APM Copter V3.1: Autotune with reflashed ESC... warning!

[Marco Robustini]

Yesterday I run Autotune on two of my heavy hexa, installing powerful motors with large carbon propellers 15x5, F40A ESC flashed with SimonK firmware (the latest version), working with 6s LiPo.
During the automatic tuning of the "pitch" two motors have completely lost sync and miraculously i recover the multicopter and land without damage.
Thinking it was a case I repeated the procedure with another multicopter, always very heavy (same propellers) and the problem occurred again (always during variations on the pitch), this time only one motor lost sync during one of the fast throttle inputs that Autotune produces, the quad is tilted diagonal, not to the right side.
This was always done during the step on the pitch because the multicopter is much more reactive on that side for obvious reasons of weight (battery, landing gear, etc.).
I try to find the logs and post them here as soon as possible, but the problem in my opinion is that Autotune produces too fast throttle change input during the procedure, from the "hovering" (quad in a horizontal position) to the max location where it's tilted, these too rapid accelerations, that only the code can produce but our thumbs would be almost impossible, causing possible out of sync with the SimonK firmware.
Now, I don't know how Leonard can solve the problem because this thing could never happen even on some setups, but I did it three times and honestly do not intend to try again because I risked crashing quad very large and expensive.
On DIY Drones I've read of someone who has had the same problem, and had not realized what had happened, the probable cause is one that I wrote, as long as the code of Autotune there is something wrong (do not think).
The problem in my opinion is certainly not attributable only to Autotune (i hope), but of course this procedure can cause crashes with fast ESC reflashed with SimonK or BlHeli firmware for out of sync, our have been warned.

Bests, Marco

[Marco Robustini]

For those who do not know what it means to lose the sync between the motor and ESC this video shows it pretty well:

https://www.youtube.com/watch?v=71eZ-ZNFeSU

With the latest firmware the complete motor shutdown is quite unlikely, but the loss of power as in the video makes it impossible to keep the quad in flight attitude, as happened to me.
Imho if Autotune can be adjusted to make sure that this fast input variations of throttle are softer perhaps the problem would disappear altogether, always however that this does not produce tuning values ​​tuning not reliable.

Cheers, Marco

[Brendan O'Reilly]

I had the same thing with some 690KV 22-pole "pancake" motors and 14" props.  I shelved these a few months ago as the ESCs were occasionally loosing sync.  When SimonK claimed that his latest release should work with any motor that can run I dragged them out again, updated the ESC firmware and gave it a try.  

Initially I did a ballpark manual tune and then tried autotune with the same results as you - one motor lost sync on about the 3rd roll cycle - but recovered before it hit the ground.  I put a bit more effort into my manual tune and the tried Autotune again - and it succeeded with no issues.  I followed that Autotune with another (just testing) in perfectly calm conditions and that also worked with no loss of sync.

This quad flew fine in stabilize for several batteries (about 45 minutes all up) but then a brief RC outage triggered it's first RTL.  The quad freaked out and wobbled its way to the ground in a big crumpled head.  Randy has triaged this and believes it was mechanical failure - perhaps one of the motors lost sync in that episode too - but hard to tell.  Logs and video of the crash can be found here if anyone's interested.

[Marco Robustini]

Here the log with the problem during "pitch step".

Marco

On Sunday, January 12, 2014 9:11:53 AM UTC+1, Marco Robustini wrote:

[john...@gmail.com]

This is not a APM problem, but a known problem with aggressive SimonK in combinations with large/heavy propellers on certain low kv pancake motors. The problem is that the high pole count pancake motors combined with large propeller leads to problems with sync when you do fast and large throttle changes. Some ESC/motor compinations handle this better then other.

Only two sure fixes.
- Use a smaller propeller or lover voltage
- Revert to the original ESC firmware

[Holger Steinhaus]

Hi Marco,

I am into copters with long flight times, 15 inch is the smallest prop in my box ;-). A very simple fix might be the "SLOW_THROTTLE"  macro in the SimonK sources, followed by careful ground testing using a 3-stage flip switch on throttle channel set to -100%, -60%, +100%. This macro limits the throttle to double of its previous value, which is still quite fast. If that is not enough to stop the trouble, you are in more serious problems that could easily reoccur in normal flight during a wind gust without any auto-tuning. I had several crashes and near-crashes due to this issue so far with certain low-kv motors. 

 

Holger

[Marco Robustini]

@John: thanks for the two possible fix, but it is a work around the problem, not solve it.
Owen e Simon say that this problem should not be present, however unfortunately with lower kv + high pole pancake motors under certain conditions turns up.
@Holger: instead this is an interesting thing, thanks Holger!
I did not understand however what happens enabling that parameter:

SLOW_THROTTLE    = 0    ; Limit maximum throttle jump to try to prevent overcurrent

Have you personally experienced with this limiter enabled?
This decreases the performance in terms of rmp speed changes, i mean the reactivity is maintained?

Bests, Marco

[Marco Robustini]

I'm sorry Holger, i've read well, "this macro limits the throttle to double of its previous value", but during AutoTune do not think that from the point of hovering on the amount of throttle that is given to tilt the multicopter can not be double the previous value here, should ask Leonard what value of throttle is used in the procedure.

Cheers, Marco

[Randy Mackay]

 

     I was having a quick look at the AutoQuad ESC32 wiki yesterday and their fancy ESCs actually have a current limiter that might prevent this kind of thing:

              http://autoquad.org/wiki/wiki/esc32/esc32-calibrations/esc32-current-limiter-calibration/

 

     These ESCs look far more advanced than SimonK…but the downside is:

1.       much more expensive

2.       require soldering before you can actually put them on your quad

 

-Randy

--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[Phillip Smith]

I tried autotune today with a similar setup.
Sunnysky 4108 480kv motors
15x5 CF props
10000mah 4S
Tarot 650 CF quad frame
30amp Quattro ESC
I had similar results. During first roll measurement a motor seemed to cog. Quad flipped and fell. I started playing with throttle to force cogging and was able to repeat cogging with extreme throttle movements. I then tried some 30A Turnigy Multistar ESCs flashed with SimonK. Less cogging but it is there occasionally on rapid throttle changes. I've read that Wii-esc had better sync than SimonK so I ordered some HK F30s that I can put Wii-esc on. I wished I'd seen those ESC32s before!

Phillip

[Pablo Lema]

Not sure if these are the same as the AutoQuad ESC32 but may be an RTF option:

http://witespyquad.gostorego.com/speed-controllers/rtf32.html

[Jesus Alvarez]

They (ESC32) look great and do not seem expensive to me taking into account what you can lose if motor loose sync.
However the problem may be the limited 5S support instead of 6S.

Big hexas looking for long flight times normally uses 6S battery with >15" props. In this case they won't be an option.

In my quad I have 2.0 AUW with 610kv Turnigy Multistar and 12x4.5 prop with 4S 3.7Ah. Flight times are around 15 min with low speed and gentle flight.
I tried Autotune some months ago and I have no problem at all with tunigy normal 20A ESC.

I recently upgraded to a Hobbywing Skywalker 25A Quattro which I loaded with BLHeli.... and I think I won't try autotune for the moment until we found a way to tests (on bench) whether this can happen in flight.

Jesus

[Randy Mackay]

 

     Those must be clones of the ESC32 hardware.  Good to see that they *can* be made cheaply at least.

 

     I totally agree that smart ESCs will very quickly pay for themselves even if they are more expensive than normal ones.

 

-Randy

[Holger Steinhaus]

I'm sorry Holger, i've read well, "this macro limits the throttle to double of its previous value", but during AutoTune do not think that from the point of hovering on the amount of throttle that is given to tilt the multicopter can not be double the previous value here, should ask Leonard what value of throttle is used in the procedure.

Depends on your weight. Using a light copter with large props you are often hovering at very low throttle. This is especially true if your motors were chosen for smaller props and are comparatively high rpm types. For 16" or 17" inch I usually run motors with 380kv on 4s. At 16" I may need to enable SLOW_THROTTLE e.g. for a wooden Xoar, but offen can run without when using slim carbon props with narrow blades. On 15", I did not need to enable it so far. 

I have to admit that I never tried Autotune so far, my sync losses all occured in weather with strong and gusty winds. Often in Loiter/AltHold mode when stopping a fast descent (throttle burst from altitude governour plus attitude correction I guess).

 

Holger

[Holger Steinhaus]

Forgot one important point: the problem is usally the inertia of the prop, not the aerodynamic load.

Holger

[Phillip Smith]

Randy,

Good find.  Looks like the shop may be state side also. I will give them a shot.

Jesus,

I've been able to test and recreate the problem on the bench by repeatedly and rapidly moving the throttle from to to bottom. It's obvious when the ESC and motor loses sync. You have to test with prop attached, though.

Phillip

[Phillip Smith]

Sorry, Pablo, you get credit for the good find.

[Jesus Alvarez]

BTW,

I would never go for a 17" prop with less than 6S. These pancakes motors develop a high torque in order to be able to move (accelerate and overcome inertia) those props.
In my mind there is the concept that the more the Voltage, the more torque can the motor develop. And if the torque is not enough, then it loose sync. I may be wrong ... :(

Jesus

[Phillip Smith]

Marco, et al,

Have you considered trying Wii-ESC instead of SimonK?  I just flashed an old 20amp RC-Timer ESC with Wii-ESC and I get zero cogging and zero timing sync issues and very linear throttle response.  I am running 22-pole 4108 480kv Sunnysky motors with 15x5 CF props.  My ESCs are crap quality but with the Wii-ESC firmware the sync issues go away.  Your F40As are flashable to Wii-ESC if I am not mistaken.

I also have 4x of those RTF ESC32s inbound to test as well.

Phillip

[Phillip Smith]

Jesus,

There are also sync issues for sure with SimonK.  Especially with 22-pole / pancake motors.  This post may have already been linked but it discusses the issue.

http://forums.openpilot.org/topic/11561-esc-firmware-flashing-and-pancake-motor-compatibility-wii-esc-rapidesc/

Phillip

[bn...@adinet.com.uy]

Hi Guys, I´ve been reading all coments on RCG regarding simonk and AfroESC with Simonk for a long time now and also I have used AfroESC, F30A and HK Blue 40A ESC´s with 4220-880Kv and 4220-650KV.

To me it seems that Sync problems is a BIG issue on muticopters. How can we expect the aircraft to be absolutely stable if we can´t really control the motors? No mater which controller we buy (I have 2 APM 2.5 only) or how much procesing power and featres we might find if when it has to execute the controls the "arms" can do their job.

UltraESC seems to be at least trying to find a solution by using more processing power, but the atmega8 used in most ESC is at it´s end of life. Someone has to come up with ther alternatives or maybe sensored esc can make a difference. 

When AfroESC with simonk´s firmware came out I thought: "well, maybe now that simon is receiving money the firmware is really going to be polished and developed to it´s full potencial" but all we got are some testing versions in an almos secret repository which seem to be working better but far from perfect.

I really appreciatte simon´s voluntary work when it was voluntary but once you start receiving money you expect another level.

By the way I also think that 3DR should be working on more robust ESC since they have the brain power and the QC to make things right (how good is an octacopter with a Pixhawk on 6s with sync problems right?

I hope I´m not offending anyone but somehow adding something to the discussion

[Marco Robustini]

Hi Phillip, many argue that the Wii-ESC has fewer problems, but personally I've had more, gray smoke during my tests, never happened with SimonK.
SimonK loses sync but perhaps more safeguards FET (with the latest release), Wi-ESC instead sometimes it crashed and burned everything.
Here my videos about this, for example, the first with Wii-ESC and the second with the version "12/09" of SimonK, same ESC:

https://www.youtube.com/watch?v=hBUBbeyLe0Q

https://www.youtube.com/watch?v=0lRTN5o4UCM

Marco

[Meier Lorenz]

Hi,

Expect some news on this topic from the PX4 dev community soon - we will be looking for feedback before attacking this topic on the full scale 8).

-Lorenz

--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com<mailto:drones-discus...@googlegroups.com>.

[bn...@adinet.com.uy]

I agree with you. I have just come from testing Wii-ESC (just out of curiosity) and the results are not good. If you go from mid throttle to WOT the motors "stays" in the same rpm for a second or two while twiching a bit and then the go to full rpm. 

Also if I use any version of simonk older than 09-11-13 it´s impossible to make them work.

I´m using Multistar 4220-650kv motor with 4s batery and F30A (I even tested Afro 30A with the same results)

[john...@gmail.com]

For future reference. It is simple to test if the motor/propeller combination has a potential sync problem. It is the same type of test that we often use for 3D airplanes where it is common to over prop the plane for short bursts of extreme performance.

Just connect a receiver directly to the ESC and do static propeller tests punching the throttle + rapid min/max throttle changes. Under no circumstance no matter what you do with the throttle, should the motor lose sync and start making weird noises. If it does, then is might also do so during flight also.

- JAB

[Marco Robustini]

You are lucky, try with 6S, all the problems are most evident... :-)
Unfortunately what is missing today on the market are good ESC that will withstand high amperage at high voltage for low kv motors.
I did a few tests with BlHeli, but I think now I will move my attention on that firmware.

Marco

[john...@gmail.com]

6S tsk, tsk..  :) My biggest acro airplane uses 12S on a 180kv motor and 26" propeller. If you don't get the sync timing correct, it will scream louder then a pig. But luckily the expensive ($400) 200A type ESC's you use for such planes have proper manual controls for everything like motor pole count, max rpm, acceleration, timing in 1 degree increments etc. so that you can match them correctly with any motor.

Funny trivial about giant scale motors setups like that. The setup I described pulls >20kg static trust. So a hexacopter using them could lift a person...

- JAB

[Marco Robustini]

I agree JAB, i've similar setup but hardly changes in the throttle on a model so they are very fast as it can give an electronic correction during the attitude of a multirotor, also the propeller of a giant rc plane is "helped" by the speed of the aircraft, instead motors of a multicopter constantly supporting his weight, so the inertia is another.
Then this expensive ESC does not go as some of the classics from $10<->$40 for here all use in out multirotor... ;-)

Marco

[Marco Robustini]

About the timing there's same setting inside the SimonK firmware for tuning the ESC with the motor:

MOTOR_ADVANCE    = 18    ; Degrees of timing advance (0 - 30, 30 meaning no delay)
TIMING_OFFSET    = 0    ; Motor timing offset in microseconds
TIMING_MIN    = 0x8000 ; 8192us per commutation
TIMING_RANGE1    = 0x4000 ; 4096us per commutation
TIMING_RANGE2    = 0x2000 ; 2048us per commutation
TIMING_MAX    = 0x00e0 ; 56us per commutation

Etc. etc., the problem is that most users don't even know flashing their ESC, let alone if they start playing with these variables in TGY.AMS... ;-)
We should create the firmware already in place for each type of motor (impossible, i think), or create guidelines that thei can follow, but unfortunately almost all buy and install ESC already reflashed, and after some flight the crash is around the corner.

Marco

On Monday, January 13, 2014 10:59:28 PM UTC+1, john...@gmail.com wrote:

[john...@gmail.com]

If TIMING_MAX is an indication of the PWM frequency then it's ~18KHZ. This is probably the max for the AVR chip and much better then the 8khz default used by most cheap ESC's. But 24khz or even 32khz would be even better for high pole count pancake motors.

It would be very interesting to see how much tuning the timing would help with your problematic motor combination.

[Leonard Hall]

Hi Marco,

I can confirm that this is the classic lost sync problem I have seen many times. 

The thing I just can't understand, why use Simon K on large copters. It only makes sense on ultra small copters. By small, I mean sub 1kg and sub 8 inch props. The whole principle of the Simon K software is to minimise the delay when the throttle is applied and in doing so greatly increase the load on the ESC.

So, Autotune results in a tune that most people consider too sharp on standard ESC's. Simon K firmware is known to be prone to sync issues. Large camera ships don't want lightning reactions, soft and smooth is desired. So why use it on an expensive camera ship?

And to put all this in perspective I will use your esc failure as an example. Your failure happened immediately after a rate test. Autotune asks for a rate of 90 degrees per second (1/4 a good acro value). Autotune found your default rate P too high and at that time had reduced your Rate_P from 0.18 to 0.145 and was still going down. However, the sync is always lost when autotune switches back to stabilise and the users generally overly aggressive settings. So in your case you have a 27 degree position error and are rotating at 90 degrees per second in the wrong direction. This will result in a roll command of (2700*4.5+9000)*0.18=3807 or 76% control output. You can get a similar control output by holding your pitch stick full over then letting the stick go. So basicly your ESC lost sync after the equivalent of letting go of the pitch stick after pushing forward all the way.

So to summarise:

1. auto tune doesn't drive the esc's harder on any single test than what we should expect from a normal flight. However, Autotune does do this repeatedly. 

2. Simon K doesn't provide any usable performance improvement on large airframes, now we can tune our copters properly (not that I have seen anyway).

3. Simon K is known to have a reliability problem in the form of sync issues,

4. Simon K pushes the ESC much harder in the form of transient currents and therefore requires a higher current ESC than a standard ESC.

I really think that if a mechanical problem happens during autotune, it can happen during normal flight.

This is just my opinion and I am happy to be educated by anybody willing to throw their two cents in :)

[Marco Robustini]

100% agree Leonard, i know the benefits and limitations of SimonK firmware, now I asked Simon if there is the possibility of deprecate the speed reaction in the firmware to make it less aggressive, on my "big birds" I think it's necessary.
I've read of others who have had the same problem, i'm sure that the same thing can happen in other flight mode but many like me are pretty careful not to create situations that require heavy attitude corrections, but unfortunately AutoTune automatically generates these corrections, the user can only watch and hope.
Knowing this problem i'm ready to recover my quad from this situation, and in fact I have not crashed (this time).
I say this only because AutoTune is believed to be the problem, when in fact it does only come to light more easily, that's all.
I await the response of Simon.

Cheers, Marco

[Luis Morales]

FYI this is not only apparent in big rigs, and 22poles motors

i was trying to run autotune last Saturday on a TBS discovery with DJI 920kv motors and afro 30A with latest simonk, activating autotune immediately brought the quad to the floor and i mean instant it dit not even look like it try to start autotune.

while i was trying to figuring this out (trying different things) on one time two of the motors lost sync while the other where doing really fast the quad went directly to the floor and broke an arm.

i can confirm that this does not happen at all with DJI 30A ESC, i have tried with both the factory fw and flashing with BL Heli, autotune works no problem with that.

[john...@gmail.com]

To understand why Simonk is/was also used on large copters you have to remember that before brushless gimbals, copter stability would have a large impact on how stable the camera was. A more aggressive tune would then help by keeping the copter more stable in windy conditions.

On Tuesday, January 14, 2014 1:30:16 PM UTC+1, Leonard Hall wrote:

[Brendan O'Reilly]

...so SimonK loaded ESCs are of no advantage (or worse) for "bigger" copters?  Randy suggested that the ESC32's might be a good option for the higher pole count motors and larger props.  If "slower" firmware is all that is required then that sounds great.  

A couple of questions:

1) Do most bigger copter (14"+ props) ESC/Motors combinations "just work" (and I've been unlucky, or perhaps too fixated on SimonK/BLHeli) - or is it actually difficult to find reliable ESC/motor combinations for these bigger copters?

2) Does anyone have any other (economical) suggestions for reliable ESCs and motor combinations?  

I apologize if this is off-topic but it seems difficult to find proven combinations and the best auto-pilot in the world is no use if the copter cant fly reasonably reliably.  For me utility flying with a payload (not sport flying) requires good range which I understand requires bigger props and lower KV.

[Randy Mackay]

Marco,

 

      If 3DR were to make smart ESCs, what’s the maximum voltage and current that you would need for your copter?

 

     I think it’s 6S batteries (i.e. 25V) but how many amps do you need per ESC?  40Amp?  30Amp?  Any idea how much current your hexacopter draws when hovering and at full throttle?

 

-Randy

 

From: drones-...@googlegroups.com [mailto:drones-...@googlegroups.com] On Behalf Of Brendan O'Reilly
Sent: January 15, 2014 9:09 AM
To: drones-...@googlegroups.com
Subject: [drones-discuss] Re: APM Copter V3.1: Autotune with reflashed ESC... warning!

 

...so SimonK loaded ESCs are of no advantage (or worse) for "bigger" copters?  Randy suggested that the ESC32's might be a good option for the higher pole count motors and larger props.  If "slower" firmware is all that is required then that sounds great.  

--

[Marco Robustini]

Hi Randy, 6S (25.2V) is ideal, it is the most widely used in multirotor quite large.
About the current 40A is right, whereas it is difficult if the system is well optimized you exceed 30A for the motor.
For example, one of my "big board" without payload (cam, gimbal, etc.) fly for 14 minutes with a single 5A 6S LiPo, therefore the consumption of each motor is very low.

Marco

[Robert Lefebvre]

I'd agree.  6S 40A would cover most of the larger machines that I see.  I'm not sure if it would be future proof or not, but at least would cover the bases today.

I find Multirotor design is kind of at a cross-roads.  There are people building bigger and bigger machines, and I've seen some using 8S.  But on the other hand, good cameras are available in smaller and smaller packages, so people are using smaller multis, for safety.

[Cornel Fudulu]

I have two machines running 22-pole pancake motors:

 

A photo-quad with an APM 2.0, AX-4008Q / 620kV turning 12x3.8 SF props, and Skywalker Quattro 20A ESC with BLHeli firmware on 4S. This one flies like a champ, went through

three Autotune sessions without any hint of desync.

 

A flat octo, with MK, Multistar 4822 / 690kV turning 13x5.5 props, AfroI2C converter and F-30A ESCs with Simonk. While three testflights in calm weather and without load failed to spot any trouble, my first fully-loaded flight in stiff winds ended up in desync and a shallow glide to a crash.

 

After that, I began experimenting. The cheapest SiLabs-based ESCs I could get fast were some $13 Turnigy AE-30A. Reflashed with BLHeli and started playing. While the motor didn't hint at desync, the ESC would get very hot and ended up with magic smoke and burned FETs. Pulling the datasheet, it became obvious that the FETs were horribly slow, and I was bleeding current before they could close. Next was a more pricey $33 K-Force 40A. Very fast FETs, with very low Rds(on). Coudn't make my motors desync. I even borrowed two more pancakes, a 4830 / 420kV and a 4114 / 360kV and went with 15x5 and 16x5.5 props and 6S without a twitch.

The trouble with the K-Forces is that they're big and heavy, each with a SBEC onboard.

So I went back to my Simonk F-30As and tried every firmware out there until I found one that drives my 4822s without any sign of problem. It's the 2013_10_29_comp_pwm.

As for the F-30A, it has a decent set of FETs, rated 90A, so I bet they'll take whatever high-current transients Simon's software throws at them. And after upgrading all the ESCs on my octo, I've pulled it through the most violent maneouvers I could imagine, and no hint of a desync showed.

 

 

[Andrew Chapman]

Ditto, 6S @ 40A for me on my Hexas.

I would *love* to see 3DR develop a smart ESC option, it would solve some issues and be a huge boost in diagnosing potential problems before people hit them. 

Currently we have an open ended system and have to guess what is happening between the control inputs and the motor outputs. Not to mention it opening the possibility of neat things like programmable reverse from the APM (which would make configuring new setups a lot easier, particularly if people want to skip bullet connectors between ESC+motors), and allow for some research into things like quad motor loss survivability.

AC.

[Robert Lefebvre]

Interesting testing.

I would love somebody to try out the YEP series (YGE clones).  They are really intended for helis, and nobody makes a firmware for them that I'm aware of.  But supposedly you can set them up to be responsive (you can drive them directly with a governor output which requires fast response).  

But the important thing about them is that they have Active Freewheeling.  This is something that seems completely overlooked in multirotor ESC's, and could give significant power and reliability benefits.  Especially since the ESC's on multis spend so much of their time at 1/2 throttle.

--

[Meier Lorenz]

Hi all,

Has anyone done actual measurements of hover throttle current and peak current? The numbers I see here indicate that you don’t crash and burn with these ESC choices, but it tells little about the current actually drawn. In fact, the battery life you achieve kind of tells me that the average and peak currents must be much lower.

My experience is quite close to Marcos - for a non-overpowered multicopter with around 800-1000 kV you will draw 5-8 amps per motor during hover. That is pretty much constant, assuming larger multicopters have larger props and higher voltages. Once you drop the kV and go to higher currents these numbers change of course.

But once you do, thermal losses due to the high currents and the relatively high weight of all cabling involved starts to burn energy unused - so from an efficiency point of view, you want higher voltages, not higher currents.

I would appreciate if anyone could share actual data / logs (with a calibrated current sensor please).

Cheers,
Lorenz

On 15 Jan 2014, at 19:28, Robert Lefebvre <robert....@gmail.com<mailto:robert....@gmail.com>> wrote:

Interesting testing.

I would love somebody to try out the YEP series (YGE clones). They are really intended for helis, and nobody makes a firmware for them that I'm aware of. But supposedly you can set them up to be responsive (you can drive them directly with a governor output which requires fast response).

But the important thing about them is that they have Active Freewheeling. This is something that seems completely overlooked in multirotor ESC's, and could give significant power and reliability benefits. Especially since the ESC's on multis spend so much of their time at 1/2 throttle.

On 15 January 2014 12:05, Cornel Fudulu <cornel...@gmail.com<mailto:cornel...@gmail.com>> wrote:
I have two machines running 22-pole pancake motors:

A photo-quad with an APM 2.0, AX-4008Q / 620kV turning 12x3.8 SF props, and Skywalker Quattro 20A ESC with BLHeli firmware on 4S. This one flies like a champ, went through
three Autotune sessions without any hint of desync.

A flat octo, with MK, Multistar 4822 / 690kV turning 13x5.5 props, AfroI2C converter and F-30A ESCs with Simonk. While three testflights in calm weather and without load failed to spot any trouble, my first fully-loaded flight in stiff winds ended up in desync and a shallow glide to a crash.

After that, I began experimenting. The cheapest SiLabs-based ESCs I could get fast were some $13 Turnigy AE-30A. Reflashed with BLHeli and started playing. While the motor didn't hint at desync, the ESC would get very hot and ended up with magic smoke and burned FETs. Pulling the datasheet, it became obvious that the FETs were horribly slow, and I was bleeding current before they could close. Next was a more pricey $33 K-Force 40A. Very fast FETs, with very low Rds(on). Coudn't make my motors desync. I even borrowed two more pancakes, a 4830 / 420kV and a 4114 / 360kV and went with 15x5 and 16x5.5 props and 6S without a twitch.
The trouble with the K-Forces is that they're big and heavy, each with a SBEC onboard.
So I went back to my Simonk F-30As and tried every firmware out there until I found one that drives my 4822s without any sign of problem. It's the 2013_10_29_comp_pwm.
As for the F-30A, it has a decent set of FETs, rated 90A, so I bet they'll take whatever high-current transients Simon's software throws at them. And after upgrading all the ESCs on my octo, I've pulled it through the most violent maneouvers I could imagine, and no hint of a desync showed.



--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com<mailto:drones-discuss%2Bunsu...@googlegroups.com>.

For more options, visit https://groups.google.com/groups/opt_out.

--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com<mailto:drones-discus...@googlegroups.com>.

[Robert Lefebvre]

I do sometimes wonder about the the high voltage thing... it has to have a limit somewhere.  Could you theoretically run on 600V and 0.01A?  

I know that electric motors have something called "magnetizing current" which is the minimum current required in a given motor to created the magnetic field in the first place.  It's the absolute minimum current before the motor would even begin to turn. Does this have any lower bounds, even as the voltage goes up?  

This is all stuff that was actually in my course of study, but I haven't gotten into it much.

Something else I'll throw out:  I recently discovered some motors called "Low Ferrous Loss" for helis. They are constructed differently, and supposedly are 5% more efficient than standard outrunners at full load, but 15% more efficient at part load.  Could also be interesting technology for multirotors.

I really want to get one to try out on my heli, but, with the required ESC, it's $700 total!  Hobby King has a clone motor for only $100, but no ESC designed to drive it (requires higher switching frequency).

To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com.

[Meier Lorenz]

Hi Robert,

Have a link to the motor and the required switching frequ?

-Lorenz

------------------------------------------------------
Lorenz Meier
Computer Vision and Geometry Group
Institute for Visual Computing
ETH Zurich
http://www.inf.ethz.ch/personal/lomeier/

Am 15.01.2014 um 20:09 schrieb Robert Lefebvre <robert....@gmail.com<mailto:robert....@gmail.com>>:

I do sometimes wonder about the the high voltage thing... it has to have a limit somewhere. Could you theoretically run on 600V and 0.01A?

I know that electric motors have something called "magnetizing current" which is the minimum current required in a given motor to created the magnetic field in the first place. It's the absolute minimum current before the motor would even begin to turn. Does this have any lower bounds, even as the voltage goes up?

This is all stuff that was actually in my course of study, but I haven't gotten into it much.

Something else I'll throw out: I recently discovered some motors called "Low Ferrous Loss" for helis. They are constructed differently, and supposedly are 5% more efficient than standard outrunners at full load, but 15% more efficient at part load. Could also be interesting technology for multirotors.

I really want to get one to try out on my heli, but, with the required ESC, it's $700 total! Hobby King has a clone motor for only $100, but no ESC designed to drive it (requires higher switching frequency).

On 15 January 2014 13:44, Meier Lorenz <l...@inf.ethz.ch<mailto:l...@inf.ethz.ch>> wrote:
Hi all,

Has anyone done actual measurements of hover throttle current and peak current? The numbers I see here indicate that you don’t crash and burn with these ESC choices, but it tells little about the current actually drawn. In fact, the battery life you achieve kind of tells me that the average and peak currents must be much lower.

My experience is quite close to Marcos - for a non-overpowered multicopter with around 800-1000 kV you will draw 5-8 amps per motor during hover. That is pretty much constant, assuming larger multicopters have larger props and higher voltages. Once you drop the kV and go to higher currents these numbers change of course.

But once you do, thermal losses due to the high currents and the relatively high weight of all cabling involved starts to burn energy unused - so from an efficiency point of view, you want higher voltages, not higher currents.

I would appreciate if anyone could share actual data / logs (with a calibrated current sensor please).

Cheers,
Lorenz

On 15 Jan 2014, at 19:28, Robert Lefebvre <robert....@gmail.com<mailto:robert....@gmail.com><mailto:robert....@gmail.com<mailto:robert....@gmail.com>>> wrote:

Interesting testing.

I would love somebody to try out the YEP series (YGE clones). They are really intended for helis, and nobody makes a firmware for them that I'm aware of. But supposedly you can set them up to be responsive (you can drive them directly with a governor output which requires fast response).

But the important thing about them is that they have Active Freewheeling. This is something that seems completely overlooked in multirotor ESC's, and could give significant power and reliability benefits. Especially since the ESC's on multis spend so much of their time at 1/2 throttle.

On 15 January 2014 12:05, Cornel Fudulu <cornel...@gmail.com<mailto:cornel...@gmail.com><mailto:cornel...@gmail.com<mailto:cornel...@gmail.com>>> wrote:
I have two machines running 22-pole pancake motors:

A photo-quad with an APM 2.0, AX-4008Q / 620kV turning 12x3.8 SF props, and Skywalker Quattro 20A ESC with BLHeli firmware on 4S. This one flies like a champ, went through
three Autotune sessions without any hint of desync.

A flat octo, with MK, Multistar 4822 / 690kV turning 13x5.5 props, AfroI2C converter and F-30A ESCs with Simonk. While three testflights in calm weather and without load failed to spot any trouble, my first fully-loaded flight in stiff winds ended up in desync and a shallow glide to a crash.

After that, I began experimenting. The cheapest SiLabs-based ESCs I could get fast were some $13 Turnigy AE-30A. Reflashed with BLHeli and started playing. While the motor didn't hint at desync, the ESC would get very hot and ended up with magic smoke and burned FETs. Pulling the datasheet, it became obvious that the FETs were horribly slow, and I was bleeding current before they could close. Next was a more pricey $33 K-Force 40A. Very fast FETs, with very low Rds(on). Coudn't make my motors desync. I even borrowed two more pancakes, a 4830 / 420kV and a 4114 / 360kV and went with 15x5 and 16x5.5 props and 6S without a twitch.
The trouble with the K-Forces is that they're big and heavy, each with a SBEC onboard.
So I went back to my Simonk F-30As and tried every firmware out there until I found one that drives my 4822s without any sign of problem. It's the 2013_10_29_comp_pwm.
As for the F-30A, it has a decent set of FETs, rated 90A, so I bet they'll take whatever high-current transients Simon's software throws at them. And after upgrading all the ESCs on my octo, I've pulled it through the most violent maneouvers I could imagine, and no hint of a desync showed.



--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com<mailto:drones-discuss%2Bunsu...@googlegroups.com><mailto:drones-discuss%2Bunsu...@googlegroups.com<mailto:drones-discuss%252Buns...@googlegroups.com>>.

For more options, visit https://groups.google.com/groups/opt_out.


--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com<mailto:drones-discuss%2Bunsu...@googlegroups.com><mailto:drones-discus...@googlegroups.com<mailto:drones-discuss%2Bunsu...@googlegroups.com>>.

[Robert Lefebvre]

There's a thread here:

http://www.rcgroups.com/forums/showthread.php?t=534261

But in brief:

Motor has no iron core.  It's an in-runner design, magnets on the rotor.  Supposedly it requires >24kHz switching.

Kontronic is the originator of the design.  They sell for $350.  A matching ESC from them is $350.  HK now make a clone of the motor for $100, but don't have any ESC's with switching frequencies that high.

To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com.

[Cornel Fudulu]

Lorenz,

 

I have most of my birds equipped with FrSky's FAS-100 current/voltage sensor.

On bench, my AX-4008Q motors, with 12x3.8 SF (carbon APC copy) on 4S top at 18A and 1730g pull, with 3.9A required for 500g. In the air, the 2,1 Kg quad eats 18A to hover, 38A for full-throttle climb and the peak I've seen was a 42.x Amps while braking from a fast descent. Mostly in Stab, driven by the APM 2.0.

 

The Multistar 4822 / 690 kV with 13x5.5 top on bench at 28A and shy of 2,2 Kg with 4.2A for 500g. The 6 Kg octo hovers at 52A and rarely breaks 85A in forward Altitude Hold flight, which is what it mostly does (aerial traveling camera). Full-throttle climbs go past the 100A the sensor can measure, and while I can switch the Hall sensor for a 200A one, I'm intelectualy challenged when it comes to modifying lines of code, to adjust OpenTX to display true current.

 

Cheers,

Para.

[Phillip Smith]

Cornel,

So am I reading correct, on your smaller quad, the most you see is about 10amps per ESC?  42amp total?  I see the same on mine, hovers around 14amp, see a max of about 35amps.  What is the logic then on 30-40amp ESCs on the average quad?  Are there situations where one motor is surging to that high of an amperage?

Phillip

--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com.

[Robert Lefebvre]

It could be because of the lack of Active Freewheeling.  The effect is worse at part load, which is what most of these machines are doing most of the time.

Try to put it simply:  The motor is an inductor.  The ESC controls the load on the motor by turning the FETs on and off rapidly. Since the motor is an inductor, the current cannot stop instantly.  This causes a voltage at the FET.  This could blow the FET, so all FET's have diodes in them to allow this current to flow.  The diode has a voltage drop associated.  This means that some of the current gets turned into heat.  The longer the FET is turned off, the more heat that is generated. And the higher the current, the more heat is generated. So, the worst possible thing you can do, is over-prop a motor to produce too much thrust, and then run a very low throttle.  

And this is exactly what many people are doing with the new powerful machines.

Active Freewheeling uses an active component like another FET to allow the freewheeling current to flow with very little voltage drop, and thus very little heat.  And also less energy is taken from the rotor, making the motor more efficient too.

I believe this is why we need to overspecify the ESC on most multirotors.  It's not that we are using that much current, but we need ESC's with that much mass to sink the heat from passive freewheeling.

Also, I find it crazy that we use these ESC's, with nothing more than a slab of aluminum mounted on them, wrapped up in shrink wrap, and call that good enough.  It's probably fine if the ESC spends most of it's time at >80% throttle, but not when they are being used the way we do.  I typically replace the aluminum with a real heat sink, and cut a window in the shrinkwrap to reveal the fins.

[Phillip Smith]

These have active freewheeling too... Anyone tried them?

http://www.ultraesc.com/index.php?ESC=20A

[Randy Mackay]

Para,

 

     Re the comment about getting the hall sensor working.  Does that sensor produce a voltage between 0 ~ 5V (or 3.3V) to indicate the current it’s sensing?  If it does then it should be possible to get it working without any code changes.  The wiring diagrams are on this wiki page:

                http://copter.ardupilot.com/wiki/common-using-a-current-sensor/

      and then to get it to produce an accurate number you might need to calibrate it as shown on this page (including a video!):

              http://copter.ardupilot.com/wiki/common-measuring-battery-voltage-and-current-consumption-with-apm/#Calibrating_the_voltage_reading

 

     No pressure of course, it’s your copter, but just wanted to provide the info in case.

 

-Randy

[Meier Lorenz]

Ok,

I worked out how I can estimate your average max current based on flight time and hover throttle. Its sufficiently linked to get me into the right ball park with just crude measurements.

So for ALL the multicopter pilots out there, please provide a few basic numbers. It takes one minute.
https://docs.google.com/forms/d/1SWySjmV2n1OtNj4uASEVMMRgLZT_l8Z_XaTypJhN94Q/viewform#start=openform

Fill the form for each multicopter / battery combination you fly. Please try to provide accurate numbers (in particular the flight time and hover throttle matters).

The more and better data we can gather here, the better we will be able to shape PX4 to suit your needs in the future.

-Lorenz





------------------------------------------------------
Lorenz Meier
Computer Vision and Geometry Group
Institute for Visual Computing
ETH Zurich
http://www.inf.ethz.ch/personal/lomeier/

Am 15.01.2014 um 20:32 schrieb Robert Lefebvre <robert....@gmail.com<mailto:robert....@gmail.com>>:

There's a thread here:

http://www.rcgroups.com/forums/showthread.php?t=534261

But in brief:

Motor has no iron core. It's an in-runner design, magnets on the rotor. Supposedly it requires >24kHz switching.

Kontronic is the originator of the design. They sell for $350. A matching ESC from them is $350. HK now make a clone of the motor for $100, but don't have any ESC's with switching frequencies that high.

On 15 January 2014 14:21, Meier Lorenz <l...@inf.ethz.ch<mailto:l...@inf.ethz.ch>> wrote:
Hi Robert,

Have a link to the motor and the required switching frequ?

-Lorenz

------------------------------------------------------
Lorenz Meier
Computer Vision and Geometry Group
Institute for Visual Computing
ETH Zurich
http://www.inf.ethz.ch/personal/lomeier/

Am 15.01.2014 um 20:09 schrieb Robert Lefebvre <robert....@gmail.com<mailto:robert....@gmail.com><mailto:robert....@gmail.com<mailto:robert....@gmail.com>>>:

I do sometimes wonder about the the high voltage thing... it has to have a limit somewhere. Could you theoretically run on 600V and 0.01A?

I know that electric motors have something called "magnetizing current" which is the minimum current required in a given motor to created the magnetic field in the first place. It's the absolute minimum current before the motor would even begin to turn. Does this have any lower bounds, even as the voltage goes up?

This is all stuff that was actually in my course of study, but I haven't gotten into it much.

Something else I'll throw out: I recently discovered some motors called "Low Ferrous Loss" for helis. They are constructed differently, and supposedly are 5% more efficient than standard outrunners at full load, but 15% more efficient at part load. Could also be interesting technology for multirotors.

I really want to get one to try out on my heli, but, with the required ESC, it's $700 total! Hobby King has a clone motor for only $100, but no ESC designed to drive it (requires higher switching frequency).

On 15 January 2014 13:44, Meier Lorenz <l...@inf.ethz.ch<mailto:l...@inf.ethz.ch><mailto:l...@inf.ethz.ch<mailto:l...@inf.ethz.ch>>> wrote:
Hi all,

Has anyone done actual measurements of hover throttle current and peak current? The numbers I see here indicate that you don’t crash and burn with these ESC choices, but it tells little about the current actually drawn. In fact, the battery life you achieve kind of tells me that the average and peak currents must be much lower.

My experience is quite close to Marcos - for a non-overpowered multicopter with around 800-1000 kV you will draw 5-8 amps per motor during hover. That is pretty much constant, assuming larger multicopters have larger props and higher voltages. Once you drop the kV and go to higher currents these numbers change of course.

But once you do, thermal losses due to the high currents and the relatively high weight of all cabling involved starts to burn energy unused - so from an efficiency point of view, you want higher voltages, not higher currents.

I would appreciate if anyone could share actual data / logs (with a calibrated current sensor please).

Cheers,
Lorenz

On 15 Jan 2014, at 19:28, Robert Lefebvre <robert....@gmail.com<mailto:robert....@gmail.com><mailto:robert....@gmail.com<mailto:robert....@gmail.com>><mailto:robert....@gmail.com<mailto:robert....@gmail.com><mailto:robert....@gmail.com<mailto:robert....@gmail.com>>>> wrote:

Interesting testing.

I would love somebody to try out the YEP series (YGE clones). They are really intended for helis, and nobody makes a firmware for them that I'm aware of. But supposedly you can set them up to be responsive (you can drive them directly with a governor output which requires fast response).

But the important thing about them is that they have Active Freewheeling. This is something that seems completely overlooked in multirotor ESC's, and could give significant power and reliability benefits. Especially since the ESC's on multis spend so much of their time at 1/2 throttle.

On 15 January 2014 12:05, Cornel Fudulu <cornel...@gmail.com<mailto:cornel...@gmail.com><mailto:cornel...@gmail.com<mailto:cornel...@gmail.com>><mailto:cornel...@gmail.com<mailto:cornel...@gmail.com><mailto:cornel...@gmail.com<mailto:cornel...@gmail.com>>>> wrote:
I have two machines running 22-pole pancake motors:

A photo-quad with an APM 2.0, AX-4008Q / 620kV turning 12x3.8 SF props, and Skywalker Quattro 20A ESC with BLHeli firmware on 4S. This one flies like a champ, went through
three Autotune sessions without any hint of desync.

A flat octo, with MK, Multistar 4822 / 690kV turning 13x5.5 props, AfroI2C converter and F-30A ESCs with Simonk. While three testflights in calm weather and without load failed to spot any trouble, my first fully-loaded flight in stiff winds ended up in desync and a shallow glide to a crash.

After that, I began experimenting. The cheapest SiLabs-based ESCs I could get fast were some $13 Turnigy AE-30A. Reflashed with BLHeli and started playing. While the motor didn't hint at desync, the ESC would get very hot and ended up with magic smoke and burned FETs. Pulling the datasheet, it became obvious that the FETs were horribly slow, and I was bleeding current before they could close. Next was a more pricey $33 K-Force 40A. Very fast FETs, with very low Rds(on). Coudn't make my motors desync. I even borrowed two more pancakes, a 4830 / 420kV and a 4114 / 360kV and went with 15x5 and 16x5.5 props and 6S without a twitch.
The trouble with the K-Forces is that they're big and heavy, each with a SBEC onboard.
So I went back to my Simonk F-30As and tried every firmware out there until I found one that drives my 4822s without any sign of problem. It's the 2013_10_29_comp_pwm.
As for the F-30A, it has a decent set of FETs, rated 90A, so I bet they'll take whatever high-current transients Simon's software throws at them. And after upgrading all the ESCs on my octo, I've pulled it through the most violent maneouvers I could imagine, and no hint of a desync showed.



--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com<mailto:drones-discuss%2Bunsu...@googlegroups.com><mailto:drones-discuss%2Bunsu...@googlegroups.com<mailto:drones-discuss%252Buns...@googlegroups.com>><mailto:drones-discuss%2Bunsu...@googlegroups.com<mailto:drones-discuss%252Buns...@googlegroups.com><mailto:drones-discuss%252Buns...@googlegroups.com<mailto:drones-discuss%25252Bun...@googlegroups.com>>>.

For more options, visit https://groups.google.com/groups/opt_out.


--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com<mailto:drones-discuss%2Bunsu...@googlegroups.com><mailto:drones-discuss%2Bunsu...@googlegroups.com<mailto:drones-discuss%252Buns...@googlegroups.com>><mailto:drones-discus...@googlegroups.com<mailto:drones-discuss%2Bunsu...@googlegroups.com><mailto:drones-discuss%2Bunsu...@googlegroups.com<mailto:drones-discuss%252Buns...@googlegroups.com>>>.

[Lorenz Meier]

Hi all,

I got some first results. I’ve used the curve below as baseline to extrapolate the numbers in this spread sheet. Its all just ballpark accuracy (who claims the full hour flight time? I’d love to see close-ups of that build).

I also would appreciate if more people could enter data into the form - takes 20 seconds

https://docs.google.com/forms/d/1SWySjmV2n1OtNj4uASEVMMRgLZT_l8Z_XaTypJhN94Q/viewform#start=openform

It does support my hypothesis though - we’re talking about an average current of 3-6A per motor (compare to the plot below which is representative for a 10“ quad), and a peak load of mostly 12-15, some up to 22A. I forgot to ask the weight in the survey, but these are not massive quads. I would appreciate if more of the „big monster“ owners could enter some data.

Non-surprisingly the worst peak current is a low-kV motor with a 3S battery.

-Lorenz

[Alex Costich]

I would like to fill out the survey but the battery capacity field does not have the capacity I am using.  Perhaps add a "Other" field to battery capacity as well?

Thanks,

Alex.

[Marco Robustini]

Damn, 20 seconds * 9 = 3 minutes, where i can find all this time? :-)
Just for fun... i compile now!

Marco

[Meier Lorenz]

Thanks!

Updated reports here:
https://docs.google.com/spreadsheet/pub?key=0AqmuaMYa3w8BdDdjS1gzN2R4ODlzeXYta2RTMlBjMXc&output=html

Still consistent with what we expected. 25A peak + short term burst capacity plus active free running would cover the range we came across here still quite well.

-Lorenz

------------------------------------------------------
Lorenz Meier
Computer Vision and Geometry Group
Institute for Visual Computing
ETH Zurich
http://www.inf.ethz.ch/personal/lomeier/

--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com<mailto:drones-discus...@googlegroups.com>.

[Marco Robustini]

I'm doing some experiments with changes to the SimonK code made ​​by a german team, very promising, i keep you updated at the end of my tests with different motors (hig poles low kv, from 320 to 420)/esc/propellers (thanks Holger).

Marco

On Sunday, January 12, 2014 9:11:53 AM UTC+1, Marco Robustini wrote:

Yesterday I run Autotune on two of my heavy hexa, installing powerful motors with large carbon propellers 15x5, F40A ESC flashed with SimonK firmware (the latest version), working with 6s LiPo.
During the automatic tuning of the "pitch" two motors have completely lost sync and miraculously i recover the multicopter and land without damage.
Thinking it was a case I repeated the procedure with another multicopter, always very heavy (same propellers) and the problem occurred again (always during variations on the pitch), this time only one motor lost sync during one of the fast throttle inputs that Autotune produces, the quad is tilted diagonal, not to the right side.
This was always done during the step on the pitch because the multicopter is much more reactive on that side for obvious reasons of weight (battery, landing gear, etc.).
I try to find the logs and post them here as soon as possible, but the problem in my opinion is that Autotune produces too fast throttle change input during the procedure, from the "hovering" (quad in a horizontal position) to the max location where it's tilted, these too rapid accelerations, that only the code can produce but our thumbs would be almost impossible, causing possible out of sync with the SimonK firmware.
Now, I don't know how Leonard can solve the problem because this thing could never happen even on some setups, but I did it three times and honestly do not intend to try again because I risked crashing quad very large and expensive.
On DIY Drones I've read of someone who has had the same problem, and had not realized what had happened, the probable cause is one that I wrote, as long as the code of Autotune there is something wrong (do not think).
The problem in my opinion is certainly not attributable only to Autotune (i hope), but of course this procedure can cause crashes with fast ESC reflashed with SimonK or BlHeli firmware for out of sync, our have been warned.

Bests, Marco

[Phillip Smith]

I've tested the following:

All with SunnySky 4108-380kv and 12-15" prop

-RC Timer 20A with SimonK -> SimonK loses sync, Wii-ESC maintains sync

-FunFly 40A -> tested stock loses sync

-Turnigy Multistar 30A -> Loses sync with SK, not compatible with Wii-ESC

-Favourite Quattro 30A (quad ESC) -> tested stock, loses sync

-HobbyKing F-30A -> SimonK loses sync, Wii-ESC maintains sync (has Complementary PWM, similar to Active Freewheeling.  This is now my flying setup.  Have not tested AT yet)

Phillip

--

You received this message because you are subscribed to the Google Groups "drones-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com.

[Marco Robustini]

Thanks Phillip for you feedback, but i think you use max 4s LiPo, right?
The "out of sync" problem is more pronounced with 5s or 6s.

Cheers, Marco

[Phillip Smith]

Marco,

Yeah, 4S is my max.

I have some ESC32 clones on the way (should be here tomorrow).  I'm really interested in seeing how they do.  But they are a 5S limit, so they may not help you either.

Phillip

[Robert Lefebvre]

How are you extrapolating the peak current?

To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com.

[Meier Lorenz]

Hi Robert,

Based on the graphs of a typical 10“ 1000 kV setup I shared earlier by looking at the 50% throttle point and the peak current at peak throttle (I do account for if someone enters hover at 30%, but I’m not compensating for the slight nonlinearity of the curve between the 30% and 50% throttle points). It won’t be very accurate therefore, but it will be in the right ball park in terms of FET specs.

Cheers,
Lorenz

------------------------------------------------------
Lorenz Meier
Computer Vision and Geometry Group
Institute for Visual Computing
ETH Zurich
http://www.inf.ethz.ch/personal/lomeier/

Am 17.01.2014 um 15:23 schrieb Robert Lefebvre <robert....@gmail.com<mailto:robert....@gmail.com>>:

How are you extrapolating the peak current?

On 17 January 2014 01:17, Meier Lorenz <l...@inf.ethz.ch<mailto:l...@inf.ethz.ch>> wrote:
Thanks!

Updated reports here:
https://docs.google.com/spreadsheet/pub?key=0AqmuaMYa3w8BdDdjS1gzN2R4ODlzeXYta2RTMlBjMXc&output=html

Still consistent with what we expected. 25A peak + short term burst capacity plus active free running would cover the range we came across here still quite well.

-Lorenz

------------------------------------------------------
Lorenz Meier
Computer Vision and Geometry Group
Institute for Visual Computing
ETH Zurich
http://www.inf.ethz.ch/personal/lomeier/

Am 17.01.2014 um 00:13 schrieb Marco Robustini <robusti...@gmail.com<mailto:robusti...@gmail.com><mailto:robusti...@gmail.com<mailto:robusti...@gmail.com>>>:

Damn, 20 seconds * 9 = 3 minutes, where i can find all this time? :-)
Just for fun... i compile now!

Marco

On Thursday, January 16, 2014 9:35:06 PM UTC+1, Lorenz Meier wrote:
Hi all,

I got some first results. I’ve used the curve below as baseline to extrapolate the numbers in this spread sheet. Its all just ballpark accuracy (who claims the full hour flight time? I’d love to see close-ups of that build).

I also would appreciate if more people could enter data into the form - takes 20 seconds
https://docs.google.com/forms/d/1SWySjmV2n1OtNj4uASEVMMRgLZT_l8Z_XaTypJhN94Q/viewform#start=openform

It does support my hypothesis though - we’re talking about an average current of 3-6A per motor (compare to the plot below which is representative for a 10“ quad), and a peak load of mostly 12-15, some up to 22A. I forgot to ask the weight in the survey, but these are not massive quads. I would appreciate if more of the „big monster“ owners could enter some data.
Non-surprisingly the worst peak current is a low-kV motor with a 3S battery.

-Lorenz


--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com<mailto:drones-discuss%2Bunsu...@googlegroups.com><mailto:drones-discus...@googlegroups.com<mailto:drones-discuss%2Bunsu...@googlegroups.com>>.

For more options, visit https://groups.google.com/groups/opt_out.

--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com<mailto:drones-discuss%2Bunsu...@googlegroups.com>.

[Cornel Fudulu]

Lorenz,

 

If you're planning production of an ESC, 25A sustained is a good choice. I've rarely seen usual motor/prop combinations that pull more then 30A peak, and that was only on bench. AXi 2814 went to 32A with a 13x6.5 prop. Then once airborne, it would idle around 7A and peak at 23.8A in serious maneouvering, on an octo with 5 Kg payload ( 2x 2.5l waterfilled plastic bottles). So they'd be OK even for people futzing around with 5Dmark3s on gimbals. Only the Beerlift contenders will have to look elsewhere :)

 

@Randy: my current sensor is not linked to the APM, it's part of my FrSky telemetry, so changing the hall sensor would mean rewriting the OpenTX to double the current it shows.

 

That being said, the postman laid at my door an envelope with some ESC32s today, so I'm eager to play with them tomorrow. Another set of PIDs, yay !!!

 

Cheers,

Para.

 

[Phillip Smith]

I just wanted to follow up that I got 4x of the ESC32 clones.  They are pretty impressive.  They have a full CLI interface for configuration/tweaking.  I was easily able to eliminate the sync issues by setting a sane max-current and the correct poles/advance timing.  So for those of you out there with 5S or less, these are a viable option for pancake motors.

--

You received this message because you are subscribed to the Google Groups "drones-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com.

[bn...@adinet.com.uy]

Hi, where did you get the clones? I am unable to find them

[Phillip Smith]

Got them here: http://witespyquad.gostorego.com/speed-controllers/rtf32.html

He is slow to ship, took about 10 days coming from FL (USA).

Phillip

On Wed, Jan 22, 2014 at 9:29 PM, <bn...@adinet.com.uy> wrote:

Hi, where did you get the clones? I am unable to find them

--

[Randy Mackay]

 

     Just remember that as with clones of 3dr products, buying these does not support the original creators who obviously put a lot of work into them.

 

-Randy

 

From: drones-...@googlegroups.com [mailto:drones-...@googlegroups.com] On Behalf Of Phillip Smith

Sent: January 23, 2014 11:33 AM
To: drones-...@googlegroups.com

[Robert Lefebvre]

Hmmm... yeah, that's a bummer.  I didn't even know anybody was cloning them.

[Phillip Smith]

Randy,

Just remember that when you open-source, a product, tested and improved by a community for free, you know and understand that anyone else can produce the product for possibly less, cutting you out of the deal.  I'd be happy to purchase the originals if they were not 2.5x as expensive and were not in Germany.  Also, you yourself are quoted as saying: "Good to see that they *can* be made cheaply at least" because I'm guessing you recognize that at $52.00 each they are overpriced especially when the R&D and testing was all done for free by the community under GPL.

Is this even the place to discuss our personal beliefs on the topic?

Phillip

[Robert Lefebvre]

*If* the R&D is actually done _for free_ by the community.  It's not always simply the case.

[Phillip Smith]

Robert,

In any case, by releasing a product even under the guise of GPL you are freeing yourself of any form of obligated warranty or customer support because you can easily hand those responsibilities off to the community or simply claim that, well, thats what you get with "free".   With that you rightfully assume the risk of someone else cloning and somehow profiting off your work.

If any one has an issue with any of that they should patent the work, keep it secret, and accept the risks going that route as well.

In any case, I don't think the politics of GPL, cloning, etc are within the scope of this thread and the guilt-trip above was unwarranted.

Phillip

[Phillip Smith]

Happy to report with my calibrated ESC32 clones I was able to successfully and without error complete autotune.  The results are excellent.  The sync issues with my 22-pole pancake motors are gone.

[Randy Mackay]

 

     Excellent news.  Hopes are climbing that these smart ESCs are really going to help with the mechanical failures.

[Robert Lefebvre]

Has anybody tried these?  I wonder if they really do active freewheeling?  I was reading the description, and it sounds to me more like they're doing regenerative braking.  It talks about being able to slow down the propeller faster, generating high power spikes, etc.

[Phillip Smith]

Before even testing this, what are the ramifications with the APM and the power module?  Is the power module prepared to handle large voltage spikes going backwards through it?  I assume it is unidrectional?

[Holger Steinhaus]

I am doing this since more than 10h of air time, have seen up to -5A on the data logger (log below, sampled @ 10 Hz). The voltage effect is not too different from normal unloading. I use the 3DR power module in this copter.

Holger

[Robert Lefebvre]

I don't think the PM would have any problem with back-current, since it's just a simple shunt resistor.  I would be more concerned about the voltage spikes. 

So I guess not much experience with these?  

I'm looking at building an affordable, practical, long-duration, high quality quad.  And I'm looking for the right ESC to do it.  So far choices are:

Tarot 650 folding frame

iFlight Multi Mate 4108 370 370kV motors

Tarot 17" CF props

2x 4S 5000 batteries should fly for ~40 minutes?

I just need to figure out the ESC as I do not want any sync problems.  And active freewheeling could return some efficiency.

[john...@gmail.com]

My go to ESC is the Turnigy TY-P1 25A. 25A is closest thing to a universal size that fit's most frame sizes. And since it uses quality FET's with minimum leakage and high burst rates it good for heavy lifters also. Also compatible with SimonK.

Largest prop I drive is a 15" using HengLi pancake motors (also recommended). SimonK problems with pancake motor sync is mostly because everybody is using default settings. If you recompile it should work with "any" setup just like when you configure the ESC32.

I've literally bought a bag of them since I use them for everything multicopter related.

- JAB

[Gary McCray]

Hi Robert, 

Just a thought, 17" props on that frame are absolute max - probably longest endurance, but a little wobbly possibly.

15" or 16" might be better for photo/video use and 15" is more normal for those motors.

15" real CF Gemfans are on really great sale for $15.00 a pair:

http://www.shop.aglhobbiesllc.com/Motors-ESCs-and-Props/Gemfan-Full-CF-15x5-5-CW-CCW-Pair.html

They also have a nice 650 quad frame.

Of course they aren't in Canada :-(

Best Regards,

Gary

[Robert Lefebvre]

Yeah, I wonder how wobbly it would be.  The use case for this machine, it might be fine though.  One of the primary roles would be a "flying repeater".  A signal relay at 300 feet.  In that case, all that really matters is duration and reliability.  I could also consider using it for some FPV where rock-solid stability isn't a big deal.  If it worked out well enough I would consider it for some mapping or AP use, but that's not the primary role.

[Marco Robustini]

Hi all, today I have done many tests with different SimonK firmware release, starting with the famous "12/09/30" until the most recent "2013/09/20".
The tests were carried out with the Hobbyking F40-A ESC (with BEFM CAPS installed), "DJI 4114-11, 320 Kv" motors with "RC Timer 15x5 Carbon Propeller", here some images of this big bird.
With all the releases I've played a lot with COMP_PWM and MOTOR_ADVANCE, but with I've always had problems with "out of sync", the motors emit the classic "jerky sounds" at high revs and with fast thottle changes, I put the throttle control on a 3 positions switch so i can have "0% - 40% - 100%" of throttle output in a flash.
The only version that has not given me sync problems is "12/09/30", where the only abnormal "know" behavior is at the motor start, the propeller tends to bounce sometimes, depending on where you stopped previously.
In my opinion in later versions we tried to remedy some problems but introducing others, perhaps also due to the excessive load on the Atmel chip, if anyone wants to test I'd like to compare my results with yours.
So for now I stay with the old release, perhaps not reliable on the sync recovery but imho perfectly stable, or rather trouble-free sync with my setup.

[Ben H]

Figured I'd chime in here with my experience.  Running genuine ESC32's on 480kv motors and 5s on a hexa, tried doing autotune after getting it dialed in manually a little and it fell out of the sky just like Marco's post.  I'll try to get some logs up later.

[Marco Robustini]

Hi Ben, interesting, can you provide a flashlog of this event?
ESC32's i think they are more reliable than ESC reflashed with SimonK, then this is a very strange thing.

Cheers,

Marco

[Phillip Smith]

Ben,

Did you run an ESC calibration successfully before trying autotune?

Phillip

--

[Ben H]

If you're talking about the qgc esc32 calibration then yes, I did both the current limiter and feedforward calibrations.

[Phillip Smith]

I think you should try a slightly more conservative initial amperage when running the current limiter calibration.

Do you have a copy of your ESC32 config?

Phillip

[Fourfingers]

Hi all,
Unfortunately I an not running Apm, but I am running 6s 16Ah with pancake motors in a heavy (9.5 to 12kg depending on payload) octo with no sync issues(touch wood).
Since this is my first multirotor, 8O, I tried to be on the safe side ESCwise. I opted for redbricks50Aopto and flashed them as Marco with sk 2012 12 30, no mods (Read quite a couple of bad notes on these ESCs, but, up to now they are running fine with my setup). Cables are run short from Esc to motor.
Motors are sunnysky4112 320kv(22n24p) 15x5 or 16x5 eating each in hover aprox 130w(5.8A 22.4v) with an AUW9.5Kg and around 200W(9A 22.4v) AUW12Kg. Maxing this bat-motor setup will take around 300W per motor.
Since I started reading about sync issues, a couple of days ago, I have bench tested some other pancakes with this ESC-firmware combo and have not been able to get any of them out of sync.
Hope everyone get their gear running safely.

Cheers

@robert lefevre, Been looking at those YEP esc a while back ago when planning on 12s, might buy one to test and let you know if they are any good.

[Adam Orens]

Hi there,

 

Had a motor sync issues as well on my large hex running Maytech Simonk 40A-Opto ESC's and T-motor MT3515-15 400KV, with a 6s battery and 13 -6.5 Props.

 

You would figure that you could tune Mission Planner for the motor size and eliminate the ESC's stalling the motors. 

 

There are sooooo many options for ESC's out there. 

 

Do you think these ESC's your talking about would work for me?  Im in fear of flying my rig till I figure this out.

 

Thanks!!

[Adam Orens]

Found a possible fix for the low KV Motor problem!!!!!! 

 

Im running a similar setup and had a motor stall from the APM asking too much of the ESC's and in the end, the motor stalled. 

 

Ends up there is a function in the full parameter list called RC_FEEL_RP - You can set this for a softer response to stick input commands. 

 

On the Throttle side, I set the radio for a small throttle delay incase I throw all the power in at some point.  Its barely noticeable to me, but the motors don't knock at all anymore.

 

Besides having to tune by hand and NOT using Auto Tune.... That seems to have fixed the issue.  

[Jason Short]

Yeah, the RC feel is a small low pass filter used to keep the tune sharp, but the user input soft. That way you can pilot a camera ship smoothly without letting the wind get the best of you.

Jason 

[Leonard Hall]

The dc feel parameter doesn't fix the problem, it does make it hard or even impossible for the pilot to cause the problem. However the problem could still be caused by flying in turbulence.

The basic truth here is if auto tune can cause the problem, the you can repeat the problem in normal flight, provided you are using the same PID gains.

The only way to prevent this problem is to keep all the gains low enough to make the esc throttle response too slow to ever cause the problem for that esc/motor/prop combination. Anything less is just flying arround the problem.

The basic problem is people are using Simon K on big copters where it isn't needed because the're is a perception it is better. For large copters Simon K like not using lock tight on your motor mounts.

Where Simon K has real value is small copters using small props. Here the reduced latency can really improve performance.

--
You received this message because you are subscribed to a topic in the Google Groups "drones-discuss" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/drones-discuss/tq8l52cwGXg/unsubscribe.
To unsubscribe from this group and all its topics, send an email to drones-discus...@googlegroups.com.

[Jonathan Challinger]

What if we add a throttle slew for each individual motor? That would limit acceleration of the prop and might run on apm2.

--

You received this message because you are subscribed to the Google Groups "drones-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com.

[Leonard Hall]

Yes it would. But all that is doing is reversing what the Simon K firmware does.

Simon K basicly removes the input filter from the esc. What you are suggesting is to add that back into the arducopter code.

[Jonathan Challinger]

But as a parameter, controlled by us.

[Leonard Hall]

Yes I agree with you. I just don't like adding tuning parameters and complexity to the arducopter code because people are purchasing and using hardware that is not appropriate to the task and a safety hazard.

As an example there would be equal cause to limit the maximum throttle response because people wanted to drive motor/prop combinations that can draw 30 amps with 10 amp esc's. I use this example because this is what we are doing.

I personally like that Auto tune causes these problems because I would prefer to see the copter crash during tuning in an isolated field from a known issue than while flying over a wedding reception.

[Josh Welsh]

Heh.. nice call Leonard.

 

In case no one gets the reference… https://www.youtube.com/watch?v=ocqB6_y71xE

 

Don’t be that guy.

 

 

From: drones-...@googlegroups.com [mailto:drones-...@googlegroups.com] On Behalf Of Leonard Hall

Sent: Wednesday, March 5, 2014 10:14 PM
To: drones-...@googlegroups.com

[Marco Robustini]

RC_FEEL_RP it doesn't solve the sync problems with low kv motors because APM Copter with automatic function remains bound to the PID.
Often I've motors sync problems during certain maneuvers very pushing, or when abort a mission engaging RTL while the quad goes in the opposite direction to the home, then forced to a screeching halt and back again.
Our code does not include a sorta of  "soft switching" between flight mode and another, and this is what often causes the problem, but also other electronic behave in this way, for example WKM is such a pain during these steps.
If we can make more soft  the switch from one flight mode to another i'm sure that there would be fewer problems, but not with Autotune, with what remains.
Only the code of ESC can fundamentally solve this problem, some as the Hobbywing FlyFly not affected because they are less driven on performance, but you also lose stability.

Marco

On Thursday, March 6, 2014 3:49:09 AM UTC+1, Adam Orens wrote:

Found a possible fix for the low KV Motor problem!!!!!! 

 

Im running a similar setup and had a motor stall from the APM asking too much of the ESC's and in the end, the motor stalled. 

 

Ends up there is a function in the full parameter list called RC_FEEL_RP - You can set this for a softer response to stick input commands. 

 

On the Throttle side, I set the radio for a small throttle delay incase I throw all the power in at some point.  Its barely noticeable to me, but the motors don't knock at all anymore.

 

Besides having to tune by hand and NOT using Auto Tune.... That seems to have fixed the issue.  

 

 

On Saturday, January 11, 2014 10:11:53 PM UTC-10, Marco Robustini wrote:

Yesterday I run Autotune on two of my heavy hexa, installing powerful motors with large carbon propellers 15x5, F40A ESC flashed with SimonK firmware (the latest version), working with 6s LiPo.
During the automatic tuning of the "pitch" two motors have completely lost sync and miraculously i recover the multicopter and land without damage.
Thinking it was a case I repeated the procedure with another multicopter, always very heavy (same propellers) and the problem occurred again (always during variations on the pitch), this time only one motor lost sync during one of the fast throttle inputs that Autotune produces, the quad is tilted diagonal, not to the right side.
This was always done during the step on the pitch because the multicopter is much more reactive on that side for obvious reasons of weight (battery, landing gear, etc.).
I try to find the logs and post them here as soon as possible, but the problem in my opinion is that Autotune produces too fast throttle change input during the procedure, from the "hovering" (quad in a horizontal position) to the max location where it's tilted, these too rapid accelerations, that only the code can produce but our thumbs would be almost impossible, causing possible out of sync with the SimonK firmware.
Now, I don't know how Leonard can solve the problem

...

[john...@gmail.com]

Like Leonard already said. If the APM is able to provoke a sync loss regardless of gains or how aggressive you fly, you have a propeller/motor/esc mismatch problem. Plain and simple.
There is no excuse. If you use SimonK, you choose to do so and are supposed to know what the deal is. You can even tune the problem away in SimonK, but everybody is just using default timings.

- JAB

[Robert Lefebvre]

John, do you think it's related to the timing advance?  Or when you said timing, you meant something else?

--