I have assembled a new APM from the diydrones store and uploaded the newest code for Jan.3 (1.01).
When I have the APM/IMU installed in my Easystar, the PPM light flashes and the A B C light never go into a flashing state for calibrating the gyros. All of the airplanes control surfaces and throttle work normal but the ailerons have a twitch that matches the pulse of the PPM light.
On the benchtop, I have a BEC powering the APM with 3 servos connected. Upon powereing up, the ABC lights flicker as normal and the aileron servos no longer twitch. When I try to use my transmitter to control the servos, they work normal but after a few seconds of inputs the APM resets into calibration mode.
I've tested the flight sim and everything works on that end. My older APM works with the newest code from Jan 3rd just fine on the Easystar.
I think the issue may be bad firmware for the PPM Atmega328 chip but I wanted to run this past ever one before ordering an AVR programmer.
I removed them one by one (Airlerons, Elevator, Rudder) and tried to boot APM from ESC power only.
Once I was down to only the throttle (which supplies power to the APM), it worked. The APM booted normally.
So, what does that mean? I have to provide a separate power source for my servos? No problem. But why? Or better - why do others not have to? (Looking for the Lessons Learned here so we can provide advice to others).
I think there must be a break in the trace from SJ1 to the +5 trace. I soldered in a jumper wire from SJ1 to the +5V pad on the GPS port. All is good now! No resets.
I'v flown a very good 13 waypoint AUTO flight with a Katana.
I'm hearing you loud and clear, Irvin. I read your prior post several times. I pulled my XBee out of the loop with no effect. I measured the resistance between my SJ1 and the GPS port and got 90 ohms. Is that enough to pull the volts low enough to where APM won't boot?
And while your jumper sounds like a viable option (that I may exercise in the future), is it the source of the problem or just a band-aid which covers the real problem? I think I'll try the upgraded ESC first for the experience. I've gained the Manager stink recently, but I'm still an Engineer at heart. And I've got time.
Otherwise, if the houses were rapidly getting bigger on short final, or we were on fire, I'd go with your suggestion in a heartbeat.
I would like to have Jordi verify this - but I believe the resistance between the SJ1 pad and the 5volt pad on the GPS should be a dead short. Any resistance between the two is the source of the problem. It may be a band-aid but it seems to be a necessary one.
I've scoured this thread hoping to answer a problem of my own.
I think I may have found a fault in the PCB layout of the ArduPilotMega?. This is a little difficult to explain but I'll try.
I have a v1.4 PCB with attached oilpan & DIYDrones GPS & magnetometer. I have a newly built quad which I was in the final stages of setting up. I have a Turnigy 9x Rx, eight channels are connected to APM. I also have four unbranded ESCs connected. The only other connection is 0 & 5v from one ESC to two pins of the APM I/O connector.
While testing the setup everything seemed to work much as I would expect then suddenly it died completely. The APM would work fine if USB powered but not if powered through the I/O connector. On close inspection of the APM I noticed that the 0v track on the underside of the board running between IN5 & IN6 had burned out.
Mystified, I did some probing, there are plenty other 'ground' tracks on the I/O connector BUT as far as I can tell these don't connect to ground!
Just to make sure, I scraped off a patch of resist on the 'ground plane' which connects these tracks, it is NOT connected to ground on my board now that the above mentioned trace is broken.
If I'm reading this right, the only power connection to the APM (and oilpan etc) is effectively the 0.25mm track between IN5 & IN6.
This is quite difficult to check. As long as this track is intact there is continuity between all 'grounds' but when it's broken none of the 0v pins on the I/O connector are grounded.
Unfortunately I can't check the layout in Eagle as I only have a Free version which (apparently) can only display two conducting layers so I can't see the ground plane.
Anyone having problems powering their APM via the I/O connector might check that the 0v pins are actually connected to 0v.
Maybe someone familiar with the PCB layout can explain where the I/O connector is connected to ground?
I do have a couple of problems with my reasoning but I think I've waffled enough for now!
APM PCB fault?
PS forgetting that this is a community project :(
If someone can send me a blank PCB I can check this, or maybe someone else can:
Connect a bench PSU with current limiting to the 0v on the I/O connector.
Set the current limit to 0.5A (the polyfuse rating)
Connect the PSU to another 0v on the APM (as far away as possible)
See what happens. It should sustain 0.5A indefinitely.
Thanks for the swift response. I'm not entirely sure that I follow you. Isolating the + rail makes sense but NOT the ground.
My point is that there is a common ground throughout the APM board but this is ONLY through the track which I seem to have burned.
As I indicated, I'm not entirely comfortable with my reasoning. It's difficult to imagine driving several servos if this is the only ground connection to them (a few hundred milliamps would probably burn it)
However, that's how it is with my APM. This track has vaporized and now the 0v has become isolated. There might be some other point on the PCB where the I/O pins are (or should be) grounded, where? A via somewhere?
Anyone with a blank PCB can check this quite easily.
I will have to disagree here with you. Jordi's v1.4 shows a common ground between the APM electronics and the servo header ground pins. There has to be a common ground between the servo header and the APM electronics or the board would not be able to receive receiver signals nor drive the servo signal lines. There is a Schottky diode between the APM VCC and the servo header +5vdc which prevents the APM VCC from powering the servo header +5vdc, but allows the servo header +5vdc, which is usually provided by a BEC, to power the APM. This is smart because the Xbee Pros can draw up to 250ma which the APM v1.4 onboard regulator cannot supply. Even the bigger +5vdc regulator on the v1.0 board could not have powered the APM, Shield, GPS, speed sensor, and Xbee Pro without approaching overload if the APM was powered separately from the servo header.
We are always learning! I always take away some new tidbit of knowledge everytime I visit the Forum. A number of times I have been wrong, was corrected, but learned something in the process.
Just a thought.