-
in over my head on power distribution design
Posted by
guardian257
on 31 Jan, 2016 09:54
-
so im building a rather large multirotor(drone). an octocopter.this is not by any means my first model ive build dozens, i pretty well know how to make them. ive 3d modeled it part by part. and i will soon have my cnc router to start working on it and i will do several projects but my 1 issue i cant figure out. i petty much have my heart set on a power distribution board. i know it works as ive used them and seen them. but how to make one. not manufacturing, i believe i can mill out a test board with the cnc, but rather the design processes is enough to stump me.
specifically, i need to pull an s***ton of amps. ether about 145amps @ 25volts or 240 @ 16volts, haven't deiced which is better.
all it need to do is transfer power to the hungry motors and things from the battery and if i where feeling confident maybe a few voltage regulators and simple circuts to make the build neat. i really want to go all out for this model as it will be a centerpiece. something i built my self that was as maxed out as possible.
note that the big number of 145/240amps disperse pretty quick. its only 18/30amps per motor and with the plug in the middle it will divide the amps at the connector
also note that i may go with the larger number. because the speed controllers are still optimized for 4s lipos at 16volts. and several drawbacks to 6s power at 25volts means dealing with massive amp draw at 16volts,
where should i start. what software should i use, and how do i determine the trace size to pass the amps i need.
-
#1 Reply
Posted by
_Andrew_
on 31 Jan, 2016 10:38
-
-
-
looks usefull. but how do i figure out how much i need. and what software should a beginner use
-
#3 Reply
Posted by
Paul Price
on 31 Jan, 2016 19:19
-
Using low RDSon N-Chan Power MOSFET's and a MCU for PWM to create buck converters, it shouldn't be much harder than feeding pigeons in the park, but like feeding pigeons, every HW/SW software solution will require a little tuning to give everyone its fair share.
-
#4 Reply
Posted by
uncle_bob
on 01 Feb, 2016 01:08
-
Hi
Consider that regardless of the minor details, you are working with loads in the 1 ohm to < 1 ohm region. For a lot of reasons, that's not a real good place to be. They all revolve around resistance (batteries, switches, wires, connectors). Coping with that adds weight without adding benefit. Moving this up to 48V or 100V motors would be a really good idea.....
Bob
-
-
You definitely want the voltage to be as high as practical - high currents = lots of copper = weight.
But not so high that you get into other problems. Definitely not as low as 12V as you'll waste a lot of power.
-
-
taking that into account. 6s may be better. i will need to test some of the motors on a quad to see if its as practical as id like it to be for what im trying. its a bit experimental at the moment.
can you explain trace thickness to amps by any chance.
-
#7 Reply
Posted by
uncle_bob
on 01 Feb, 2016 12:42
-
Hi
Trace thickness and amps relate in terms of how hot you will allow the trace to get. If a trace that is heating up by 100C is acceptable, you get 10X the current through it as you do if only a 10C temperature rise is allowed. The concern is things like fire and smoke.
Trace thickness also depends on the "weight" of the copper used. For historical reasons, copper is often described in ounces rather than a thickness number. Things like one and two ounce copper are fairly normal.
Using a 10C rise and 2 ounce copper, a 300A trace comes out at 4" wide. If it's a foot long, it will dissipate about 7 watts. A trace 1/10 that wide will have a 100C rise. It will dissipate 700 watts while carrying the 300 amps.
Bob
-
-
Of course on a quadcoipter you could make creative use of the downdraft and duct some to cool the insides.
Though remember any heat is wasted battery power.
Might be interesting to calculate the optimum compromise between energy lost due to lifting additional copper weight and power losses with thinner conductors.
-
#9 Reply
Posted by
uncle_bob
on 01 Feb, 2016 23:07
-
Of course on a quadcoipter you could make creative use of the downdraft and duct some to cool the insides.
Though remember any heat is wasted battery power.
Might be interesting to calculate the optimum compromise between energy lost due to lifting additional copper weight and power losses with thinner conductors.
Hi
Since the high current phase is brief, you probably can get away with some loss. The normal problem there is you are looking for maximum lift at that time. If you can get away with less lift, dropping the size of motors is the best thing you can do.
Bob
-
-
That is a good point. Didnt even think about the fact that its pulsing the motor instead of continuous current. Though its still pretty fast.
The thing aboit motor size though. Im building a model designed to go fast. With 30 second burst of full throttle being a common occurrence on a model like this. And my motor size was selected to push the limits of this rig
How do i determine the nesasry trace size. Can anyone point me towards a write up or video please
-
#11 Reply
Posted by
uncle_bob
on 02 Feb, 2016 18:05
-
How do i determine the nesasry trace size. Can anyone point me towards a write up or video please
Hi
Is there a problem with the numbers I posted above?
You probably are better off using wire rather than a PCB trace.
Bob
-
-
Well. I would really like to sell a few of these at events to help fund my hobbies.
That cnc router was expensive and to top it i just really love yo build high quality and share.
And high quality model often include power distribution board. Every ounce of weight matters to people. But mostly it jyst looks clean an nice to not have wires everywhere.
As well. Its kinda like teach me to build fire vs making me a fire and saying goodbye. Im have trouble figuring out trace sizes and its something i need to learn for projects like these. Which i will probably do often.
-
#13 Reply
Posted by
uncle_bob
on 02 Feb, 2016 21:58
-
-
-
Thats the one i keep finding. But it says near the bottom that its only accurate to 35 amps
-
#15 Reply
Posted by
uncle_bob
on 03 Feb, 2016 01:06
-
Hi
The real point is "nobody ever does this once you go over 35 amps". PC boards simply are not ever used for this sort of thing (many hundreds of amps). The calculations still work at higher currents if you punch in the numbers. You probably should multiply them by 1.5 to be safe.
Here's why you don't do it this way:
You need a 2 inch (or 4 inch) wide trace for the current.(one for + and one for -) How do you connect that to your battery or to your motor without necking (narrowing) down the trace? If you narrow it down, you generate a hot spot.
Much easier to use wire or busses of some sort. That's how it's been done every time I've seen anybody do it.
Of course if you need to control the current you are back to the same problem. PC boards have the same issues there when you neck down to a control device.
Bob
-
-
Note: edited heavily to explain some things
Sorry. It just dawned on me that i asked the question wrong. I should have been more clear on my circit. Ive been out of my hobbies to long and im forgetting steps. 240 is the total input. And thats a worse case scenario. i hope to pull less with a more optimised build. Stock power boards normaly support 30 amps per motor on an fr-4 board.. But custom boards are so rare because there unessasary expenses in most cases. However i see tons of custom hobby frames are poping up and i want to stand out by going all out. And this something i like from common models like the qav and arris x-speed frames. And ive seen it on bigger models as well like my 550 flame wheel. Which has potential to pull 30 amps per motor with a high level build. And its a stretch goal as someone told me i could mill my own boards with my new cnc router.
I completely forgot to consider my circit in trace design has some room to play with. And i should have stated more clearly. But no really ever tries to help me on the forums ive been on in the past. So i didnt go into as much detail as i normaly whould have. I hope that clears the air a bit.
Each motor draws max 30amps.
8 motors x 30 =240
But there all split. 240 amps is to much for a single battery cable, so parallel 2 batteries that can combined handle the juice.
If i bring them seperatly in on 2 sides of the board i have to inputs at 120amps. Then split the source 4 ways and only the plug needs to be beefy enough to take 120amps and the traces can take a much healthier 30amps each.
Then bridge them across the middle for a parallel circit so the batteries can still balance with each other during the flight. And i can break out the small bits of draw for flight control and camera gear.
If you think thats plausible. I need to look at a 2 sided pcb to sort out all thoughs traces. But im not sure how to create an 2 sided pcb. I dont know if a cnc router is the only tool i need to pull that off.
Note. If it can handle 240 amps. Nobody will ever over draw this power board. But scaling back a bit can be an option as a 6s build at 25 volts pulls like 136amp @ 17 per motor but has draw backs. But a more standard build would pull 15amps per motor at 4s@16volts.
-
#17 Reply
Posted by
uncle_bob
on 03 Feb, 2016 01:30
-
Hi
Why are you doing this with a PC board rather than a wiring harness? The weight and volume of a harness will be less than the board.
Bob
-
-
I edited that last statement pretty heavy to explain.
-
#19 Reply
Posted by
pmbrunelle
on 03 Feb, 2016 04:57
-
A product I'm working on draws some 50 A, and there's only a short distance to go, and we have copper bus bars in the gadget.
You can't pass this amount of current in a PCB.
On an aircraft, aluminium may be better.
Copper has better conductivity per unit of cross-sectional area, but aluminium has better conductivity for a given cable weight per unit of length.
-
-
-
-
Remeber you can get thicker copper - typically 4 oz (4x standard), probably more to special order,and augment if necessary with multiple layers and solder tinning
-
#22 Reply
Posted by
uncle_bob
on 03 Feb, 2016 17:43
-
Im not trying to be difficult. But it hard to convince me it cant be done when im holding one in my hand. I see this being done all the time. Every major multirotor manufacturer does this. And the 3rd one on this list i know for a fact that catalyst quads clamed he pumped 140amps through it without any problems. And my 6y tri-hexacopter has one half its size that takes 6x20amps=120amp!! I know it can be done.and claims it can handle more. Drones are power hungry.
I want to make my own for my model. Guys i may be insaine, but im not crazy.
http://ep.yimg.com/ay/yhst-90268561309754/power-esc-distribution-board-for-multicopter-3.gif
http://i1290.photobucket.com/albums/b527/gadgetsalon/FPV/PDB_1M_resize_zpsf1b3bfe5.jpg
http://abusemark.com/store/images/afropower_mini.jpg
Hi
The reason it is done that way in a high volume production device is to reduce assembly time. In low volume (one to a hundred) production, the assembly time gain is not as much. You don't have the money for the fixtures and the amount of work to keep several people busy.
If you go the PCB route, be sure to take a close look at vibration and it's impact on the connections to the board in your design. That is generally the biggest issue in terms of reliability. You do not want a vibration node (a lot of flex) right at one of your connections.
Bob
-
-
No Im not trying to mass produce them, but if i can make a few with my own cnc router ill do it just because i think it will be fun. And its a cnc router. The process is automated, and i need the software practice anyways.
All i wanted was to get some tips. But i think ive just decided that ill just cut out a bunch of different sized traces and pump amps through them till they burn and then again till i find something safe.
-
#24 Reply
Posted by
Brumby
on 03 Feb, 2016 23:46
-
You seem to have a vision of what you would like to end up with.
Can you sketch it for us - and can you provide physical dimensions? This might allow people here to give more directed thoughts...