DesignSpark Electrical Logolinkedin
Menu Search
Ask a Question

6 Apr 2020, 6:43

DIY Photon Propulsion – Innovation meets DesignSpark’s maker sense

A typical spacecraft achieve propulsion by reaction engines. They expel gas at their back end. The spacecraft gains the same momentum which the expelled gas mass has gained. This “jet propulsion” is used by most airplanes. NASA has started to experimentally use photons instead of gas. But they have also reversed the effect and used a large mirror foil as a solar sail. A traditional water wheel starts turning because the momentum of the water mass hitting the paddles is transferred into a turning moment of the wheel.

The same principle is used for a solar sail. Light has what is called a “radiation pressure”: If it hits a mass, the pressure exerts an accelerating force to the mass. Would it be possible to use this last principle for a DIY light wheel?

A physicist would start to calculate, but the engineer says “let’s try!”.  So I was thrilled about the idea to use the newest ultra-high-power LEDs to make a “light wheel” turn. Summing up the cumulative costs, I thought it would be better not to waste my money but to learn my lesson from the physicists and do some calculations first.

The Plan

So what was the plan? I would take four 1CoB LED arrays with nearly 17,000 lm each. These cumulative 68,000 Lumen should shine onto close positioned mirror-paddles mounted to a vertical axis. The axis has needles on both ends, which are pivoted in bearing jewels to reduce the friction to a minimum. The overall weight of the wheel with the mirror paddles is critical. So I would try to reduce as much mass as possible. The second force which needs to be drastically reduced is the air drag. Therefore I would place the whole setup into a vacuum container and evacuate as much air as possible. This will make cooling nearly impossible, so I would only use a 1-second light pulse. Hopefully, this tremendous pulse of light pressure would result in a high enough reaction force which would turn the paddles around the axis.

Please do not mistake Crookes' radiometer (also called "light mill") for what I planned to build.

Although people often think that this device would work the way my apparatus is planned to work, it does not. The wheel of a Crookes radiometer turns with the reflecting surface toward the light source because the force turning it is a thermal effect. The black surface of the paddle does absorb more heat than the reflecting side. Therefore it is much warmer than the reflecting side. The vacuum needs to be about 5 Pa so that there are still air particles that hit the surfaces. When they hit a hot surface, they will leave the surface with a higher momentum compared to when hitting a cold surface. Therefore the black (hot) surface receives a higher reaction momentum than the reflecting (cold) surface. This can be proved by eliminating the remaining air of the glass container. Without air particles hitting the paddles' surface, the wheel is no longer turning. 

But why should light hitting a mirror should exert any force on it? Because light can be thought of as a constant flow of photons. Although a photon does not have a mass, it does have momentum. The momentum is transferred to the mirror twice when reflected: The first time when it hits the surface (momentum transfer by a collision) and the second time when it is catapulted away from the surface (repulsion principle).

The Theory

Now it gets a little fancy. We need to calculate the force exerted by the photons on the mirrors by the radiation pressure, which is accelerating its mass:

How many photons will hit the surface when I fire the LEDs for a second? Let’s assume we have a total light energy of

Etot = 300 Ws

(we use nearly 600 W electric power for a second but not all light will hit the mirrors, and the LEDs do have an efficiency below 100%). The average solar power of sunlight shining vertically on the floor is about 1100 W/m². We will light an area of

A = 0,04 m * 0,04 m = 0,0016 m²,

so we have nearly 200 times more Power per area as sunlight. The light intensity of the sunlight can be up to 150000 lm/m². We provide about

68000 lm / 0,0016m² = 42.5mega lux. (I think I will need my sunglasses ;-)

Assuming a mean wavelength of 663 nm (just to make calculation easy) each photon has (according to Einstein) the energy

Ephot = h*c/l = 6.63 * 10-34 * 3.00 * 108 / (663 * 10-9 ) J = 3 *  10-19 J.


Etot/Ephot = 300 / (3*10-19) = 1021

gives us an estimation of the number of photons being reflected during a 1-second light pulse.

The momentum of a photon is

p = h/l

During reflection, it transfers

2 * h/l

6.63*10-34 / (663 * 10-9)  Ns = 2*10-27 Ns

to the mirror. As 1021 photons are transferring their momentum to the wheel, we get a total momentum of

2 * 10-6 Ns.

This is all we have to accelerate the mass of the wheel from zero to a resulting speed. The calculation is

(end velocity minus start velocity) = momentum

used for acceleration. So

p = m*v – 0 (for no motion at the start)

The resulting velocity is

v = p/m (assuming the direction of force is the direction of the velocity, which only at the beginning is the case in our setup)

If we want a velocity of 1 mm/s the mass must be 2 g. In reality, we must take the friction force into account. Using the velocity, we gained in a second we calculate the accelerating force to be

F = v*t * m = 1*10-3 m/s2 * 2*10-3 Kg = 2 * 10-6 N

Any friction force should be far less than this small force. That sounds difficult but also feasible.


So let’s get practical. I’m using a wooden plate resting on four stands. The plate is covered by a sealing mat. An acrylic cylinder works as a vacuum container. I mounted an outlet fitting with a valve and a  vacuum meter on top of it. As I needed to get the electricity inside the container, I also mounted eight through-hole banana jacks on the plate and sealed them as tight as possible. This type was a little more expensive, but it has no holes going through from top to bottom. So getting it airtight was not too difficult.

The paddles are cut of 0.3 mm mirror foil. The axis is a 2 x 2 x 60 mm 3D printed part with needles pressed on both ends. When I put the assembled light wheel on a scale, it showed 2.2 g. So my light wheel is indeed a lite wheel and I'm just by 10% over my assumptions in the calculation.

The needles rest in tiny 1,5 mm bearing jewels (you get them from watchmaker shops) which I have glued on 3D printed parts that hold them in place. The bottom bearing holder is glued to the base plate and the top bearing is held in place by a bent 2mm brass strand which is placed in a dead-end hole of the base.

Please note that we only get the full acceleration force at the starting conditions when the beams are not reflected onto the backside of the neighbouring mirror, and the beam hits the mirror at a 90° angle. So when trying to get a continuous motion, we would need to pulse the light for a short moment when the mirrors are parallel to the LEDs (continuous light would overheat the CoBs). We would also need to focus the LED beams to a very small emission angle. But tests did show that the distance from LED to mirror was close enough to get most of the emitted light onto the mirror.

Luckily having good connections to a lab, I could borrow an extremely powerful vacuum pump (rotary vane from Agilent) allowing to get down to pressures of 10-3 mbar (hPa) which is close to HV (high vacuum). Greasing the seal, I was very curious how high my vacuum would be. Just in case I would get down to a high vacuum I thought it would be wise to put some packs of Silicagel into the container. I did not want to get fogging of the mirrors (remember school physics when they taught you the relation between boiling point and pressure?).

Okay, enough words. Let’s watch the video of this astonishing experiment:

Before you will praise me for saving the world by inventing a CO2 free urban propulsion… The sum of momentum changes in a closed system always adds up to zero. So mounting the light source on a vehicle and trying to get it moving by letting the light shine on a drive wheel will not work. And positioning massive light towers along the roads may work but seems pretty impractical. But if we would forget about the mirror stuff and go for the NASA photon propulsion by emitting a million lumen of light from the back of your car, well… If you manage to reduce the car’s and your weight to some grams and build a vacuum tube around the road… could work!

And for all you who think they could build a Perpetual motion machine using this idea and sending the reflected light back again: You cannot create motion energy from nowhere. So this is NOT a solution for the world's energy hunger. The energy of the turning wheel comes from the light. It does not lose its intensity, but with each reflection, the wavelength is reduced. This was first observed by Compton in 1922, who could prove that photons can be treated as particles having momentum.

Nevertheless, although this project will not save the world, it was a great experience and a cool setup, so I needed to share it with the DesignSpark community ♥.

Volker de Haas started electronics and computing with a KIM1 and machine language in the 70s. Then FORTRAN, PASCAL, BASIC, C, MUMPS. Developed complex digital circuits and precise analogue electronics for neuroscience labs (and his MD grade). Later database engineering, C++, C#, hard- and software developer for industry (transport, automotive, automation). Ended up designing and constructing the open source PLC / IPC "Revolution Pi". Today engaged in advanced development as a service.

6 Apr 2020, 6:43


April 7, 2020 08:38

  • Moderated

hi Vdh your experiment is really good. Looking fwd to joining your team... is that possible? my email is please inbox me if i can be of any help.

0 Votes

April 8, 2020 07:49

@ManthanManojDixit Sorry if we fooled you... But when Andrew and I will start the 2021 April's project (ion thruster) I'll get back to your offer :-)

April 6, 2020 06:39

FANTASTIC!! I absolutely loved your genuinely surprised reaction :•D

0 Votes

April 8, 2020 07:49

@RacingtoMars Thnx! Took a week of drama school to get it genuinely ;-)

April 2, 2020 15:12

Very cool indeed! Could you please demonstrate an ion thruster next :D


April 2, 2020 15:13

@Andrew Back Tha's scheduled for next year. Would you like to join the team?

April 6, 2020 06:38

@VdH why not! We have two rotary vane pumps: one is used as the roughing pump for our SEM's turbomolecular pump, while the other is much larger and was used for pumping down industrial refrigeration systems before filling with gas. Been looking for a nice chamber for various high vacuum experiments... :o) Also some interesting open source work being done here:

April 2, 2020 15:14

Hi VdH - I am also working on a new system, it's for teleportation. I just need to wait for Starfleet Command to be established so I can sell it to them, bests regards, Montgomery Scott (just remember you can't mix matter and anti-matter cold!)

0 Votes

April 2, 2020 15:14

@HarryO1 good to know! Please publish your ideas at DS. It could raise the interest and therefore the price you can achieve.

April 2, 2020 15:14

Your calculations appear to be off by a factor of 10,000.

0 Votes

April 2, 2020 15:14

@SheldonCooper Where exactly can you see this mistake? Estimations or calculations? Which equation?