Access

Here is a video that demonstrates cogging (or lack thereof) in three different BLDC setups. The first is a novak GoaT crawler which handles super-low speeds and doesn't cog at all. The second is the mamba max E-firestorm which can still go pretty slow and barely cogs in any noticeable fashion. The third is the rustler VXL which really can't be driven in any controllable fashion below ~20cm / sec and, if induced, will cog like crazy. You don't have to take my word for it, the (otherwise boring) video pretty much says it all. Past the end of the video the VXL ESC locked up and had to be power cycled to recover. Blinking red light and not throttle at all. Traxxas really needs to work on their firmware a bit for the VXL. As for the mamba max, people who have problems with cogging just need to set it up correctly. As you can see, it barely cogs at all, and can be reasonably controlled even at super-slow speeds.

http://www.youtube.com/watch?v=pKdRFDuyoVo

Always Dreamin

well, novaks shouldn't cogg atall, especially a goat. most of the cogging in sensorless motors is due to the batteries. the batteries not being able to give the motor/esc what it wants. thanks for the vid[8D]

snellemin

Good Vid man. Another proof that the mamba controller has a better firmware over the Vxl. But then again the VXL was made for a basher. I noticed you were running them with a lipo, so battery power was not an issue. The cogging is barely there with the 5700 and 7700kV mamba motors in my setups. You just have to program it properly like you said.

fcazzol

Are you sure this is cogging? I've seen definitions of cogging which do not correspond to what we see in the video.

What you showed is "stuttering" (I think that's how it's called) which is not caused by the batteries. Batteries do cause cogging however.

I'm sure some of the guys who created that stuttering vs cogging thread will respond soon to clarify the situation (ACCESS? Slo-V?)

Francois

fcazzol

I checked CC's website. I made some mistakes here!

Cogging (or stuttering): caused by rotor-magnet relative position. Batteries have no effect.
"turbo lag" (CC's terminology): lack of punch caused by batteries not being able to provide enough current early in the acceleration phase.

So, I was mistaken, what we saw was cogging indeed! ah ah I just re-learned something correctly today!

sloppyG

neat video,

but I doubt many of us drive our RC's that slow

Phelansa

I think this video illustrates the point perfectly.

Power is NOTHING without control.

Good vid Access!

Access

Yeah I think the proper lexicon is like you said. You have to differentiate between what is caused by a sensorless ESC (especially one with mediocre firmware) and the 'battery voltage dropout' that can happen with any setup where the power required is greater than the battery can provide. They are different problems and there are different solutions. In this video I tried to only demonstrate only cogging, not the dropout thing.

"Cogging" is the 'sensorless startup' problem. At rest, there is no back-EMF on the motor so getting the rotor spinning depends on the algorhythm (method) used and how well that algorhythm is implemented in firmware. Logically one would expect castle, who has been doing sensorless for many years, to be better in this respect than traxxas where the VXL is their first foray into brushless. I'm not trying to rag on the VXL, it's a solid ESC in its own right and who in all seriousness tries to drive a ST at speeds under 20cm/sec.

"Dropout" or whatever you want to call is the stuttering caused by when the ESC pulls too much current out of the battery, causing the battery voltage to drop significantly where the BEC cannot provide enough voltage to keep the receiver (and maybe even the ESC itself) functioning. Receivers typically need about 3.7V to work, and the ESC, that depends on the MOSFET and gate drivers (if it uses drivers), microcontroller, anything else inside that the ESC internally is dependent on. If either one goes undervoltage, the vehicle loses power. This can be cyclic, and it can happen at any speed, not just startup from rest. Depending on how much current drawn (multiple servos, long-range receiver, lights, etc.) the dropout of a linear regulated BEC may be anywhere from 0.5V to 2.5V, it just depends on the design and tolerances. A switching step-down BEC may have less dropout or it may have more, again depends on the BEC itself (the design used and how much current is drawn). A BEC that used a SEPIC circuit
http://en.wikipedia.org/wiki/SEPIC_converter
could in theory have negative dropout, but I don't know of anyone using one (and there are other reasons not to). So the cure for dropout right now is to make sure you use a good battery, one matched to your application, check your whole electric 'power-train' for any bad connections / connectors, and consider using an independent receiver pack or a power capacitor. But even the latter two do not gaurantee sucess, especially if the ESC is internally failing due to low voltages.

As for nobody driving that slow, I'm with you there. Unless it's a crawler, the low-speed control isn't a feature that's going sell many ESCs.

http://ww1.microchip.com/downloads/e...tes/00857a.pdf

If you want a comparison of the relative complexity of sensored vs. sensorless firmware, look at Appendix C and D of the above document.
On the PICmicro in assembly language code, sensored firmware (C) is barely even 3 pages not including the comments. The closed-loop sensorless firmware (D) is around 10 pages of assembly language neglecting the comments.

Yub, yub, cmdr!

Does Microchip really supply the procs in most of our ESCs? That's cool, because I spent a while working with Microchip's C compiler and associated libs. Could I write code for a Mamba Max?

I'm sorry, but in the Mamba truck, I cracked up every time you ran it into the wall.

mrbonez

from a basher prospective i personaly would never drive a brushless that slow but i guess thats important to some folk

Access

No, from what I can tell they are 8051-based.
Look at the IC on a pheonix-ESC, that's a silicon labs 8051-core micro.
The other IC, the Texas Intruments 50L05C is a LDO.

Slo-V Flyer

I see my name brought up here. I'm no genius or techy or even electronics guru like Access or RURC. I just go by common sense (whatever I have left) and experience.

And thanks for the vid and clarifying what "cogging" (sensorless phenomenon) and "stuttering" (voltage drop) are. I think too many people are not informed enough to tell the difference or for that matter know that they are two different issues.

Takedown

My Feedback: (23)

I still prefer the VXL over the MM.

Always Dreamin

It might just be that I never drive that slow, but I don't think my VXL coggs that much atall. In full speed starts I don't even notice it at all.

Takedown

My Feedback: (23)

IMO the MM can be turned in a nice smooth running ESC if you have just the right settings but the VXL is the smoothest running ESC out of the two and stays the coolest while running right out of the package. Thats a big plus when I look for a new ESC.[8D]

snellemin

Well it will run the coolest since it's only 3500 kv....low amp motor.

Access

On normal 'hard' starts (ie. just throttle and go) you barely notice it. It's only if you lightly touch the throttle and try to go real slow that you get that constant cogging like I was getting. That's why I used the term 'induced' for the video "3) VXL which cogs like crazy if induced". With the VXL, you are either going to be moving faster than ~25cm/sec or not at all. Whereas the mamba max seems to handle this range reasonably well. Mamba max may not be up to a true crawler system, but it's very good compared to other sensorless ESCs.

In newtonian/classical physics, it is impossible for an object to go from a speed of 0m/sec (rest) to 20m/sec without passing through every point in-between on the speed scale. When starting from rest, the VXL is going to spend at least some time in its 'dead zone'. Though this amount of time might be so close to zero (instantaneous) in most common cases you can't even perceive it. A digicam in movie mode shoots roughly 30 times a second so it's not really going to show up on a hard start.

Yub, yub, cmdr!

Do ESCs really need 25 MIPS of processing power?