HobbyKing.com New Products Flash Sale
Reply
Thread Tools
Old Mar 26, 2012, 07:16 PM
Al Ducharme
photronix's Avatar
Orlando, FL
Joined Jan 2005
735 Posts
Quote:
Originally Posted by Arx View Post
While that's probably true, there are some differences between the reflashed and stock ESCs, and it seems pretty hasty to just dismiss the reflashed ones as no better than stock. If it were that simple, Simon would be kind of a dumbass to be wasting his time writing firmware.

There are differences that go beyond what the test you're doing can resolve, and some of them are quite evident in peoples empirical testing (Definitely not just their imagination).

That's not to say that your measurements are wrong, or useless, just that it's not quite as simple as your original post implies.

Keep up the good work on this stuff, just don't throw away everyone else's testing and experiences as invalid.

With that test with the bouncing motors, I believe the "throttle range" was pretty much just controlled by a servo, I interpreted that bouncing as oscillation due to the lag. Sure, the gain might just need to be tweaked downward to stabilize it as well, but that is undeniably a useful result. That said, I don't really understand what the deal was with the 3rd one.
What I would really like to do is find a solid repeatable test that allows us all to drive the ESC vendors.

I don't disagree that the video is interesting but I would like to figure out a way to measure and document what is really happening.

I agree with your previous comments about TF by definition being applied only to LTI systems. It almost never is actually, typcially assumptions have to be made concerning the regimes where the measurement is linear. I chose 25% based on many other tests. I have done the tests for other values such as 5% and 10%. I'll post more results.

What I would really like to do is find a good test case, two ESCs to compare. I'll measure it with my test stand and others can also measure the same two. Then let's all decide.

My intuition is that the momentum of the propeller limits realistic update to about 50Hz with very very minor corrections up to 100Hz. Changes in RPM greater than 10% can only occur lower than 30Hz.

I don't disagree with comments like "well it just flies better with X brand ESC" I disagree with "feel" being representative of fact which seems to be the prevailing sentiment.
photronix is offline Find More Posts by photronix
Reply With Quote
Sign up now
to remove ads between posts
Old Mar 26, 2012, 07:41 PM
Destroyer of G-10
askman's Avatar
United States, OR
Joined Jul 2004
10,085 Posts
if you look at the KK's test, biggest thing we do notice about the esc power ramp is this.

turnigy esc takes about 8 clock cycle to go from 0 to full throttle. which means for your 25% change, it take 3 cycles to effect full change. since the cycle is totally depended upon frequency, at 50 hz, 3 cyc is 60ms while at 400hz, it is 7.5ms. anything less than about 10% throttle change, it will happen in 1 cycle. Now, I believe simonk's firmware actually take this ramp out, which means power change is much quicker. I believe this is the biggest reason for better performance.
askman is offline Find More Posts by askman
RCG Plus Member
Old Mar 26, 2012, 08:44 PM
Tri-Quad-Hexa-Octo-copters!!
United States, TX, San Antonio
Joined Feb 2007
14,668 Posts
Gents,
There have been some good ESCs tests discussed and results shown here without words like "feel" and "intuition". However even after all the engineering calculations, graphs, 3d level analysis have been done and the metrics presented, it is the operational flight test that proves and is the final analysis of the outcome of the data. No doubt we eventually can improve these tests to match operational reality

@Al,
Your graphs are worthy, and it is great that you are providing some data. WE can use all we can get The only bit I would question is your observation and test result of the RCTimer ESCs which perform the same as the OEM Turnighy Plush's in all my multicopters and is frequently used successfully by many other multicopter flyers. Of course both perform even better with SimonK firmware as evidenced by being able to increase the gyro proportional by several points and by noticeably better performance in winds greater than 10kts.

From the ESC recommendations and converting ESCs to I2c and to faster PWM section in the Quadrocopter and Tricopter Info Mega Link Index as reported by successful multicopter builders:

--TIP 10: ESC recommendations for standard PWM: Bidproduct HW25a (www.bidproduct.com), eRC (http://www.hobby-lobby.com/erc_speed...ls_470_ctg.htm), DYS ESCs, ExceedRC Volcano/Proton series, HKSuperSimple (the programmable type not the basic), HK Blue Series, HobbyWing, Mystery Blue series, www.quadroufo.com 30a, Maytech Harrier-Supreme Series(these are the same as HW Blue series at 3x the cost) http://montorc.com/MT30ABEC.aspx, RCTimer (www.rctimer.com), Towerpro, Turnigy Plush, YGE, ZTW. *Note: The following ESCs are known to be poor performers when used on multicopter's due to some or all of the following issues: the lack of manual endpoint calibration and the use of automatic endpoint calibration, timing lags due to throttle smoothing, firmware that is optimized for helicopter use is not compatible with multicopter flight controllers, low update rates, non-linear throttle curves. The ESC's with those features will cause multicopters to have drifting and oscillation issues. Again the following have been tested and will not perform well on multicopters and are therefore not recommended for multicopter use: Castle Creations, Dynam Denton. Hercules, Scorpion Commander, Thunderbird.

Cheers,
Jim
jesolins is offline Find More Posts by jesolins
Last edited by jesolins; Mar 26, 2012 at 10:33 PM.
Reply With Quote
Old Mar 26, 2012, 09:58 PM
Arx
Registered User
Joined Mar 2012
18 Posts
Quote:
Originally Posted by photronix View Post
What I would really like to do is find a solid
repeatable test that allows us all to drive the ESC vendors.
Yes, this could help, but there's a certain amount of matching involved that
will probably prevent a single test from telling you everything. Someone
with a very lightweight copter that's going for high speed stunt flying is
probably going to be needing different things than someone who wants to lug
a heavy SLR. There are things that can certainly be improved which will
help across the board, but there's others that will probably help some
cases, and hurt others.

Quote:
I don't disagree that the video is interesting but I would like to figure
out a way to measure and document what is really happening.
Yes, It would be good to see a timing comparison of the output PWM vs the
input i2c/ppm. Unfortunately, not everyone has the equipment or knowledge
necessary, and misinterpretation of a single result can lead to a lot of
"This one is best" type things, which don't usually tell the whole story.
Quote:
I agree with your previous comments about TF by definition being applied
only to LTI systems. It almost never is actually, typcially assumptions
have to be made concerning the regimes where the measurement is linear. I
chose 25% based on many other tests. I have done the tests for other values
such as 5% and 10%. I'll post more results.
Yes, when you're testing the limits in that fashion, you will definitely run
into "attenuation". The problem is that depending on the type of slew
limiting that is being done, your results aren't necessarily going to say
anything that is useful. You'll get a good idea of the control bandwidth,
but you run into the point where it becomes non-linear, and that's where
things can easily get misinterpreted. It may be more important to know how
well an esc responds to hitting its limits, as opposed to just how high
those limits are. You may get one that seems to respond really well with
the prop, and conditions you're running it in, but which may behave
differently under different load, or different input conditions (non
sinusoidal)

Quote:
What I would really like to do is find a good test case, two ESCs to
compare. I'll measure it with my test stand and others can also measure the
same two. Then let's all decide.
Yeah. It would be cool, but I think there will need to be a suite of
standard tests for this to be useful.
Quote:
My intuition is that the momentum of the propeller limits realistic update
to about 50Hz with very very minor corrections up to 100Hz. Changes in RPM
greater than 10% can only occur lower than 30Hz.
That may well be true. I guess it depends on the dynamic range of the PWM.
There may actually be a high speed "dithering" occuring between a couple
steps of pwm which might turn into a slower vibration at lower speeds.

That's just hypothesis on my part though. I haven't tested it at all to see
if it's an issue or not.

Quote:
I don't disagree with comments like "well it just flies better with X brand
ESC" I disagree with "feel" being representative of fact which seems to be
the prevailing sentiment.
Yeah. A lot of people don't understand enough to know why it "feels
better", they just know that they have an easier time flying it, etc. In
some cases people exaggerate, but if many people are saying it works way
better for them, It would probably be a good idea to try and figure out why.
I doubt it's a shared delusion, and the few before/after videos I've seen
seem to show some definite improvement.
Arx is offline Find More Posts by Arx
Reply With Quote
Old Mar 26, 2012, 11:54 PM
Registered User
Australia, QLD, Brisbane
Joined Nov 2011
949 Posts
Quote:
Originally Posted by askman View Post
if you look at the KK's test, biggest thing we do notice about the esc power ramp is this.

turnigy esc takes about 8 clock cycle to go from 0 to full throttle. which means for your 25% change, it take 3 cycles to effect full change. since the cycle is totally depended upon frequency, at 50 hz, 3 cyc is 60ms while at 400hz, it is 7.5ms. anything less than about 10% throttle change, it will happen in 1 cycle. Now, I believe simonk's firmware actually take this ramp out, which means power change is much quicker. I believe this is the biggest reason for better performance.
Just because the ESC does it's speed change in 7.5ms or less if modified, doesn't mean the motor and prop are going to do theirs. There is not an infinite supply of current to accelerate the motor/prop instantly by 25%. You need to measure how fast a given motor prop combination can accelerate/decelerate. I would be surprised if would keep up with an ESC.
Erknie is offline Find More Posts by Erknie
Reply With Quote
Old Mar 27, 2012, 02:04 AM
Arx
Registered User
Joined Mar 2012
18 Posts
Quote:
Originally Posted by askman View Post
if you look at the KK's test, biggest thing we do notice about the esc power ramp is this.

turnigy esc takes about 8 clock cycle to go from 0 to full throttle. which means for your 25% change, it take 3 cycles to effect full change. since the cycle is totally depended upon frequency, at 50 hz, 3 cyc is 60ms while at 400hz, it is 7.5ms. anything less than about 10% throttle change, it will happen in 1 cycle. Now, I believe simonk's firmware actually take this ramp out, which means power change is much quicker. I believe this is the biggest reason for better performance.
I'm pretty sure the ramp isn't completely gone. I think he's just using a different algorithm for it.
Arx is offline Find More Posts by Arx
Reply With Quote
Old Mar 27, 2012, 03:50 AM
let's brm together!
Czech Republic, Prague
Joined Nov 2011
390 Posts
I'm pretty sure there will be "some" improvement in removing the ESC average filtering. But then you have a rather large momentum of motor and prop.
So in the end it will be like improving car performance by driving naked - it will improve it, you can measure, but I doubt it's worth it.
BUT if there were some vendor (like Turnigy) who would do "Multicopter line" of ESCs for the same buck as the regular ones with SimonK firmware or just a regular firmware with faster reaction - they would get my money too, of course. Because what if...
zviratko is offline Find More Posts by zviratko
Reply With Quote
Old Mar 27, 2012, 04:14 AM
Registered User
Australia, QLD, Brisbane
Joined Nov 2011
949 Posts
A human eye can't even see oscillation much faster than 25Hz, most pilots notice a multicopter oscillating when its a few Hz, and ignore it when it's much higher. When you have people applying things like gyro LPF set to 42Hz to remove noise, you have to wonder where the input to drive these these 400Hz ESC micro corrections is coming from. I mean it's not the pilot, not the gyros, so where is the stimulus at least 400 times a second that requires the FC to make ESC adjustments that often? I just don't see a viable feedback loop to justify it, perhaps I am missing something.
Erknie is offline Find More Posts by Erknie
Last edited by Erknie; Mar 27, 2012 at 05:56 AM.
Reply With Quote
Old Mar 27, 2012, 06:10 AM
let's brm together!
Czech Republic, Prague
Joined Nov 2011
390 Posts
Erknie: exactly what I was wondering - Multiwii race for cycle time looks silly to me - you want a smooth flight with fast reaction to user input, not copter correcting at 400Hz.Position change vector and factors influencing position (like wind, CG, different motor RPMs) are not changing at xx Hz either so trying to level it hundreds of times per second with strong changes in ESC inputs makes no sense.

BUT! On the other hand, I am using Turnigy Plush with my copter, and when I tried other ones I was _not_ having that good time. I can't put my finger on the ESCs, but I think they were very contributing factor to the overral feel (ability to use much higher PIDs with Turnigy ESCs and less filtering on gyro was one symptom).

But even if ESCs react in an instant, the props don't because you don't have "infinite power" to change the rotation - there's always momentum with "change dampening" properties orders of magnitute greater than any SW filtering in ESC will do. (And it's "analogue" )
zviratko is offline Find More Posts by zviratko
Reply With Quote
Old Mar 27, 2012, 06:51 AM
Registered User
Australia, NSW, Wollongong
Joined Aug 2011
310 Posts
for me its easy to see that simonk flashed escs are heaps better.

I had turnigy plush on all of my quads/octo's/hex's. They all flew great and I did and still do love turnigy plush, although hey had the usual slight wobble when descending etc and just weren't perfectly, silky smooth.

I started off just buying 4 simonk escs for a quad. I did a direct replacement, did not change any gains or anything else. When I launched it for the first time I was convinced straight away. It flew up smooth as silk, hovered perfectly, descended without wobble.

So I ordered more, put them on my octo and same result. With gps, wookong m is so much smoother, it fights wind better etc.

Best way i can describe it is lthat the escs are like a 1/8 second quicker, instead of responding to wind from the right by banking left and then smoothing that motion by banking right and back and forth which gives a seemingly steady motion but a slight wobby, the simonk's react that much faster so it seems to be a perfectly smooth reaction and counters that wobble.

If you doubt it, it's so easy to try. Just order 4 x 30A DYS esc from quadroufo.com @ $18 ea and try it on a quad yourself!

Just to clarify I have no technical/engineering experience. I'm just stating what I've seen from so much trial and error with these beautiful machines
slidins15 is offline Find More Posts by slidins15
Reply With Quote
Old Mar 27, 2012, 06:59 AM
Registered User
Australia, QLD, Brisbane
Joined Nov 2011
949 Posts
Quote:
Originally Posted by slidins15 View Post
for me its easy to see that simonk flashed escs are heaps better.

I had turnigy plush on all of my quads/octo's/hex's. They all flew great and I did and still do love turnigy plush, although hey had the usual slight wobble when descending etc and just weren't perfectly, silky smooth.

I started off just buying 4 simonk escs for a quad. I did a direct replacement, did not change any gains or anything else. When I launched it for the first time I was convinced straight away. It flew up smooth as silk, hovered perfectly, descended without wobble.

So I ordered more, put them on my octo and same result. With gps, wookong m is so much smoother, it fights wind better etc.

Best way i can describe it is lthat the escs are like a 1/8 second quicker, instead of responding to wind from the right by banking left and then smoothing that motion by banking right and back and forth which gives a seemingly steady motion but a slight wobby, the simonk's react that much faster so it seems to be a perfectly smooth reaction and counters that wobble.

If you doubt it, it's so easy to try. Just order 4 x 30A DYS esc from quadroufo.com @ $18 ea and try it on a quad yourself!

Just to clarify I have no technical/engineering experience. I'm just stating what I've seen from so much trial and error with these beautiful machines
I have read lots of anecdotal evidence that flashed firmware is quicker, but I am looking for scientific evidence as to which actual changes in the firmware make people perceive it's as quicker.
Erknie is offline Find More Posts by Erknie
Reply With Quote
Old Mar 27, 2012, 07:04 AM
Registered User
MIRV's Avatar
United States, NM, Las Cruces
Joined Jun 2011
185 Posts
Quote:
Originally Posted by KC Flyer View Post
I just received 5 of these HK SS 18-20a Card Programmable ESCs which I plan to reflash with SimonK's tgy.hex firmware.

I probably wouldn't have considered it but read about n3m1s1s's problems with these HK SS 18-20a Non Card Programmable ESCs. It seems that the firmware recently changed and now they release the magic smoke when connected to some flight controllers (not sure what he is using).

Then I read Warthox's recent bad experiences with his favorite Hobbywing ESCs where they recently switched to a SIL chip.

So maybe there is not much of a difference between "good" stock firmware on a HK ESC and SimonK's.

But...I don't really trust what fw version they loaded at the HK factory. Also, I understand that the motors run quieter (less whine) with SimonKs fw and most report more stability after reflashing.
Does Turnigy make these as well? HK ss 25/30 And is there an update flash for them? Would it be worth the flash? I've only seen ESC flashing mentioned a time or two, maybe someone could direct me to the relevant discussion(s)? Thanks!
MIRV is offline Find More Posts by MIRV
Reply With Quote
Old Mar 27, 2012, 07:54 AM
Suspended Account
Joined Dec 2010
11,764 Posts
Not hard: http://www.rcgroups.com/forums/showthread.php?t=1513678
timecop is offline Find More Posts by timecop
Reply With Quote  (Disabled)
Old Mar 27, 2012, 08:39 AM
Registered User
MIRV's Avatar
United States, NM, Las Cruces
Joined Jun 2011
185 Posts
Quote:
Originally Posted by timecop View Post
Thank you. Almost 200 pages to read. Oh well, its windy season here, anyway...
MIRV is offline Find More Posts by MIRV
Reply With Quote
Old Mar 27, 2012, 10:00 AM
Tri-Quad-Hexa-Octo-copters!!
United States, TX, San Antonio
Joined Feb 2007
14,668 Posts
M,
Nek has done a fine job indexing and continuously updating all the info you need in his first post of his ESC firmware flashing thread. No need to read and travel all the road it took to get there Read the first post and then go and read the last page for any updates or to ask questions.
Cheers,
Jim
Quote:
Originally Posted by MIRV View Post
Thank you. Almost 200 pages to read. Oh well, its windy season here, anyway...
jesolins is offline Find More Posts by jesolins
Reply With Quote
Reply


Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
Mini-Review BD5B or BD-5B Comparison of Aeroworks vs Hobby King FRENCHSTAN Electric Plane Talk 4 Jul 01, 2012 05:14 PM
Mini-Review 3 DVRs comparison (FPV-Japan, Foxtech D1, Wireless DVR 5.8GHz) sharonx FPV Talk 8 May 23, 2012 11:31 PM
Discussion X-Air FC1212-P / DJI Wook Kong WK-M Comparison dansparks Multirotor Talk 6 Mar 02, 2012 10:08 PM
Discussion Upgrade ESC instead of engine AND ESC. Dynasti Electric Plane Talk 15 Sep 23, 2011 10:42 AM
Wanted hobbi co. NexSTAR trainer rtf paintballer#1 Aircraft - Fuel - Airplanes (FS/W) 8 Mar 21, 2006 01:44 AM