Shortly after obtaining the 310A tubed scope, I became interested in how tube technology works. After learning the basic principals of how vaccum tubes and cathode ray tubes work, I wanted to build a clock using a CRT. I have seen scope clocks in the past but I lacked the knowledge of how to program and it required a fairly high voltage power supply. After reading Riad's and Thomas's webpage about their scope clock project, I already started thinking of ideas on how to write the program. Shortly after, I ordered the parts to get started.
The scope clock uses a vector based graphics display. It is controlled by an Atmel atmega168 running at 8Mhz and a AD7302 dual 8-bit digital to analog converter. Timekeeping is kept by a DS1307 real time clock and a DS32Khz temperature compensated crystal oscillator for very accurate timing. My clock is heavily based on Riad's but the main differences are that I used my own processor and software and a different CRT (3RP1A). Since I am using an 8-bit DAC, it is possible to have 256 different voltage values between 0 and 5 volts in 0.0195v increments. The CRT displays a small dot, so to control the dot, the X controls the vertical and the Y controls the horizontal deflection. The X and Y high voltage amplifiers have 0-5volts input and +380 to +1100volts output to their respective deflection plates. The font is a set of (x,y) values programmed in the processor. You can think of the display as only operating in quadrant 1 of a graph. The value (0,0) is in the bottom left hand corner, (0,256) is in the top left hand corner, (256,0) is in the bottom right hand corner, and (256,256) is in the top right hand corner. Now you can see that each number and symbol is just a combination of many (x,y) values. To save time, I used a simple DOS program called Targa92 that uses a .tga picture file to generate vector points. I made the colons by physically plotting it out on graph paper. There is also a screen blanking circuit to blank the screen in between writing numbers. This is necessary or else you would see a line going from one number to the next number. At this point, it's only of a matter of displaying the correct numbers at the right position on the screen and blanking the screen at the right time. The power supply consists of several different voltage values: -380, 0, +380, +780, +1100. This is possible by using a few resistors, diodes, capacitors, and a Hammond 270x transformer that outputs 5v, 6.3v, and 240-0-240v (480v) AC. The +5v for the logic is provided by the +5v AC line rectified to around 7v DC and then the 7805 provides the regulated 5v. The 6.3v AC provides power for the heater in the CRT. The high voltage power supply is basically a few Greinacher voltage doublers in parallel. There is a magnetic shield over the CRT to eliminate the magnetic fields produced by the transformer. Since the CRT uses an electron beam to display the image, it's affected by any neighboring magnetic fields which causes the display to be unreadable.
As you can see here, there are "droops" at the beginning of every number. I have tried tweaking the blanking timing for hours to try to reduce it but this is the best i've gotten. The droops are only caused from the beam turning on (unblanked) but when turning off the beam, there are no droops. Take a look at this oscilloscope screenshot with the same x,y outputs from the DAC, there are no droops. I believe that's because the scope has way more bandwidth than the X and Y amplifiers used in the clock. I will be doing some more tests in the future to try to improve it. I also had another strange issue: The clock was powered on for about an hour and suddenly I hear a "pop" and a see a blue flash above the HV transistors. The clock was right next to me at the time and I think I saw a fly above the clock when this happened but I wasn't able to find a carcass. I believe a fly flew in between the HV transistors, which I know have about 800v across them, and blew it to smithereens! There wasn't any magic smoke or any signs that anything had happened but of course the clock stopped working. I figured a few transistors got internally blown but after some testing, one half of the DAC stopped working and the focus pot was stuck at 200k ohms. I definitely thought that those would be the last parts to die but apparently I was wrong. The clock works fine after replacing those parts. At least I know that it's one hell of a fly killer!
I was able to "hide" the droops in the vertical parts of the numbers. This was done by modifing the starting points, ultimately placing them in the vertical section. It doesn't completely get rid of them but it definitely looks better. The modified font can be seen on the first picture below and the video. All other pictures were before the modification. More will be added soon.
I was able to get the numbers to look even better but there are still some problems that need to be worked out. I hope to eventually rewrite the code in C++. I finally built a plexiglass case for it. I think it looks nice but it was a real PITA. Next time I'll find somebody that can easily cut plexi with higher precision. All new pictures are above.
Time selection is done by two buttons, one selects (underlines) which digit you would like to change and the other advances the number. To adjust the screen size and voltages, there are 8 blue potentiometers. Going from top to bottom: Anode/Astigamatism, Cathode, Focus, Brightness, Y position, Y gain, X position, X gain. I also have a "screensaver" programmed to move the whole display slightly every minute (there are about 120 different positions). The screen will eventually burn if you don't do this. I also plan to add the date above the hours and minutes. The clock is going to be in a handmade 1/4in thick plexiglass case. A schematic will be available soon.
Contact: Tai Oliphant
Copyright © Tai Oliphant 2010