DIY Head Tracker For A Tenner

I have a 9150 in one. The DMP does *not* process the magnetometer readings and fuse them, or at least the firmware available doesn't and it's not documented how to get it to, so you have to manually read them and fuse it yourself. At least this was the result of my initial investigations - Pocketmoon can correct me if I'm wrong, the nav6 libs might be better with the newer MotionApps version....?
 
I have a 9150 in one. The DMP does *not* process the magnetometer readings and fuse them, or at least the firmware available doesn't and it's not documented how to get it to, so you have to manually read them and fuse it yourself. At least this was the result of my initial investigations - Pocketmoon can correct me if I'm wrong, the nav6 libs might be better with the newer MotionApps version....?

Invenses have backed away from offering full tripple sensor fusion either in the DMP (magical secret on-chip processing) or in any example code.

Magnetometer calibration is a whole world of interesting. Much trickier than acceleration and gyro. There is example arduino based code out there on github etc that does full sensor fusion but my worry is that it would introduce a lot more latency.

I do have a concept to try (when mine arrives from HK) that doesn't require mag calibration or additional fusion.
 
Invenses have backed away from offering full tripple sensor fusion either in the DMP (magical secret on-chip processing) or in any example code.

Magnetometer calibration is a whole world of interesting. Much trickier than acceleration and gyro. There is example arduino based code out there on github etc that does full sensor fusion but my worry is that it would introduce a lot more latency.

I do have a concept to try (when mine arrives from HK) that doesn't require mag calibration or additional fusion.

I've got some code I've been working on based on the arduino (and a couple of other) libraries that perform fusion of all sensor data for the 9150, even one that uses the DMP I believe although I can't confirm that right now.
 
Sounds cool. My plan was to simply take a reference value at dead-ahead, and monitor it over time along with the DMP x-axis value. Whenever the EDTracker spots the magnetometer value at dead-ahead, it checks the DMP x-axis and if it's way off the original value, it adjust it back towards centre - not an outright "reset to zero" but, say, modest adjustment towards it - a percentage adjust of say 1%. Over multiple clock cycles this should trend it back to zero, so should be accurate enough for what we need, but not too affected by magnetic variation or any external influences in peoples households (magnetic headphones remember, people.....)
 
Finally soldered up the PCB - Board all works great!

Wondering if I'll be able to squeeze it into the earcup of a headset...

Thanks again guys. :)
 
Finally soldered up the PCB - Board all works great!

Wondering if I'll be able to squeeze it into the earcup of a headset...

Thanks again guys. :)

Give me a shout. I'll make it fit in yer ear… ;)

So how does a poor exiled Irishman out in Yankeeland get his grubby wee paws on one of those PCBs? The gyro just arrived on the slow boat from China, and I just happen to have an Arduino Micro in the parts bin downstairs… and a bluetooth module, I might add… ;)
 
For anyone looking for a simple way to case the final ED Tracker, this is what I came up with using a "mini ABS" project box from Maplins.

The part numbers for the boxes are "SC782 for the grey and "N78BQ" for the black

IMG_20140518_155130.jpg

The push button hole is about 8mm and tapered just enough to allow me to reach the switch.

IMG_20140518_155205.jpg

The USB plug fits into the box so you might need to take out more of the sides than I did for my cable. Also removed the two PCB board mounting post in the main box to allow clearance for the unit.

IMG_20140518_155220.jpg

Had to remove the opposite two corners to allow the main PCB board to fit in between the two screw mounts. You could cut them away if you plan to glue the cover on.

IMG_20140518_155304.jpg

Used double sided sticky foam to hold the unit to the cover to stop it moving about.

The final thing will be to use Velcro on the lid which is to be the base of the unit so it can be attached to my headphones.
 
Last edited:
  • Like (+1)
Reactions: Rog
@ Omegahunter that looks very good idea, beats my current cellotape to headphones config lol

I been playing around with the new version of the tracker and while it is certainly easier to use, i have a current problem where by i look as far right or left as i can and the view shoots across to the opposite side?

anyone else seen this? Also i find when looking left and right to look at the panels i have to move to far left and am not really looking at the screen as i would like and looking down ever so slightly.

i calibrated on the table flat and let it monitor for just over an hour.

any ideas?


Edit : and using the latest version which includes the calibration fix, also EXPONENTIAL seems to make no difference to me whether in or out

here is a screen shot after an hour of monitoring

v0y0W2n.jpg
 
Last edited:
I been playing around with the new version of the tracker and while it is certainly easier to use, i have a current problem where by i look as far right or left as i can and the view shoots across to the opposite side?

anyone else seen this? Also i find when looking left and right to look at the panels i have to move to far left and am not really looking at the screen as i would like and looking down ever so slightly.

i calibrated on the table flat and let it monitor for just over an hour.

any ideas?


Edit : and using the latest version which includes the calibration fix, also EXPONENTIAL seems to make no difference to me whether in or out

here is a screen shot after an hour of monitoring

v0y0W2n.jpg

First question is did you run the "EDTrackerIICalibrate" sketch first?
That is needed to set the default offsets for the three axis's. It saves the values it gets to the EEPROM ready to be used in the main sketch.

A drift of 15 units in an hour is very high but your drift but not having the offsets set could cause it.

Would also be useful if you pressed button 2 "Get Info" so we can see what values you have set, ( they will get listed bottom left of the UI interface.
 
The flip view has been know to be caused by moving the ED Tracker in the first 10 seconds while the MPU6050 chip is being warmed up.

Its worth waiting the 5 minutes for everything to settle then " 3 Reset View/Drift "

then after another 5 minutes "8 Save Drift " and " 3 Reset View/Drift "

See how it goes using the new values for about 30 minutes ( should only reach about 0.5 drift ). If it a lot over 1.0 then resave a new drift comp. and reset view.

Once it looks right just quit the UI dont resave.

You can end up chasing the drift if you keep trying to hard, and only take being out 0.1 on the drift value plus or minus to have the yaw drift to fast.
 
Getting better results this time by a long margin, odd that the first run i did yesterday ended up the way it did?

JwRPzlO.jpg


thats after 15 minutes doing it the way you advised above
 
Its not that odd at all, we saw a lot worse than this during the alpha testing with the early code.

Just remember to take your time when setting up so you don't miss a step, and if you get stuck all you need to do is ask, most of us from the alpha forums that built it or help to get it working check this thread and are more than willing to help out.
 
Been using for the past hour or so and im finding i have to reset it during gameplay quite often, every 5 minutes ish ?

JnfM5ND.jpg


looks like this sat on my head set on my head keeping still
 
Last edited:
Back
Top Bottom