AQ6 makes Magic smoke - Phoenix rises from ashes

[jabram]

Before performing the Static Data Gathering I ran flashed AQ6 with my Team 16GB class 10 uSD card to make sure it could reliably gather data for over an hour, examined resulting graphs, they looked fine.

I made an insulated food container, checked the freezer temp, it is at -17c
Placed AQ6 inside the open container in the freezer
waited 60 minutes.
closed container while in freezer
placed it with one marked corner pointed toward earth
attached alligator clips to a 12V 12AH sealed lead acid battery, didn't want the battery going flat.
heard some crackling inside the sealed food container
quickly disconnected the battery connection
opened the container to see a lot of smoke and strong acrid smell

Oops I had connected battery polarity reversed, bad move

removed the uSD card, tested it, that has survived

My AQ6 is toast
connecting AQ6 to battery now looks like this and does not smell good


Lesson learnt
I should have placed a diode and a pico fuse in series with the main battery lead. too late now, hindsight is always 100% accurate and Murphy is out there just waiting for me to stuff up.

I suggest for future revisions of AQ6 board a Schottky diode and a pico fuse in series with main supply positive lead, on the board side of the pico fuse another reversed diode between positive input voltage from the pico fuse and GND so it conducts through the pico fuse if reverse polarity power is applied and saves damage to the board.

EDIT:
Details of Phoenix like rising from ashes further down in this thread

[kinderkram]

Oh dammit!
Looks like a complete loss -who knows what else was fried except the STM.
Do you see any chance to repair it yourself?

Sorry for your loss,

Norbert

[jabram]

[jabram]

You poor people on this forum, just when you thought you had got rid of me - I am back

2 days after the infamous "Magic smoke" incident.






I decided to have a closer look at my smelly AQ6
I cut the 3 legs on both RECON and removed them, yuk!!
I cleaned up the black liquid mess with alcohol a brush and a microfibre cloth.

What I originally saw as damage to the corner of the MCU and the adjacent capacitor was actually black goo spewing out of the split side of the RECON. Same for the black goo spewing towards the GPS module. All cleaned up quite nicely with a brush and alcohol.

I thought why not power it from the FTDI connector which has 3.3V signal levels yet supplies regulated +5V.

Connected jumper J3, connected FTDI cable.......NOTHING, no LED.
Then I remembered there are 2 DC supplies from 2 RECONs, soldered a wire between the 2 pin 3 [+Vout] of the RECON pads and voila - LED lit, a short while later, green LED flashing once a second. No uSD plugged in, will try that later.

Checked voltage, a nice steady 5V
Checked for anything getting hot, no signs of any distress.

Ran QGC , connect, AQGCS, select Head Up Display, all sensors working, of course reported battery voltage is zero as there is no voltage going to normal DC input.

I soldered another wire from +5V to the power input +pin and now QGC reports a nice steady 5.0V and reports the battery at 00%, no wonder.





All I have to do is obtain 2 decent 5V power supplies that can cope with MORE THAN 18V and are mounted nowhere near the sensors. Did not like the idea of 2 switch mode DC/DC converters so close to the sensors anyway.

The only thing I like about the 2 RECONs is their suicidal tendency when reverse voltage is applied.

Yes, my new power supplies will have a a diode in series and a pico fuse.

I inserted my Team 16GB class 10 uSD into the slot and the green LED is blinking fast, seems all may have survived, amazing

[rwijnhov]

Nice to have you back . So only thing that got smoked were the 2 recon? good to know.

[JussiH]

Hi jabram

Good to hear that you was able to solve it. On the one i reverse plugged only the recoms died and i was able to revive the board by replacing the recoms.

The offer still stands. I am enjoying some holiday in italy but will take of it when i am home again on the 30nth. I also have spare recoms.

The reason a switching regulator was used was to avoid the usual heating and noise associated with linear regulators. However, a normal to220 7805 will fit the footprint of the recoms and you might get away with it if you keep the input voltage low. Bill knows more about this than i do, so i would wait fornhim to comment.

Best regards

Jussi

[Kisssys]

jabram

The switching regulator is used because of the heat created by a linear regulator. A 7805 would in theory work but it will generate a lot of heat with a 4 cell lipo. I tried to power a Mbed board that way and it was not acceptable.

The Recom's have been used since AQ5 with good success.

The board has digikey 945-1054-ND which is rated at 18 volts in and 1.5 amps.

The 945-1052-ND is the same footprint and is rated at 32 volts in and 1.0 amps. This should be fine if you don't put too many Xbee's, blue tooth modules and camera servo's.

http://www.digikey.com/scripts/DkSearch ... =0&cur=USD

Give them a try and let us know.
Steve

[jabram]

Jussi and Steve,

Thanks, I understand the need for a switching power supply to avoid heat when converting down from voltages higher than say 9 volts, been there done that on numerous designs.

Is there any detriment in having the switching DC/DC converter that close to the sensors ?
Has the board been tested using an external power supply away from the board ?
Was there any difference ?

I was going to try an equivalent which can handle a higher input voltage of 36V [7 to 36V], same output specs 5V 1.5A

MURATA POWER SOLUTIONS - OKI-78SR-5/1.5-W36-C - DC-DC CONVERTER, 7.5W, 5V


At Element 14 it is half the price of the RECOM, not that I would select it based on price only.

Further testing confirms the AQ6 has survived, all sensors work, GPS works really well, nice sensitive receiver.

Joe

[jabram]

[Kisssys]

The board is 4 layers and it's laid out with considerations for noise, it's also obvious it's got to be very compact to put the
equivalent of 3 MK boards in one. Bill as spent a lot of time ensuring that his analog readings are good and as noise free as he can get them. Just take a look at the analog data logged from the baro's and compare them to another board and you'll see why it holds altitude better. Look at the GPS accuracy readings and compare with another makers board and you'll see he's spent a lot of time getting maximum performance out of the ublox.

You could try out that Murata and see. The Recom have not shown any noise problems.

Get it flying and you'll forget about the bad solder job you got.

Steve