Author Topic: EEVBlog #230 - Arducopter  (Read 11406 times)

0 Members and 1 Guest are viewing this topic.

Offline Thermal Runaway

  • Contributor
  • Posts: 48
EEVBlog #230 - Arducopter
« on: December 29, 2011, 02:59:25 pm »
Regarding Dave's tale of woe with the 5V rail short in EEVBlog #530, I have the following to add:

It is indeed possible to find a short by supplying it with a crap load of current and letting the short blow out.  Often this will result in the physical destruction of a component, but at least you will have found the culprit.  In the trade we call this procedure "tune for maximum smoke".  The dangers are that you can vaporise PCB tracks in the board if you're not careful.
Sometimes you can get away with a safer approach.  If you supply the board with a smaller amount of current, but enough to cause some heat dissipation in the short circuit, you can have a careful feel around for suspiciously warm components.  The warm component is often the culprit.  Careful though, you can burn yourself - as I have done many times.  OUCH!!! B*stard, that's the culprit!!!

At work we see this kind of problem quite regular.  And a short on the supply rail can often be the trickiest to pin down.  Especially when you've got a bunch of BGAs on the board - nothing worse than that.  So I've ordered in an AIM i-prober 520, which is a non-invasive current probe for measuring current flow in PCB tracks.  I haven't had time to try it yet, but check it out:

http://www.aimtti.com/go/iprober/index.htm?gclid=CJihscbAp60CFUIMfAodZHMjoA

I also wanted to take this opportunity to share a similar tale of woe with the forum members.
The other day we had a problem at work with a board that was drawing excess current.  It wasn't a dead short, and curiously the board was otherwise fully functional, but it was drawing about double the expected current.  We didn't have any good boards around us, so we had to troubleshoot it on its own.  We were suspicious of the CPU because it was getting warm, so we lopped it out and then the current consumption problem went away.  Of course, we weren't fully satisfied with this because with the CPU removed the board was basically not doing anything, so it was no guarantee that the CPU was really at fault.  The easiest thing to do was change it and see what went on.  We did that, and the problem was still there!!!
We couldn't tune for maximum smoke because we weren't dealing with a dead short, so we began removing components from the supply rail one by one.  Eventually we got all the way down to pretty much nothing on the board.  Only the CPU was left.

What the hell is going on? Well, we powered the CPU up off the board and what do you know, it was drawing excess current! Tried another 10 components from stock, and they were all doing the same.  We only had one batch code of component in stock, but we went through them all and only found 2 that were good.  The two that were good had the same batch code printed on them, but I noticed that the packaging was slightly different! The good ones had a pin 1 indent, and the bad ones did not.  How can this be, if they were the same batch?

It turned out that purchasing had got them from the grey market.  Bar-stewards!!! They were almost certainly counterfeit.  I lopped the top off one of them and nothing looked untoward, but nothing else explains the problem to me.  I still don't know how the boards were otherwise fully functional though.  Bit of a weird one.  But the problem was definitely the CPU in the end.

Brian H.
« Last Edit: December 29, 2011, 03:01:18 pm by Thermal Runaway »
--------------------
Electronics Engineer, Land Rover enthusiast, Amiga Computer fan and general GEEK
--------------------
 

Offline The_Penguin

  • Regular Contributor
  • *
  • Posts: 103
  • Country: ca
Re: EEVBlog #230 - Arducopter
« Reply #1 on: December 29, 2011, 03:18:32 pm »

It is indeed possible to find a short by supplying it with a crap load of current and letting the short blow out.  Often this will result in the physical destruction of a component, but at least you will have found the culprit.  In the trade we call this procedure "tune for maximum smoke".  The dangers are that you can vaporise PCB tracks in the board if you're not careful.

I've done it, not always pretty. Did find a shorted cap that I might have spent hours trying to find. "quick, where'd that puff of smoke come from"

In Dave's case, because the short was a physically bent pin on a connector, I suspect it would have some some serious damage, probably melting a trace or worse.
 

Offline joelby

  • Frequent Contributor
  • **
  • Posts: 634
Re: EEVBlog #230 - Arducopter
« Reply #2 on: December 29, 2011, 03:21:49 pm »
It turned out that purchasing had got them from the grey market.  Bar-stewards!!! They were almost certainly counterfeit.  I lopped the top off one of them and nothing looked untoward, but nothing else explains the problem to me.  I still don't know how the boards were otherwise fully functional though.  Bit of a weird one.  But the problem was definitely the CPU in the end.

Maybe the grey market parts were a previous revision of the silicon that has different specs?

One of the chip teardown articles I read turned up an unreleased engineering sample of some unrelated chip, much to the surprise of someone who'd had a hand in designing the chip. Presumably some fab ended up with a bucket load of these scrap parts from an earlier spin, and rebadged and sold them as something more desirable.
 

Offline metalphreak

  • Frequent Contributor
  • **
  • Posts: 815
  • Country: au
  • http://d.av.id.au
    • D.av.id.AU
Re: EEVBlog #230 - Arducopter
« Reply #3 on: December 29, 2011, 03:24:41 pm »
What sort of motors is Dave using on his copter? They look like Turnigy ones from hobbyking and the radio system is from there :) There's so many choices for motors that I'd have no idea what to choose. I've heard 1000rpm/v is a good spec to aim for with the appropriate prop size?

Offline Rufus

  • Super Contributor
  • ***
  • Posts: 2094
Re: EEVBlog #230 - Arducopter
« Reply #4 on: December 29, 2011, 03:45:12 pm »
If you can pass some limited current through the short without causing further damage it is easier to probe around the board looking at voltage drops than trying to measure resistances.

Variable probe contact resistance means you are never really sure of what you are seeing. I have a ToneOhm with sharp kelvin probes and usually struggle to make sense of what it tells me despite covering the board with stab holes.


 

 

Offline McMonster

  • Frequent Contributor
  • **
  • Posts: 417
  • Country: pl
    • McMonster's blog
Re: EEVBlog #230 - Arducopter
« Reply #5 on: December 29, 2011, 03:48:29 pm »
I have to test this, I got a perfect test case sitting on my bench, but I already narrowed the problem down to some of the four electrolytics on the motherboard of my function gen. Googles and other safety gear on and lets see which one will go off.

It's time to buy a camera to document my electronics experiments.
 

Offline Rutger

  • Regular Contributor
  • *
  • Posts: 201
  • Country: us
Re: EEVBlog #230 - Arducopter
« Reply #6 on: December 29, 2011, 04:07:35 pm »
Hi Dave, I would suggest you paint the front of the copter red or something really obvious that you can see from a distance. That way you can fly the Helli much better, otherwise you have no way of orienting it correctly and you will crash...

Also to make it more stable have you incorporated any giros in the design?

Rutger
 

Offline steff

  • Contributor
  • Posts: 30
Re: EEVBlog #230 - Arducopter
« Reply #7 on: December 29, 2011, 04:17:51 pm »
The IMU board will have (at least) accelerometers and gyros on board, combined with something to combine the readings into attitude and movement. You can even get gyros and accelerometers in a single component now: http://invensense.com/mems/gyro/mpu6000.html
 

Offline Thermal Runaway

  • Contributor
  • Posts: 48
Re: EEVBlog #230 - Arducopter
« Reply #8 on: December 29, 2011, 06:42:02 pm »
In Dave's case, because the short was a physically bent pin on a connector, I suspect it would have some some serious damage, probably melting a trace or worse.

Yes, Dave broke the age old rule that you should always inspect before you diag.  Everyone does that at some time or other, and it doesn't seem to matter how experienced you are either.  Engineers have a tendency to always think of the problem as being complex, when sometimes it's really simple.  You have to be very disciplined to make yourself always fully inspect a board first, before you do anything else!

Brian H.
--------------------
Electronics Engineer, Land Rover enthusiast, Amiga Computer fan and general GEEK
--------------------
 

Offline Thermal Runaway

  • Contributor
  • Posts: 48
Re: EEVBlog #230 - Arducopter
« Reply #9 on: December 29, 2011, 06:45:14 pm »
Maybe the grey market parts were a previous revision of the silicon that has different specs?

Well, I don't think so because these things really were drawing excess current.  Instead of 10mA, they were drawing 300mA!!! But otherwise they seemed to be functional.  It's possible that there was a real fault that our application didn't see because we weren't using the faulty feature or something like that.

But I definitely think they were re-packaged on the grey market.

Brian H.
--------------------
Electronics Engineer, Land Rover enthusiast, Amiga Computer fan and general GEEK
--------------------
 

Offline Zad

  • Super Contributor
  • ***
  • Posts: 1013
  • Country: gb
    • Digital Wizardry, Analogue Alchemy, Software Sorcery
Re: EEVBlog #230 - Arducopter
« Reply #10 on: December 29, 2011, 07:52:51 pm »
It reminded me a lot of my Sanity Clause post! www.eevblog.com/forum/index.php?topic=5985.msg78282

I was ready to jump in and start stripping things down, having made an assumption that the mains supply had to be fine. After all, I had just had it working 2 minutes ago! You make these mental assumptions - you have to otherwise nothing would ever get done - which sends you down the totally wrong route. In Dave's case there was no reason to believe a robust mechanical component that was previously known to be good, would fail in any way. I thought in this instance that it was going to be a small piece of wire under the connector shorting it out, but I have seen connectors fail like that before. It is so easy to try and plug in the wrong connector, or to hit it with something and bend a pin over, and you just don't notice you have done any damage.

One non-destructive method of testing that can be done is to power it via a current limiting power supply. Feed it 50mA (or whatever you feel to be comfortable) and meter the voltage on the board. This is one thing where a good analogue meter is actually better than a digital one because you get a better feel of where the current is going, and in what proportion.

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29257
  • Country: au
    • EEVblog
Re: EEVBlog #230 - Arducopter
« Reply #11 on: December 29, 2011, 09:41:31 pm »
Yes, Dave broke the age old rule that you should always inspect before you diag.  Everyone does that at some time or other, and it doesn't seem to matter how experienced you are either.  Engineers have a tendency to always think of the problem as being complex, when sometimes it's really simple.  You have to be very disciplined to make yourself always fully inspect a board first, before you do anything else!

I did inspect it before I started, I just missed the pins shorted pins inside the connector, probably because my subconscious was shut off to the possibility because:
a) we had never used that connector
b) the pins were protected deep inside the connector (and not easily visible)
c) the board was working just fine before we seemingly plugged in the USB cable again, which pointed toward some sort of electrical fault.

If this was a brand new board out of the box, I very likely would have found the physical fault inside the connector much quicker.
It's a case of being lead up the garden path.

Dave.
 

Offline firewalker

  • Super Contributor
  • ***
  • Posts: 2321
  • Country: gr
Re: EEVBlog #230 - Arducopter
« Reply #12 on: December 29, 2011, 10:09:24 pm »
Dave, are you going to place a wire frame around the "blades"?

Alexander.
Become a realist, stay a dreamer.

 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29257
  • Country: au
    • EEVblog
Re: EEVBlog #230 - Arducopter
« Reply #13 on: December 29, 2011, 10:23:49 pm »
Dave, are you going to place a wire frame around the "blades"?

We will have something to protect the blades, yes.

Dave.
 

Offline firewalker

  • Super Contributor
  • ***
  • Posts: 2321
  • Country: gr
Re: EEVBlog #230 - Arducopter
« Reply #14 on: December 29, 2011, 10:35:45 pm »
More likely to protect fingers e.t.c.  :P :P :P

Actually what damage such a motor/blade can cause to a leg, knee, thigh, finger e.t.c.?

Alexander.
Become a realist, stay a dreamer.

 

Offline Thermal Runaway

  • Contributor
  • Posts: 48
Re: EEVBlog #230 - Arducopter
« Reply #15 on: December 29, 2011, 11:20:37 pm »

I did inspect it before I started, I just missed the pins shorted pins inside the connector, probably because my subconscious was shut off to the possibility because:
a) we had never used that connector


Yep, that makes sense.

How much of troubleshooting involves being lead up the garden path? Quite a lot of it, I reckon.  And there's probably nothing that annoys you more than when you spend some time trying to locate a fault and then find something silly like that.  Easy to miss for sure, I'm not saying you should necessarily have spotted it.  But I bet you still kicked yourself when you found it  :o
--------------------
Electronics Engineer, Land Rover enthusiast, Amiga Computer fan and general GEEK
--------------------
 

Offline Teknotronix

  • Regular Contributor
  • *
  • Posts: 145
  • Country: au
Re: EEVBlog #230 - Arducopter
« Reply #16 on: December 29, 2011, 11:40:46 pm »
Dave, are you going to place a wire frame around the "blades"?

We will have something to protect the blades, yes.

Dave.

Protection is nice, but it will also increase drag.

In the video it was pulling to one side and looked like it needed an excessive amount of trim to hover level. Have the blades been balanced? You can pick up a blade balancer very cheap at Hobby King, def worth it.
Don't drone me bro!

 

Offline Lightages

  • Supporter
  • ****
  • Posts: 4295
  • Country: ca
  • Canadian po
Re: EEVBlog #230 - Arducopter
« Reply #17 on: December 29, 2011, 11:51:19 pm »
Why not shroud the blades and gain lift and efficiency as well? This would then give better battery life too. Sure it will ad some drag to lateral motion but these things are hardly what one could consider aerodynamic in the the first place.
 

Offline Teknotronix

  • Regular Contributor
  • *
  • Posts: 145
  • Country: au
Re: EEVBlog #230 - Arducopter
« Reply #18 on: December 30, 2011, 12:00:34 am »
Why not shroud the blades and gain lift and efficiency as well? This would then give better battery life too. Sure it will ad some drag to lateral motion but these things are hardly what one could consider aerodynamic in the the first place.

The shrouds will produce some additional turbulence from the blades in addition to the lateral drag, more and more as the design is poor. The fact that they are not already aerodynamic is even more reason to spend time improving it. Quad copters are nearly always travelling at an angle of attack where something like a shroud will always reduce the efficiency. Inefficiency adds to the poor range a lot of these drones have, the more efficient you make them the bigger the potential radius you can operate them at.
« Last Edit: December 30, 2011, 12:06:27 am by SimRacer »
Don't drone me bro!

 

Offline mikeselectricstuff

  • Super Contributor
  • ***
  • Posts: 11940
  • Country: gb
    • Mike's Electric Stuff
Re: EEVBlog #230 - Arducopter
« Reply #19 on: December 30, 2011, 12:20:18 am »

I find that probing with a resistance meter  isn't sensitive enough, and contact resistance too inconsistent - a better way is to set the bench supply to normal supply voltage and a current limit of a few hundred mA into the supply rails, then probe around with the meter on millivolt range, starting at the supply point and moving downstream - voltage decreases as you get nearer, and if it levels off, you took the wrong branch.
Quote
So I've ordered in an AIM i-prober 520, which is a non-invasive current probe for measuring current flow in PCB tracks.  I haven't had time to try it yet, but check it out:
I just got one - will be doing review/teardown as soon as I get time. First impression is that although it works down to DC, earths magnetic field makes it hard to use at high sensitivities, and it's a lot easier if you power the circuit from a pulsed supply (e.g. 50R outut of siggen) so you only see an AC signal.
Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 

Offline sonicj

  • Frequent Contributor
  • **
  • Posts: 752
  • Country: us
  • updata successed!
Re: EEVBlog #230 - Arducopter
« Reply #20 on: December 30, 2011, 01:24:53 am »
Hi Dave, I would suggest you paint the front of the copter red or something really obvious that you can see from a distance. That way you can fly the Helli much better, otherwise you have no way of orienting it correctly and you will crash...
i like the red reflective safety tape, like scotchlite, etc... if you lose your model, catching even the smallest bit of that tape with a flashlight results in brightly returned light. also, i always write my contact information on/in the model in UV resistant indelible ink or even etched on the frame.

dave, did you calibrate while powered solely by usb? or while powered by the on-board BEC?
-sj
 

Offline PeteInTexas

  • Frequent Contributor
  • **
  • Posts: 344
  • Country: us
Re: EEVBlog #230 - Arducopter
« Reply #21 on: December 30, 2011, 09:14:37 pm »
Yes, Dave broke the age old rule that you should always inspect before you diag.  Everyone does that at some time or other, and it doesn't seem to matter how experienced you are either.  Engineers have a tendency to always think of the problem as being complex, when sometimes it's really simple.  You have to be very disciplined to make yourself always fully inspect a board first, before you do anything else!

I did inspect it before I started, I just missed the pins shorted pins inside the connector, probably because my subconscious was shut off to the possibility because:

You sure it was not one of your mates mucking about with the board?  I thought I heard him giggling as he watched you take out one tantalum after another. ;)
 

Offline Thermal Runaway

  • Contributor
  • Posts: 48
Re: EEVBlog #230 - Arducopter
« Reply #22 on: December 30, 2011, 10:37:05 pm »
You sure it was not one of your mates mucking about with the board?  I thought I heard him giggling

I used to work with this guy who took great delight in stuff like that.  He was particularly fond of hiding behind you with a segment of that plastic air-pouch stuff used in packaging, and he would watch as you probed various points of a board with your meter.  He'd wait patiently until you decided to probe something scary like a mains point or something, and then he would pop that damned air pouch and make you jump out of your skin!

Hilarious for him, really annoying for you.  On one occasion I remember that he made me cause a real bang because he popped his air pouch, I jumped, and then in so doing I shorted out a reservoir cap which vaporised my multimeter tip and almost gave me a heart attack.  I was not amused, which seemed to make it all the funnier for him.  Good old Andy.

It's funny, the things we used to get up to.  I don't think you'd get away with that kind of fooling about these days - health and safety will have made it a sackable offence for sure!

I just got one - will be doing review/teardown as soon as I get time. First impression is that although it works down to DC, earths magnetic field makes it hard to use at high sensitivities, and it's a lot easier if you power the circuit from a pulsed supply (e.g. 50R outut of siggen) so you only see an AC signal.


Great, I shall look forward to your review I'll check it out.  I ordered it mainly to help the Techs when they're troubleshooting supply rail short circuits where high current can be flowing, so the problems at high sensitivities shouldn't be too much of an issue for them.  However, your suggestion about using a pulsed power supply makes sense, I'll keep that in mind.  I'm not sure it'd be much use on a supply rail short though because you'd just be shorting the sig gen output down to ground.

Thanks for the tip!

Brian H.
« Last Edit: December 30, 2011, 10:44:31 pm by Thermal Runaway »
--------------------
Electronics Engineer, Land Rover enthusiast, Amiga Computer fan and general GEEK
--------------------
 

alm

  • Guest
Re: EEVBlog #230 - Arducopter
« Reply #23 on: December 31, 2011, 12:31:42 am »
I just got one - will be doing review/teardown as soon as I get time. First impression is that although it works down to DC, earths magnetic field makes it hard to use at high sensitivities, and it's a lot easier if you power the circuit from a pulsed supply (e.g. 50R outut of siggen) so you only see an AC signal.
I'm not surprised, even shielded DC current probes are orientation sensitive at the lowest (down to mA) range.

Note that if you're using AC, the old HP 547A current tracer is also an option. It was designed to find shorted logic gates in combination with the HP 546A logic pulser. It only gives a quantitative measure for current and has one fixed range, but this may be enough for finding shorts. On the used market (it's discontinued) it's a lot cheaper than the TTI probe. Not sure about the current/frequency range, I believe there's a manual available on the net.
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29257
  • Country: au
    • EEVblog
Re: EEVBlog #230 - Arducopter
« Reply #24 on: December 31, 2011, 07:14:03 am »
Quote
So I've ordered in an AIM i-prober 520, which is a non-invasive current probe for measuring current flow in PCB tracks.  I haven't had time to try it yet, but check it out:
I just got one - will be doing review/teardown as soon as I get time. First impression is that although it works down to DC, earths magnetic field makes it hard to use at high sensitivities, and it's a lot easier if you power the circuit from a pulsed supply (e.g. 50R outut of siggen) so you only see an AC signal.

I'm getting one of those in Jan sometime too.

Dave.
 

Offline Psi

  • Super Contributor
  • ***
  • Posts: 7128
  • Country: nz
Re: EEVBlog #230 - Arducopter
« Reply #25 on: December 31, 2011, 07:16:32 am »
More likely to protect fingers e.t.c.  :P :P :P

Actually what damage such a motor/blade can cause to a leg, knee, thigh, finger e.t.c.?

Alexander.

Friend of mine had his finger sliced down the middle from end to end and his fingernail torn off by an RC prop this Christmas.
Although, he was using a prop which had sharp edges (for cutting through the air), so it wasn't a typical prop.

He was holding the plane and testing the engine when he moved his hand and..... bang..... blood everywhere and Christmas night in hospital having surgery.

So yes, they can do ultra damage depending on blade sharpness and speed.
Also the angle at which they hit you various the seriousness as well.

« Last Edit: December 31, 2011, 07:23:37 am by Psi »
Greek letter 'Psi' (not Pounds per Square Inch)
 

Offline IanJ

  • Supporter
  • ****
  • Posts: 976
  • Country: scotland
  • Pro EE guy many years ago, now a hobby/home biz.
    • IanJohnston.com
Re: EEVBlog #230 - Arducopter
« Reply #26 on: December 31, 2011, 04:20:51 pm »
I did inspect it before I started, I just missed the pins shorted pins inside the connector

There's something wrong if any electronics engineer hasn't been there........so you are in good company Dave.

Ian.
Ian Johnston
www.ianjohnston.com
Manufacturer of the PDVS2
 

Offline Kilroy

  • Regular Contributor
  • *
  • Posts: 85
  • Country: ca
Re: EEVBlog #230 - Arducopter
« Reply #27 on: January 01, 2012, 03:55:33 am »
Note that if you're using AC, the old HP 547A current tracer is also an option. It was designed to find shorted logic gates in combination with the HP 546A logic pulser. It only gives a quantitative measure for current and has one fixed range, but this may be enough for finding shorts. On the used market (it's discontinued) it's a lot cheaper than the TTI probe. Not sure about the current/frequency range, I believe there's a manual available on the net.

The 547A current tracer is an outstanding tool for locating elusive short faults. Sensitivity is specified at 1ma - 1A but actual performance exceeds this...it has no problem sensing CMOS current less than 1ma. The low noise frequency bandwidth is out to 20Mhz and it can be powered with any convenient DC source between  4.5v - 18v. Performance is particularly good when injecting pulsed current into the circuit using the 546A logic pulser.

These are the best tools I have found so far for fast, accurate short fault detection. HP's peerless attention to engineering detail and quality execution is clearly evident in their construction and performance.
« Last Edit: January 01, 2012, 06:14:47 am by Kilroy »
The fool generalizes the particular; the nerd particularizes the general; some do both; and the wise does neither.
 

Offline Lightages

  • Supporter
  • ****
  • Posts: 4295
  • Country: ca
  • Canadian po
Re: EEVBlog #230 - Arducopter
« Reply #28 on: January 01, 2012, 06:58:15 am »
The shrouds will produce some additional turbulence from the blades in addition to the lateral drag, more and more as the design is poor. The fact that they are not already aerodynamic is even more reason to spend time improving it. Quad copters are nearly always travelling at an angle of attack where something like a shroud will always reduce the efficiency. Inefficiency adds to the poor range a lot of these drones have, the more efficient you make them the bigger the potential radius you can operate them at.

I don't understand how the shrouds add turbulence from the blades. The shrouds should counteract the losses at the blade tips and add thrust. The shrouds also do not need to have cylindrical walls on the outside. They can be aerofoil  or round nose shaped on the outside so as to be more aerodynamic in somewhat horizontal movement.

Please forgive as I have not researched quadcopters, I am just a nerd who also is a pilot.
 

Offline IanB

  • Super Contributor
  • ***
  • Posts: 9489
  • Country: us
Re: EEVBlog #230 - Arducopter
« Reply #29 on: January 01, 2012, 09:13:38 am »
I am not an aerodynamicist either, but from observation it would appear than fans (small cross section, high air velocity) tend to have shrouds, while props and rotors (large cross section, lower air velocity) do not. So I would guess the importance of shrouds to the quad-copter depends on the extent to which the blades are fans rather than rotors...? In general engineering terms, I think open bladed impellers tend to be less efficient than shrouded impellers due to the way that shrouds and casings can channel and direct more of the air in the right direction rather than the wrong direction.
I'm not an EE--what am I doing here?
 

Offline Hobgoblin612

  • Contributor
  • Posts: 47
  • Country: au
    • Youtube Channel
ooooohh... what does this button do???
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf