V power to ground with the input to an inverter stage on the 7414 in the middle at 2.5V, feeding in the OCXO signal, capacitively coupled and then feeding the output of that stage into another inverter on the 7414 for a buffer, I was able to get a clean square wave at 10MHz. This I coupled to the board oscillator using a 370 Ohm resistor. The resistor is more or less a protection in case I touched it to something that I shouldn’t and I don’t want too much current to be able to flow in either direction. I also temporarily drove an LED via a 370 Ohm resistor off of the same output as a check that I was actually getting output, the LED is lit at half the brightness when it is excited by a square wave.
Initial testing with the breadboard shows that the clock now tracks time very well as compared to another clock I have that is synced to the WWV 60KHz signal.
About 10 years ago I built a clock built out of 7490 TTL decade counter chips. It was based off of an article that I found on How Stuff Works, that I could no longer find, but I have it available here. I also have posted the schematics for the version of the clock that I built that uses a 10MHz timebase and not the 60Hz one that the How Stuff Works version of the clock uses. Schematics for the 7490 TTL Chip Clock
The clock was in my shop for a number of years, then it got packed away when I moved in 2013. Meanwhile, In 2016 built a well made 24 hour clock from a kit from MTM Scientific, Inc that has a TCXO and drifts only a few seconds between the times of the year that it needs adjustment for daylight savings time. It has a nice bright readout and works well in my bedroom. But, it also got me thinking about the TTL clock again and I thought that one might be worth revisiting.
One issue with the 24 hour clock that I built from TTL chips is that it occasionally required adjustment of a trimmer capacitor to keep the oscillator running in time. I also noticed that occasionally it would flake out completely and get extra counts, causing it to run fast by about 1.5%. Sometimes just restarting it would fix the issue, it was a mysterious. It first occurred when I moved it to a different spot on the work bench after initially getting it put together and pinned to a wood board. My first thoughts were that I had a flaky chip. At one point I had accidentally connected the clock to a 12V power supply and blew out one chip and I speculated that there might be others that were working but damaged. Or it could have been a bad solder joint, with all the points of connection that was certainly plausible.
The Real Issue
One thing that I didn’t have access to in 2007 that I have in hand now is a frequency counter. When I built the clock my frequency counter was missing. I had it packed away somewhere in 2003 and only found it years later (2013) as it was packed in a box that made no sense at all. But, it would have been handy at the time to track down where I was getting these extra counts. In early 2017 with the frequency counter in hand, I was able to quickly determine that it was not only extra counts, but noisy extra counts as the lower digits on the counter were fluctuating. Something was ringing or going into an oscillation, that was my initial thoughts. I relatively quickly suspected a buffer chip, 7400 inverter, that I had added to provide some reference outputs at 1MHz,100KHz, 10KHz and 1KHz. The chip was the only one on the oscillator and divider board that I had tacked on after the clock was working, so it seemed likely that it might be part of the issue. Plus, it was the only one that I had neglected to put a bypass capacitor across from power to ground. Removing power from the chip solved the problem, good counts, a perfect multiple of the 10MHz clock were now coming out of the board to drive the dividers on the second board with the digits. The chip might not be worth using, if I need lower frequency references I can always take another oscillator I have and use that.
Calibrating the 10MHz crystal oscillator by tweaking the trimmer, against a 10MHz OCXO ( Oven Controlled Crystal Oscillator ) and then watching it keep time for a few days, it now holds reasonable time.
The Next Plan
The next plan for this clock is to drive it with the 10MHz OCXO to overcome the limits of the simple 7400 Inverter TTL chip 10MHz crystal oscillator with a trimmer capacitor. This simple oscillator is limited, temperature will make it swing along with any proximity of metal or whatnot that changes the frequency. If you touch the crystal or any part of the circuit around it, I imagine the frequency is pulled off target.
By feeding it with the closed box OCXO, powered through a decent regulator, frequency variations due to temperature changes, voltage changes and capacitive changes due to proximity of conductors will be minimized. It will be interesting to see how stable the clock can be.
Also, it would be nice for this thing to finally wind up in a decent box and be powered by something other than a 12V to 5V regulator tacked to a spare breadboard.
I plan on writing more on this as the project unfolds, along with the schematics, when I find them, all hand written and might be lost when I moved, but I might just have to recreate them!
It starts with a 10 MHz 7400 inverter oscillator and divides down using 7490 decade counters set up to divide by 6 or 10 as needed. Some AND/OR logic appears in the design as well to provide a pseudo WWV time code, 1kHz second ticks, minute and hour marker. This output is provided as an amplified audio output. This is done using a small 1 stage transistor amplifier driving the 2 inch speaker, with series resistor to limit volume. A 1/8in jack is provided as well for driving a larger speaker. The marker is also able to modulate a 1MHz output for a test signal. Three modes of output are provided, 1MHz carrier, 1MHz modulated with steady 1kHz signal and 1MHz modulated with pseudo WWV time signal. This signal and 100kHz,10kHz,1kHz are provided as buffered outputs for off the board use.
The display itself is an array of dual 7 segment common anode 0.75 inch elements, with appropriate 7-segment drivers. Setting is via 3 pushbuttons. Two provide speedups of seconds 1000X and 10X to roll the clock ahead faster than real time, a third button is a halt button for syncing with another clock source. A 10Hz ‘heartbeat’ LED is provided for debugging purposes. This is connected at the junction between both boards.
One board is the oscillator and divider to 1000Hz and the test outputs (1MHz,100kHz, 10kHz and 1kHz. Pseudo WWV 1MHz and audio) and the other board divides down further and has the display and the drivers for the 7-segment LEDS.(3/2007)