Projectile Sensor

[mribble]

A long time ago I build this projectile sensor to measure the speed of a bullet.  That version was a pain to use because you had to shoot the bullet directly between the two LEDs and IR detectors.  That said if you had the bullet lined up properly it worked quite well.  http://www.cameraaxe.com/wiki/index.php?title=Sensors#Projectile_Sensor

I'm thinking about improving the design.  Does anyone have ideas on good ways to detect projectiles?  Some bullets are lead and some are steel.  Bullets are generally moving at 500-3000 ft/sec you have a max of 3 microseconds to detect the bullet at the high end.

My idea moving forward unless I hear a better one is to see if I can get IR LEDs and diodes working as a reflective sensor.  I'd like to have a target region that is 1.5x1.5 inches.  To get that I've done the trig and think 3 emitters/detectors for the front and back would be best.  I was looking at IR emitters/detectors with an 110 degree field of view.  I'll use a differential op amp to increase the gain from the light diode and then use an op-amp comparator to a reference voltage set by the users.  Seeing the reference voltage would just be dialing the pot until it it is close to, but not quite triggering in the current environment.  The reason a comparator is needed is because the micro needs to be reading digital values since the ADC is too slow.

One concern with the current system is I don't know if enough light will reflect off a bullet 2 inches way to be detected even after my amplification of the signal.

Any thoughts on this or ideas on how to make it better?

[woodchips]

Shoot it through some aluminium foil, the impact will be detectable using sound, vibration, reflective light etc

[Vtile]

There were about 15 years ago in Finnish gun magazine a DIY chrono which did use two regular headphone speakers hooked as microphones. This solution needs a bullet with a comparatively high flight noise though, so anything transonic or sub-sonic will be hit or miss.

[mribble]

I have a sound solution already and it works ok, but it is not good at measuring the speed or knowing the position of the bullet.  This is supposed to be an improvement and that requires knowing both position and speed so you can place the bullet in a photo (the link I gave to my earlier solution knows the bullet position to within 1 mm).  I could see sound working for bullets faster than the speed of sound since then you could measure the bullet whizzing by, but the most common bullets for my project are pellet guns and those are slower than the speed of sound.  This would mean I'd need to filter out the initial sounds and that would be very difficult depending (ambient noise, gun noise, not having time to use an ADC).

Could you explain how you'd use reflective light on tinfoil?  I'm not seeing an optimal solution for that.  I could see maybe a pizo sensor attached to the tinfoil working (basically your sound solution).  That's an interesting idea.  I would be concerned about my pellet guns triggering it with air before the bullet, but perhaps an extra sheet of paper could prevent that.  My concern here is how fast is a pizo element at reacting?  I'd need something like 1 microsecond response, and I have a feeling pizos might not be that fast.

I did consider printing conductive ink to paper and shooting through that would be easy to detect, but I don't think people want to have to order consumables like that.

[Kleinstein]

Piezos can be very fast, however the sound inside a tin foil also has a finite speed and will thus add to errors. So it would kind of need two piezos on both sides.

There will be reflections from the projectile, and this can be rather tricky. Working in reflection effectively double the distance. Thus focusing and adjustment get more difficult. A retro-reflector helps a little - otherwise the angle of the mirror is important too.

One could modify the shown circuit by adding a second detector, thus making targeting less critical. One could likely still use a single IR LED. Even if the detectors a separated a little, the sensitive area can be overlapping and without a gap.

Even if using a comparator for triggering, one should consider also capturing the amplitude, as a partial hit could also cause a wrong timing.


In school we did an experiment with two thin strips of Al foil, that get separated and detected by interrupting a current flow. So it needs disposable materials, but that is cheap.

[richard.cs]

I have a 1950s book that talks about doing it with foil, in this case simply 2 closely spaced layers at each detection point and looking for the short circuit between layers. Mind you it also talks about counting a 100 kilocycle clock with neon-lamp based bistables

[mribble]

I've experimented with the tinfoil strips in the past (I got the idea from a wire setup they used in the 1890s to photograph horses running).  Setting that up between each shot is pretty annoying which is why I made that original sensor which go rid of the setup.

The length of tinfoil would be quite short and sound should be traveling through it much faster than air.  So I think that latency is fine (though I can't be sure without testing).  Even though setting this up would be a little better than the thin strips of tinfoil since you could just put up a big sheet, it still involves some setup and isn't ideal.

Two detectors would be a nice improvement, but it still means I'd need 3 PCBs.  I was looking forward to getting this down to a single PCB.  Your retroreflector idea is very interesting though.  I wonder if I could put the emitter and detector on the same PCB and just have a sticker stuck to a brace on the other side.  Have a single emitter and two sensors.  Then look for a drop in light values since the bullet would block the retro reflector.  However, if the bullet itself is too reflective this wouldn't work.

[rstofer]

In a perfect world, the electronics is on the bench.  The only thing downrange is the sensors and they work from light reflected from the bullet.  Commercial sensors work well with both lead and jacketed bullets.

I use this one:
https://www.cedhk.com/ced-m2-chronograph

You can buy the screen set separately:

http://www.brownells.com/shooting-accessories/range-gear/chronographs/infrared-screen-set-sku399100002-6889-17431.aspx

The above screen set includes an IR emitting surface above the sensor.

There is also a skyscreen kit (included with M2):

https://www.cedhk.com/ced-m2-millennium-chronograph-sky-screen-set

The sensor, by itself:

https://www.cedhk.com/ced-m2-replacement-sensor


As I am spending more time measuring rifle rounds, I use the MagnetoSpeed.  It has the advantage that I don't have to try to align the screens with the target (or whatever I am aiming at) during a cease-fire period.

http://www.magnetospeed.com/

[mribble]

The first one you linked seems to be using some kind of light sensor (though I suppose it could be using a reflective system like we discussed).  I looked at one kind of like that a long time ago and it wouldn't work for my use case because the photography is done in the dark and the one I looked at depended on ambient light.

I'm not sure how the second sensor you linked works.

[mribble]

I started thinking about alternate approaches and thought maybe a magnet plus a hall sensor would be an option.  However, it looks like hall sensors are generally pretty slow.  Does anyone know if there is a fast type of hall sensor?

[rstofer]

Almost all chronos have a problem with sunlight.  It can't be early in the morning, it can't be late in the afternoon and winter is iffy, high north or south latitude may always be out of the question - that kind of thing.  So, sky screens don't always work.

The IR screen is an IR light source over the sensor and replaces the sunlight skyscreens.  Sometimes it works better to not use the skyscreens.  I don't have any experience with the IR screens.

There is at least one example I have seen where the IR screens and sensors are placed in a wooden box to eliminate any outside interference.  There is every possibility that the camera could be inside the box and triggered by the sensors.

There might be some reason to mount the sensors at the top of the box and the IR sources as the bottom.  The interior would be painted flat black.

[rstofer]

I started thinking about alternate approaches and thought maybe a magnet plus a hall sensor would be an option.  However, it looks like hall sensors are generally pretty slow.  Does anyone know if there is a fast type of hall sensor?

Can you rig something up using the MagnetoSpeed ideas?  The bullet has to be about 1/4" over the sensors.  This is easy with a rifle since we can just use spacers to mount the head.  I'm not sure how to do it for a handgun but there ought to be a way.  It's probably easier for a revolver.  Still, some kind of rigid mount could be fabricated.

The sensors on the M2 are 24" apart so a 3000 fps bullet takes about 1/1500 of a second or 667 microseconds to trigger both screens.  So, if the sensing occurs in zero time, the camera has to respond in 333 microseconds if the shutter is being used.  Pretty fast!

There's another sensor approach here:

https://petapixel.com/2016/11/10/teaching-art-capturing-bullets-flight-classroom/

[mribble]

Microcontrollers are really fast.  If they are using 24" apart they are either needing really good accuracy or wanted to use a very power efficient micro.  My previous sensor had the gates 2 inches apart.

The link to petapixel is basically the kind of sensor I linked in the original post.

[Beamin]

What if you measure the sound of the shot vs hitting the target? On second thought this would be kind of hard to fully automate. Unless it just told you the time which you could figure out with a known distance. How hard would that be to make anyways? Could it be done all analogue or would you need a raspberry pi and software? 

[mribble]

I have a mic system which doesn't detect hitting the object, but detects firing the gun.  This is more work to setup because you need to do some trial shots to get the bullet placed where you want.  However, the real reason this isn't ideal is because the velocity of the bullet changes from one shot to another.  This results in me only being able to place the bullet to about 15mm in the photo, and as I mentioned before the optical trigger that measured the bullet speed per shot could place the bullet within about 1 mm.

Now if I could detect sound of the bullet hitting the object that would improve the accuracy over the 15mm I had before, but there are things that make this hard like the bullet is sometimes faster than sound and other times slower than sound.  Also sometimes you want to take a picture right before the bullet hits the object because that is cool looking too.

[tautech]

I was always under the impression that ordinary photocells were used, are they not ?

The simplest shooting chronograph is the good ol' Chrony for ~$110 new.
http://www.shootingchrony.com/introduction.htm

Mention them in a shooting forum and most think they're a POS however my experience with them has been good.
We set up two units one day, one behind the other and results agreed within a few fps.

On overcast days with full cloud cover the light diffuser screens can be dispensed with but for blue sky days they must be used. For indoor use they also offer a mains powered light fixture:
http://www.shootingchrony.com/products_ACCESSORIES.htm

The Chrony's are the most affordable and I hankered for one for a while, had the $ to get one when an appreciative shooter on my range gave me his. 
Few bits missing....one segment of the wire uprights....some #8 gauge wire was the fix and it was minus the diffusors but some plastic road marker offered a good white background for the photocells to spot the bullet passes. 
All works a treat and gets plenty of use.

[mribble]

I'm not sure what a "photo cell" is, but if you my an photo diode with some kind amplification then that is how the one I looked at worked.  However, as I mentioned before that won't work for me since I need to do the photography in pure darkness.  It was mentioned that you can have an LED canopy, but then it's essentially what I made and linked to in the original post.  The device I made had a BOM of about $15, and I'm sure I could cut the cost by using a less expensive photo diode and an op amp.

Thanks everyone for the ideas.  I have ordered a few parts where I'm going to try and do some testing with the reflective properties of some bullets and of retroreflectors.  Those were my favorite ideas mentioned here so far.

[StillTrying]

"My concern here is how fast is a pizo element at reacting?  I'd need something like 1 microsecond response, and I have a feeling pizos might not be that fast."

I've measured the response time of cheap metal disk piezos by tapping them in the center with a small metal object so I could electrically detect the metal to metal contact, and the time the output rises to a detectable 0.2V, the time between contact and 0.2V output varied a bit between 2us and 5us.

About the only way to get 1 or 2us timing on a bullets position is to have it break a thin strand of wire or tinfoil. Then you're just left with the problem that a photo flash tube discharge start varies between about 5us and 20us after it's triggered depending on it's voltage, temperature and a few others.

In the dark you can use a visible red laser as long as the dot is not shining on anything in the pic, some of the photo triggers can turn the laser off right after the trigger so that a red dot doesn't appear on the photo of the smashed to bits object.

[Brumby]

I was thinking something along the lines of your original idea - but increasing the sense window size.

Something like this, perhaps.


(Projectile is travelling normal to the page)

So long as the angle of the laser beam is shallow enough to ensure the projectile breaks it somewhere within the detection aperture, the actual angle isn't too critical at all.  It just has to traverse the mirrors and hit the detector when no projectile is in the way.

[mribble]

StillTrying, thank you very much for that data on the pizo elements.  I never had numbers for response times.  The system you described sounds good to me.

I have oversimplified the sensor response time when I say 1 us.  With a 3000 fps bullet it will travel .036 inches in 1 us.  I figure that is a good target for a sensor duration (and many photo diodes are much faster), but it really could be a little more time than that if the delay is consistent.  The real limit on sensor response is making sure the projectile doesn't make it past the sensor before the sensor detects the projectile.  The smallest projectiles I use are probably 0.1 inches.

As for flashes a standard xenon flash tube with typical electronics in flashes today takes about 20-25 us like you said.  But even worse is the minimum duration of those flashes is also 20-30 us so they result in blurry pictures.  The solution is to use air-gap or specially designed LED flashes.  The air gap flashes I designed have about a 3 us lag and 0.5 us duration (I suspect there is lots of room for improvement here by changing flash tube geometry, but I have moved on to LED flashes).  The specialized LED flashes I've used are about 2 stops dimmer, but have <1us lag and a minimum 0.5 us duration.

Yup, I have a visible laser trigger that turns off the beam during the photo.  THe issue there is most of the time you are trigger flashes and not cameras so it doesn't help, but sometimes it's a nice feature.  One issue with these systems is alignment.  I came up with a really nice trick here.  If you put a diffuser (basically some thin white plastic) that's a few inches big in front of your laser detector it makes alignment much easier.  It sound like you have a lot of experience with these kinds of systems.  If you want to work with me or just review design docs for my open source trigger (Camera Axe) feel free to contact me.

Brumby, that kind of system is interesting for the reasons that are obvious with your great picture.  The issue that has always scared me from trying it is alignment would be a pain.  There are some really nice alignment and inexpensive alignment rigs being used on cheap laser cutters these days.  Maybe I should look into this someday.  The alternative are bars of emitters and sensors which are easier to align, but probably more expensive.

[Melt-O-Tronic]

Radar is the latest thing in personal ballistic chronography.

http://mylabradar.com/

Other than that, nobody seems to have come up with much improvement over the tried-and-true photocell method in the last 40+ years.  Magnetospeed is invasive and alters the bullet's trajectory, so I don't consider it a viable option unless velocity is the *only* parameter of interest.

[mribble]

Thanks for the link since I had no idea anyone was doing this with radar, but I suspect radar won't work for my use case.  First thing that kills it is cost (I want to be under $20).  Then from what I know radar has a lot of signal processing which probably means I'd miss the photo.

[Melt-O-Tronic]

Ahhh, so photography is the goal . . . in that case a Magnetospeed-type device should work OK.  But still, I'd recommend either making your own photocell sensor or picking one up as a replacement part from any of the numerous chronograph manufacturers.  That may be over $20, but shouldn't be by much.

CED replacement sensor - $37
PACT replacement sensor - $15

You'll still need a sky screen, but that's simple.  My choice of sensors would be Oehler Research because I already have a 35P and I suspect that the signal processing is dead simple.  Unfortunately, Oehler's web store is down right now, but you can email them at sales@oehler-research.com

This sounds like a really fun project to tinker with.  Lots of data collection, measurements, etc.    If you're near Central Texas, you're welcome to do your tests on my private shooting range.  It's nice to not have anyone else to negotiate cease fires with and no inane rules.   

[mribble]

Ah, if only I lived in Texas, but I live in the Northeast so that won't work.

I have a working photo trigger system, and will be doing some experiments based on all the great ideas here to see if I can improve it.  Bullet photography is always done in the dark because you use the flash to freeze the photo so the sensor needs to provide it's on IR LED lighting if using photodiodes.

[Raj]

I was thinking something along the lines of your original idea - but increasing the sense window size.

Something like this, perhaps.


(Projectile is travelling normal to the page)

So long as the angle of the laser beam is shallow enough to ensure the projectile breaks it somewhere within the detection aperture, the actual angle isn't too critical at all.  It just has to traverse the mirrors and hit the detector when no projectile is in the way.

what a coincidence.I am working on same stuff using 50mW laser and making it reflect multiple times off a mirror making a laser 'wall' that is sure to get blocked by the bullet.I'll be using atmega8,and i'll have 2 laser walls, and ill use the formula, speed=frequency of cpu x distance between 'walls' /cycles the clock of the cpu counted

i'll be using aditional analogue circuitry that'll trip the interrupts of the atmega8

[Raj]

I started thinking about alternate approaches and thought maybe a magnet plus a hall sensor would be an option.  However, it looks like hall sensors are generally pretty slow.  Does anyone know if there is a fast type of hall sensor?

hail sensor or any other sensor is ok, as long as you have seperate ones as 'start measuting time' and 'end measuring time'.
The deal is,both need to be independent of the cpu and need to be EQUALLY SLOW.