The NBTV Mixamp
This article also appears in the NBTV newsletter Vol. 45 No. 2
Years ago I began a long term project attempting to recreate the BBC's facilities for broadcasting 30-line television circa 1935, as well as for 32-line NBTV. The eventual goal is to produce television programs using mechanical equipment and demonstrate both the quality and limitations of the original system, albeit with modern components.
There are three major pieces of equipment to build. The mirror drum projector-scanner (yet to be built), the caption scanner, and the electronics that control and process the video signal. The caption scanner (which also works very well as a general purpose flying spot scanner) is detailed here and in the NBTV Newsletter Vol 43. No. 3. This article covers the mixing and processing electronics. The mirror drum projector will be the most difficult apparatus to construct, so naturally it will be built last.
The processing electronics are housed in a box called the Mixamp, short for Mixing Amplifier. It is essentially a four channel video mixer, a timing generator, and a few other mechanical television circuits all in the same box. I wanted it to look at least a little like the original; using knobs instead of linear faders, so I was excited when I came across an old Russco Studio/Master 505 audio mixer in a radio station's junk pile. The circuitry had already been modified by the previous owners so there was no shame in gutting it for my project. The prominent, oversized knobs and plywood body/aluminum face made it perfect for custom modifications.
Above: Finished Mixamp, Below: Unmodified Russco Audio Mixer
Douglas Birkinshaw at the controls of the original vision mixer at the BBC in 1932
Circuit Description
A master timing generator in the form of the club's Triple Waveform Generator circuit feeds up to two scanners or cameras with the same signal through a pair of 1/4” phone jacks on the back panel, providing a "genlock" function. Since this machine is able to handle both 30-and 32-line television systems, the adjustable RC oscillator on the TWG is used instead of a fixed crystal. A multi-turn trimpot allows fine tuning of the required 375 Hz or 400 Hz line frequencies. It also retains the ability to produce simple, repeating line waveforms (as well as a white field to act as a transition, see NBTV Vol. 36 No. 2) for setting up a monitor and calibrating the downstream processor. Following that is an adjustable gamma corrector and sync inserter in the form of a slightly modified Karen Orton NBTV Standards Processor described in Vol. 38 No. 1. Using this circuit a gamma curve can be applied when necessary and the signal limiting indicator is a very useful feature for adjusting levels dynamically.
Rear of the Mixamp. From left to right: AC power input, XLR plugs for photocells (grey one is AUX in), 1/4" phone jacks for sync out, RCA audio pass through, RCA NBTV out (in red)
The inside of the Mixamp. A hinged lid allows for adjustments to the trimmer pots on the circuit boards. The Triple Waveform Generator is on the far left. The mixer's original circuit board was stripped of components and used as a base plate for my additions.
The inputs are separated into two buses. The "A" bus consists of the four photocell panel inputs (the first four large knobs) and the "B" bus accepts the output of a switch for selecting either the TWG patterns or an external auxiliary input meant for the caption scanner. A crossfader allows the mixing of the two (the effect of superimposing the studio and a scenery card is described in period magazines).
After the mixer there is an amplifier that makes up for the losses in the mixing resistor network and provides a bit more boost to the overall signal. This is a simple emitter follower which inverts the signal and is paired with another unity gain inverter that rights it again. Because of this configuration, I included a board-mounted switch that provides the option of producing negative video simply as another bell and whistle. The mixed and amplified signal is then sent through a master level pot and into the Standards Processor, where gamma is applied and sync pulses are inserted to create the composite video output. This is fed to two RCA jacks on the rear panel, one for a monitor and another for a recording device.
A front panel line level volume control for an external microphone setup is included as a convenience and consists of only a potentiometer and an input/output jack in the rear. For now I'm using the audio channel from an old camcorder on a tripod off to the side to pick up sound for testing and demonstrations. Finally, there is a 12 volt BNC jack in the top left corner for a gooseneck lamp so the controls can be seen in the dark. The front panel layout was designed in Sketchup. I don't have the facilities available to etch aluminum, so the front panel itself is made of thin plywood with a vinyl "brushed aluminum" sticky paper finish laser-etched with the label markings.
The Pickup Panels
Since this is a flying spot system, the lighting panel is actually an array of photocells picking up the light reflected off of the subject. The panels are modeled after those used in the BBC's studio, and two have been built so far. The bodies are made from 1U rack mounted aluminum enclosures that were composite video switchers in their previous lives. The photodiodes and the XLR jack are mounted on what was originally the bottom of the panels due to their thick, flat bases. After the holes for the wires and mounts were drilled they were covered in grey vinyl craft paper to hide the ugliness of the underside and the previous mounting holes from the original innards. The panels are bolted to a custom made yoke mount and set on a pair of old lighting stands.
Unlike the BBC panels which contained only phototubes and no other active electronics, my panels each have a high gain op-amp based amplifier attached to a large area photodiode. The amplifier was designed by Jeremy Jago and published in Newsletter Vol. 38 No. 1 as a test platform for the suitability of different photodiodes in NBTV applications, and in practice works quite well as a general purpose head amp. I've substituted the CA3130 with a TL072 and, being a dual op-amp package, it can accept two photodiodes per chip. Therefore, there are a total of eight head amps in each panel with their outputs combined via a 10k resistor. Using this many head amps in parallel provides two advantages. One is that the lighting appears softer as the sensors are spread out over the panel's face. The other is that any random noise generated by the individual amplifiers partially cancels out. I 3D printed a plastic seat to hold the photodiodes in place and to interface with a removable LED reflector for collecting more light in distance shots. The sensors used are Silonex (now Advanced Photonics) SLSD-71N700s. I chose these rather large rectangular cells to make the most of the available real estate within the reflectors. Despite their large size, the SLSD-71N700's internal capacitance does not seem to be a limiting factor. I did, however, have to add a 3pf capacitor around the 2M feedback resistor of the head amps to combat an issue with ringing. During development I did also try using an unbranded Chinese-made 10x10mm photodiode that was in a much more convenient potted ceramic package rather than free hanging leads, but even though its surface area is smaller, the speed response was so poor that it was completely unusable for this purpose.
Pickup Panels with removable reflectors
Head amps as seen with the rear panel removed.
Results
Personally I am very excited by the results this system has delivered so far. However, due to the overall complexity, reliability has been an issue and repairs and adjustments are difficult on a system that can only be operated in total darkness. It is also very sensitive to interference from anything from streetlights to indicators on electronics to infrared proximity sensors on mobile phones. I will also be reminding myself in the future to stick by the KISS principle (Keep It Simple, Stupid). It works to an advantage in the long run.
Here is a description of the lighting setup used in the following off-screen photos. One pickup panel is placed on top of the scanner tilted down towards me, the other is on a stand off to my left facing the wall which is about six feet behind me. By illuminating the wall you reduce the spotlight effect that you see with many flying spot systems. The side panel is just in front of me so it is also able to catch some light reflected off of the side of my face. Overall this lighting arrangement works well. I am excited to try other lighting techniques since I read that at one point there were as many as nine banks of photocells in place in the original studio to choose from!
Both panels on (both face and wall lit)
Side panel only
Side panel only, behind me facing the wall
Shadows from the front panel. Remember that no light is shining from the panels, but shadows will still appear if the subject blocks light from reaching the photocells.
Front panel placed below me for a firelight effect
Superimposed "Pulse and Bar" Waveform