New transport speed, tension, and framing control approach

Below is a system diagram showing how I intend to control film speed, tension, camera shutter (framing) and backlight flash. It is not a scale drawing, and only shows the system components in the context of the tasks they need to perform. In particular, the film gate area is only conceptual, but it is my intention to keep the film path as straight as possible to reduce drag and reduce film stress.


The heart of the system is a Phase Locked Loop (PLL) device that will probably be implemented in an Arduino, along with the tension controls and user interface.


A Phase Locked Loop is a simple device capable of synchronising two signals so that not only do they have exactly the same frequency, but they also have an accurate timing relationship between each other (known as ‘phase’). A PLL can be constructed in both hardware and software, and many millions of them are in use constantly all over the world empowering such diverse things as communication and entertainment devices, to automobile systems, sattelites, and heart pacemakers.

These devices can use either analogue or digital techniques, depending on the application. In the case of the Kinograph, I am using digital which is normally used to synchronise two pulse generator circuits. However, in the Kinograph one of the pulse generators has been replaced by a combination of the take-up motor, the film, and sprocket hole sensor. In this way, the speed at which the sprocket holes pass the sensor is synchronised to the rate at which pulses are produced by the pulse generator (our speed controller). Further more, frame alignment in the camera can be adjusted electronically by changing the phase between these two pulse streams. Because of the inertia of the take-up motor,and film reel and smoothing within the PLL, this system will act like a ‘flywheel’, and be tolerant of missing or broken sprocket holes. For more information, there is a detailed technical discussion on PLLs on Wikipedia.


Both take-up and supply motors are DC permanent magnet motors and they are driven by constant-power circuits. This is to aid in maintaining fixed tension regardless of the diameter of the film on the reels. Some simple maths illustrates this :


So, if the film speed is doubled, then the supply reel power must be doubled to restore tension. If film speed is fixed, then the power can be pre-set, but if a range of speeds are required then I show a tension measuring device in the form of a ‘dancer’ which feeds back to the motor to maintain original tension.

There is an additional facility to stop the system if there is a film break, jam, or end of reel is reached. For this to happen, both the dancer will have sensed unusual sustained tension and the PLL will have dropped out of lock.

Because the two motors are identical, it should be possible to reverse the film direction, and fast rewinding should also be possible if the film is passed through the centre rather than around the dancer rollers. Also shown are two Particle Transfer Rollers (PTRs) which attract particles of debris from both sides of the film.


The ‘dancer’ rollers are a particular challenge because they need to have virtually zero-torque-friction of their own, which means measuring their angle cannot easily be done with a simple potentiometer. To overcome this, I propose using a cheap magnetometer chip, like those inside cellphones to provide a compass. The dancer will have a small low-field plastic magnet fitted to it, which will rotate above the chip providing a continuous stream of digital serial angular data. The dancer mechanism itself will be 3D printed to operate over 90 degrees, and will have an integrated helical (spiral) spring as part of its design. I do not yet know whether the mechanism will need any form of damping.

I have chosen to use a reflective laser method for detecting sprocket holes. This provides the greatest contrast ratio, makes no difference if the substrate is black or transparent, and is the most suitable method for 9.5mm film with central sprocket holes.

I am still undecided about the design and illumination of the backlight. My preferred approach is to try using an integrating light sphere, as it promises to give even illumination and significantly reduce the visibility of scratches, but it may be difficult to find space for it.

At the moment I am only interested in films with single perforations per frame (8, 9.5, 16 mm), but I have a modified scheme for 35mm (4 perfs per frame) which should be equally viable. Matthew is currently previewing it, and I will add it here in due course.

This system I have just described is a bit of a Rolls-Royce solution, but I am also looking at a very much simpler version that has no control elements as such, but still maintains constant tension at a fixed chosen frame rate. The downside is that everything may need adjusting manually, including framing, for each film. When I have more details I will update this thread.

1 Like

Question: If you’re able to maintain constant torque via electrical means, couldn’t you eliminate the dancer rollers?

Yes, but there would be no way of knowing what the tension was, or the ability to adjust it. Also, constant tension (provided by constant power drive) only overcomes the varying diameter of the film on the feed reel. If the film speed is changed, the tension will change. The dancer will restore it.

If the film is only going to be run at one fixed speed, and you are happy with the tension (because it looks and feels about right) then the dancer can be removed.

BTW this system maintains constant tension, not constant torque.

1 Like

Yes, makes sense. The only thing I find confusing is changing the film speed. When running a telecine, I would expect the film speed to be constant, so I’m still unsure why a dancer would be necessary at all, unless it were to take up shock in case of a sudden stop or some unusual condition like that. Or to put it another way, what would be a reason to vary the film speed?

The requirement to be able (or not) to choose a transport speed is entirely up to the user. The control system I propose lends itself very well to making this adjustable, in fact it is almost a no-cost option.

If your Kinograph will be handling a range of film guages, then variable linear speed would be welcome. Also, like many systems, there will probably be a ‘sweet spot’ where performance vs transfer speed will be at an optimum (considering camera/capture-software constraints). Variable speed will help you find this.

Of course, adding a little compliance in the film path can be beneficial as you pointed out, and the dancer conveniently provides this.

I quite like your design, especially trying to keep the number of rollers to a minimum, and keeeping them in an almost straight line.

Do I understand it correctly, that the dancer is used to both measure and control the tension?

So at times, the film would pass through 2 additional rollers (which on your not-to-scale model seem quite small, I imagine they will be as large as the others?) Wouldn’t it put less strain on the film if you’d replace the two rollers with just one roller that could push the film, and thus increase or decrease the tension?

@VitalSparks do you know what chip you’ll be using for the PLL? I’d like to learn more with some hands on experimenting.

Hi, new member. I joined as I’m interested in this project. I bought a 3D printer and would like to explore building a telecine for 8mm film with no audio. A family member has some old film that was transferred to VHS, but I like to see what this project could do.
Looking forward to more updates.
There is a Teensy forum that I visit that have a digital PLL project, which I may investigate for vitalsparks’ project. I also have some '4046 CMOS PLL chips that can be used. However, I first need to understand the process some more before I can contribute.

Welcome, Peter. Glad to have you!

So the force of the spring is responsible for the tension.
A constant angle of 90 = constant tension.

Q: what is an appropriate tension for a 40 years old S8 film?

This guide has some helpful ideas on tension, dancer, PID that may be useful to a telecine transport builder,