It's a long story that I will try to summarize. The latest revision of the magnetometer that had some design improvements including making it easier to manufacture to boost production using the pick and place machine, but that also came with a weird issue.
When I was developing the APM Oilpan we had some noise issues on the 3.3V line caused by a missing 200K resistor on the Voltage Reference pin on the 5V side of the I2C translator (obviously this board never saw the light), if you don't populate this resistor it will introduce terrible noise on the line and cause the analog sensors to fail, this bug was found by Jose Julio (Thanks!).
A curious thing is that the first version of the magnetometer didn't have this resistor (it worked fine anyway), in the second revision i tried to do the right thing and follow what the datasheet suggested so I added the 200K resistor (same that caused terrible issues on the Oilpan because we didn't use it), expecting some improvement.
When the boards arrived from the fab, we assemble some and they appeared to be fine, Cindy (our testing lady) added the observation that the sensors were slightly noisier, but we didn't listen (Sorry Cindy!). For the pressure of months of long waits we just released the boards. In just matter of days we received tons of complains.
After days trying to solve the problem we finally discover that the 200K was the issue. The 2nd issue we have discovered but is not critical is that the voltage regulator will not perform well if you don't suck enough current from it, the magnetometer uses very, very low power and the best practice to solve this is to add a 200 ohms resistor from the 3.3V to GND to add enough load to it (Thanks to Nathan Siedle for this trick), we are not doing this right now, so you will see a power supply output of 3.7V, but the performance is perfect.
All new boards have this fix, but for those who already got affected boards I would like to give a sincere apology to all our custumers and I would like to offer two resolutions to this problem:
1.- Send it back and we will repair it for you. We will cover all the shipping expenses. Please contact custumer support for more info: help[@]storediydrones.com
2.-Repair it yourself (DIY) quickly and safely by following the following instructions.
NOTE: Before you repair your board, check if your board has a resistor marked "204". This means it has not been fixed (see the first two photos of the repair steps for reference), and if you see instead a resistor marked "0" your board is fine.
If you choose to repair it yourself and something gets messed up, you cannot then return the board. If you are unwilling to take this risk or are feeling a little uncertain about your soldering ability, do not do this. Just return the board for replacement.
1- Locate the 200k resistor, it is marked with "204."
2- With a soldering iron, heat either of the resistor's solder points
and wait a few seconds for the heat to reach both solder points. Weak solder irons may not work very well.
3- Without applying too much force as no not damage the pads, gently
nudge the resistor with the tip of the soldering iron.
4-The resistor
should stick to the tip, so just lift the tip off the board to remove
the resistor.
5- Make sure you the pad are still there.
6- To create the solder bridge place the tip of the soldering iron
over both pads and apply enough solder to brigde the pads.
7- If the solder ball is touching both pads then your are done.
Comments
Hello!I just bought couple of days an HMC5843 - Triple Axis Magnetometer Rev 1.1 and I don't know if my magneto is concerned by this issue. When I try to test it in CLI mode, it tells that the magneto is not active ...
Help please
P.S Great work with this Arducopter!!!Thanks
Guys,
I am lucky enough though the mag board looks ugly but it has worked.
Taylor Cox helped me check the mag onboard. I can see in the configurator now in art horizon window that the bottom compass reading moves as I rotate the quad. clockwise +ve readings from 0 to 180 and anticlock -ve from 0 to 180 starting from o degrees as base and offsets also looks ok. Tilting the quad 45 degrees in each axis does not show variation of more then 7-8 degrees. So I am done.. Ready for a test flight , I guess.
SID
I have soldered Mag sensor on OilPan, chip side down. I just removed breakout board and change register to very short copper wire. Then back again on the OilPan. Soon, I found the loss of communication with IMU and configurator. Remove again and found that not shorted for these two pins. I managed to short these two pins as well. Then, the communication come back again. Those were happened to me and exhausted.
Jiroh
Guys,
I tampered the board and teh pads came out but finally I managed to solder those two pins and create a bridge with fine wire to the 200k resistor pad and now it looks ok..Still don't know how to figure out whether the mag is working onboard or not.. But please do not do it yourself unless you are sure. It's worth waiting or I suggest go to a friend or a pro to get it done if you are in hurry or else you can always send it back to the store free of charge.
Sid
Hello Jordi
Now I´m sure that the problem as I had discussed at http://diydrones.com/forum/topics/hmc5843-3-axis-magnetometer?id=70...
is the 200KOhm resistor.
After I removed the 200KOhm resistor and jumped the pads, I turned back my board to 5V and it is working fine now.
best regards
Luís Mauro