PX4 baro problems needs case.

[Gary McCray]

Hi Guys,

At this point I am pretty well convinced this is not just my anomaly, this is a significant problem for PX4's and can easily result in a bouncy copter.

This is a response I sent on the Forum to somebody else with a PX4 who was having bouncy copter syndrome.

Hi,
I am currently having trouble with PX4 in Alt Hold modes and it strongly seems to be related to prop-wash back-pressure resulting in auto-throttle correction bouncy pulses.  

Switch into Alt Hold mode from stabilize and copter immediately goes from nice steady hum to loud - quiet - loud oscillation while moving up and down over about a 3 second period in a 6"+ range.

Nothing dramatic, but it appears that the props pressurize the baro which thinks it is going down and so it increases the throttle (which both increases prop back pressure and the copter also now senses it is going up, so the baro now responds by decreasing the throttle.  

You can see and hear this while the copter is flying in Alt Hold: Whrrr-UP -- Quiet-Down -- Whrrr-UP and so on. Switch to stabilize and nice constant easy hold prop sound.  

I have a piece of medium foam heavily compressed sandwiched over the baro between the PX4FMU and PX4IO board and this has only a small improvement.

My Loiter is not working well, but with constant throttle oscillation for the Alt Hold part it isn't worth trouble shooting that until the Alt Hold part is working better (and is probably related).

Stabilize is working excellent, would rate a 9 out of 10.

Bottom line is un-encased PX4 can definitely be a problem and foam may not be an adequate solution.

Probably depends on characteristics of copter and whether prop down wash can affect local air pressure where PX4 is.

I also have an APM that does not have this problem at all:  

On the APM the baro jitter on the log is about 1/10 to 1/5 whole division on the PX4 it is about 1 so 5 to 10 times as bad.

My Accels are all in the range of < 1/10 G in fact < 1/20 G thanks to Kyosho Zeal Gel mounting.

I have tried a partial top cover and this did improve the situation on my Hoverthings Flip, but still not adequate. 

I really think that the PX4 needs a small enclosure to let these pressure pulses even out.

Best Regards,

Gary

I also included the log above which definitely shows the baro jitter on my PX4 and which is 10 times less on my APM (in a case).

Obviously some copters can have this problem because of their design, others whose prop-wash doesn't impinge on the central flight controller holding platform may not have it or at least not as badly.

On my APM there is no problem on a copter of very similar design, but there is a box around the APM which would act as an excellent damper for this phenomenon. It has the same jitter at the same frequency, just 5 to 10 times lower amplitude, thus not a problem.

The constant throttling up and down screws up Loiter as well.

I thought this needed to be here, because I am pretty sure it is a real - genuine problem associated only with the unboxed PX4.

The Pixhawk and APM in a box can even out most pressure pulses before it gets to the baro, thats my story anyway and I'm sticking to it. :)

Best Regards,

Gary

[Randy Mackay]

Gary,

 

      Indeed that attached graph shows a very noisy barometer and I think the lack-of-a-case could definitely be the problem.  Others do have the px4 working on their copters like Leonard and Rob though so it’s not impossible to fix it DIY style.  The quickest route to a more formal solution might be a user designed 3d printed case.

-Randy

--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[Tom Coyle]

Hi Gary,

My solution was to use my PX4 only in a rover. No altitude measurement issues with that kin of chassis:-)

I have my PX4 FMU/IO and associated components stuffed in a Traxxas 1/16 Fiesta chassis which I might try to run in the AVC only it is too big for the Micro Class:-)

On the other hand I suspect your issue with the PX4 baro is very real, but trying to get a custom case made for it might be futile since the Pixhawk has now become available.

Regards,

Tom C

[Gary McCray]

Hi Tom and Randy,

@ Tom, I agree, Pixhawk is the best solution, of course keeping on the ground solves the problem too.

A Randy, are you running a PX4 on any of your Flamewheels and if so do you see this?

Should be possible to cobble something together good enough for this though, I have a pretty decent shop.

Copters where the prop radius is far enough away from the central hub to not impinge on it probably won't even have this problem.

But for little copters like the Flip and Flamewheels, which also have somewhat flat arm tops, the diversion of some of the prop wash to the top of the central plates is inevitable and an exposed flight controller gets the full "benefit" of that.

Especially when your running the biggest props that will fit in the name of performance and efficiency.

Mostly just wanted to mention it here, because it is showing up as an issue from others specifically regarding the PX4 and the traditional "piece of foam" is insufficient to fix it.

And at this point I've isolated it enough to make a pretty good case that what I have said is actually the problem, (A couple Dozen flights and logs, with various "modifications" in place.

For us, the best solution is undoubtedly the Pixhawk, but a homemade enclosure can likely be made to suffice as well, (The foam by itself will not).

I will post whatever enclosure solution I come up with here along with a new Log. 

(Love the New Log browser stuff in 1.2.9.3 MP by the way - Thank You Michael.)

[Robert Lefebvre]

Hey Gary,  I haven't tested the PX4 on my quad yet, because I never could get it to stop glitching. Is yours still glitching or what's going on with it?

And I haven't noticed this behavior at all on my helicopter.  Maybe the reduced downwash velocity (pressure), or less change in the pressure with collective, or the fact it's completely awash in the air column, I'm not sure. 

There's still plenty of reasons to use the PX4 over the Pixhawk, so hopefully we can still work on getting them running properly.

Rob

[Gary McCray]

Hi Robert,

Glad you responded,

I'm getting in the new 1.2.9.3 MP log window a strange "Stabilize" roll failure which could be related to glitchng, still checking on that though. 

So much performance on this little Flip Guy you don't even really notice it though in actual flight.

Stabilize mode is superb.

The 380 sized Flip has more "deck" than the Flamewheels and the end of the 10" props come very close to the deck at the corners, my 450 Flamewheel has far more prop to deck clearance than the Flip even with 11" props, so it doesn't surprise me if you don't encounter this problem.

I suspect this phenomenon is specifically caused by prop-wash being in an optimal location to leak onto flight deck.

And of course it would be much worse on a multi where each prop tip prop-wash thrust asynchronously impinges on the flight deck twice rotation - Blecch! as opposed to a nice helicopter.

Also, likely your Heli's rotor is producing most of its thrust / prop-wash / pressure pulsing towards the tips of the prop as opposed to the central hub area where the flight controller would be likely to be located.

Don't think I'd stick it out on the tail though. (probably not an issue).

I think PX4 can work for a lot of stuff, this is just an issue you need to be aware of and ready to ameliorate if it rears it's ugly head.

Once you understand the problem you can deal with it and although this could be sort of compensated for in firmware, it would not be ideal, so I will develop an appropriate hardware solution for this.

Prefer hardware when its appropriate anyway.

Need a PX4 box.

Best Regards,

Gary

[MB]

Gary,

i have a px4 on a small F330 quad, without no problems. Just cover the sensor with some foam and shrinking tube.

https://www.dropbox.com/sh/lh5qhhgtcv014i8/QyLGhNkAyh

-Marco

[MB]

Aehmmm without any problems :)

-Marco

[Gary McCray]

Hi Marco,

My other PX4 is on an F330 and has foam over it, it also has a much lower pressure pulse and barely perceptible vertical oscillations.

I have tracked the problem further and it considerably relates to the proximity of the propeller blade tip to the multicopters central frame plate.

On my Flip, the 10" propellers, tips pass directly even with the corner of the frame plate permitting prop-wash to actually hit the frame plate directly resulting in the higher pressure pulses.

I have tried several varieties of various density foams and even compressing them double or more between the boards over the baro. To no avail and barely changing the pressure oscillations.

The F330's 8" prop tips are still over an inch away from the frame plate so it does not build up nearly as much prop-wash induced pressure oscillation as the Flip, in practice negligible.

This is apparently very much a frame based phenomenon and only those frames with the blade tip having a close proximity to the central frame plate are likely to encounter it.

Fortunately, 9" DJI compatible props are available and I believe that will minimize this problem.

Of course putting it in a box would help smooth out the pressure oscillations as well, but foam alone will not fix it.

Best Regards,

Gary

[Gary McCray]

Hi again Marco,

There is also not anyway to use heat shrink tubing on the baro on the PX4 and because of connector access issues it can also really not be used over the whole PX4FMU / PX4IO stack.

An enclosure of some sort should act to smooth out the pulses and it could be vented advantageously to further increase the benefit (on top perhaps)

Most copters have more prop tip to central frame plate clearance, so this is a highly specific problem related to un-encased PX4's and those copters with minimal prop tip to frame top clearance only.

But it is the first thing to look at if you have a PX4 and bouncy copter syndrome.

Basically this should be a rare enough phenomenon so that it can be dealt with remedially.

Best Regards,

Gary

On Sunday, January 19, 2014 8:32:23 AM UTC-8, MB wrote:

[Gary McCray]

Further developments,

I just noticed tried yet another unsuccessful test using electrical tape to block out the suspected air pressure pulses when I noticed that the baro was showing bad jitter even before the motor was turning.

I just ran an arm static test with the copter sitting on my desk and the main battery not even hooked up.

The baro is showing the same unacceptable jitter without the motors turning at all.

In use this does cause the copter to bounce up and down like it is on springs and pretty much seems to sabotage loiter.

Looks like I might have an actual bad baro.

I'm going to swap my PX4FMU out for another one and take another look.

Best Regards,

Gary

[Robert Lefebvre]

Just out of curiousity, you might want to check the Vcc (3.3V) on that thing.

[Gary McCray]

Hi All,

Just switched PX4 FMUs and Alt Hold works fine.

(Solid as a rock, no vertical movement at all.)

The original PX4FMU has a bad barometer!

Looking at Terminal CLI baro, movement is much, much less on baro now than it was with the previous PX4FMU.

Also Altitude on the Quick display in Flight Data now only moves 3 or 4 hundredths of meters.

Unfortunately I can't get any valid logs.

No matter what I do I get a 89 byte log with the date and time zeroed in Terminal CLI mode logs download.

If I record a second log it is just log 2 with exactly the same content as this one.

1 logs

Log 1 in /fs/microsd/APM/LOGS/1.BIN of size 89 1979/12/31 0:0

I'm guessing the SD card needs to be reinitialized somehow, but if anybody knows, please tell me or point me to the correct wiki page.

Best Regards,

Gary

[Craig Elder]

run chkdsk on the card

[Gary McCray]

Hi Craig,

Aleady did that based on your previous same suggestion to another user on the forum.

Chkdsk'd fine but still records Empty old date logs though.

Also tried Windows format, same thing, also tried new card.

And tried reload of AC3.1.1 rc1.

Figured that it might be a problem with current trunk firmware relating to PX4, may go back to old stable version and see where that gets me tomorrow.

If nothing else turns up.

At least it flies a whole lot better with a functioning baro.

In any case, no worries, time has a way of fixing this kind of thing.

Best Regards,

Gary