FreeTrack Forum

Hi. I managed to find wide angle IR LEDs locally with a vendor that keep spec sheets around but the viewing angle was only 75 degrees as opposed to 80 degrees on the SFH485P. The rest of the specs look pretty similar to the SFH485P. 1.7v max forward voltage (1.5v typical) and 100mA max current so I'm expecting similar performance. I'm using a single point model (I'll build my 3 point setup once I work out the kinks on my current system), 2 rechargeable batteries that churn out 2.4v-2.5v total and 10 ohms for the resistor, giving the about LED 90-100mA. I sit pretty close to the webcam as I use a laptop and I have the webcam mounted on top of my monitor so the webcam is just a bit beyond arm length from me. I didn't bother removing the webcam's IR filter as there's nothing in between the lens and the sensor so I guess it's painted on the lens itself. Though the lens doesn't appear to have a red tint if I look through it so maybe the manufacturer skipped it altogether to cut down on costs.

I'm using absolute mouse output and for the most part, tracking is pretty good. My (very cheap) webcam runs at 25fps at 640x480 resolution and I get very little jitter if any. When I try to reach the edges of my screen, however, the LED gets very weak and the mouse cursor gets jumpy. I've adjusted my webcam and FreeTrack's threshold settings just high enough so that any "noise" disappears and only the LED shows up. I think I can get slightly better performance if I fine tune those settings further, but I'm not expecting large improvements. I'm currently using a piece of floppy disk media placed in between the lens and the sensor to filter out visible light and I'm guessing that's the source of my problems. I am looking for some developed film to but these days it's kinda hard to find those. Also, my webcam has 4 white adjustable intensity (via potentiometer) LEDs that I plan on replacing with IR LEDs if I get good results after replacing the "floppy filter."

So anyway, I have a few questions. Is there anything I can do to the floppy disk media to improve IR reception strength for now or perhaps some other cheap readily available material that does better? Will the extra 5 degree viewing angle of the SFH485P make a big difference? How much improvement will I get if I switch to a PS3Eye (or some other webcam, maybe a true night vision webcam) instead? Is there anything else I should be thinking off doing to improve tracking?

edit: I looked at my webcam's lens assembly closer and I didn't see it at first but i finally found the IR filter. The problem now is that it's going to be a pain getting to the IR filter as the lens assembly appears to be molded plastic.

What is the part number of the IR LEDs?
Do not operate at 100mA! Use 50 mA to 60 mA max for longer life of LED.

What is the camera model number?

Post some pictures of IR-LED and camera parts.

Well looks like I made a few critical mistakes during my single point build.

Apparently I underestimated my webcam's IR filter and now that I managed to remove it (after almost crushing 1 end of the lens assembly), using 10 ohm resistors are overkill and just floods the cam with IR.  I'm using 18 ohm resistors now.

The receipt is handwritten so I'm not sure if this is the correct part number: IE-O57SHC6. I also should have photocopied the spec sheet because it appears that 1.5v isn't the actual forward voltage of the LEDs. Now I'm not 100% sure I did this right but I measured my circuit's voltage without the LED with a multitester and I got about 2.5 volts.  I measured it again with the LED and got about 1.3 volts.  This means that my LED's forward voltage is actually 1.2v correct?  If that's the case, I was running my LED past 100mA and now I'm running it at roughly 70mA.  I already have a 27 ohm resistor that I am about to test on a 3 point model that I just finished building.  That should bring me to around 50mA.  If my webcam picks it up fine, I'm going to replace the resistor on my single point setup as well.

Also, for additional info, the LED didn't heat up while I was using a 10 ohm resistor. I didn't use it for more than 10 minutes at a time though and I have airconditioning.

edit: At 27 ohms, the leds get picked up fine.

I managed to find wide angle IR LEDs locally with a vendor that keep spec sheets around but the viewing angle was only 75 degrees as opposed to 80 degrees on the SFH485P.

Half beam angle for SFH485p is 20°, so 40° in all.
See: SFH485p Data sheet

If yours got 75° they will be better. If not, 5° less is not critical.

Steph @ 01/02/2012 - 12h15 a dit:

Half beam angle for SFH485p is 20°, so 40° in all.
Please read the correct spec sheet and note 80 degrees total.
See: SFH485p Data sheet
That link is showing the non P version.
The P version has a flat lens.

You are right:
This is the right Datasheet: SFH485P

Beam angle +/- 40°.

After some more testing, I conclude that it was definitely the IR filter. No more tracking problems right now. Also, I recommend not doing any math while doing any math or multitester work. I just realized now that I completely botched the forward voltage measurements. If the diode test mode on this multitester I'm using is correct, the LED's forward voltage is 1.3v. I will be doing more measurements later on. I also need to get more or less accurate specs on the LEDs as my camera has 4 white LEDs built in that I plan on replacing with IR LEDs later on.

I really can't say what my camera's model is as this is one of those cheap china cams that pop up under different fly-by-night brands. I got the webcam from a not-quite-reputable-but-very-cheap local reseller called CDR-King. I do have pics but I don't have pics of the IR filter in place because as I was trying to remove it by cutting open the lens assembly, the part of the lens assembly were I was cutting crumbled apart. That's cheap manufacturing for you. I did get the IR filter out and replaced it with a "floppy filter" and glued as much of the crumbled plastic back onto the lens assembly as I can. I'll take more photos and post them later.

Anyway, many thanks for the advice so far. Now I'm putting the finishing touches on my 3 point clip model. I've played Evochron Mercenary with my single point model already and it works great as a mouse replacement. Considerably more accurate and responsive than the face tracking alternatives I've tried so far.

Is the camera on this link?

http://www.cdrking.com/index.php?productstype=All+Products&searchvalue=web+cam&mod=products&type=search&x=14&y=6

dewey1 @ 01/02/2012 - 20h32 a dit:

Is the camera on this link?

http://www.cdrking.com/index.php?productstype=All+Products&searchvalue=web+cam&mod=products&type=search&x=14&y=6

ACM-410 on the third page.

I recommend staying away from a reflective type setup using IR LEDs on the camera for a head tracking system.

The discrete clip or hat with IR-LEDs is much more stable and less prone to "false" head tracking signals.

I have both a TrackIR 4 Pro Clip and relective head band. I prefer the Pro Clip with my webcam over my TrackIR 4 camera using FreeTrack software.

Actually I plan on replacing the LEDs just for IR photos but it turns out that the resistance on camera LED circuit is too high for the LEDs that I plan on using so I'm still re-evaluating that. The original LEDs had a forward voltage of 2.7v to 3v and I estimate the max current to be between 20mA to 30mA.

I did plan on trying a reflective setup much later down the road to avoid burdening my head with the weight of the model but I quickly decided against that. While I was testing my single point setup I noticed that my glasses reflected IR well enough to get in the way.