I have noticed in the forum a number of people advising the connection of ground (negative power) of all their ESCs to the APM.  In some cases some people are also connecting multiple 5 volt lines to the APM in the hope of providing redundancy of the 5 volt line.   In my opinion  both these practices can lead to intermittant issues that could cause your copter to occassionly behave erratically, and at worse damage your APM (as connecting multiple ESC grounds to the APM can create ground loops). Ground loops will create a noisy and unstable ground for the APM, which is the last thing you want with sensitive gyros and accels.  The random issues that ground loops/bad power can lead people to incorrectly assume that there is something buggy with the software, causing unnecessary work for the developers and testers.

Ideally a seperate, single BEC (5 volt regulator) should be used to power the APM as the 5 volt line from an ESC that is also powering a motor will get very hot, causing the 5 volt line to become noisy and possibly brown-out/shut-down in flight.  The BEC should not get so hot that it is uncomfortable for you to keep your finger on it for more than a couple of minutes with APM+GPS+Xbee etc connected.  If so, a BEC with a larger heatsink  or a switch-mode BEC should be used. In addition using a seperate BEC also ensures that the APM does not share a noisy ground return of an ESC.

I think a lot of the intermittant and unexplainable issues people have had can be resolved by ensuring the power supply and powering wiring topology is sound to begin with.

Does anyone know who contact about having documentation updated to reflect the above?

Views: 9511

Reply to This

Replies to This Discussion

I have done just that, connected with two diodes that block if one power blow

From what I understand,(correct me if I'm wrong) linear reg's (78xx) are very inefficient.  They burn off- as heat- the difference in voltage times the current flowing through them.  Running a 1A load from a 12v battery through a 7805 passes 5W and wastes 7W as heat  These units are undoubtedly reliable, tough, almost ESD proof, but inefficent.  I stay away from them whenever efficiency matters- which is always for me- I live off of solar power.

The oscillations you mention are what cause so much issue.   The rule I always heard was one large regulator is better than competing parallel smaller reg's due to the fact that voltage will never be perfectly matched and one will work itself to death, followed by the next one- OR they will try to compensate for each other and end up with voltage spikes killing the reg's or the circuits they drive.

As a side note- I have an 8A switching BEC I just bought, but will not be needing.  Turnigy MegaBEC if anyone wants it... Its brand new- tested @ 4.98V or 5.97V (5v or 6v jumper select) <.2mv AC @ 4A.  I considered it to be more noise I don't want to deal with.

I measured a lot of noise at altitude sonar/baro, I disconnected the wire for measuring at apm (volt divider)thus all the noise was gone. I put on a feritt core and plugged the cable into apm again, the problem is solved.

Theres a good link for a power line filter for the Maxbotix sonar using a 100 ohm resistor in series and a 100mfd cap right at the sonar unit- somewhere... Seemed like a good idea to me, so I did it.  A choke works equally well usually.  Just don't include your output signal from the sonar in the choke- you may filter out the signal you want fed into the APM.   Or did you put the ferrite on the V.divider cables? 

The battery + BEC with diode protection is a cool idea. Weight being a major concern, has anyone used a big cap instead of a battery?  I have one that is rated 1fd @5.5v that I had pondered soldering to the APM as a backup.  Still pondering... 

I put the ferrite on the volt divider cables, because it seems that is a lot of noise at this wire. I have also choked power line to sonar with ferrite, not the signal wire. I have made separate cable for +5v for sonar stright from the bec. The filter you mention 100ohm/100mf, works best with 10ohm and 50 mf, this is also debated here in the forum.

My configuration is two different bec 6V, one linear and the other switching, both connected to APM with two diodes (the output is 5.3V).
I use this system to all my VIP model (like maxi plane and big heli), zero iusses at the moment.
For the gimbal servo power supply i've 8 bec inside the esc on my drone, and the rule is "one bec for one servo".

Thats the one Michael.  The point is that you can't filter DC too much- until the weight of ferrites, caps and resistors drag your craft back to earth.  Without exception, pure clean DC is crucial.  The cleaner the better. Even a couple millivolt ripple (AC component) has caused me issues with some circuits- I built a tachometer driver circuit for my Cummins diesel and alternator noise was making the tach twitchy- same s#*t different circuit... filter filter filter... 

But please note, some ESC's need the power connected!  My 100 amp ESC is opto-isolated.  The board has two sides.  One is powered by the motor battery, the other is powered by the reciever.  So I need that power wire connected to the reciever.


I know this is not the case with most quads.  I just wanted to point it out in case any heli guys come across this.

FYI, the MegaBEC 8 is linear, not switching.




I'm using an L7805, but I've got well more than enough battery power so I'm not too concerned with the inefficiency.  I like it's simplicity and reported reliability.

Just as a new point of discussion...


Yesterday I learned that bigger caps are not always better for filtering DC noise.  It resulted out of a discussion on the L7805 requiring 0.33uF on the output and 0.1uF on the input, in order to stabilize the voltage and prevent it from making noise (yes, apparently even linear voltage regulators can make noise).  I thought I was fine since I had a 16,500 uF cap bank on the APM. 


Not necessarily.  So what I learned was that caps have inductance and generally speaking, the bigger the cap, the more inductance. This prevents larger caps from being able to filter higher frequency noise.  Smaller caps are better.  And in particular, small ceramic caps are best.  So I procurred the required small caps, and will add them to my L7805.  However, there is nothing wrong with running multiple caps in parallel.  So I can still use that cap bank which is intended primarily as a brown-out preventer.


The information I read is that it's quite common to use large and small caps in combination for the best effect, some circuits even have intermediate sizes.

Yup- Hobbyking says its linear.  If it is linear, they went way overboard in circuit design.  I just opened it up to satisfy my own curiosity, and its got a 16 pin controller IC operating a couple of FETS with their tabs cut off.  Looking at the traces and component arrangement, I'd say its a pretty standard buck converter circuit.  Glad I opened it up too- one of the FET's almost fell off the board cuz it was barely soldered on.  The only thing holding it on was the heatsink- sheesh!  I'll get the FET soldered back on and put a fat load on it with ammeters on in and out and see what the Pin vs Pout is.  @ 4A out I noticed no heat indicating its fairly efficient.

And yes the 78xx series regulators have built-in short circuit, reverse polarity, overtemp and over current protection- in other words electrically bulletproof.  78xx series regulators date back to the early 70's(?) and have been used almost religiously since then.  Most 78xx circuit designs encourage filter caps out to ground from both input and output since they will react to noise from either end.

Again the filter tuning thing.  Smaller value caps for higher frequency noise.  Speaker crossovers are a great example- the tweeter is usually driven through a 6.8mfd cap (high pass), the mid through a 10mfd and sometimes small choke (notch filter) and the woofer is driven through a big choke which absorbs anything above woofer frequencies in ferrite core heating (low pass).  Careful with multi caps- they can introduce anther layer of noise.  Its all too easy to make an oscillator.

Reply to Discussion


© 2018   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service