Solid State Argus

It was the beginning of the coronacrisis. I got a tip that some old equipment was available for pickup. A good friend drove me to the location. Amid the stuff was a type 62 indicator unit. At first I did not know what it was for but it turned out to be part of the Gee navigation system used in WWII for the bombing raids into Germany. A documentary about the Dambuster (on UK Channel 5) rekindled my interest because it very briefly showed this very unit. Which was not used on those Dambuster raids... On Antique Radios forum and on Buizenradioclub forum (in Dutch) I posted some messages documenting my new project. The 62 unit had been completely stripped and while people suggested I follow the excellent rebuild as done by Dr. Hugo Holden I decided to go solid state.

In the early fifties these 62 units were widely available in army dumps and the like. Many people followed the instructions to build the Argus TV in Practical Television (see Dr. Holden's article). There also was a Dutch version from Radio Bulletin called Oog in Al (here, here, here and here). Interestingly, while PT's design was based on a whole new chassis, Radio Bulletin's TV actually was based on the 62 unit. As I had one now myself I decided to go the RB route.

A big project like this is always broken down in stages and this one is no exception. I started with the power supply, then the deflection and ended with the video processing. As this TV has no tuner it is video only which simplifies matters significantly. There are no analog TV transmissions anyway. At the time there was only one video source: the single TV channel on air. In RB's case channel 4 Lopik (first Netherlands TV network).

Anyway. On to the project. From my junkbox I chose a nice toroid:

This would provide the raw input for the SMPS. For that I decided on the good old UC3843 SMPS controller. This drives an IRF740 MOSFET which is connected to a flyback transformer that generates all high voltages (click to embiggen):

Because the PSU has to provide so many voltages, I popped it in LTSpice. At the same time I built it:

The various voltages are:

I also wanted to use the UC3843 for the horizontal deflection but that turned out to be problematic because of power supply ripple from the transformer (raw 11 V) and the switching spikes of the MOSFET. I eventually used a separate sawtooth generator for that. The transformer is an old Siemens E-E core with an air gap:

The sim in LTSpice was very useful to reliably design the transformer. I still don't understand why I needed R12 (2.2k). Eventually the PSU turned out like this:

I could have designed it better but this works. The filament of the CRT gets fed in the same fashion as modern color TVs: with a winding on the transformer. This works great, I tried it with a dummy load and it gets exactly as hot as with a DC supply. The sawtooth generated on pin 4 due to the JFET current source is not used but I left the circuit in place. The JFET can safely be left out. The PSU SYNC is an absolute necessity to avoid interference in the image.

Now the power supply was working it was time to attach the CRT to the circuit:

The 300 V for the astigmatism control is actually the +252 V from the PSU. Which is used nowhere else... Once it was done I hooked it up and was rewarded with first light:

Note the piece of foam plastic under the PCB. My workbench is ESD safe and the high resistance of the bench top is not friendly for high voltage circuits.

The deflection circuits were next. Quickly I kludged together the horizontal and vertical circuits:

And was rewarded by first frame!

But when I started on the video output it soon became apparent that the horizontal deflection was seriously flawed:

Ugly lines ran from top to bottom due to RF ripple in the horizontal sawtooth. Only when I employed a dedicated horizontal line oscillator the lines disappeared:

The whole video processing now looked like this:

The BD115 was left over from a Philips K9 color TV and posed some bandwidth issues. Again, I turned to LTSpice to solve this:

Still, I could not get a better frequency response than this without the transistor slewing on me. The point is moot though because the CRT's resolution also isn't all that great.

The DC restoration at the wehnelt also posed a challenge. I wanted to do a proper back porch DC restorer but the optocoupler that I intended for this purpose was too slow. I discovered that not only there were no faster opto's in my tray than the PC817 I had used, a faster one still would not have worked properly. So I resorted to the good old symc tip clipper. Works fine as long as there is video, then the screen goes ultrablack. Which doesn't really matter anyway.

I then wrapped things up. By now the layout was complete:

I put everything on the chassis and wired it up:

The video goes into the white BNC on the lower right and the figure 8 power plug goes in the socket on the lower left.

But when I applied video to it I was underwhelmed:

I had overlooked the tiny little detail that my power supply is not stabilized which wreaks havoc on my image! What I should have done from the start is use an extra 78L08 regulator to feed the timebase circuitry. Or maybe even feed it from the 5 V output provided by the PSU chip although that may be too low for using the JFET current sources. Also, picture brightness is terribly low because of the age of the CRT. And there is more: the green filter has a piece broken off at the bottom and should be replaced. The vertical linearity is pants. Still, this was a great project to spend time on during the first months of the corona crisis.

But in case you are interested in the period correct Philips test pattern, here it is:

Date: 21 December 2021