Introduction: Accelerometer-logger With SD-card Memory

Picture of Accelerometer-logger With SD-card Memory

A logger unit for measure forces on a roller coaster and save them to a SD-card.
It's also possible to modify the software in the unit so it can measure other things if it can be connected to a i2c-bus.

Top Thrill Dragster

Step 1: Introduction

This is the fifth version of this unit (I know it's named v3, the fourth was called v3310 :) , it has also been a v0).
I have been working on this project in my spare time over two years now.

I use a LIS3LV02DQ from ST-Microelectronics to measure the forces, a uALFAT from to manage the SD-card, a HCMS2915 display and a PIC16F876 to control every thing. I also use a Canon NB-4L battery ( the same model i have in one of my cameras) to power the unit.

BOM (aka what you need) :

a PCB (a gcprevue-file to download)
a piece of Plexiglas 72.3x57x8 mm to the battery compartment
a piece of Plexiglas 3 mm thick to the front window
soldering iron with a small tip 0.4mm
some drills
some mill bits
thread cutter M3
Super glue and some regular glue stick
a X-Y table from Proxxon
a bench mounted drill
a battery (I use a CANON NB-4L).
a Microchip PICkit2 programmer.
a aluminum case 1455B802BK
a broken camera to get a battery connector from
and some surface and hole mounted components (BOM-file below).

Step 2: Prepare the PCB.

Picture of Prepare the PCB.

The PCB is in three parts and have to be separated before they can be populated with components.

(The new PCB layout has every thing cut out if it's ordered from a board-house and just needs to be broken apart.)

Step 3: Populate the PCB.

Picture of Populate the PCB.

Start to populate the three PCBs. The 2 32KHz crystals has to be super glued to the PCB.
The trickiest part to solder was the sensor, that have all the connections on the bottom of the component (I borrowed the x-ray @ work to check that the soldering was perfect :) ).

Step 4: Mill Out the Plexiglas.

Picture of Mill Out the Plexiglas.

Start to glue a printout of the mounting drawing on to the Plexiglas using a glue stick.
Then start to measure and mill out space for the components that are to high.
Drill and cut threads in the four holes for the PCB.
Mill out the space for the battery ( you can use almost any camera battery, but it should not be bigger than 45x38x6.5mm).
Then mill out space for the battery connector ( the connector comes from a broken CANON camera :) ).
And finally mill out a canal for the sensor cable so when you pull the cable the connector don't damage.
Create a front window of 3mm thick Plexiglas.

Step 5: Final Preparations

Picture of Final Preparations

Connect the battery and check that +3.3v_1, +3.3V_2 and +1.8V is correct.
Load the character map in to the external memory, then load a testing software to the uP to see if the unit is starting.
Prepare the sensor-board, it just need the sensor, some capacitors and a couple of wires.
Then connect the cable between the sensor and the main unit and check that the sensor is working.
The last thing to do is mount the uALFAT-chip and connect the SD-card.
Mount the PCB to Plexiglas with 4 screws.
Now you are ready to measure the forces in your car, bike and of course roller coasters. :)

Step 6: Conclusions

Picture of Conclusions

So far my brother and I have measured about 200 roller coasters in Europe, Asia and USA.

The unit is versatile platform to use for logging data that is collected from sensors that is connected though a i2c-bus, the only modification to do is in the software logging-loop.
The files the unit is creating is csv-format that is simple to create graph from in ex. Excel.

The next thing to do is make a new sensor that can measure more that +/-6g in the Z-axis.

ThemeParkReview forum post
Ride Guide


NathanGriffin (author)2014-08-24

Impressive DIY!

JHSTUBA (author)2013-03-18

Both files won't download, could you reupload them please?

JHSTUBA (author)2013-03-18

Do you happen to have a schematic for this? I'd like to try remaking it with mostly through hole components.

enSYS (author)2010-10-25

How did you get these graphs ? Matlab ?

lond (author)enSYS2010-10-25

From a custom made QT4-software.

enSYS (author)lond2010-10-26

How did you make that ? Coz I'm looking for an observational graph of the logged values that I've but don't have any platform for it.Can you help for the same ?

evanwehrer (author)2010-04-25

 I really like this project, but my skills arent good enough to make this. If I paid you, would you build one of these for me?

chew socks (author)2007-12-29

Which direction is the x, y, and z axis? PS: this is awesome

technodude92 (author)chew socks2007-12-29

X is Forward Backward(roller coaster car speeding up/slowing down) Y is Left Right(turning etc.) Z is up down(falling sensation) P.S. someone correct me if I'm wrong

Just to follow up with one accelerometer you will only get reliable displacement not rotation(turns), you can try with one but you will get better results with a gyro, compass(if you only care about rotation with respect to the ground it makes the math simpler) or second accelerometer.
To further explain imagine two paths of travel; one strait along the x axis with rotation about the Z axis the sensor will read this as altering acceleration and reverses along the X and Y axises(plot the sin and cos of x to visualize the sensor data), the second starts traveling along the X axis then begins to decelerate to a stop at the same time acceleration along the Y axis begines and reaches maximum just at the point travel on the X axis stop.

So given your accelerometer data did you just rotate a quarter step while riding the train or did you just get hit by it while traveling down the road?  I say this from the perspective of having played with an accelerometer and a holonomic drive system at the same time, both paths of travel are equally likely, sometimes you just need another sensor/data point.

lond (author)cokebottle tuque2009-10-22

The graph is from this ride:

// Marcus

cokebottle tuque (author)lond2009-10-22

The graph dose no seem to be showing up in Firefox v3.5

lond (author)cokebottle tuque2009-10-22

I'm using FF 3.5.3 and the graph and the YouTube-clip is working.

lond (author)technodude922007-12-29

You have completely right. To explain some more: Red: Linear G Chest to Back is denoted +gx Red: Linear G Back to Chest is denoted -gx Green: Lateral G Right to Left is denoted +gy Green: Lateral G Left to Right is denoted -gy Blue: Positive G Head to Foot is denoted +gz Blue: Negative G Foot to Head is denoted -gz Marcus

civiles97 (author)2009-08-19

Hi, I´m civil engineer from mexico city and I do structure health monitoring, could you tell me if this great idea can function in the things that I do, I have a lot of questions, I appreciate if you contact me to, thank you and well job

Alienum (author)2009-05-08

Hi lond, i make me best compliments. where you buy the special project pcb ?

GorillazMiko (author)2007-12-28

Wow... this thing looks really complicated... that 1st picture is awesome by the way. :-)

lond (author)GorillazMiko2007-12-28

It's not that complicated if you are not afraid of soldering smd components.

Alienum (author)lond2009-05-08

Hi Lond, i want make my compliments for the project. where i can buy a special Pcb project board?

ste5442 (author)2009-04-12

Nice project and a very good idea to make the large hole under the QFN package - I guess this is to make hand soldering of the centre pad possible :-)

PCB Police

lond (author)ste54422009-04-12

And when you solder a QFN you have to raise the QFN ~0,25mm from the pcb, I use a strip from a label that I have removed the glue from. Then use lots of flux to solder 2 sides of the QFN, inspect your soldering, if you are satisfied then remove the strip of label and solder the other 2 sides after that you can solder the center pad of the QFN. I usually inspect my soldering on my QFN in a x-ray machine at work to be on the safe side :)

SpaceShipOne (author)2009-02-26

Nice project. I've been wanting to make something like this, but it looks like a lot of work, especially since I don't have access to a milling machine. is ther a simpler way to make this that doesn't use an SD card?

lond (author)SpaceShipOne2009-03-01

My milling machine is just a drill press and a x-y-table. My first versions used 3 i2c memories, one for each axis. The drawback for using memories to store the measurements is that you can only store one session and then you have to transfer the data to a computer or some other memory backup. Because of this i use SD-memory card, i can have more than 65000 logs on the card. // lond

Lyleeby (author)2009-01-15

Is there any way that I can contact the originator of this (Accelerometer-logger with SD-card memory)project? Or can they contact me at

icrashrc (author)2008-11-04

What is the record time of this unit? Seconds, minutes, hours? Thanks!

lond (author)icrashrc2008-11-04

The viewer software have a max limit of $FFFF (65535) rows of data in the files. So i have set the unit to stop logging when reaching $FF00 (65280) rows. That means that i get ~10min of data, but i can put a delay between each measurement if i want to measure longer, 1sec between each measurement gives ~18h so i could log my suitcase from Sweden to Japan.

Toulouse (author)2008-11-03

If you have a Mac laptop, give this a try:

(I wrote it a while back, and am planning on making some major improvements)
(written in java)

lond (author)Toulouse2008-11-04

I have sent the link to my brother thats owns a mac and he tested the software. His iBook G4 didn't like it, the screen flickered and the software didn't fit the screen, What he could see was that the software uses the internal g-sensor of the laptop and the laptop don't like riding rollercoaster :(

Patrik (author)2007-12-28

Nice work! Very professional. Of course, now you need some decent signal processing on this data as well. One place to start may be to smooth all three signals to more clearly see the overall movement of the rollercoaster cart (in principle, you could even use these signal to reconstruct the entire layout of the rollercoaster by "dead reckoning"). Complementary to that, you can also quantify the amplitude of the high-frequency signal you've smoothed out, as a measure of how jittery/bumpy the ride is. The first depends on the layout of the rollercoaster, the second on the quality/smoothness of the track and the cart wheels.

lond (author)Patrik2007-12-28

I have a ECC-member that are helping me with this. He have created a viewer for the log files. The viewer can rotate the axis, create screen shoots and even subtitles for your POV-movie :) . Next step is to build in noise reduction etc.

bert612 (author)lond2008-02-20

A very interesting project! Can you make the viewer software available for download please?

ariesta (author)2008-01-06

may i have the schematics, please... : )

Geordiepom (author)2008-01-03

Hmmm... I could put this in the car and find out how the kids drive it when I'm not with them!

Kiteman (author)2008-01-02

This is a brilliant project!

You could set yourself up as a coaster-guru: Oh, this coaster is the best because the g-forces are such-and-such...

lond (author)Kiteman2008-01-03

Precisely :)

perkinsb1024 (author)2007-12-29

Very nice circuit board, where did you have this made, and how much did it cost?

lond (author)perkinsb10242007-12-30 10 boards for 124euro

hardburn (author)2007-12-29

Very cool. I've been thinking of doing something like this with the addition of a GPS chip ( sells a GPS chip the size of a postage stamp) and fit it inside an altoids tin. It'd be useful on go-karts, motorcycles, etc. to get telemetry data and generate a complete map of your course. The accelerometer should be accurate enough to get the 0-60 time and the GPS gives current speed with an acceptable error level.

mrbob1000 (author)2007-12-28

sweet... so you use this to log the intensity of roller coasters that you have been on?

lond (author)mrbob10002007-12-28

Yes. My brother and I have logged almost 200 roller coaster now. Next year it will be 80 more coaster logged :)

thearchitect (author)lond2007-12-28

Tough job, man. One roller coaster per year is enough for me. 200? Ridiculous!.. :-))) Great work! I admired your neat job, very well done!.. K.

About This Instructable




More by lond:How to convert a IKEA Jonisk lamp with LEDsAccelerometer-logger with SD-card memoryNokia 3310 acceleration logger
Add instructable to: