How important is tension control?

I have given this site a quick perusal, and haven’t seen much about tension control. If you use constant tension control, the centre of the roll will be wound tighter than the outside due to film slippage, trapped air, etc. That would be alright if you were to unwind/rewind again soon. If your intention is to archive the film, I believe it could lead to problems. Opinions?

I don’t have first-hand experience, so my first question is - is there some sort of “standard” or value that has been scientifically researched and determined best? Is it the same for all films, or does it vary by gauge, type of stock, reel size, etc?

If there’s some published data, then we could try to build that into the design.

@Maker @digitap all good questions. I’m unaware of any standards. I would assume that the user will want to rewind right away and probably doesn’t have a dedicated rewind station. In that case, the Kinograph would have to rewind and the tension setup can be built specifically built to handle that tension problem separately from the capture film path. A great point that will need addressing at some point down the road.

Let me know if you find out anything or have concrete examples we could try to learn from.

Man, I miss @Peter


I have always assumed that constant tension is a desirable thing to have, but what value I would personally be happy with is something I have not determined yet, until I complete the experimental rig that I am currently building to test the motion control system that I have outlined elsewhere. My gut feeling is that it should be somewhere between these two absolute limits -

  1. Not so loose that the film moves easily on the spool when shaken.
  2. Not too tight that it periodically slips and screeches when being

I think it should be closer to 1. than 2. I am anticipating somewhere between 20grm and 60grm, but I could be a way out - time will tell.

If tension is not controlled, what other parametric option do we have? I can’t think of one offhand.

BTW why should constant tension wind tighter near the hub than the outside? That is certainly true with constant torque, which is not a good idea in my opinion.


I am anticipating that it will need to travel at a constant speed to match the shutter timing. That leaves torque control.
When you first start a reel, it is very heavy and each RPM feeds a lot of film. The unwind motor will be working very hard to keep speed.
While the rewind roller is very light and each RPM winds very little film.
Therefore, I understand that each motor needs to be controlled separately, in either open loop or closed loop configuration. That is not even considering the centre drive motor.
If you just set the rewind at a certain torque, and consider the holdback, it would be like pulling on the end of a film while holding the reel. It keeps slipping and getting tighter, which is multiplied as the role gets heavier and bigger, and at any great speed, would be trapping air.
What ends up happening,in my experience, is that the centre of the roll is way too tight, potentially leading to emulation problems and the outer half is too loose causing a telescopic effect.
The only solution I can think of is to set up a tapered torque. Which needs to keep the circumference of the rewinding reel in the equation and adding dancer rollers.
It seems very clear in my head, but hard to explain.
I can come up with the formulas, but I’m not much of a coder.
I have pages and pages of info. It is a very hard process to get correct. If you want, I can put up some links to the information.
BUT…maybe someone else is more knowledgeable about this technology or knows how to find some open source code?

What would you say, is the difference between tension and torque?

Quite so, and that can only be achieved by constant tension, and that means the take-up torque needs to be varied as the diameter of the film on the reel changes. This can easily be achieved without resorting to any complex coding, or even any coding at all, as I have explained in several posts on this forum recently.

This would only be of relevance if intermittent motion was used, when the inertia of the reel would put a large strain on the sprocket holes, and the film would tend to run-on after the claw had completed the pull-down, possibly causing a ‘snatch’ on the next frame which could rip a sprocket hole to shreds.

In projectors, this is overcome by constantly feeding and unloading the gate via sprocket-driven rollers into holding-loops. When constant motion is used, it is only necessary to gently accelerate the film to its operating speed to limit the strain on the film. Even here, if large heavy reels are used, some form of compliant mechanism may be needed in the film path to ease any transient stress during acceleration. These aspects are a good example of, and well explained, by Newton’s laws of motion.

Which motor is that?

This is what constant torque will do. It should not happen with constant tension, because the torque will be low at the reel hub, and gradually increase as the film loads on the reel. Think of this like raising a bucket of water up a well. If the winding drum was small diameter, then the torque required to lift the bucket would be low, but if the drum was large then much more torque would be required to lift the bucket at the same speed. Of course, the winding speed will be different between these two, but the power needed will still be the same as Power = Torque x Angular-speed. This makes sense, because why would the power required to lift the same bucket of water over the same distance in the same amount of time change because the winding drum diameter had changed?

They are related, but not the same thing: Torque = Tension x Radius.



Great explanations. Thank you. I missed some of the info because I was always checking latest which was set to 2 days.
I was planning to build using only PTR’s. Now I understand that you need at least one or more sprocket roller.

Great thread, you two! Thanks for the info. @VitalSparks I know you cautioned against using a stepper motor due to the constant torque mentioned here, but that’s what I have working right now. I’m post a video within the hour of the current setup. Would love to get your feedback as it relates to this conversation regarding tension/torque - and anything else!


I cautioned against stepper motors because they are not designed to replace conventional analogue motors (as many believe), but rather as a means of providing accurate and repeatable angular position control which used to be the domain of the synchro motor.

Stepper motors rotate in discrete angular steps unlike dc motors which make them unattractive for driving things like film which should really be transported smoothly and without vibration. Also the torque is very high, and constant over a wide range of speeds which is not what is wanted. A third disadvantage is that they are power-hungry, even when stationary. This kind of motor is brilliant at doing what it was designed to do, and makes possible a lot of things at very low cost (flat-bed scanners, page printers, 3D printers, etc.) but I don’t see something like the Kinograph being one of them.


Hi, I am Andreas, my first post here.
What about frequency converter (with internal PLC) driven servo motors and a dancer position control?
They are often used in industrial winding systems.
I have programmed some of them a few years ago.
By the way, the discrete angular steps of stepper motors are not that problem at all.
Due to the inertia of the rotor and the mass of the load, they have no effect, when the motor is controlled proper. Actuelly stepper motors are on the rise in industry applications.

Hi Andreas,

I join everyone here in welcoming you to the forum. I am sure your participation will be much appreciated.

Too complex for a small power machine like a Kinograph, and difficult to match speed with tension.

Inertia can be both friend and enemy to a system driven by impulses. whilst it can ‘slug’ the transient effect of the steps, it can also produce overshoot which can lead to ‘ringing’. This is very evident in 3Dprinters and CNC machines where imperfections in surface quality can be the result unless run at reasonably low speeds and acceleration control is employed and optimised between steps. I have specific practical experience of this in my own 3D machine designs that I have worked on.

Now to consider what a stepped drive means in terms of the peripheral movement of the film on the take-up reel. Almost universally stepper motors have 200 steps/rev. At the periphery of a 250mm reel of film, one step equates to about 4mm of film. Microstepping the motor can reduce this somewhat, but at a considerable loss of torque which may mean a much larger motor needs to be used. If a motor is microstepped by factor 32, 90% of the motor’s torque capability will be lost, yet the resolution in peripheral movement per step is still 0.125 mm (125 microns) which could relate to about 16 pixels at the image sensor for narrow gauge film. Yes, a gearbox could be used together with microstepping to virtually eliminate this problem, but now another problem emerges - backlash, which does not bode well with stepped drives because overrun caused by inertia is no longer checked.

Another problem with steppers is that their torque cannot easily be controlled. Even if it was, there a limits to how low it can be set before the motor starts to miss steps. If a gearbox (or belt-drive reducer) is used, then considerable high torques can be produced which could stretch or snap a delicate film if its free movement was impeded in any way.

If anyone considers using a stepper motor to drive something like a Kinograph on the basis that ‘this is the modern way to do things’ could be making a big mistake unless they are prepared to put some effort into understanding the principles involved, doing the calculations, weighing up the pros and cons, and backing this up with a lot of experimentation. If not, there could be a long road to trying to figure out why their machine is not the silky-smooth thing they expected it to be. They should also ask the question if a stepper would be be ideal to drive a record deck or a cassette/tape machine? (yes, there have been direct-drive turntables, but these were sinusoidal synchronous drives, not digital stepper motors).

In my view the Kinograph is best served by a transport drive which can be categorised as a ‘steady state’ system, where smooth speed and tension control are the main requirements over a relatively large range of reel speeds and varying inertia. My opinion is that DC motors (brushed or brushless) have so much going for them in terms of their inherent characteristics that for me the choice is simple.

And quite rightly so. They bring to the table a level of positional accuracy that a few years ago could only be achieved with extremely expensive servo control systems whose reliability was always questionable. Also, their digital nature aligns well with micro-controllers and their attendant software algorithms. The meteoric rise in production assembly robots is a testament to the use of stepper drives.

Last year I was involved in the design of a sun-tracker for solar panels, and a garden-watering robot that learned exactly what to water, for how long, and which varied its jet pattern and aimed it over a full 360 degrees in azimuth and 90 degrees in elevation according to what the user had demonstrated previously. None of this would have been feasible at reasonable cost without stepper motors (and an inexpensive Arduino to control them).

Yes, I am a big fan of steppers, but alas not for my own version of Kinograph I’m afraid.



Thanks for the generous reply. I’m still having trouble finding the right kind of DC motor set-up. I’d love to test it and am wiling to buy parts to make that happen and compare the results to the stepper motor setup I already have working.

Can you help me choose a motor and the necessary components I need to get it working? We’re working with an 8mm shaft coming through the backside of the Kinograph right now. Let me know if you need any more info.

Thanks! Your advice is great - now I’m hoping to get to brass tacks and build something with specific parts. But I need some help!


Hi Matthew,

Sorry I missed your email about this. I was on an extended Easter break with the family, and when I returned my inbox was overflowing and I only discovered your mail today. I was going to reply, but then saw your post here so this is where I will respond.

I see from the photos you sent that the motor does indeed seem inadequate, and deviates somewhat from the one I suggested in the ‘take-up motor’ thread earlier this year. You really need one with a gearbox, especially if you are running 35mm film. I have looked on US Ebay site and found this one which looks to be well worth experimenting with. This seems to be a common, if not standard, design so continuity of supply should not be a problem. I understand these are manufactured in a range of gear ratios - this one on Ebay looks as though it is 75:1 and should suit 35mm film up to about 8fps.

If using a 10" reel, then the take-up will rotate between about 12(full) and 78(empty) rpm at 8fps. Assuming a tension around 50 grm, the maximum torque required would be less than 0.6KgCm which is well within the capability of the motor. The power required should be less than 1W.

The output shaft is 8mm dia. which seems compatible with your needs, and may be long enough to couple straight to your reel without the need for any intermediate belts or pulleys.

I hope this gives you something useful to play with…

I am currently calculating all the speeds, torques, tensions, frame-rates etc. to cover the full range of gauges to see if a single motor model like this could cope without changing ratios. My gut feeling at the moment is that whilst 8-16mm may be OK, 35mm may require a different ratio unless lower frame-rates can be tolerated.

Anyway, I will be presenting the results of these calculations in a table sometime soon, when I am happy with them.


That looks very similar to the motor I used on the original Kinograph. A geared DC motor. DC Gearmotors | DC Gearmotors | Electric Motors |

This worked well but if I used PWM to control speed the torque was not enough to pull the film after a while. This could probably be fixed by finding a motor with a different gear ratio.

I’ll wait for your recommendations before buying another. In the meantime I’ll post specs about this and another motor I bought around the same time when I get home.

Here is something important that I must comment on relating to the selection of suitable take-up motors. I cannot put any real numbers in this discussion at the moment (but will be doing shortly as I mentioned above), but rather I will try to add some sense to the complications and confusions that surround this difficult subject (include me in this as well, as it has taken me some time to get my head around the relationship between all the variables that have to be considered in making a suitable choice). As with many things, it is not just sufficient to know that something works, but so much more reassuring to know why.

When I first suggested using a DC motor for take-up, it was based on my experience of having used an anonymous 12v motor (that I had lying around in my workshop) to replace the spring-belt drive for take-up on a projector I was converting. I did this because originally I considered the film was too tight at the hub, and too slack when the reel was full. I adjusted the voltage until the tension seemed about right, and found that it was fairly constant as the diameter of the film on the reel changed. I knew that this was because of the inverse-linear relationship between torque and speed of a PM DC motor, but I hadn’t realised how lucky I was to have used a motor that was such a good match for what I was doing. When I began looking at using the same technique on my version of the Kinograph, I was dismayed to find that, when I did the sums taking into account all the variables (speed, tension, gauge, etc.) that choosing a suitable motor for everything was going to be nearly impossible. I had many sleepless nights over this, but was convinced there must be a simple solution and, voila!, after much head-scratching and deliberation here it is -

Note - this is for the ‘open-loop’ method that Kinograph currently uses

  • Select a motor/gearbox combo that exceeds the speed/torque range that
    is needed to cover all eventualities.
  • Add a suitable value resistor in series with the motor so that the slope of its torque/speed
    characteristic is very close to that required of the take-up reel.
  • Adjust the voltage of the motor until the two torque/speed characteristics coincide.
  • That’s it!.

If a closed-loop control is used using a PLL, as I described elsewhere, then things change slightly, because the value of the resistor is now chosen to provide a limiting torque if the reel is stalled (film-safety feature).

If I get time today, I will draw some diagrams to make this easier to visualise.



This is a symptom of not having matching torque/speed characteristics of film/reel, and is exacerbated by operating open-loop. A PLL control system would overcome this because tension is set elsewhere, and all the take-up motor is doing is supplying enough torque to feed the film at the chosen speed.

This is the motor I was using in Kinograph 1.0:
RPM 30
Voltage 24 DC
Amps 0.68 no load
Torque 47 in. lbs.
Rotation Reversible

Duty continuous
Shaft 11 mm diam. x 12 mm w/dual flats
Mount 4 bolt 2-1/2" x 2" pattern
w/ 10-24 threaded mounting holes
Size 4-1/2" x 3" x 4-3/4"
Shpg. 5 lbs.

This is the other I bought but never used:

when i built my scanner, i got motors from here, good sdk and specs

Here are the illustrations I promised regarding tuning the take-up motor for correct Torque/Speed characteristics when applied to an open-loop drive for the Kinograph.

Take a moment to absorb the principles shown in A and B, then the following 4 illustrations should make sense.

I get the impression that once this tuning is carried out, there will be no need to adjust the resistance again as requirements of speed and the tension employed are changed. It will just be a matter of readjusting the motor voltage. I make this assumption because the relationships between all the variables (speed, tension, reel diameters, etc.) all appear to be linear, there being no obvious second-order effects. The only thing I do not know is whether in the current Kinograph design the tension changes with speed. In the version I am building, the tension is controlled and held constant electronically via the supply reel motor, so the problem does not exist.

In a short while I will post the table of values that I mentioned earlier which will enable the true significance of that red line in the illustration with respect to the motor characteristics. BTW the motor characteristics are usually stated by the manufacturers as two values - no-load speed and stalled torque.

Have fun

This makes a lot of sense. Thanks for the diagrams! I can see how we would dial in a beefy motor for our needs.

There is also tension in the system as a result of rollers and the gate. We will have to factor that in, too - and may vary depending on the design.