Alternatives for sensing current?


#1

I managed to pull the current sense data from the cape through to the cockpit display easily enough. But it appears directionally the approach moved from having all power come through the cape to having the motors pull power directly... which means the cape is only showing current draw from the servo and the beaglebone itself. Which is also good because the Vsense on the cape looks limited to measure 5Amps.

There appear to be lots of break out boards that we could put inline between the batteries/tether and everything else. Is anyone already working with a cheap current sensor break out for that purpose?


#2

Brian,

Right you are-- during the design phase, it seemed that 1) there would not be enough real estate on the Cape to have the ESC power go through it, 2) routing wires through the cape would require extra wire harnessing that we didn't want to use, and 3) handling the kind of current the ESCs would draw would require very large traces and hefty components which we could not afford the room for.

Now that the EChassis layout has solidified a bit, I'd say that if we could free up enough room on the Cape, it would be worth revisiting the idea of having the ESC power go through the Cape.

Since we also want to distribute power from the single source too three separate ESCs (a job that currently done by several bulky screw terminals on the back of the EChassis), I've also thought of making an entirely separate "Power Distrabution" board that would have less layers and thicker copper. Such a board could be made cheaply and could also have on-board current sensing for each ESC which would be linked to the cape through something like an I2C connection. I envision that it could be placed either above the set of ESCs or on the back of the EChassis where the wire harness comes in from the endcap. For the above-ESC version, here's a preliminary drawing I made to show the layout.


In this layout, holes are spaced so that the leads coming from the ESCs could go directly up to the board above them and be soldered in place. "PWR IN" (coming directly from the battery pack) would be split and routed to PWR 1, 2, and 3 respectively as well as to the screw terminal which would power the Cape, but these traces could be interrupted by current sensing chips like this one for each channel, which would report their data to the Cape over I2C (after going through an on-board multi-channel ADC). Motor outputs from each ESC (labeled 1A, 1B, 1C; 2A, 2B, ..etc) are routed to the top side of the board where the wires can be grouped with the same spacing as the Molex connector. (Note however, that the current Molex connector used on the ROVs may later be switched out for something else that is less bulky and easier to use)



#3

are the measurements for the pins on the esc in your sketch precise? and just to confirm, the numbers are in millimeters?

if yes, i can _easily_ make a board design for you. i've used the ACS712 earlier on a 1.5kW frecuency converter. so i have the layout ready in my cad software.

for an prototype / first edition i would recommend to put on 1pcs of acs712 on the main power input, so we can measure the amp-draw in total on the system. this is easily adapted to the cape with removing R60, and populating R32 with the same 0-ohm resistor or a jumper.

then its just a matter of doing the right calculation of the output voltage from the acs712 into the ADC3 on the arduino. this is as easy for the software guys as the layout is for me ;-)

for connector replacement, i would use a phoenix contact rail, these come in 3.5 or 3.81 pitch that can take some amps.

something like this: http://uk.farnell.com/phoenix-contact/mc-1-5-10-stf-3-81/plug-screw-flange-3-81mm-10way/dp/1792853

if you have other wishes to try out, send them over, and ill see if its easy to implement.


#4

Thomas,

Sorry for taking so long getting back to you- I must have missed this earlier.

YES! THAT WOULD BE AWESOME!

The dimensions are correct, and yes, all values are in millimeters. Holes should be sized for 18AWG wire, and traces from the power input should be able to handle something like 8 or 10 amps.

Please let me know if there is other information you could use, or if there is anything further we can do to help make this happen.

Thanks, man!

Eric


#5

I'll get on it as soon as possible,this board shouldn't take long to design.

some small clarifications:

20mm width seems a bit narrow, i think 25-30mm would help the design a bit, but would cover the cooling-fins more.

i can do a first edition quite fast equal to your drawing.

and when tested, we can expand with new connector or similar. i have some thoughts about mounting the esc ontop of a board that rests on the base in the electronics structure, and the connector infront. i can do a mockup of this when i've started the design.

I just have some small concerns.. how is the temperature on the ESC's after a while? if we stack this board ontop of the cooling fins, could this lead to overheating? what are your experience?


#6

Thomas,

I'm really excited about this!

You're right- the 20mm seems narrow... I think I had some reason for that dimension, but now I can't recall what that was.... hmmmm.... I agree: maybe we should make it wider.

I don't think the ESC fins get *too* hot, but more experimentation is needed there. I'll get back to you with more information there if I get a chance to test that.

Thanks again!

Eric


#7

No problem, i've pondered myself on how to organise the wires between the esc's. so why not produce a board... we have an expression in Norway (direct translated) "Help for self-help" ;-)

schematics done.(see attachement)

I've reused the esc-caps on the schematics. its easier to solder them to the board, than to keep them connected to the solder points on the esc.

1647-Openrovpowerdistschematicrev0.jpg (290 KB)

#8

Started on layout.(see attachment)

I'm having some problems placing the caps. they are quite high when placed on the board, so i'm thinking of taking them of the board. the caps can be placed really anywhere on the 12V rail.

the board is now 39.5x75mm and i think to just do a copper pour on top and bottom with +power on top, and 0V/GND on bottom. this will give some areas where the copper can be a bit narrow between the motor outputs, so i'll probably make som extra space between each motor section.

The reason for the board beeing 39.5 is because when the motorleads to the connector is soldered, they need to be away from the heatsink,

thats it for tonight, i'll check in tomorrow for comments. do not hesitate to comment!


Last thoughts before bedtime: i think the approach with the esc's ontop of the board is better than the board ontop of the esc's.

1646-openrovpowerdistlayoutprelim.jpg (697 KB)

#9

Thomas,

This is looking pretty good!

I think the board will still need to be on top of the ESCs because 1) the wires coming out of the ESCs go in that direction so they could be routed directly into the board, and 2) the acrylic barriers in the current E-Chassis design would not allow a board to be placed underneath the ESCs.

As for the capacitors, sure! I think we can play around with ways of not having them be so awkward. In the mean time, it's not that much of a problem and it's nice to not need to do additional solder. I guess we'll do it whatever way you come up with!

Thanks again,

Eric


#10

Hey Thomas:

Nice work! The only thing that really catches my eye is that I'm wondering if connector P2 shouldn't be pushed further away from terminal 3C, so that there is no interference if you want to put some kind of terminal block in P2.

Is there a preferred orientation for connector P3? I don't have an ROV in front of me, so I don't know whether this is going to need right-angle pins, and if so whether the right-angle pins should be heading in some preferred direction.

Here are some cool ideas I had. They would just be icing on the cake, but are something to think about.

- How about adding a small voltage divider (say, 2 0402 resistors) between power in and ground, with the tap going to P3. That way we have the option of monitoring the real battery voltage, as opposed to Vbat on the cape, which is downstream of some ferrites, a schottky diode, and a polyfuse.

- If you have room to expand P3 even further, will a 5-place connector fit? That way you could make it symmetrical, so that if the user installed the harness backwards, he wouldn't fry anything. I'm thinking a pinout something like this:

Gnd--Current Monitor--Vcc--Voltage Monitor--Gnd

If you plug this in backwards, the telemetry is reversed but otherwise no harm. Let me know what you think!

-W


#11

Eric:

My thought with the board under the ESC, would be based on removing the acrylic barriers between and infront of the escs.
And mounting the wires in an upsidedown U down trough the board infront of the esc. the connector would then be infront of this,
and the power distribution traces would be in the board under the escs. this would make the board probably equal big, but a better layout.
I will sketch it up.

I will drop the caps for now.

Thank you for great input!


Walt:

P2 connector: my thought is for the wires to come from the right side of the board and into a 3.5mm pitch screw terminal. its as close as this to 3C to prevent the terminal from faulting with the acrylic tube.
but i could probably give it a extra millimeter or two. I'll to a "PADV"(paper assisted design verification) of the placement before before routing the traces.

P3 connector: i havent really placed this yet, and i thought this would mainly be hardwired, but i will look at the possibility to have a right angle header.
I will try with a 5-pin header. that was a really good suggestion with the voltage monitor aswell.
there is "alot" of space in the board as long as you dont build high.

Thank you aswell for great input!


I just came home from overtime at work, and i'll probably be doing some more 12-16hour days this week , so i dont think i will get time to layout that much. but i'll probably be finished by sunday.


#12

Thomas:

Nice catch on the interference between P2 and the acrylic tube. I'm anxiously awaiting delivery of my kit so that I can see how all this is going to fit together.

-W


#13

(Starting a new reply thread to get width...)

Evening #2 of designing Power distribution board for OpenROV:

-Removed Caps

-Added voltage divider equal to the one on the cape, but directly at power input.

-Changed P3 3-pin to 5-pin 2.54mm pitch

-P3 placed in middle of board to be able to use right angle connector.

-PADV performed - 39.5mm height is to much, the board will be to tight to place. changed to 38mm (see attachments)

-tried to place 1206 resistors and caps far apart so its easy to solder.

-tuned a bit on the silkscreen print - more to go

Schematics:


Layout:


Todo:

-Tracing

-copper fill (PWR+ on top, GND on bottom)

-Silkscreen description on all connectors (P2/P3)

-Place designators for R* and C*

-public review

1642-PADV12013031319.45.30.jpg (407 KB) 1643-PADV22013031319.45.19.jpg (396 KB)

#14

I got some extra time to work on the design tonight, so the first layout is now finished for review.

i've also done a IDF 3.0 export that should be able to import into 3D cad software. i know for sure that solidworks "eats" this file.

please review and comment:

Component layout and silkscreen:


Copper top Layer:


Copper Bottom Layer:



#15

Hey Thomas:

This is looking really good!

So I took a close look at the schematic, and there's an issue with the voltage divider that's troubling me, and I'd like to get your opinion on it.

So in the first schematic you released (with no voltage monitor), the current monitor was electrically isolated from the battery buss. The current sensor gets its power from P3, presumably good, clean power.

Now, with the voltage divider, the voltage has to be referenced against something, so the P3 ground is, by necessity, tied to the battery buss ground.

Are we setting ourselves up for some kind of ground loop issue here? Perhaps with stray currents flowing through the ground wire of the P3 wiring harness?

I still like the idea of having a voltage divider that measures true battery voltage. But I'm wondering if it would be forward-thinking to isolate the P3 ground from the battery-buss ground with a 0-ohm resistor, so that if there is some sort of ground loop issue, we can cut that lead, or install a ferrite bead, or what have you.

I'm trying to think of an easy way to keep the two grounds separate, but still measure battery voltage. Short of bringing two voltage pins back to the A/D for differential measurement, I haven't come up with anything good.

Your thoughts?

-Walt


#16

HI Thomas,
It looks very good.
Is it possible to manufacture a round board?
To replace this part of this e-chasis with the board.



#17

Walt,

Sorry for the late feedback, i have practically been living at work since yesterday morning.

You are very right about the ground connections on P3.i think we can get some foreseen issues with this, probably a jumpy current reading and other adc measurements.

If i remember correctly, the ADC should normally be smack down in the middle of a star connected system like we get with having.the current and voltage reading on the power board.

this i quite right within the cape, but when we get wires up to the esc's and the power board. we are putting unscreened wires directly ontop of HF emitting devices.

How about mounting 0-ohm resistor on both the ground connections on P2, and remove the link between, then the ground is separable on both. we could then also see the difference on current reading via P2 ground, or P3 ground down to the cape.

with this connections just being "jumpers" i'll use 0805 resistor layout, these are solderable without a microscope or really good eyes. and you can easily make a solder jump between.

I can easily implement these resistors.

i've also seen that the clearance between the motor connections and the battery voltage was quite small(0.5mm) this is extended to 1.0mm)


#18

Holger,

Thank you!

yes, its possible, i've drawn something this before, so it shouldnt be a problem.

Normally does circuit boards come in thickness of 1.6mm(the most common), 2.4mm and 3.2mm

i saw your thread here in the electronics section. i would be cool to make the e-chassis in FR4(the pcb material) its quite strong, i'm not an mechanical engineer but based on tensile strenght the FR4 seems to be 5times as strong as extruded acrylic.

we could possibly get some more room aswell.chinese production companies like iteadstudio.com is perfect to produce cards like these,


#19

Latest update:

Schematics:


Layout - silkscreen


copper top:


Copper bottom:



#20

Thomas-- this looks SO good! I'm really excited about how clean and capable this will make everything.

Another thought I just had and wanted to mention before I forget: since the trim on the motors may be off, it's possible for the motors to end up moving even when the "stop" command is sent (which is a safety issue). Do you think it would be possible in later versions of the distro board for there to be a high power FET that could cut power to the ESCs all together to assure no motors move if an "stop" command is sent?