Originally Posted by PhracturedBlue

Using the DFU it should not be possible to brick the device. the worst case is a 'reboot loop'. You need to pull the battery then power back up with 'Ext' pressed to get back to the bootloader.

Sweet, thanks!

Originally Posted by thloh85

I did use summon-arm-toolchain to get the arm-none-eabi- cross compiler, my build works for the Devo8 repository, failed for the Devo10 ones...

I have no idea how to help you then. There is very little difference between the two. You could try compiling with TARGET=devo8 and see what issues you get (it won't finish because the BACKLIGHT functions are new, but it may give an idea what is different between the two platforms)

Originally Posted by thloh85

I have no idea. I did not disassemble the TX. Can you share the circuit board picture here?

I was assuming you'd know by using the Devo10 firmware, not by visual inspection.
The thread in question is here:
http://www.rcgroups.com/forums/showt...1606210&page=9
(bottom of the page)

Originally Posted by PhracturedBlue

I was assuming you'd know by using the Devo10 firmware, not by visual inspection.
The thread in question is here:
http://www.rcgroups.com/forums/showt...1606210&page=9
(bottom of the page)

Yes I know it in the code, just wondering how you get the info when you write the code

PB,
I pulled new changes for the Devo8 and the undefined reference now happens to the Devo8 repository as well. Something changed in your build?
Revision 482:483 is where the undefined reference started to happen on me. Any idea?

Well, I was trying to debug why I can't reliably shutdown my devo8 in Deviation, so I traced out the circuit.

I have one of the early Devo8s that have the silk-screen showing all component values, so I know exactly which resistors capacitors and transistors are used.
Following is the circuit diagram.
The power transistor that I and Tripert blew was an IRLML6401. The AO3401 is a good replacement part.

Originally Posted by thloh85

PB,
I pulled new changes for the Devo8 and the undefined reference now happens to the Devo8 repository as well. Something changed in your build?
Revision 482:483 is where the undefined reference started to happen on me. Any idea?

You definitely do not have the proper newlibc compile then. That is the commit that imports the new printf (which pulls in a bunch of newlibc code) and would cause this issue if the unwind code is not disabled.
What flavor of linux are you running?and what is the git version of your summon-arm build script?

also, in the future, the short numbers are not a reliable way to mention hg versions, you need the hexadecimal value after the colon to uniquely identify a version if there are multiple contributors.

Edit: you can stub out the 3 functions in question, and the code may build. I was able to build it once in that way.

Useful schema. I think there are two values that are not right. At the left side, the values of 1M and 100K need to be interchanged, I'm not sure because I haven't the transmitter opened. I have had a problem with powering off that sometimes fails and I have solved it by reducing a little the value of the 1MOhm resistor till 950KOhm. Also I have added a diode of 0.5V to the P-Mosfet and now the transmitter advises when voltage is low for around 2 or 3 minutes. I'm going to try again not including the diode, and also with a germanium diode of around 0.3V.

Originally Posted by Tripert

Useful schema. I think there are two values that are not right. At the left side, the values of 1M and 100K need to be interchanged.

No, I am quite sure about them. They constitute a voltage divider which is the reference for the Tx detecting the off switch. That value needs to be driven high when the switch is off. The voltage measures 4.5V at the divider which is exactly what the circuit would predict. This is also the circuit I am interested in, since Deviation has difficulty switching off. I think there may be some internal leakage in the STM32 that is causing it not to work. The circuit essentially assumes zero current to work properly. I need to experiment with powering ff some of the STM32 clocks to see if one of them affects the power switch detection.

EDIT: Slowing down the poll rate did improve the power-switch detection in my firmware. The most likely culprit is the A/D conversion, but I didn't spend time to narrow it down.

Originally Posted by PhracturedBlue

No, I am quite sure about them. They constitute a voltage divider which is the reference for the Tx detecting the off switch. That value needs to be driven high when the switch is off. The voltage measures 4.5V at the divider which is exactly what the circuit would predict. This is also the circuit I am interested in, since Deviation has difficulty switching off. I think there may be some internal leakage in the STM32 that is causing it not to work. The circuit essentially assumes zero current to work properly. I need to experiment with powering ff some of the STM32 clocks to see if one of them affects the power switch detection.

There was/is a design flaw in Devo 6 circuit so it consumed alot of battery when turned off.
Coud you have a similar problem in Devo 8 that only shows in Devation or when trying to use a lipo?

Originally Posted by PhracturedBlue

You definitely do not have the proper newlibc compile then. That is the commit that imports the new printf (which pulls in a bunch of newlibc code) and would cause this issue if the unwind code is not disabled.
What flavor of linux are you running?and what is the git version of your summon-arm build script?

also, in the future, the short numbers are not a reliable way to mention hg versions, you need the hexadecimal value after the colon to uniquely identify a version if there are multiple contributors.

Edit: you can stub out the 3 functions in question, and the code may build. I was able to build it once in that way.

I did stub out the 3 functions to make it build earlier today, but had to go out for family events, so I left it there untested.

Originally Posted by thloh85

I did stub out the 3 functions to make it build earlier today, but had to go out for family events, so I left it there untested.

PB,

I tried the deviation10 branch with stubbed out _exit, _getpid, and _kill functions but after flashing the DFU the Devo10 just gave me blank screen and not able to turn off (cannot get back into flashloader) unless I pull the battery. Could you possibly upload your Devo10 DFU for me to try out?

EDIT: Alright, I removed some function calls (particularly the all the printf) and now it works... Must be the newlibc stub didn't quite work out afterall...

Thanks

I have started documenting the Devo10 hardware here: http://deviationtx.com/articles/9-de...re-connections
I've found the Analog controls and how they map (and I did find the pin to read the battery voltage on the Devo 10 as well), and have documented them.

Here is a devo10 dfu. It corresponds to build: cf092cb16295
It will power on the screen, beep, and write button and ADC values to the UART.

Originally Posted by thloh85

EDIT: Alright, I removed some function calls (particularly the all the printf) and now it works... Must be the newlibc stub didn't quite work out afterall...

If you are running a 64bit Linux OS, I can tar up my build ENV for you. I have it built for 64bit Ubuntu 10.04 and 12.04.

FYI, my summon arm toolchain was built using git:4e2f939919a036ced5bac61f00e86c82c4027e78

EDIT: Also YAGARTO works on windows, so may work on Linux too as an alternate

Hi PB,

What's the CYRF_MaxPower check for? What information does the address 0x80001007 contain (code in cyrf6936.c)? It appears my Devo10 reports the power of 10mW, but the datasheet of CYRF6936 doesn't say there's a limitation, it just says write the power output to offset 0x3 (PA setting), is that correct?

Thanks