I’ve uploaded the design files for my CRT deflection yoke driver board. This works for CRTs that use magnetic deflection. For a complete design, you will need the following boards:
ScopeVideoOnly – Video amplifier, focus chain, and nothing else (the ScopeDefl electrostatic design combines this amplifier with the electrostatic deflection amplifiers)
Gerbers are in the repository but you can get them directly from OshPark by clicking the links for each board above.
This board, unlike the others I’ve developed so far, requires both +12V and -12V. My projects typically use an Artesyn NFS40-7608J but it is now obsolete and a bit expensive, so you probably should use something else for power.
There are places on the board (C6/R9, C10/R20) for coil compensation components. You can figure out the values that you need with a little experimentation.
The board is designed for a vector-style yoke, not the far more common raster scan yoke that has a high inductance vertical deflection coil with lots and lots of turns. It can usually drive the horizontal coil no problem, but you’ll want to modify the vertical winding to reduce the number of turns. Check out my blog post on winding deflection yokes for more information on making your own.
The LM4765 audio amplifier, which drives the deflection coils, will dissipate a lot of heat so you will need to bolt it onto a good-sized heat sink (at least 3″ x 3″ aluminum with fins, not some dinky little TO-220 clip-style heat sink).
Circuit operation is pretty straightforward. CRT electron beam deflection is proportional to the magnetic field which is proportional to the current in the coil, so the LM4765 controls the coil current (measured through R14/R25) instead of the voltage. The current is therefore directly proportional to the input voltage (X or Y). An extra gain stage facilitates the width/height and left/right offset adjustments.
At the electronics flea market I recently found something particularly interesting…
This is a vintage Philbrick K2-W vacuum tube operational amplifier! Turns out they have quite the following. Sadly the original 12AX7 tubes were gone–somewhere, someone probably has rare GAP/R marked tubes in their guitar amplifier. I put in some generic replacements.
I decided to build a little jig to try it out.
This is based on the inverting amplifier schematic given in the K2-W datasheet I linked above. I added a simple linear power supply to generate the +/-300V rails. If you build your own supply, be sure to add bleeder resistors so you don’t get a nasty surprise after you turn it off and try to work on it.
After connecting a 10K series and 100K feedback resistor to the op-amp, I ran a 1KHz 5Vp-p square wave from my function generator into the circuit and saw this:
Neat! The top trace is the input and the bottom trace is the output. The bottom trace has a magnitude of 50Vp-p, as expected.
It’s really interesting to see how the short paragraph of specifications at the bottom of the first page of the K2-W datasheet developed into the formal electrical characteristics tables you can see in more modern op-amp datasheets, like the 741.
If you build either of those and try it out on an STM32F407 Discovery board, you will notice that no sound comes out. The reason is that the onboard audio codec has a LRCK line (left/right framing clock for the I2S audio stream) wired up to GPIO PA4, which happens to be one of the DAC outputs that we are using for the vector graphics generation. You’ll need to disconnect LRCK (U7 pin 40) from PA4 and wire it to PA15. I did it by cutting a trace (indicated with a black arrow) on the back of the board to isolate the PA4/LRCK net from the STM32:
And by adding a wire (indicated in red) on the top of the board from R48 to pin 77 (red arrow) of the STM32 (PA15):
Oh, and one more thing. To keep the audio mixer marching along, we had to connect PC6 to PC7. This ties the I2S MCLK to timer 3 channel 1, which triggers a periodic interrupt that mixes new samples into the audio output buffer. The easiest way to do this is to connect a jumper across pins 47 and 48 on the GPIO header P2. They are clearly marked on the board and the pins are right next to each other so this is easy.
If you follow me on Twitter, you’ve probably seen the picture of a CRT deflection yoke undergoing surgery. Well, it’s now installed in a mini Battlezone arcade cabinet that I built for Maker Faire this year.
It uses a 5-inch B&W CRT taken from a broken security camera monitor. I had to rewind the yoke to make the vertical windings fast enough to handle vector graphics. The horizontal windings were already pretty close, but they did need to be rewired to operate in series to reduce the drive current.
The CRT is a bit different than most of the old-style CRTs I work with. For the curious, it uses an EIA E7-91 style base which has an additional grid for acceleration purposes. The lower circuit board in the photo is a potentiometer resistor divider that develops a ~300V bias voltage to drive this additional grid. The upper board is a Cockcroft-Walton multiplier circuit to step up the 1KV power supply voltage to the 4KV necessary for the post-deflection acceleration (PDA) anode. I’m running this CRT a bit under voltage, and it works fine because I’m not trying to fill the whole screen with a solid raster.
The metal enclosure houses the power supplies–it is made out of soft steel with the idea of containing the stray magnetic fields generated by the power supply switching inductors. It works a bit, but I’ve had better luck in the past with mu-metal shields around the CRTs themselves. There are three power supplies: one steps 120V down to +12V, +5V, and -12V. Another steps +12V to +1KV, and the final one steps +12V up to +60V. These last two power supplies are detailed here. They are open-source hardware so you can build your own.
Outside the shield you can see the video amplifier board at the lower left, which is the same circuit used on my electrostatic deflection board but without the deflection amplifiers. At the bottom right is my new magnetic deflection amplifier board which drives the deflection yoke. It is based around an LM4765 audio amplifier IC. I’m not totally happy with the design but eventually I will release the board design. The green board in the upper middle is an STM32F4 Discovery board. The board to the right of it is an audio amplifier that drives the speaker.
The joysticks I machined out of small aluminum blocks. It took some experimenting to figure out how to arrange the springs to get the joystick to return to the center position. The joystick handle on the right is hollow to allow the wires for the fire button to pass through.
I also build a second unit with the same driving circuit but a different CRT–a 5-inch round 5AXP4. This CRT would originally have been used by a TV technician to look for problems with the chassis or the full size picture tube in a customer’s set. The glass is extra thick because the tube was designed to be handled more. This is the same tube that I started experimenting with last year.
