Quadcopter flight control board routing

On 17.10.2016 12:42, Peter Newman wrote:

I am in the process of redesigning my quadcopter flight control board from running brushed to brushless motors . This necessitated me adding a 5v buck boost regulator and some other minor changes as well as completely relocating or re-orientating virtually every component on the board. As you can no doubt tell from my board layout I am a novice when it comes to PCB routing but I have done the best that I can so far. The circuit is proven so it is just a matter of finalising the routing so I can get some test PCB's made and build them out at home which I am completely comfortable and experienced at doing.

 

Even with the first successful brushed version of this board the Iteadstudio DRC would flag clearance errors on the Atmega, MPU6050, and Molex connectors and this has not proved to be an issue as I have made around twenty of these boards with no issues and they perform flawlessly.

 

With this in mind I am wary of believing some clearance issues being flagged by the DRC on this board but would just like someone to cast an eye over what I have done and point out any glaringly obvious omissions or mistakes that may stop the board from functioning from what they can see. Being a novice I realise that a seasoned professional would make a much more elegant job of the routing but I am not after perfection or anything close to it, just something that will work and that I can hand assemble. When I run rats nest I can see a lot of un-routed airwires but if I run a "run length ulp" it only shows two un-routed wires. I have not checked through all the clearance errors but as I said the only ones that I would be concerned with are trace to trace and/or via clearances.

 

Thank you for any help or advice of a constructive nature and thanks for your time,

Pedro

 

-The schematic is missing, so I can't look at the circuits or even

identify the components functions, wich is somewhat important for

layout, especially when it comes to smps regulators.

-When I do ratsnest, there are 26 serious airwires left. Maybe you need

to add gnd vias around here and ther?

-A quick look at clearances tell me these are probably fine.

-A quick look at the +5v net suggests that you have been sloppy with

decoupling capacitors. Since this board contains a xyz sensor, I'd be

careful with power- and ambient noise. A smps reg just next to it may

have undesired effects.

If I were you, I would add another two layers (dedicated gnd and

+5v/vcc) and get this manufactured in China at very low cost.

For some reason I could not attach the schematic as well as the board file in my initial post, presumably because of size restrictions so I am now doing so.

@CadSoft Guest thank you for your response. Unfortunately I am limited to two layers as I only have access to the Freeware version of Eagle.

Might you be able to point me to any information on how to connect these remaining air wires to the ground planes please. I will also look at the possibility of some decoupling caps on the 5v net and what might be considered a minimum distance between the IMU and v-Reg thank you.

On 17.10.2016 21:08, Peter Newman wrote:

For some reason I could not attach the schematic as well as the board file in my initial post, presumably because of size restrictions so I am now doing so Thanks

Some comments on the schematics (yea, I know you said the circuit is

proven, but still):

-I am not sure if you can plug the battery direct to USB like that. That

is outside my knowledge, and I would have to check it out. Maybe the

battery type allows this, or maybe you dont care if its overcharged, and

maybe it doesn't matter if it catches fire during charging

-For easier BOM management and to optimize volume pricing, I'd consider

replacing the U3 (TPS63001) and U$3 (TPS63002?) with the same adjustable

output voltage chip (TPS63000). (For tidyness, you may want to rename U$3)

- How are the brushless motors driven, and what current do they need?

Driving with a logic output sounds harsh. If the motors have a speed

control input, at least I would assume you need some direct voltage.

And, I would also expect some reverse voltage protection diode over the

motor drive.

Back to the layout

-Make the VCC track routing wider. Make sure there is a thick wire from

the battery to the regulators. Also make sure the gnd wire/fill is just

as thick. The other power tracks could do with some improving too. Some

tracks take totally wild directions, like 5v below u$3.

-You default class 0 clearance has been set to 10. Set it to 0 and use

general 6mm clearance rule. Also, the 10mil clearance set on the vcc and

gnd is impossible to meet because of pad sizes and placement. Get rid of

them, or set them to min (8mil seems to do). The clearances left now can

be resolved by fine tuning.

-There is an excess of layer crossings. You can optimize that if you want.

-Personally I prefer to route stuff to grid and 45deg angles. That helps

me keep tidyness and I know if I see something off, I know something was

moved and needs some cleanup. I see yours is a bit messy, although it

will work if the DRC and airwire issues are cleared.

-Check the position of the xyz sensor. If its in a quadcopter, I would

assume the ideal position would be at the center of the copter, but

maybe that can be offset in software? Or maybe the pcb is slightly

offset on the copter itself? In any case, make sure its far away from

noisy components (maybe there is a note about that in the datasheet?) L2

looks a bit threatening.

-And, like last time, check the need for decoupling caps.

Before you go on fixing the routing, I suggest you fix the airwires. To

fix them you need to add several vias, and they may need you to move

stuff around, creating new clearance issues.

Thanks for your advice and suggestions Morten, I appreciate your time.  The battery and USB  powering is used only during fIrmware and bootloader flashing. Thanks for the suggestion on the adjustable buck boost I'll look into that but these are only limited production runs done at home. The logic level output is only used as a signal into the Electronic Speed Controller for each motor and the motor is independently powered from another battery.

Regarding your suggestions on adjusting my min clearance settings in the DRU check, I am not certain of exactly where to adjust those settings but I assume it may be in Edit > Net Classes? Any chance that you could be more specific as there are quite a few different places to adjust things in all those tabs.

Thanks again you have been a big help.

On 19.10.2016 05:15, Peter Newman wrote:

Thanks for your advice and suggestions Morten, I appreciate your time.  The battery and USB  powering is used only during fIrmware and bootloader flashing. Thanks for the suggestion on the adjustable buck boost I'll look into that but these are only limited production runs done at home. The logic level output is only used as a signal into the Electronic Speed Controller for each motor and the motor is independently powered from another battery.

 

Regarding your suggestions on adjusting my min clearance settings in the DRU check, I am not certain of exactly where to adjust those settings but I assume it may be in Edit > Net Classes? Any chance that you could be more specific as there are quite a few different places to adjust things in all those tabs.

 

Thanks again you have been a big help.

Yes, you are right regarding the location of the net classes edit.

(The general clearance stettings are under a tab in the DRC settings

box). In the net classes box, you will find the last "clearance" column

for class 0 set to 10mil. Set it to 0, and for class1 and class2, set it

to 8.

Now rerun a DRC. The clearance errors that are left can be fixed by

modifying the routing slightly.

But as I mentioned, first fix the airwire issues. Read the lower left

status message after a ratsnest is done. It will tell you how many

airwires are left. Unfortunately Eagle doesn't report those in

the DRC issue list, although it has been a topic on the suggest forum

for a while.

Thank you so much for your help and advice Morten. I set the clearances as you suggested and got them down to 30 odd and them slowly worked my way though them by re positioning traces and some GND plane vias until I now have no errors at all. I have taken your other suggestions on board and will also incorporate upsizing of some power traces and decoupling of the 5v supply as well.

I realise that my routing is far from an elegant solution by seeing as it is the first circuit of this complexity that I have routed totally myself, I am very pleased with the outcome. Of course until I submit the boards to a fab house and assemble them I will not know if my project is a success but I am extremely confident of the outcome

People like you are a Godsend to beginners like myself and once again Thank you for your time, help and encouragement

Pedro

Just as a final update I had these PCB's manufactured and I successfully.assembled some at home. They have performed flawlessly so far which is a very pleasing outcome for me and I once again thank  you for your help and advice Morten.