I believe I have a problem with my Pixhawk and was hoping for some help. Also posted this on rcgroups.
The issue is that when my speed controllers are hooked up to the Pixhawk they exhibit a bad sync issue. However when they are connected to my old APM (3.1) or just a receiver directly they do not exhibit these symptoms. This has resulted in 4 crashes now until I just tonight narrowed down the issue to the Pixhawk.
Can anyone advise what is going on? Here are the specs of my setup:
Afro 30A ESC's with SimonK firmware (have tried all versions, and made my own with various parameters)
Tiger MT2216-9 1100kv motors
4S 3700mah battery
I sent in the combo to Craig @3DR from this video: https://www.youtube.com/watch?v=u_KZ5K0R7eY&list=UUYZ_ZGYHxt_CX.... He still has it.
He was not able to replicate my issue though, which was odd since I think the issue was pretty clear in the video. In any case, my issue was resolved when I bought 3DR speed controllers so I did not pursue it further, aside from following the thread occasionally.
I would like to add this here, not as a criticism of the hardware (lord knows I don't have a clue) but for the information of those who might find it useful.
ESC: Afro 20A ESC (each ESC is grounded to the Pixhawk)
Motor: iFlight MT2212 1000kv
Props: 11x4.5 CF (I fly at 1,200m, around 10% humidity and current temps around 90F)
Power System: 3s
This was my second rebuilding of the tricopter. Due to the weight I had upgraded from 10x4.5 to 11x4.5 carbon fiber props. I had a couple of test flights on the prop/motor/ESC combination before the rebuild and I thought everything was okay except for some tuning issues. I tore down and rebuilt (6% weight savings) last week and took it out for a test flight. After putting a pack through it I decided to try an autotune. On the first roll, the tricopter simply rolled over and fell to the ground. Two broken CF booms and some severed wiring will not be a big deal to replace, but I am now concerned about this issue.
Before this thread, I hadn't heard of sync problems with a high kv, low(?) pole count motor. I knew that high pole motors could have issues and that low kv motors could have sync issues, but before I found this thread I didn't know this was an significant issue. I assume that I have created this sync issue by using an oversized prop on the 2212 motor. However, looking back on past flights, I can think of two other times that the copter rolled unexpectedly using the 10x4.5 props before this incident. One time the 'copter was high enough that it recovered, however, the second led to a crash into a nice soft ornamental tree. However, outside of those two instances I had probably 20 successful flights on those props.
The way I see it, I have to change something. I probably have to accept that the 11" props are too much for these motors. However, that alone will not solve the issues that I have had with the 10" props. I just purchased a set of iPower MT2217's and 18A Kiss ESCs that I could divert from a quad build, but ideally, I would like to find a way to make this setup work, so if anyone has some suggestions, I am willing to learn.
I have considered changing this build from a AV rig to a more performance based FPV tricopter. Does anyone have an opinion on whether I would be more or less likely to experience sync issues on a 4s system with 9" props on this build?
I'm no expert but I have used much of the same kit you list, so FWIW here are my thoughts :
1) Did you re-flash your AfroESCs? (See my overly length post earlier about that being the root cause of my similar issue with a 3DR Y6 + PixHawk + Afro). This is an absolute must for all AfroESCs. Manual calibration alone is not enough in my opinion as you just don't know what firmware is on it.
2) Save the Kiss ESC for smaller Acro or high performance FPV quads. Ideally, save them for a build where you're going to use a MultiWii and can enable OneShot. Bare in mind these ESC are not reflashable, and bare in mind that while the reports on performance are incredible, they also have their teething issues and a few have come down in smoke. A reflashed Afro (to my knowledge) has never done this and is more of a known quantiy while you're ironing out issues with a new build.
3) Always test on the ground up to full throttle ( I use strong zip ties that are loose enough to see that it is off the ground and in control)
4) If possible measure the current draw at 1/2 and full throttle for each motor. I appreciate this isn't everyones cup of tea but well worth it and easy to do with a cheap HK Turnigy amp meter.
5) While I've never flown a tri, I would say 3s / 10x4.5 / 1000kv is not going to be sufficient for a AV system. On 3DR style UAVs I always run 4s / 11x4.5 / 880kv which would be significantly more thrust given I have 4 or 6 motors vs your 3.
Hope that's of some use?
Dan thanks for your reply. Sorry that it has taken more than a month to put this into use but parts and life sometime get in the way of a ground-up rebuild.
I flashed all of the ESCs and took the prop down to a 10x4.5. Test flights went very well and I was seeing no signs of sync issues on the ground or in the air through the first couple of packs. Loiter and AltHold tested fine without issue (in the past AltHold has caused a sync issue) so I thought we might actually be good to go.
Now, I had used autotune with the 10x4.5s several times without issue. I suspected that I was pushing my luck with the 11's so with that delusion I put the tricopter in AltHold and hit the channel 7 switch. This time I didn't even get a twitch, it just rolled over and into the ground. Fortunately, I only popped some zip ties for once instead of shattering carbon fiber booms and breaking frame plates.
Honestly, at this point I am going to trash out the Afro escs and go with something else, probably the Hobbywing xrotors for now.
Just another datapoint to this. Last night I did some testing of my own. I decided to pull out an old school Futaba 72MHz receiver, and see what it was doing for PWM voltage levels, to determine what if any "standard" there is. What I found was that the PWM level coming out of the Rx was simply whatever voltage it was supplied with. Whether I used a 6V BEC supply, 5V, or even 3.3V from a benchtop power supply, the PWM level was the same as the input. I even tried to see how low the Rx would operate, and it went down to 2.44V, at which point it happily output a PWM signal at 2.44V level.
I also tested another old Rx, this time a JR 2-channel unit, and had the same result.
Conclusion: There is no such thing as a standard PWM voltage level. Certainly not 5V. In the distant past, the level was simply whatever the Rx was supplied with. In the recent past, the level was set by the internal operating voltage level of the Flight Controller, which meant 5V for Atmega-class systems. This, since the PWM was generated internally by the micro-controllers. Now with Pixhawk it is 3.3V since that is what it's microcontroller operates at. But this is nothing new, as Lorenz' points out that modern digital receivers also output signals at 3.3V levels.
This latest kerfuffle is just another example of Kendall and Joe's backwards scientific method that goes something like this:
Problem: Something doesn't work right.
Conclusion: 3DR/Arducopter must have screwed something up.
Hypothsis: I found something I don't like about their design or program.
Testing: The first confirmation bias I can find.
@Papi78: Why would you assume that all the other boards are outputting 5V levels? Have you measured them all? It would be interesting data, so please share. Here is the schematic of the CC (the CC3D doesn't seem to have any shifter onboard, too, as you will easily be able to tell from the photos of it):
Its 3.3V straight off the 3.3V MCU (some might argue having protection resistors there would be a good idea). I wouldn't be surprised if that applies to more of the autopilots you listed than you assume.
Please keep in mind though (before now doing a full market research on signal levels) that 3.3V is a perfectly valid PWM pulse with margin and that a cable (or missing GND connection) able to introduce a 1.5V drop easily also introduces a 3.5V drop. You can't build a safe system relying on particular resistance values or floating ground - a decent system setup has to be one to two orders of magnitude (a factor 10-100) away from a badly conditioned signal (so 0.01 to 0.1V drop as absolute maximum, and never anything like 1V).
Problem Solved.... Cargo Cult Reveals All....More to Come
May I suggest you provide scope shots of your preferred Futaba, Spektrum or FrSky receiver (assuming a model released in the last ~3 years) as well?
The issue is, you stuffed up and and continue to try to avoid responsibility. I'm not Playing your Game
To consider these scope shots as data, they need to show the complete picture. The claim of 5V levels being necessary all builds on the assumption that this would be the standard / required - if we make that assumption, we have to prove this first. Needless to say, all big RC brands output 3.3V or even 3V levels, which are perfectly fine high levels at TTL.
Even if we ignore what the big RC brands do, we will need to adhere to the definition of TTL levels. Please consult a data sheet from a 7400 series (which is *the* TTL logic family and essentially an industry standard): It does not output more than 3.5V as high level and considers a 2.0V as high input.
On your RC groups post: You got the part number for the level shifter wrong. We're using the TXS0108, not the TXB. The TXS data sheet says this about the center pad: "The exposed center pad, if used, must be connected as a secondary ground or left electrically open. " As the shifter is connected to massive wires and can sink heat effectively through its outputs, it has not been connected, perfectly in accordance to the data sheet.
The margin argument between a 5V level and a 3.3V level dropping to less than 2V is not one I would dare to make for a flying vehicle. If the wire resistance is so bad and high enough to drop enough voltage to drop 1V or more, its essentially a broken wire. This is not something anyone can fly on safely. Which is probably the same reason Spektrum, Futaba and FrSky are considering 3.3V signal levels the better choice.