paparazzi-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Paparazzi-devel] crashed, 5V supply overloading issue


From: Adam Spence
Subject: Re: [Paparazzi-devel] crashed, 5V supply overloading issue
Date: Tue, 11 Aug 2009 00:50:55 +0100

Sounds very good, anyone want to knock one up? I was going to do the half way house and still use the signals from PPZ but it may be cleaner to have the demux on board like you've suggested.

2009/8/10 Yves Touchette <address@hidden>
I personnally think that the best solution is not to power the servos from the AP, period.

How about designing a PCB that would contain a servo demux (like a 4017), be powered by the ESC BEC and have pads for both .1" connectors (regular servos) and for pico blade connectors?

Many of the newer good quality ESC have switching BEC's that are more then capable of driving a reasonable server load.

Running of the BEC will also prevent power related crashs. The BEC is already in most planes. Those that don't have one (glow engins, etc) could use an external BEC.

This should be pretty easy to build without having to reroute a complete AP.

The two sets of connectors would alow users to choose which they prefer and install only those.

Does this sound good?

-yvest

> Date: Mon, 10 Aug 2009 08:59:29 -0700
> From: address@hidden
> To: address@hidden
> Subject: Re: [Paparazzi-devel] crashed, 5V supply overloading issue

>
> Yes, I am referring to the PPM signal from the RC receiver to PPZ.
>
> It was pretty clear that the fault was due to an overloaded 5V supply.
> On the ground I was able to faithfully re-create the AP crash conditions
> simply by rapidly exercising all the servos. I also did fairly extensive
> testing to debug this problem, including measuring current drain with
> all servos operating, which is what clarified the issue.
>
> I have suggested in the past that PPZ be updated with a higher capacity
> 5V converter, and given what you wrote below, it might also be a good
> thing to source the 3.3v line off the battery, rather than the 5V
> supply. However, the "temporary" solution is to simply use a separate 5V
> supply for external devices like servos, modems, etc.
>
> Adam Spence wrote:
> > John, what are you referring to with the RC RX cable? Is this the PPM
> > cable attaching the receiver to PPZ?
> >
> > There's been a couple of crashes like this discussed on IRC recently.
> > I'm not sure if the fault was identified but the overloading of the 5v
> > PSU was suggested. This could easily happen when the servos are driven
> > to the limit and stall due to the hardware set up. Where ever possible
> > the currents drawn by the servos need to be measured when all control
> > surfaces are at the limit. If there are any stall conditions the setup
> > needs to be adjusted either physically on the air frame or by reducing
> > the control throws in the config file.
> >
> > The 5V PSU is not only responsible for supplying the servo power but
> > also the 3v3 regulators power! Therefore if the 5v line drops due to too
> > much current being drawn then it may take out the main CPU. For my
> > particular setup with an 868 modem, the CPU, GPS and Modem can take upto
> > 500mA of current. The limit of the 5V PSU is 2.25A so roughly there may
> > be something in the region of 1.75A left for all of the servos. That
> > only needs a couple of servos to stall and you could be in trouble.
> >
> > Due to the above, I will be removing the 5V connection to the servos and
> > providing another regulated 5v directly from the battery. This will have
> > a limit of maybe 5-10A which is complete over kill.
> >
> > I've just noticed the current-limiting switch on the hardware, this
> > means the servos can only draw 1A and then the switch kicks in and
> > notifies the CPU using GPIO. Does anyone know what the firmware does in
> > this instance? That therefore means that there's even less current
> > available for the servos and anything else which you may have connected
> > on the 5v rail... hmmm. I see this can trigger from 0.5A to 1.25A,
> > anyone ever tested this? I'm wondering now what would happen if this
> > were to fire, the servos would be left in their current position (or
> > driven back slightly from mechanical pressure). The AP might (depending
> > on what the GPIO signal does) try and keep driving the servos at their
> > current position which would mean that when the current limiter recovers
> > it then goes straight to the limit again?
> >
> > Anyway I've rambled enough this morning!
> >
> > Adam
> >
> > 2009/8/9 John Burt <address@hidden <mailto:address@hidden>>
> >
> > Martin,
> >
> > I crashed an Easystar just like that during early flight testing.
> > The AP (Tiny 2.11) went nonresponsive and dived into the ground,
> > despite my having immediately flipped the switch to manual. It would
> > seem that the AP processor just locked up - truly a worst case
> > scenario. The cause of the behavior was an undervoltage spike
> > condition due to sourcing the four servos and RC RX off the 5V
> > supply. In case you haven't already, you should always power your
> > hardware off a separate BEC (which begs the question of why the AP
> > has a 5V source anyway).
> >
> > AP generated noise on the ground wire was other cause of bad AP
> > behavior that I experienced. This causes glitching and reduced RC
> > range. It has been covered ad nauseam in other posts, but I'll
> > repeat it: it's especially helpful to put an RF choke on the RC RX
> > cable. Try to wrap the wire 4+ times around the choke.
> >
> >
> > Martin P wrote:
> >
> > Hello everyone!
> > Last weekend I had a good AUTO1 flight with my Minimag. It
> > reacted to pitch and roll commands, and I was able to control
> > the throttle. I had an attitude command in my flightplan which
> > was never left (as I know now).
> > Yesterday I tried again, without the attitude statement and
> > without a holding point block.
> > In AUTO1, it killed the throttle. Switched to manual and landed
> > to change battery because it was not fully charged.
> > Took off again. Switched to AUTO1 and it immediately stabilized
> > the plane. Strangely, this time it did not react to the roll
> > command. I flew around using the rudder. When I used the
> > elevator stick, the autopilot controlled the throttle. Good, so
> > far.
> > Suddenly the plane dived down vertically from an altitude of
> > roughly 40m. No reaction, just kept going vertical and slowly
> > spinning around the longitudinal axis until it crashed.
> > There was not as much damage done as I expected. These foamies
> > are awesome. All RC components including Tiny and Xbee are OK.
> > Except for the Lipo which is badly bent out of shape and will be
> > replaced for safety reasons.
> > Unfortunately I can't see much from the log as I had lost the
> > link probably minutes before. I was in stage "HOME" because I
> > had left the fly-too-far circle before and came back manually.
> > The switch was on manual so I probably tried to rescue. The Tiny
> > was unplugged which can have happened at the crash or before.
> > One aileron was ripped off and the horizontal IR was missing (I
> > found it in the grass).
> > So far, my report. The Minimag is already repaired and I'll keep
> > trying.
> > Greetz, Martin
> >
> >
> > _______________________________________________
> > Paparazzi-devel mailing list
> > address@hidden <mailto:address@hidden>
> > http://lists.nongnu.org/mailman/listinfo/paparazzi-devel
> >
> >
> >
> >
> > _______________________________________________
> > Paparazzi-devel mailing list
> > address@hidden <mailto:address@hidden>
> > http://lists.nongnu.org/mailman/listinfo/paparazzi-devel
> >
> >
> >
> > ------------------------------------------------------------------------
> >
> > _______________________________________________
> > Paparazzi-devel mailing list
> > address@hidden
> > http://lists.nongnu.org/mailman/listinfo/paparazzi-devel
>
>
> _______________________________________________
> Paparazzi-devel mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/paparazzi-devel


Stay in the loop and chat with friends, right from your inbox! Learn how!

_______________________________________________
Paparazzi-devel mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/paparazzi-devel



reply via email to

[Prev in Thread] Current Thread [Next in Thread]