Timing advance with Sidewinder esc
 

I have been wondering this for a long time, just never bothered to ask. Can a Castle Sidewinder sensorless esc advance the timing on a sensorless motor? At first thought it would be no but maybe it can. Obviously the esc is not getting any feedback from the motor but couldn't it still advance the timingly blindly by setting the amount with the program card since (I think) all esc's control when the pulses's fire?

Rich

Great question. Im interested in the answer too.

I am not familiar with Castle products specifically but I will share some of my experience with many other budget brands of sensorless ESC's. There are 2 types of timing, digital and mechanical... many sensorless ESC's have the ability to adjust the digital timing, but according to this manual here, I don't see any support from Castle for this feature:
https://dzf8vqv24eqhg.cloudfront.net...95-0314-01.pdf

Mechanical timing is controlled on the end bell of a sensored motor and requires a sensored ESC to leverage this feature, I have operated sensored motors without the sensor wire connected, which is effectively sensorless operation, but not really sure if this negated the mechanical timing or not... I typically disconnect a sensor wire for troubleshooting purposes to isolate if the wire has gone bad.

If you are just bashing around, I would not mess around with digital timing regardless, more often folks introduce more problems with overheating when they jack with the default timing settings in their ESC. I would just leave well enough alone with what you have :)

Thanks for jumping in Bill. Always like to hear your advice. From what you said some esc's can set the timing ahead without a sensor wire and others will only do it with a sensor wire. The Castle programming card seems to imply that their non-sensored esc's can do it. Best way to tell I guess is try different settings, run the car for 5 laps, and check the motor temp. If temp changes then the esc is probably changing the timing.

Rich

so long as you have a temp gun and keep a close eye on your temps, then you will do just fine... good luck with the experiment!

I looked at the manual for the sidewinder. It has electronic timing adjustment but its not a value its a description. Low, medium and high. So lets talk about sensorless commutation or timing. Lets say you have a merry go round and your job is to push it slow, fast, hard, efficiently. There is 6 bars spread out evenly that are each only big enough for your 2 hands to spin it. 6 people that you are spinning (riders) and telling you when to push because you are blindfolded. If you are late on the push one hand is off the bar and you get a half push.

your riders can say get ready, push, and rest. As they go faster its harder for them to consistently say what they need to tell you. As you push faster it gets harder for you to time the push. And the guy sitting at the bar you are going to push has to stay silent when you are going to push his bar. Only the other 5 can speak.

you can adjust timing mechanically and electronically. Manual timing is like a global delay or pre warning by a fixed amount. If you push is always off by .1 degrees then timing will help with that. If you want a little more power then push the merry go round a little sooner so that when you get up to a higher speed your timing will have a period of time where both hands are being timed properly to push the motor along but at the expense of bad pushes at low speed and really high speed.

so with 5 people yelling at you about when to push and they arent evenly spaced how much resolution could you have with electronic timing? I wont bore you with the math but there isnt much resolution, linearity, or repeatability so better or more timing increments would be really great.

Just for clarity, we add sensors. Now you have blinders on and can see the rail approaching, when its in the middle, and when its leaving. 3 snapshots to help you get you timing right. But lets not forget that you still have 5 people yelling at you which creates allot more data to time your push even closer.

you get better at predicting the the differences in angle between the push bars and have twice as much data.

dont take my comments as saying sensorless tech or even castle has something less. There are allot of things that can be done to the esc as in code to get the pushers prediction better. One of them being a motor with more magnets that induce the back emf. Or a math model based on a commutation test to find the neutral plane and remember it.

my personal opinion is that whether the motor has sensors or not all esc can take timing out of the rc car drivers hands and the motors will get smaller and more powerful than anything that anyone can do manually or through current electronic timing tunes. The esc just needs to do a .5 second test on the motor and store the position. Then compensate when it pushes the merry go round by a simple math function as the rpm increases. All done automaticly and consistently.

Great visualization of how esc works. So once the sensorless esc has the rotor position stored by using the back emf it could add timing but not in relation to rpm since it doesn't have a sensor wire. Correct? Could it add timing in relation to throttle position?

Rich

the answer is yes and no. You know that 6 back emf pulse is a rev (for 2 pole which I dont believe castle does). The motor could count six pulses and multiply it into seconds. Does it do it? I dont know but it could. Just like it could store the magnet positions and the coil positions or hysteresis of field expansion and collapse per rpm. In fact it doesnt even need to know how to commutate if it rotates on of pulses at an increasing rate around the coils. Im guessing but i think castle limits you to low medium and high because they are adding timing that makes sense but doesnt require people to understand degrees per rpm and how the field distorts differently from 1 rpm to the next but yes its possible to add timing based on rpm and trigger position. Its just better when an esc can calculate how much to add at a given moment because the math is better than the majority of people deciding how much timing to add without the ability to see what is happening. Brushless motors combined with modern processors and experienced programmers can do amazing things. Castle knows there stuff. Ive talked to a few mfgs. Im not going to bash anyone but I got someone who really knows motor control on my first call. He also knew RC way better than i ever will. Every company can do more and i obviously want them to take advantage of all the new tech (old now) but if there arent people asking, suctioning, or buying it there arent millions of dollars waiting for a research project. You’re logic is good and it is possible but your intentions to make something better needs support. Otherwise the MFGs have to remain silent because most of them know that more can be done but dont want to get in a battle over whether or not they can be a business that keeps the people they love employed. Not that they dont love the community but we can be pretty back biting when someone decides to invest everything they have and they wont create something new. Maybe an open source esc would be a compromise they would make. I dont know.

At high RPM, finding the rotor position sensorless is actually dramatically easier and more accurate than it is with sensors. On our sensored systems like the Mamba X, we stop listening to the sensors over about 1000 RPM unless you program the controller in sensored only mode. The windings on the stator cause a lot of magnetic interference at higher RPMs and actually cause the sensors to stay triggered for longer than they should, sometimes trigger early, and sometimes not at all. In your merry go round analogy, sensorless would be like pushing it around based on what you can see with your eyes and not listening to anything and sensored would be like being blind and pushing when a person tells you to and and having another person standing right next to him yelling louder and louder as RPM increases. With the same amount of timing advance, sensorless is normally 5-10% more efficient than sensored at full throttle.

A motor test is not required, having fixed timing results from a test would fail to compensate for changing load and conditions. It is much easier to dynamically alter the timing on the fly and get it to perfectly match what the motor is doing. We don't need to see 6 back EMF pulses, we compensate every single pulse to keep the motor running perfectly in time regardless of any outside forces on the motor.

well, well, well. Thank you for joining the party. I suspected low medium and high wasnt a literary shortcut based on some of the choices you guys have made on other elements of your systems. You’ve confirmed it. I dont want to take you into your IP so feel free to comment or not. If I really have to know I can call and figure it out. you probably have a fairly good understanding of servo drives for robotics and ... I assume you provide a range of timing so because you are using a model rather than a CToroid? The development of a dynamic model is obviously my choice if its being compared to the other option of manual or timing by degrees.

I dont code around Hall effects or back emf anymore. Commutation is either stored from a test cycle in perpetuity or we run the test every cycle with 4M ppr (sin wave encoders or resolvers if your that kind of guy). Since you can model your motors thermally you might want to consider a duty cycle model. Predicting an overheat condition is pretty awesome. We used it to move peak torque from 3 times continuous to 4 times. The thermistors can confirm the thermal dissipation of the setup before you allow someone to unlock it.

I dont want to keep playing make a wish because i know what that leads to but here is my last concept. You know or can determine peak torque and efficiency of a motor (your motor). On chassis you can log a lap on a track. You can calculate the average rpm to run a lap. A ratio calculator based on average rpm or maybe a motion profile generator (draw a track) would probably be helpful to guys that struggle with picking a fast or efficient ratio for a track.

Thanks for the info. i wont be offended if you would prefer to talk about these things. Analogies are always full of holes if you know too much but you make great points and im sure it created some interest. That’s why I post anything to begin with.