Replies to This Discussion

Permalink Reply by Rick Eis on May 22, 2012 at 10:06pm

R_Lefebvre,

Hmm...interesting. I have so much to learn.....:)  

I was with you till your number 5 and 6 :)

I was trying to figure out how your ESC was "COMPLETELY electrically disconnected from the rest of the system" which gave me the "opportunity" to learn a little about Opto-Isolated ESC's.

So, if I am understanding things correctly, though you plug your ESC into your receiver, no power is transferred to the APM through the jumpers from your receiver to the APM because there is actually no electrical contact with the jumpers and your ESC because your ESC is Optically Isolated....am I getting it right?

Rick

.

• ▶ Reply

DeveloperPermalink Reply by R_Lefebvre on May 23, 2012 at 9:44am

Yes, that's right.  So in this case, I did not disconnect the power wire from the servo connector of the ESC.  That is because that wire is intended to send power TO the ESC, not the other way.

Basically, the Opto-isolated ESC has two sides to it.  The two sides are electrically isolated.  The output side is powered by the motor battery.  The input side is powered by the Rx power supply, and all that it does is to convert the PWM from the Rx into an optical signal which is what jumps the gap to the output side.  I don't think even the ground passes through.

You'll notice that Opto-isolated ESC's don't have internal BEC's, since obviously that would bridge the gap which defeats the purpose.  

Technically, you could power your Rx power supply from the main flight battery, but I think this partially defeats the purpose of the Opto-isolation.  That is why I have two batteries on my 600, the main one for the motor, and the smaller battery for the Rx/APM/Servos.  There is ZERO electrical noise from the motor with this system.  Also, it allows me to change the motor battery, without removing power from the APM, so I can retain GPS lock, MP connectivity, etc while changing the motor battery.  This works out well because the electronics battery lasts upwards of an hour, while the flight batteries last about 12 minutes typically.  I can go through several sets of motor batteries without having to reboot the APM, reconnect MP, or wait for GPS lock.

Also with this system, I have a 7A switching BEC to power the servos at 6V, and it's obviously seperate from the 7805 powering the APM.  Obviously the advantage here is no servo power noise ends up on the APM (or it's greatly reduced), but the other big benefit is that if there is a momentary power spike from a servo, it can cause an APM brown-out.  This has absolutely taken out other people, and again, why I recommend the APM have it's own supply.

On my little 450 heli, I'm not sure I'll do this, since it's supposed to be a simpler system.  But what I am doing is powering a single BEC through the balance port on the main battery.  This again, allows me to boot the APM safely without powering up the motor.  When I'm ready to fly, heli on the launch pad, I then connect the power to the ESC.  

The ESC I'm using in this case is not Opto-isolated, so I did disable the power pin on the servo plug.

I don't have the benefit of zero electrical noise, nor brown-out resistance, but it's a trade-off for simplicity.  

As for 7805 vs. BEC.  As I said, the 7805 isn't the be-all end-all, but it's one option.

I originally purchased a CC BEC from the "LHS" which is about an hour away.  Got home, and realized that it outputs 6V by default, and requires another damn piece of electronics just to change it!  Why they didn't simply use a jumper to select voltage like the Chinese, I'm not sure, but not wanting another 2 hour round trip to the LHS, I simply decided to use the 7805 I had in a drawer.  I haven't looked back.

It's not hard to construct a neat little package for it, basically some wires sticking out of a shrink-wrap bundle, which is basically what all the "professional" BEC's are also.  So no disadvantage there.

Also, I'd seen enough comments about CC BEC's and ESC's failing to know that the idea that spending $50 on a BEC guaranteed reliability was false.  

• ▶ Reply

Permalink Reply by Rick Eis on May 23, 2012 at 10:01am

Well, yes I am learning to be very careful indeed.

Being married with 5 kids, single income, and helping my wife go back to school,  

I do not have the $ to be experimenting with $200 here and there, thus this thread :)

I used to do a bunch of cad stuff, maybe I will pull the program back out and draw some of this up just to be sure I understand things....

Maybe others can show some drawings of their ideas to facilitate this learning ....yes?

Rick

• ▶ Reply

Permalink Reply by Ron Curry on May 23, 2012 at 9:47am

Rick,

The opto isolation refers to the ESC signal lead - the white lead. If you remove, tie back, and insulate the read lead then you will disconnect the ESC BEC from the rest of the system.

Robert's configuration is fine if you feel you must run your servos at higher than 5 volts but in my experience it's not necessary. As well, even in a traditional heli (bicopter) with big fast servos (overkill for an APM application in my opinion) it's not going to draw much more than 5-6 amps amps including the servos. So there is no reason not to run APM, servos, and RX (and even telemetry) all from the same 5v supply.

I'm always a fan of KISS (keep it simple sir) and the less complex the better. There are plenty of NASA whitepapers showing that the more components there are the greater the odds are that one of them will fail. So why not just put one good quality BEC in the system and run everything off of that? The Castle BEC is good for 10 amps. It's small, reliable, and simple.

If you are doing FPV you might need a different voltage or more current to power a camera and video transmitter. In that case you could add nother CC BEC that's been programmed for the voltage you need. I think it goes up to 8 volts as I recall and the CC BEC PRO goes up beyond 12 volts (assuming you have greater than 3c batteries).

Yes, a CC BEC is more expensive than a 7805 and some caps but the APM alone cost $200 not to mention the rest of the heli. Do you really want to add possible failure nodes to save $15? - just a thought. ;-)

Anyhow, lots of good advice here but lowering complexity should be considered as well.

• ▶ Reply

DeveloperPermalink Reply by R_Lefebvre on May 23, 2012 at 10:12am

Well, the optical coupling does occur on the signal lead.  But from what I have seen, the Opto-ESC designs incorporate complete separation of all the electrical wires.  Not even ground passes through, and all the opto-ESC's I've looked at don't have BEC, so the power side is isolated.

I also like simplicity where it is warranted, but there is a problem with what you propose.  The APM should be supplied with a nice clean 5V, you can go up to 5.4V, but that can lead to clipping of some of the Analog signals.  Really, it should be 5V, exactly.

However, anything below 4.8V can cause a reboot of the APM. That's very very bad.  Unrecoverable with a copter of any sort.  It's very possible for the servos to momentarily drop the BEC voltage below 4.8V.  It has happened to people, Randy in particular, and that was with an expensive BEC on a little 450 with tiny servos.  I think he's now running a CC BEC Pro rated at 20A on the little 450 to solve the problem.

I think my approach is better.

If you don't like the 7805, there are plenty of other 5V power supplies available.  Even 7805 replacements that are actually switching regulators inside and don't require any external caps, if you want.

I just feel that in anything with servos, there's a good case for the APM to be on it's own supply.  Yes, it's extra complexity, but then, if we are afraid of complexity, why are we using the APM?

At the end of the day, there are many more things that are more likely to bring down a copter than a 7805 failing. One good example, is all the un-retained servo connectors we use for signals, and power supplies.  No matter what power supply you use, it's probably 10x more likely that the power supply connection vibrates off the APM than it is the power supply fails.

For me, the possibility of of the 7805 failing is very far down a long list of things that can go wrong.

• ▶ Reply

Permalink Reply by Ron Curry on May 23, 2012 at 1:31pm

It would be useful to have some clarification on your voltage data. I'm wondering about getting clipping on some analog signals if 5v goes to 5.4v. Why would there be clipping on some analog signals. Last I looked at the data sheet and schematics we were using the internally regulated analog reference which is not affected by VCC.

As well, all the 5v parts on the board have a spec of a safe maximum operating voltage of 5.5v or greater and all the 3.3v parts are on an on-board regulator so it seems it would be perfectly safe to run up to 5.5v.The 2560 at 16mhz is spec'd to run correctly up to 5.5v but can run up to 6v without damage.

On the reset issue, if it really is 4.8v and you can't run more than 5.0v then we have a big problem because that kind of variance can happen due to heat, connections needing to re-seated etc. Respectfully, I don't believe that data. Particularly because the 2560 at 16mhz is spec'd to run correctly down to 4.5v with margin and the RESET pin requires a maximum of 0.9 x VCC (about .45 volts) to initiate a reset. So, the only thing that could cause a reset is to go below 4.5v (probably wouldn't happen until 4.2v). This is according to Atmel's ATMEGA2560 datasheet.

And... the only way to get the 5v down that far is from either a deficient power supply (not enough current capability) or wiring. I was a beta site with Skookum on the SK720 and we were having brown-out problems on some configurations so we instrumented several large TREX700's with both JR 8717 servos (big, fast servos), and Futaba BLS253's and found that with with the low amperage BEC's internal to the ESC we could get brownouts down below 4.0v where the servos hit spiks of over 6 ams. However, with proper wiring and a 10amp CC BEC voltages never dropped below 5.0v for a 5.2 volt system.

I've designed a prototype 32bit APM and have studied every data sheet for every component in the APM2.0 and I can confidently say that the safe operating range is 4.5v to 5.5v and that we should be running these APM's at 5.2v's to ensure sufficient brownout margin. If there is credible data that says otherwise I'd love to see it and will change my position accordingly for now, I'm listening to the people who make the chips since that is always where our powersupply data is derived from

I think my approach is better.

Well, I wouldn't have expected otherwise. I don't dislike your approach - no emotion here. It's just that all data says it's not the safest, most efficient, way and very subject to one's soldering and building skills. For my flying machines that always seem to cost $1000plus and most are $2000plus I tend to go with the data - but experimentation is good and if you like it go for it so long as you are comfortable and informed.

At the end of the day, there are many more things that are more likely to bring down a copter than a 7805 failing.

Yes, very true but a big part of our task as hobbyist/builders is to eliminate as many obvious failure nodes as we can within budget and knowledge/experience limitations. Talk with any commercial manufacturer of electronics for RC hobbyists and they will all tell you that poor power supplies and wiring are by far the number one issue they deal with in customer support everyday. Even the best electronics are junk without a sufficient power supply.

For me, the possibility of of the 7805 failing is very far down a long list of things that can go wrong.

Actually, these little linear regulators have been the bane of the electronics industry since the '70's and are generally known to be a common failure point but actually, I wasn't referring to that as the possible issue. More likely a poor solder joint or insufficient mounting integrity or insulation would be the failure point.

Anyway, nice to see you and others experimenting and if it gives you pleasure to do so that's cool. In the spirit of information sharing I think it's good to offer alternate approaches and information so we can all make informed decisions on our projects. I also think it's especially important to know where data is coming from and what's driving it so I'd really like to understand where your voltage info came from because it's very different from the manufacturer data sheets.

BTW, I think we should give you an award for THE most active guy on DIYDrones! You are everywhere! I just don't know where you find the time.

Cheers,

Ron

DeveloperPermalink Reply by R_Lefebvre on May 24, 2012 at 7:34am

I can't speak to authority on the 4.8V limit, it was Randy and a few others who experienced it.  I think it has to do with the Atmega328p on the APM1.  I don't have the details, but I am passing on the warning.

As for the AD clipping issue, I believe the issue with some of the analog sensors using Full Scale = Vcc, but when the Atmega does the A/D conversion, is assumes Full Scale = 5.0V.  So if you supply the board with over 5.0V, the full scale does not match up and you have an error.  I think this is particularly true with the sonar.

Permalink Reply by James Y. on May 22, 2012 at 4:43am

Hold on a moment.

When you say the ESC produces 6 volts, are you measuring this open with a voltmeter?  The ESC regulation circuit may be a switching type, and you can't rely on the voltage output being sane on them when you're not drawing a load through it.  They float horribly when they aren't under load.

Hook up a 50 ohm resistor across the output of the BEC, then measure the voltage across that.  Keep in mind that the resistor will dissipate half a watt, so pick an appropriate sized resistor.

• ▶ Reply

Permalink Reply by Wesley Chuen on May 22, 2012 at 7:20am

Servos for the larger copters need 6 volts.  ESC that came with Align's also output at 6 volts.  My 450 helis did.  I guess his ESC is also in the same category.  so he was mentioned the ESC produces 6 volts.

Besides, all the semi-decent ESC/BEC allow you to program exactly how much voltiage output from the BEC.  Very accurate as well.

I measured my APM2 had a 0.3V drop across the output and input rail.  So I programmed mine to output at 5.3V and powered it from the output rail before.

• ▶ Reply

Permalink Reply by Ron Curry on May 23, 2012 at 2:09pm

Servos for the larger copters need 6 volts.

Not absolutely true. A lot of servos have a 6 volt spec because they put out a little more torque and are a little faster vs. 4.8 or 5.2v. However, there are lots of full size big, powerful servos from JR and Futaba that are spec'd at 4.8 to 5.2v. The JR8717 is one and it's arguably one of the fastest and most powerful. Most Futaba brushless servos for large aircraft are 5v however in the last year both JR and Futaba have come out with "HV" servos which allow you to directly connect the servo power leads to a 2C (7.2v) battery bypassing the BEC completely. Makes a lot of sense since the servos don't need perfectly regulated power. If I was really concerned about having enough power for the servos and keep the electronics power really clean I'd run a seperate battery for the servos and run everything else from a regulator or BEC. Actually, that's what a lot of the pro 3D guys do.

On APM2 VCC, all the chips and stuff that connects to the APM are rated 5.5v and above and the only place I've seen the APM2 power spec is one notation in the Wiki that says "if you run more than 5volts your burn your APM up" or something to that effect. I've been suspect of that for a long time since there is no data to support it whatsoever. All data says the spec should be 4.5v to 5.5v +/-0% or 5v +/-10%. I think a better spec would be 5.2v. I'm going to drop a not to Jordi and see what he has to say about it.

• ▶ Reply

Permalink Reply by Chad Frazer on May 22, 2012 at 10:43pm

Clever way to get 3 power sources, I would not of thought to only hook up the ground and the signal wire from my ESC and leave out the RED 5V wire from the connection.  TYVM, for redundancy in aircraft is a beautiful thing.

• ▶ Reply

Permalink Reply by Brad Hughey on May 23, 2012 at 8:26am

I'll just be another voice in the 7805 chorus, with the caveat that I'm a capacitor snob (from my old audiophile days) and hyper about stable supply rails. 

Most application notes suggest something tiny at the output, but I suggest a 4.7 uF tantalum cap vs. a cheap electrolytic.  If you choose the latter, go 47 uF.  A similar or smaller cap at the input is recommended to protect against input noise.  A 47 pF ceramic in parallel with both seals the deal for noise suppression.

• ▶ Reply

• ‹ Previous
• 1
• 2
• 3
• 4
• 5
• …
• 8
• Next ›
• Page