Search This Blog

Sunday, March 30, 2014

2nd quadcopter practically RTF...

Other than having the recently ordered flight control board (Multiwii SE) mounted and configured, everything else was completed today with the addition of the custom made landing gear:

While both trying to save money and reuse scrap parts, I came up with a simple solution which apparently will do the job, instead of spending 10-15 EUR on a standard landing gear. Using the four carbon rods that were bundled with my first helicopter I made a small fixture in each of the quadcopter arms, to secure the rods:

In case of a hard landing this fixture also retains the rods preventing these from sliding into the path of the propellers.

Another advantage is the weight savings, as the four together total about 30 grams (nylon fixture, nuts, bolts and washers not included), which compared to the almost 200 grams of a conventional landing gear is close to nothing. Also the clearance above the ground is the same as that of the conventional landing gear (about 250 mm), which will allow me to consider installing the gimbal in this frame if I get confident enough...

Tuesday, March 25, 2014

Brushless gimbal almost complete

With the arrival of the Mobius action camera, which is a Full HD (1920x1080 @ 30 Hz) little animal (very nice video quality for the price), my aerial filming addition to my #1 quadcopter is almost complete. The only missing element is the gimbal controller board, which I have ordered through ebay, but my mistake (or not) the seller have sent the MPU6050 instead (which I had at the same time ordered from a different seller). Result: I have two MPU6050 gimbal sensors and no gimbal controller. As such a dispute is being solved, and the seller decided to (supposedly) ship me the controller board. Otherwise I will escalate to ebay, get refunded, and the seller downgraded (or banned, who knows..).

It is a bit of a hassle, as I would otherwise have a complete working gimbal by this weekend. This way it will be another month or so of wait..thank you idiot seller (and me thinking that a $6.5 controller board would be a simple deal).

Tuesday, March 18, 2014

Improved quadcopter - more "avionics" and increased run time

In the several weeks that followed a crash caused by an ESC failure, plenty of work have taken place (some of which documented here). Several things have been done since:

  • Replacement of the ESCs with a 30 Amp quad ESC;
  • Upgrade of the flight controller firmware;
  • Building and programming the PPM Sum encoder;
  • Building a BEC/filter with 12 V and 5 V outputs;
  • Adding landing gear;
  • Adding voltage and current sensor;
  • OSD firmware upgrade;
  • Change of LiPO battery from 11.1 V (3S) to 14.8 V (4S);
After some effort tuning the PIDs, this machine is becoming very obedient and stable, which is suitable for aerial filming.

While I haven't yet went through a full battery discharge, still I managed to attain 17-18 minutes from each battery, until I had to leave and do other things (e.g. parenting). This is a very promising figure, taking into account that the full weight of the quadcopter is currently around 1.9 Kg. I expect it to get past the 2 Kg after adding the gimbal with the Mobius camera.

Below is a video onboard the helicopter, with the current measurement working (for some reason the voltage display disappeared before this recording, to never reappear - this is not the first time it happens, and I suspect the MinimOSD extra firmware might be corrupting the EEPROM under some specific circumstances).

And here is the home-brew BEC in operation, powering the Video TX, OSD and camera:

Monday, March 17, 2014

Induction Stove

Some days ago my father asked me to check if I could repair this single element induction stove. I disassembled the device to inspect its inner workings. I was a bit curious because I have never taken apart this kind of apparatus.

I must say that in spite of its relative simplicity it is a pretty elegant piece of electrical engineering. The first thing to notice is the use of conventional materials such as regular ABS plastic found in consumer electronics such as DVD players and other devices we don't expect to fry an egg on top of. Even the main element, the induction coil is held by a very conventional plastic support. The magic behind the thermal insulation is for sure the glass-ceramic material that is used on the top of the stove, which keeps the unit cool in spite of the high temperature formed at the surface of the pan.

The unit comprises two boards:

One featuring the LED display and a couple of Elan EM78 series microcontrollers:

And the power electronics board featuring transformers, inductors, mains voltage rated polyester capacitors, optocouplers, a 20 Amp bridge rectifier, and an IGBT (Insulated Gate Bipolar Transistor):

The largest IC in this board is a LM339 quad comparator.

For tracking the fault (which consisted of the unit starting correctly but when trying to initiate the cooking it would display the error E01), I started with the first candidate (assuming that the control board electronics were ok, given the apparently normal behaviour), the IGBT transistor.

This consisted of a H20R1202 device. Its rated threshold voltage was according to the datasheet 5.7 Volts, so I unsoldered it and tested against my bench power supply providing voltage to the gate, and my multimeter testing for current flow between the collector and the emitter. First I checked that the body diode was OK by testing with the multimeter in the opposite polarity, and next I checked if the device would turn on, by gradually increasing the gate voltage in the power supply. As I passed the 5.8 Volts margin, the device quickly switched from the off state to full on. With this result I found very unlikely that the IGBT would be faulty. Next I looked at the bridge rectifier (a KBJ 2010) attached to the same heat sink. no physical damage would be apparent by visual inspection of the package, but after measuring with the multimeter, every terminal pair appeared to be open circuit in any polarity. This would not be the expected behaviour for a working bridge rectifier. I first considered if the multimeter would not be providing enough bias voltage to cause the diodes to turn on, but after reading the datasheet, I found that this would be unlikely, as the forward voltage would be of only 1.1 Volts, which is perfectly within the test voltage used by the multimeter.

Given this result, I decided to order a replacement bridge rectifier, and if nothing else is damaged, it seems like an easy and cheap fix.

Monday, March 3, 2014

Smoke test after a lot of changes

It's always a stressful moment when after many changes something is put to the test. This was the case, taking into account that since the last flight there was a replacement of all ESCs, upgrade to Arducopter MPNG 3.0.1 R4, installation of PPM Sum module, addition of voltage and current measurement, and use of a higher system voltage (4S instead of 3S LiPo).
Fortunately the test went well, with the quadcopter having no problem taking off from the bed at less than 50% throttle. It demonstrated to require further work with the PIDs, as the stability only became close to rock solid when the rate P parameter was dialed to nearly the maximum. On the other hand it seemed a very nervous and responsive platform, which impressed me positively, taking into account the much greater inertia of the rotors compared to the previous configuration, and the increased weight of the quadcopter. Still I will stress some more on ground tests to make sure the PPM sum solution is reliable. This is an important component, because like most elements in these machines, it is a single point of failure. And from my observation, the flight controller firmware doesn't seem to handle loss of sync with the PPM signal very well (it appears to only get back on track after a few seconds). And of course in this context, a few seconds are the time between flying and crashed on the ground.

The ESCs also impressed me well, because these were barely warm after a few minutes hovering over by bed. Comparatively the other ESCs would heat up more in a similar run.

In a way during this moment I felt like these guys must have felt here (at a different scale though :) ):

Sunday, March 2, 2014

Busy quadcopter setup weekend

With the arrival of some of the parts I have been waiting for, more work accumulated with the finishing of my 2nd quadcopter. The parts were the landing gear and the quadruple ESC: The landing gear arrived with all the necessary parts as expected, including the lower crossing bar and the foam rods. The bar and other rods are all made of carbon fiber. The 4 support rings (for attaching to the frame) all had a rubber ring which helps to further dampen the vibration to additional items such as a camera:
The ESC arrived in good condition, however the motor leads were too short. I had to replace these with others twice as long:
It was a good opportunity to look at the hardware in detail:
As expected it had four Silicon Labs C8051F330 microcontrollers, a switching regulator for the 5 Volts, and a 12 Volt linear regulator for the internal circuitry (probably to drive the FETs). The exact FETs were the unknown for me, but I was not disappointed. It came with 24 N-FET transistors, the LR8726, from International Rectifier. This means that each ESC has 6 of these FETs, one for the high side and one for the low side of each phase.
This particular FET has nice specs such as a continuous drain current of 86 Amps at 25ºC, and 240 Amps pulsed (see datasheet at\irlr8726pbf_4532756.pdf). This gives me some comfort, at it is an indication that the manufacturer was conservative in the selection of the part, giving a good margin between the capability of the components (86 Amps) and the rated max current of the ESC (30 Amps, 40 Amps peak). Of course that the switching nature of the application makes it more important to play safe in the selection of the part, as datasheet values illustrate a DC scenario. For the landing gear I had to make four aluminium support bars to integrate it with the helicopter frame:
This was the solution that allowed me to have space under the frame to accomodate the ESC. Finally I flashed the 4 ESCs of the 2nd quadcopter with the latest release of SimonK. Was a boring and painstaking process of soldering the ICSP connector into the tiny pads of each ESC, repeated four times. At the end I put everything together, but while bolting the flight controller after reorienting it in the frame, I touched one of the piezo gyros with the wrench, and its lit popped off. I put the lid back and added hot glue on top, but during the first test running the motors, I realized roll sensing was lost. Perhaps air contamination or some mechanical contact damaged it. This is probably a justification to replace this KK board with a new flight controller.. And here a view of the quadcopter inventory, with the two machines in a near ready condition:
Next effort will be to take care of the power supply for the video system, and make further tests to ensure everything including the PPM Sum are reliable enough for a test flight. The 2nd quadcopter will have to wait for a new flight controller.