No-Silicon digital clock

I noticed that the dekatron that is counting the tens of minutes sometimes seems to miss a step. It seems to be geting worse when the clock is left on for a longer period of time (several hours). I guess I will have to look into that. I remember that this particular tube has shown some problems before, when I was building. Because it appeared to have a problem in a certain step, I decided back then to have it run in the opposite direction, e.g. counterclockwise in stead of clockwise. Which is kind of strange for a clock, admittedly. But it is as easy as swapping the G1 and G2 connections. 

I have already tried to change the triode that is driving it, but that doesn't solve the problem. So the problem is either in the resistors and capacitors generating the phase shift G1/G2, or it is in the GS12C itself. I still have one or two of these lying around, but still it is a lot of work, because this particular tube does not sit in a socket, but the wires are soldered straight on. Maybe it's best to start with some measurements of the G1/G2 waveforms and then some tweaking.

Yesterday, I noticed a problem with the No-Si digital clock. The hours figure show something unreadable. On closer inspection, it looked like the '5' was always burning regardless the actual hour, through the intended figure. I thought a possible cause might be the triode tube, so I exchanged the responsible E90CC with another one in the clock. Bingo, now the '3' was constantly burning. So yes, the tube was the cause. Luckily, I have some spares laying around, so I put in another E90CC and the problem was solved.

Flash back to a few years ago.

I have been experimenting and building circuits with dektrons for years now. Absolute highlight was a project I did for Elektor, the printed magazine in 5 languages. The original title of this was: I finally found a useful application for dekatrons. It was published in 2022 under the title Audio Spectrum Analyzer With Dekatrons. Which is exactly what it is: it uses 7 dekatrons to display the levels of audio in the different frequency. Stereo, also. You can still read this story at https://www.elektormagazine.nl/labs/i-finally-found-a-useful-application-for-dekatrons

In 2023, to get inspiration I visited the WITCH computer in the National Museum of Computing, near Bletchley Park, which is comprised of hundreds of dekatrons. The people working there have managed to get the machine running, which I think is an incredible feat. They do a daily demonstration.

In July 2024, I got the opportunity to buy a couple of 12 step dekatrons (GS12C), about which I didn't hesitate a second. I had no clue what I was going to use these for, but I knew I needed them urgently. They had been laying around for a couple of months until the idea to use them in a digital clock gradually matured. 

In my previous projects, I had been using dekatrons only as display devices. I did not use them for their original purpose: as counting (up/down) device. That was because all the "counting" was done in a microcontroller and the microcontroller directed the dekatron. This bothered me for a while. I wanted to do something where the dekatron was used for its original purpose, to count things and not be subordinate to a microcontroller.

I did some experimenting on how to couple the 'carry' on the output of one dekatron to the input of the next dekatron. Two document showed how to do this: the Sylvania brochure Decade Counter Tubes second edition and the Ericsson book Electronic Counting Circuits by J.B. Dance (chapter 4). but of course, the circuits were intended for different types of dekatrons, for different types of triodes and for different supply voltages then I intended to use. So I had to do a lot of experiments before I got it right. Luckily, the documents described very well the intended time-shifted waveforms on the two guide electrodes G1 and G2. So I could measure the waveforms on the oscilloscope and choose component values that gave the intended waveforms.

I took the idea from Sylvania to connect G1 through resistor-and-capacitor-parallel and G2 through an RC-filter-network, as illustrated below.

In the Ericsson document, they use a different RC topology, e.g.:

I found out that dc coupling the cathode of a dekatron to the grid of the triode means that the triode is conducting for a longer duration, which means that the power consumption is relatively large. This also leads to the necessity of a high-watt anode resistor. This can all be avoided by introducing a small capacitor between dekatron-cathode and triode-grid, which has the effect that only a brief pulse is given to the triode.But this technique cannot be applied for driving the nixies. So in the end I used both AC-coupling and DC coupling.

In chapter 4.4.9 of the Ericsson document, it is described how to drive nixie tubes (Ericsson calls them Digitrons) from a dekatron. Unfortunatly, the use a GCA10G type of dekatron for this puropose, which has 10 auxiliary anodes, that can be directly tied to the figures of the nixie tube. So I had to design a way to solve this for a normal dekatron. That proved to be quite easy, it only requires a lot of triodes: 10 triodes for one digit. 

I found out the existence of the E90CC double triode tube, a tube specially intended for digital purposes. Fun fact about the E90CC that it clearly states in the datasheet that this tube is 100% useless for audio applications, because of microphony and hum. I really like this, because it means that not all of them have been taken by the audiophiles.  I found someone in The Netherlands that would supply me with the desired quantities of tubes and sockets at a low price. In fact, the tubes cost no more than the sockets.

I decided to publish the on-going project on Hackaday.io and submit it to their 1Hz challenge. This meant to speed up the process. I went on to create an aluminium base plate, which was a large amount of work, given the tools that I have at hand. Some woodwork, find transformers for the power supply. Enfin, you can read this all in the project log.

Presented the No-Si Digital Clock to the members of the family. Wife and kids (age 20+) were quite amused, I think they even liked it. My son wanted to know exactly the rotation time of each dekatron. 

The clock is standing in the living room next to the tv now. We'll see how annoying the continuosly spinning dektrons are when viewing tv. Especially the very first one, rotating five times per second.

I think I will leave it there for a week or so and then it'll go into storage. Just to get it out sometimes to impress friends and family.

The No-Silicon Digital clock is ready! It is working correctly now. You can set the minutes by turning the "time adjust" knob counterclockwise and adjust the hours by turning it clockwise.

It consumes a little over 80W (mainly consumed by the filaments) and keeps the room nice and warm.

No, not tubeless, just one tube less... 

The filaments of the rectifier tubes and triodes consume a lot of power. In total, 70W. In my transformer drawer I found a big transformer, with secondary taps of 23V, 24V and 25V. Therefore, I decided to work with strings of 4 tubes in series. Being all E-series tubes, 4 x 6,3V = 25.2V would fit well enough with the 25V output of the transformer. So, 4 strings of 4 E90CC's and one string of 4 EY88's. Although I needed only 3 EY88's in my design, I added one extra, doing nothing, just hanging around in the corner as a dropper tube. 

While gradually realizing the No-Si Clock, I realized that I would need some kind of mechanism to set the time initially. I came up with a switch with three positions: one normal position, one position in which the minutes tick switfly by and one position in which the hours pass by. But where am I going to put the switch? then I remembered the useless extra EY88. If I could do without that tube, I would have a place to put the switch. The solution I ended up with is to use not EY88's, but PY88's. These were used in series configuration, directly connected to mains voltage in tv sets of the day. They officially require 26V filament voltage each, so instead of stringing them in series, I could just use 3 of those, connected in parallel to the transformer's 25V. They will also work on a volt less.

Now I have room for my switch and additional advantage: the power consumption drops by almost 10W (6,.3V×1.55A).

Photo: The time adjust knob sits where a useless EY88 used to sit. You can see remaining three PY88 rectifier diodes. The switch on the far right may be the one I am going to use.

I was working on the part of the circuit that deals with the tens of minutes. I had the dekatron wired up. Every time it gets an impulse, it should jumps to the next stage. I have used a GS12C, which has 12 stages. I have connected stage 6 to stage 0, stage 7 to stage 1, etc, thereby halving the number of outputs to 6, which is good, because it's the tens of minutes and after 5 it is supposed to go back to 0. So we will count 58, 59, 00, 01...

The particular problem I had is that the dekatron stepped through the stages, but got stuck on 0. It would not move from 0 to 1. I checked all solder joints, resistors etc. but didn't find a problem. The dekatron's inputs are two guides, G1 and G2, which are supposed to get an impulse, briefly shifted in time after one another. If you give an impulse to G1 and then to G2, it moves one stage up. You can also reverse the effect, by giving an impulse first to guide G2 and the to G1, which will cause the dekatron to move one stage down. 

As an experiment, I reversed G1 an G2 and the dekatron started to move in the reverse direction and - a little to my surprise - did not get stuck anywhere. So for now, I think I will leave it this way. I have not connected the nixie tube yet; in this case I consider it an advantage, because I will have to reverse the nixie wiring to compensate for the reverse dekatron.

It looks a little like cogwheels now, like in a mechanical clock. Instead of looking at this as a problem, I might see this as a feature. I'll have to reverse some of the other dekatrons also, I guess.

Major milestone: I've got the 1 Hz working. Mains frequency is given to the rightmost dekatron, which spins with 50 steps per seond, that gives 5 full rotations per second. The outputs from states '0' and '5' are taken, inverted by an E90CC triode. This gives a 10Hz pulse train, which is given to the second dekatron, which makes 10 steps per second and therefore one full rotation in a second. See the attached movie in the files section.

Mixed day. After I blew up my function generator two weeks ago, by applying 300V to its output, today the magic smoke escaped for no reason from my 450V home built power supply. My indestructible 300V Delta Zierikzee power supply unfortunately just doesn't give enough for the Dekatrons to fire up. thankfully I found yet another home grown power supply in the attic, up to 400V.

This helped me to find one problem why the Dekatron wasn't spinning on first try. It appeared that for some reason, a large DC component appeared on the gates of the dekatron, where I would expect only 50 ~ 100V AC, phase shifted between G1 and G2. I still haven't pinpointed the cause, but I solved the problem by adding a 390nF capacitor, blocking whatever DC component there is.

Had it working. Then it stopped working again. This turned out to be a loose solder joint with one of the pins of the dekatron. With that solved, I have a spinner. Next stage is to get the output signal from the spinng dekatron. That should be 10Hz, to be inverted by one of the triodes and fed to the second dekatron, which should churn out the desirable 1Hz.

Yesterday and today I worked on a wooden base plate, to put everything on including the heavy transformers. I glued three pieces of wood together, cut it and painted it. The aluminium base plate will sit on the wood using a hinge mechanism, so that it is easy to inspect the back where all the soldering should take place.

I soldered the rectifier circuit with EY88's. so I now have something like 200V and 400V. The EY's and all other tubes that a filament voltage of 6.3V will be connected in chains of 4, connected to 25.2V. Or something in that neighbourhood. 

I preliminaruly connected the nixie tubes so that display something (it's currently 12:06 all day). I started wiring the first dekatron circuit. It gives a sign of life but it isn't spinning yet.