There is an old saying that goes something like this "A man who has a clock always knows what time it is. A man who has more than one clock is never sure what time it is." This page is devoted to my attempt to solve that problem through the use of a Jeff Thomas and John Miktuk designed GPS controlled Nixie clock.

Now, if you don't know what Nixies are (originally from Numeric Indicator Experimental-1) they are a product of the mid 50's to early 70's. They predate the light emitting diode. They look like a vacuum tube but are in fact a form of neon light. Nixies consist of a sealed glass tube (like a vacuum tube) with a single internal cathode plate at the rear and, in front of that, 10 successive formed wire digits. Each anode digit and the common cathode has a pin at the bottom of the neon filled tube. To light a digit you simply apply approximately 170VDC between the common cathode and the appropriate digit pin. The energized digit causes a warm glow of neon to light up around it. Nixies however, unlike vacuum tubes, have no heater and operate cold.

I'm sure most people are familiar with the Global Positioning Satellite (GPS) system and that with them you are able to determine your position and speed (even walking) very accurately. What makes it all possible is that each satellite in the system has four atomic clocks on board, and those clocks are compared with those in Building 78 at the Naval Observatory twice a day to make sure they're accurate to the billionth of a second (if GPS used time scales less accurate, say to a thousandth of a second, its margin of error would equal roughly the distance between Ottawa and Montreal). This Nixie clock makes use of that extremely accurate GPS satellite time output so that in essence I have a cesium time reference in my house.

 

I was originally "turned on" to Nixie clocks by an article in an IEEE Spectrum magazine passed to me. In the article there was a GPS version designed by Katsushi Matsubayashi of Tokyo that inspired me to think about designing my own. The first thing I did however was secure 6 beautifully large NOS (New Old-Stock) ZM5680M Nixie tubes. My thinking here was the supply of Nixies is fixed and limited so I should get them first. If I sat on the idea for a while I'd at least have the tubes while any design I came up with to run them, even if delayed for years, could evolve with the times and technology. My hope was to use them to create a clock driven by a GPS receiver unit so that it was, in fact, an in-house cesium time standard. Using momentary switches I would be able to selectively display sidereal time, equation of time, sunrise time, sunset time, latitude, longitude & elevation. I would also be able to set alarm time on it through rotary switches and have that alarm output drive a high fidelity sound board. Different alarm sounds were also to be selectable by a rotary switch. Additionally the clock would command the sound board to generate selectable clock sounds for seconds, the quarters and on the hour. Possibly the sound card could be amplified by car audio components. I did a fair amount of research into the algorithms needed to convert exact time to the outputs above. The following was my basic idea...

The more I got into it the more it seemed the hardware would be trivial and that the real challenges were in the machine code software that would allow the PIC microcontroller to convert the serial GPS output from the GPS receiver into the display outputs listed above. As I stated on my Propeller Clock page my abilities in machine code programming are very limited! It was about this time that I joined an online Nixie clock group only to have Jeff Thomas (above) announce, within a week of my joining, he was coming out with a GPS clock kit that, whoa and behold, used the very same tubes I had already bought. After some deep thought and a few inquiries I decided that there was no point in re-inventing the wheel and that his kit could possibly be the purchased heart of my original idea and so joined the waiting list for the kit. In a few months Jeff emailed me that the kits were ready and I took the plunge and ordered one.

Above you can see the kit and other peripheral items not included such as the tubes etc. The building of the kit was very straight forward as it was all well documented. It went together quite quickly and without any issues at all.

The pictures above show the board near completion. The colon indicators have yet to be done on the primary side of the board. The GPS unit can be seen on the secondary side of the board. There are a few mods yet to be done which get described below.

The next item I had to resolve was the base. Jeff offers a good one for a very reasonable price but I wanted something a little different. I wanted to capture a tinge of "retro" feel. A lot of different (even wacky) ideas were explored. Some of the more esoteric ideas were definitely going to be prohibitive in cost and time as they involved casting complex shapes. After many sketches I came up with a basic idea which I simplified a bit to allow for ready machining. The idea was a black anodized aluminum block with soft corners (like some 50's appliances), a bit of heatsink-look and a back-lit logo. I designed the base on Pro/E.

The base also had to accommodate the glass cover. I went locally to Northern Art Glass and inquired about starphire crystal. starphire crystal is essentially extremely clear glass. Even the edges are clear (not green - see below).

While they could special order the cut & polished glass and put it all together they could not guarantee holding to machined (3 decimal) dimensions. They could work to +/- 1/32". This meant I had to get the glass done first so I could measure its actual dimensions before finalizing the machined base design. The starphire crystal was always referred to as 6mm but I was told it was actually 1/4". I designed a cover on Pro/E accordingly allowing me to give them the piece dimensions for front, end and top. They did a wonderful job on the cover but there is a slight overhang (0.023" front and back) on the top piece. It turns out that starphire crystal is not 1/4" or even 6mm but rather 0.227" thick. While this could have been avoided had I known the actual crystal thickness the very slight overhang is not readily noticed and does provide a sure hand grip when lifting the glass.

With the glass received and measured the base dimensions were finalized and prepared for machining. This is a case where I didn't feel my mill/drill was up to the job. The raw material alone was going to be over 30lbs and the rounded corners meant machining a curve (CNC) something my etch-a-sketch mill/drill couldn't do. I had previously submitted my paper drawing to several machine shops and went for the best quote. That turned out to be Alzar Industries on 77 Auriga Drive Nepean Ontario. Transfer of the design was easy. From my computer I simply dropped the Pro/E model into their FTP dropbox and they were off and running. It would take 8 business days to complete. While this was going on I turned my attention to the logo/LED board PCB.

The actual logo design itself came to me fairly quickly however the rest involved a bit of thought as it was to have 3 back-lit sections, blue, red and blue. The logo was etched out of 0.010 stainless shim stock. I had fotofab quote the etching but it was going to be over $1000 even with me supplying the phototools. As a result this part is currently not finished. This is why there is a bright glow in the lower center of the clock photos.

                

The logo sits against the inside of the front glass with both held in place by a 1.375 aluminum tube. The length of the tube was designed to allow the 30° light from the LEDs to properly spread. The tube is held in place by the front center mounting screw used to secure the clock PCB to the base. Inside the tube are 0.003" stainless steel shim stock "section separators" that keep the different colour lights within their own sections. By threading screws from the inside of the tube outward the 0.003" stainless steel shim stock "section separators" could be held in place by using the sloted heads. At the far end is the LED board providing the illumination. I designed and laid out this PCB for mounting the LEDs, connector and a brightness adjusting potentiometer. Power for the LED board is tapped from the +12V supply although a larger switched mode power supply (wall adapter) was required as the one supplied for the kit was near it's limit (850mA). The LED board is keyed to the tube so that it can't rotate out of alignment and is held in place by spring clips.

When the layout was complete the LED board PCB was sent out by ftp to APCircuits of Calgary Alberta for their "P1" service. These guys offer an amazing service for small quantity PCBs. Within minutes of dropping the files off and emailing them they contacted me and acknowledged receipt and that all files were OK. Periodically after that you get automated emails telling you where your boards are in the production run. If you get the files in before 11:00am the boards are shipped (Fedex rush) the afternoon of the following day and I had them in hand the morning after that. Amazing!

The first thing one should think about when putting up an antenna is lightning. Why? You are in essence, without any other precautions, building a lightning rod with the antenna mount acting as the rod while the coax cable is acting as the grounding conductor. A coax cable does not make for a good lightning ground conductor and dangerous and damaging potentials could develop. Take a look at the picture above of a typical storm passing over a town. I wasn't about to throw the dice and say that one day one of those 20,000A bolts of electricity wouldn't hit my so-so lightning rod! At the base of each of those hits above are people who had to live with the consequences of their individual decisions (good or bad). If you want to get an inkling of how likely are you to get hit check out this map. I can summarize the 3 pronged rational of the National Electrical Code (NEC) as follows.

By doing all these things you give an easy path for the majority of the hit to get to ground, you keep the voltage developed at the antenna low and if the ground potential does resultantly rise then all things are referenced together and they will tend to "ride the wave" together (a good thing). That's why it's important to ground the coax at the hydro ground point! Remember that the GPS is ultimately powered by hydro and if the coax & hydro ground points are different they may try to equalize with large currents (a bad thing). Apart from all the above having the antenna sitting on a good ground plane enhances performance too. The funny thing is having followed all of this I now notice all the satellite dish installations (done by cable providers) which are not done to code.

The next thing I had to consider was how to protect the antenna from the elements. The antenna as received has the coax entering its housing through a hole which did not appear to be sealed (it may in fact be sealed inside but I couldn't see it). Since I wanted high availability this meant the antenna had to have a good view of the sky which meant it couldn't be placed on a window sill but had to be placed outdoors and hence protected. This is where one must become a "McGuiver". I came up with the basic design fairly quickly but had a fair amount of difficulty finding an appropriate workable environmental cover. The first three iterations were rejected designs. I finally bumped into the solution at Home Depot while waiting for my 3rd cover to arrive. It is a lightweight white lamp glass which I could use inverted. The opening on the lamp glass was large enough to allow the antenna to fit through while the domed shape would allow snow and rain to be shed. Putting all the above requirements together led me to the following Antenna Installation Design. With this done the only thing left was to get on with the job and build it.

The design I came up with was based on materials I had in the garage so it was merely a matter of firing up my MIG, lathe & bandsaw and putting it all together.

The only item that I could not do was a special "D-punch" for the surge protector bracket. The surge protector I used was panel mount and needed a "D" hole so it wouldn't rotate when tightened to the panel. Here Forel Metal Products was able to help me out (photo below) as they had the 0.501" x 0.469" D-punch.

The rest of the bracket was pretty straight forward.

The pictures above show the result before mounting. In the left picture you can see all the parts relative to one another. The rear of the surge protector bracket with the black lug brace can be seen in the lower left. The right picture shows it with the environmental cover on. The #6AWG grounding connection can also be seen. In the lower left of the right picture the front of the surge protector & bracket can be seen. Since the 19ft antenna coax was not long enough to get all the way inside the house a weather tight exterior junction box (lower right) is used to protect the BNC connection to the extender cable. An AMP crimper (see the clock kit photo above) with RG174 hex crimp dies were used to make the BNC connections while a Paladin CST coax stripper was used to prepare the cable. All crimps when completed were protected with heatshrink while the connection inside the exterior junction box, after mating, was further covered with 1" heat-shrink and the ends were ty-wrapped tight to the cable. The #6AWG connections for the surge protector were simply made with a T&B lug crimper.

The following photos show the antenna installation (left), the cable run into the house and grounding (right).

Having installed it on a warm sunny day it snowed that very night (lower left) and rained all the following day. Below right shows the snow covered grey coax junction box above the green phone entrance box. The cables can be seen entering the house right next to the phone box.

Taking advantage of this test opportunity I fired up the clock PCB and within 15min it had completed a cold acquisition of the overhead GPS satellites. Everything worked and in the months since, no issues.

Final assembly presented no issues at all and the clock has worked flawlessly since I've finished it. I did make one modification to the board and that was to put a 1MΩ resistor to pull up the enable pin (4) on the MAX 771 power supply chip (that generates the high voltage for the Nixies) then cut the track to the PIC micro which normally holds it to GND and put a switch in series with that cut track. The switch is mounted on the back of the base. This allows me to keep the clock running at all times but be able to switch off the Nixies so as to prolong their life. Even thought the Nixies are off there is a tiny on board green LED which indicates all is well with it's 1 pulse per second heartbeat. Whenever I need a time reading or want to put it on display I switch on the Nixies but otherwise I generally leave them off. Given that there is a lovely soft hi-lo beep on the hour this works fine.

   

When they are on they are such a fascinating delight to watch. The combination of metal and glass is exquisite. It's delightful to watch the thick orange glow magically appear on the fine wire digit then instantly disappear as a different wire takes its turn. The oscillating change in depth as each digit lights up in turn is captivating. Within each tube each lighted wire lights up its neighboring wires and the glass just like neon bar sign would light up the night. Yet for all the nostalgia there is also a certain thrill to hear the NRC time signal countdown broadcast on CBC radio near 1:00pm. Sure enough at the exact sound of the last long beep the clock rolls over from 12:59:59 to 1:00:00. Now, despite the old saying, this man who has more than one clock is always sure what time it is.

Back to 

 

Back