Jump to content

Tilt-back, power demand myth, and THE life saving reflex


Mono

Recommended Posts

Tilt-back is a mechanism to incentivise the rider to slow down. The mechanism is simple: the neutral inclination angle of the shell and hence the pedals is changed from horizontal to negative, tilting the pedals back ;) This gives the rider the incentive to initiate a slow down (see also below). Here I discuss my understanding of the energetic (and a few other) consequences of tilt-back. 

Remember the feeling to lose the ground under your feet when the tilt-back sets in? Here is why. Simple geometric consideration reveals that if the riders feet stay in contact with the pedals, tilt-back raises the riders front feet and lowers the riders heels. Lowering ones heels feels like losing support and means that the riders body lowers as well if the heels remain grounded.

The effect from the centre of mass: most of the work to raise the riders body (or the toes ;)) is done by muscles, hence the energy comes from the food the rider has digested. (Lifting 102kg by 1cm loss-free needs about 10W=102*9.81*0.01W for one second or 100W for 0.1s thereby adding 2.78mWh=10/60^2Wh=0.0024kcal to the potential energy of the rider). However not only the rider needs to work: when the wheel pushes the rider forward (or backward), straightening up or raising the riders body adds momentarily to the riders perceived weight and hence to the power demand of the wheel. Vice versa, bending the knees or lowering the body gives the wheel a short period of decreased power demand (perceived decreased rider weight). Lowering by 5cm would remove the entire rider weight for 1/10 of a second. This is definitely something one should exploit in any critical situation: the reflex of bending the knees to keep or restore the wheel under the rider is a life saver!

I had two or three quite surprising saves from intentionally going rapidly-almost-falling-like deep into the knees. Unfortunately, going deep into the knees is particularly difficult and somewhat physically limited under tilt-back. Yet, soft knees are our suspension. Soft knees get us over bumps and out of potholes. Bending knees is THE invaluable reflex when riding an EUC. But I digress...

The effect from the change of tilt angle (here I stand corrected): because changing the tilt angle backwards increases the speed of the motor traveling relative to the shell, changing the angle requires energy. The amount however seems to be rather miniscule. If we travel 20km/h=5.6m/s with an 18" EUC and change the tilt angle from 0º to -10º in 1 second (pretty scary, IMHO), the shell position changes over the wheel circumference by 4cm = 10/360 * 18" * π. Hence, the circumferential rotation speed increases for 1 second by 0.72% = 0.04m/5.6m, i.e. by less than one percent. I am actually not sure what the power demand of this mechanism is (between 0 and 1.4% seems a good guess), but to all I can tell it must be negligible. Tilting the wheel also lowers its centre of mass. Lowering 20kg by 1cm in 1s may deliver 2W for 1s at most.

For the remainder, the simple but conclusive approach is to considered energy conservation: any consumed energy from the battery must be converted into kinetic energy or potential energy or heat.

After the tilt angle has changed, from the energy balance perspective nothing is different to the situation before. If the wheel consumes additional energy, it produces more torque. More torque leads to acceleration (hence energy is converted to and conserved as kinetic energy), just as it happens without tilt-back or while the tilt-back sets in.

Some people feel that under tilt-back they apply more pressure to the front foot, or equivalently, that the wheel applies more up-pressure. This means that the wheel produces more torque to provide this counter pressure. Torque however invariably leads to acceleration of the wheel (or the wheel and the rider). The other way around, if the wheel does not accelerate, this feeling is a perception due to the uncomfortable foot position but not actually an increased up-pressure.

Finally, slowing down the wheel, with or without tilt-back, can be accomplished by initially accelerating the wheel to the front of the rider. Tilt-back is the invitation to do exactly this. In particular, if the rider does not adapt to the changing neutral tilt angle, the wheel accelerates (without the rider and quickly). This acceleration requires some additional power (less than the acceleration of wheel and rider). With the knee-bending trick applied immediately, the additional power to initiate braking can at higher speeds probably be reduced to zero.

Link to comment
Share on other sites

It just occurred to me that by using the Wheellog app, where it logs current vs time, maybe a simple test could be conducted. Set the tilt-back to 10km/h and slowly increase speed until tilt-back kicks in and then look at the data to see if there's a noticeable surge in current.

Because although I like what you've written, it's still theoretical. I'd like to get some hard numbers, and maybe my test idea would work. I'm going to try it.

Link to comment
Share on other sites

10 hours ago, Marty Backe said:

Because although I like what you've written, it's still theoretical. I'd like to get some hard numbers, and maybe my test idea would work. I'm going to try it.

Go for it! As Igor Sikorsky, Russian immigrant & airplane and helicopter designer, said:

In the course of your work, you will from time to time encounter the situation where the facts and the theory do not coincide. In such circumstances, young gentlemen, it is my earnest advice to respect the facts.
 

Link to comment
Share on other sites

9 hours ago, Marty Backe said:

It just occurred to me that by using the Wheellog app, where it logs current vs time, maybe a simple test could be conducted. Set the tilt-back to 10km/h and slowly increase speed until tilt-back kicks in and then look at the data to see if there's a noticeable surge in current.

Because although I like what you've written, it's still theoretical. I'd like to get some hard numbers, and maybe my test idea would work. I'm going to try it.

Hi Marty,

I have tested it recently with my ACM, because I had my tilt back set at 42kph but some buddys said it was more dangerous than not having it at all, and that I should just respect the final alarm instead. In a way it made sense to me because I'm not going to change my tilt back speed when my battery get's lower in voltage, and I know that under 30% (under load) the final alarm kicks in at 37kph so the tilt back is pretty much useless.

So I still wanted to know if the tilt back was really draining some power, I set the tilt back at low speed, and from what I saw, there was a sort of power pic, unless I immedialy start decelerating when the tilt back kicks in (then the power might get down as low as 50W, must be regenerative breaking as I don't think 50W is enough to balance me), so I can't say that the theory is wrong, but I decided to skip the tilt back unless I really want to limit my EUC to a low speed (under 30kph with the ACM because then it'll have enough power reserve). This is in no way a scientific measure btw

I think the greatest danger comes from accelerating and hitting the tilt back speed while still accelerating, because you're already leaning forward, and even if the tilt back doesn't overpower the motor, one might loose his balance in this particular situation, leading in a dangerous face planting.

I would recommend keeping the tilt back speed for low speeds or "slow" EUC (such as the MCM series, KS series and all of the EUC that don't go above 30kph)

 

Link to comment
Share on other sites

6 hours ago, Pingouin said:

I don't think 50W is enough to balance me

I think we have plenty of measurements suggesting that one can go at 10km/h with 10Wh/km consumption. This equates to 100W at 10km/h speed. I have a hard time to believe that more than 50 of these 100W are used for balancing only

See e.g.

 

Link to comment
Share on other sites

3 hours ago, Pingouin said:

Hi Marty,

I have tested it recently with my ACM, because I had my tilt back set at 42kph but some buddys said it was more dangerous than not having it at all, and that I should just respect the final alarm instead. In a way it made sense to me because I'm not going to change my tilt back speed when my battery get's lower in voltage, and I know that under 30% (under load) the final alarm kicks in at 37kph so the tilt back is pretty much useless.

So I still wanted to know if the tilt back was really draining some power, I set the tilt back at low speed, and from what I saw, there was a sort of power pic, unless I immedialy start decelerating when the tilt back kicks in (then the power might get down as low as 50W, must be regenerative breaking as I don't think 50W is enough to balance me), so I can't say that the theory is wrong, but I decided to skip the tilt back unless I really want to limit my EUC to a low speed (under 30kph with the ACM because then it'll have enough power reserve). This is in no way a scientific measure btw

I think the greatest danger comes from accelerating and hitting the tilt back speed while still accelerating, because you're already leaning forward, and even if the tilt back doesn't overpower the motor, one might loose his balance in this particular situation, leading in a dangerous face planting.

I would recommend keeping the tilt back speed for low speeds or "slow" EUC (such as the MCM series, KS series and all of the EUC that don't go above 30kph)

 

Just to reiterate, I do ride all my wheels with tilt-back off. I never attempt to ride at the rated max speed and always slow down when I hear the 3rd alarm. There are a lot of people who keep riding while the alarm is beeping away - not good in my opinion.

I'm still going to try my little experiment this weekend - I'll post something later.

Link to comment
Share on other sites

As some practical input to the tiltback topic my graph from accelerating to fast and by this hitting the 20 km/h limit quite hard on my 9bot E+. Tiltback set in in two stages. The first at a pitch angle of about 7° and the second at about 10°. The second stage imho set in since i could not balance fast enough after the first stage begun - I could not put my weight "neutral" on the pedal, i stayed more on my toe tips and by this did not decelerate the wheel enough to stop the tiltback. After some endles seconds in the second stage i managed to put enough weight on the heels to brake the wheel enough to stop the tiltback:

sjcSruZ.png

As seen in the graph the power demand is quite low compared to intense accelerating. The one current spike (almost 10A, ~600W) once the second stage set in could be the power needed to raise me standing on the toe tips from the ~7° to 10° pitch angle.

I assume that the "rumours/stories/accidents" from faceplanting after tiltback kicked in are not primarily related to the tiltback it self but to too fast acceleration towards cut-off speed. Or also the (small) acceleration needed to tilt-back could be the last bit "needed" to reach the cut-off speed?

Link to comment
Share on other sites

30 minutes ago, Chriull said:

I assume that the "rumours/stories/accidents" from faceplanting after tiltback kicked in are not primarily related to the tiltback it self but to too fast acceleration towards cut-off speed. Or also the (small) acceleration needed to tilt-back could be the last bit "needed" to reach the cut-off speed?

That would be in either case incredibly irresponsible controller design :(, though I guess it wouldn't be a first.

Link to comment
Share on other sites

1 hour ago, Chriull said:

As seen in the graph the power demand is quite low compared to intense accelerating.

AFAICS all current peaks correlate with acceleration, which is somewhat what we would expect. Is there any chance to get these data in some ascii format to play around with them?

Link to comment
Share on other sites

I'd like to give my statement also....just one single one....

i think this discussion is as old as the arche Noah :-) 

I have seen graphs that tiltback needs power, videos where the tiltback clearly lead to a voltdrop and also a wheel acceleration, there where plenty reports of cutouts on highspeed when tiltback sets in.....and had an own faceplant where the tiltback on highspeed/low batterie has made the wheel weak and mushy.....

Sure tiltback needs not as much power as an acceleration, but some it needs, and as all movements/energy can not been produced out of nowhere....

So seams i am an faithful member of the myth/rumour/stories gang then B) 

 

Link to comment
Share on other sites

9 hours ago, Mono said:

Go for it! As Igor Sikorsky, Russian immigrant & airplane and helicopter designer, said:

In the course of your work, you will from time to time encounter the situation where the facts and the theory do not coincide. In such circumstances, young gentlemen, it is my earnest advice to respect the facts.
 

Since we're tossing quotes:

In theory, there is no difference between theory and practice.  In practice, there is.   - anon

Link to comment
Share on other sites

3 hours ago, KingSong69 said:

I have seen graphs that tiltback needs power, [...]

So seams i am an faithful member of the myth/rumour/stories gang then

If this gang wouldn't exist, I am pretty sure this thread wouldn't exist :D I'd delighted if you would post links to the graphs you have seen!

3 hours ago, Slaughthammer said:

All this is, as @KingSong69 already stated, an old debate, and there can be only one conclusion: If your wheel has not enough power left to enforce a tiltback, it should have tilted back when it still hat that power.

Agreed, that is a perfectly valid conclusion. The good news: the power to enforce tiltback is seemingly zero. The bad news: the power to initiate braking can be a decisive factor. However, under no circumstances should a wheel cut off if that power is not available, never, ever, ever.

3 hours ago, JimB said:

Since we're tossing quotes:

In theory, there is no difference between theory and practice.  In practice, there is.   - anon

Scientifically speaking, we don't expect a theory to hold up in practice under any circumstances. Some theories may, but these are rather great exceptions.

Link to comment
Share on other sites

10 hours ago, Mono said:

That would be in either case incredibly irresponsible controller design :(, though I guess it wouldn't be a first.

I agree with your first post as it mirrors my wheel's behavior. I think the rear up behavior is exactly correct; one wants the rider to crouch in fear. Fear is one way of making a rider slow down. I mean, it feels like you're about to die the first couple of times rear up happens. I hesitate to use tilt-back and rear-up interchangeably as I think the wheel (or at least my Inmotion V5) feels like it's doing two distinct things. It really isn't but doing the same thing fast vs slow gives it an entirely different feel.

I will say the engineers who designed these wheels must be geniuses for thinking out wheel behavior. Amazing.

Tilt-back is the wheel maintaining speed but tilting the pedals back. Yeah the wheel increases speed but it's so slight I cannot feel it. Tilt-back occurs gradually as speed is slowly increased. To me, tilt-back is never scary and it's an elegant way of telling me I'm approaching the never-exceed speed.

Rear-up is when the wheel suddenly speeds up while it rears its pedals to the maximum angle. There is always an audible warning on my wheel. I postulate the wheel's thinking thusly, "ok, I'm nearly maxed out and at this final stage we are going to crash. I'm going for a last gasp burst of energy to put myself in front of the rider and I'm going to try to scare the hell out of him by using max tilt-back. Because I know if I don't we will crash, but if I do this we might crash but there is still some chance of my rider saving it. And so some chance is better than none, so I'm speeding and tilting, and hopefully he won't get chucked off." Maybe that's the reason for the odd controller design.

I hit the rear-up on purpose several times or many times per day because I rather enjoy the wheel going past me. Wheel feels alive, you know. In all cases the wheel is working hard at slower speeds when that resistance is suddenly removed, as it would say cresting a hill or going down an ever steepening hill. Suddenly removing resistance seems to make the wheel "stumble" into a faster speed, and quickly too.

I really really recommend an athletic stance when riding these wheels at higher speeds. Knees bent, leaning forward at the waist, arms bent at the elbow. This stance gives you the best chance of recovery as the wheel goes sailing past you in max tilt-back mode.

Interestingly I've never detected tilt-back during slow off-road hill-climbing or descending. I would think the power demands of that are high but maybe that doesn't seem to be the case. At least, I've never had lack of power problems; I've fallen when I've got stuck in a rut or the wheel spun uselessly.

Link to comment
Share on other sites

There is one useful view point I forgot to mention. When tilt-back is setting in, we can consider a constant angle scenario and a constant pressure scenario. In the constant angle scenario, the rider keeps the pedal horizontal by putting additional pressure to the forefoot. Consequently, the wheel accelerates quickly, because the horizontal angle is now a "push forward" angle (the neutral angle is negative). This is an unstable situation and leads quickly to a bud-plant. In the constant pressure scenario, the rider follows the changing tilt angle of the pedals with the feet while keeping the pressure constant. In this case, speed remains constant and nothing really changes but the angle of the feet and pedals. This is the desired scenario, but if tilt-back sets in quickly and catches the rider by surprise it may be difficult to achieve.

I think that most wheels use tilt-back to indicate speed limit, but not to indicate torque limit. The KingSong KS 14C seems to be an exception and there might be others. In case of torque limit, the wheel may get rather soft and seems to tilt-forward (when pushed forward). When both happens at the same time it might become very confusing for the rider.

Link to comment
Share on other sites

23 hours ago, Mono said:

AFAICS all current peaks correlate with acceleration, which is somewhat what we would expect.

+1.

More exactly current is proportional to the torque which is proportional to the force. The forces to overcome are for acceleration (m*a), air drag, rolling resistance and "height changes".

As you already stated, tiltback is just some acceleration to establish the different pedal angle. Some current (power) is needed for this depending on the "speed of this change==acceleration". Depending on the "riders reaction" while this tiltback is established (standing on the toe tips or still balancing his weigt evenly on the pedal) additional power could be needed to "lift" the rider.

 

My post cited again for @Mono resoibse: "I assume that the "rumours/stories/accidents" from faceplanting after tiltback kicked in are not primarily related to the tiltback it self but to too fast acceleration towards cut-off speed. Or also the (small) acceleration needed to tilt-back could be the last bit "needed" to reach the cut-off speed?"

23 hours ago, Mono said:

That would be in either case incredibly irresponsible controller design :(, though I guess it wouldn't be a first.

Instead of cut-off speed i should have used "max torque-speed limit".

Imho cut-off speed is not (easily) reachable while driving. Cut-off speed is normally reached with very low load (like lifting the wheel). While driving/accelerating to higher speeds the wheel has to overcome a much higher load (accelerating the drivers mass, air drag and rolling resistance + balancing) so the max torque speed limit _should_ be reached way before cut-off speed *2). Maybe with extrem acceleration one could hit cut-off speed by the build up momemtum while already "going along the max-torque speed limit" - but this "going along the max-torque speed" limit means that the wheel is already in an unstable state (no balancing possible anymore), so a overlean/faceplant/crash is already unavoidable (for non acrobatic riders *1) and the cut-off by firmware just "triggers" the crash at a little bit lower speeds as it would happen anyway.

As @KingSong69and @Slaughthammer stated the tilt-back vs accident discussion are old an ongoing, but imho there are still too much false rumours around and no (for me) satisfying resolution accepted in this forum. So for me this discussions will go on, as long as i see an resolution to be found sometimes, somewhere...

So starting with normal tilt-back conditions/situations which never should bear any problem:

- Tilt-back comes slowly and nice so it is easily to be recognized by the rider and one can distribute ones weight on the pedals and shift the weight to the heels to decelerate and stop the tilt-back situation: No hard accelerations are happening, so no current spikes, voltage drops, overpower situations can happen and tilt-back does together with the rider what it's intention - warning and deceleration.

Any overlean/over-power cut-off happening while this "kind of tilt-back" occurs should be imho just in combination with an sudden incline, bump, wind gust, etc and triggered by this "extra" force needed. The tilt-back just happened at the same point in time but has no correleation with the incident.

A situation i do not consider or care about:

- People driving with the wheel is in tilt-back state. So the are nearer to the "max-torque speed limit" and every additional power needed (incline, bump, wind gust, imbalance) could lead to an overpower/overlean situation - they took their decision so they should also take the accident. There is nothing more the wheel/firmware could to to warn the rider.

So "dangerous" tilt-backs are imho the fast ones, which come within fractions of a second. If they do not throw the driver off the wheel the danger comes from the rider regaining balance on his toe tips leaning forward and by this further accelerating the wheel. In my case shown above in the graph the tilt back pushed me back resulting in a deceleration (negative, regenerative current spike) and i recovered my balance on the toe tips forcing the wheel to accelerate to the second tilt-back stage (~10A current spike), followed by some positive negative current spike (de and acelerations) until a regained balance and normaly decelerated to a safe and stable state.

If i would not have lost balance and immediately stayed on my toe tips with the sudden tilt-back happening this situation would have ended in a nasty accident. After my already strong but still sane acceleration (~10-15A on a level road) for a 9Bot E+ my balancing on the toe tips would have increased this acceleration to an not (by the 9Bot) managable overpower situation. With whichever of voltage drop, overpower cut-off, overlean by hitting the "max torque speed limit", etc being the final accident trigger.

If in the same situation without loosing the balance one does not balance on the toe tips but immediately stay with slightly more weight on the heels one just decelerates and drive on like nothing happened...

That's an acceptable situation for test-drive dummies, beta testers, adrenaline junkies and others who like to live on the edge of disaster - and after some spectacular incidents they should imho learn to manage all this situations. But for customers who just want to drive, have fun and willing to pay good money for this wheels this "fast" tilt-backs are just an irresponsible and not acceptable firmware issue!

In my above real world example the first tilt-back stage came within one second (from 26m30 to 26m31) - this does not sound to fast in the first moment, but while accelerating it comes quite surprisingly. So something like ~2-3 seconds should be the absolute minimum for a tilt-back warning. So the tilt-back should have started slowly and nicely at 26m28-26m29 quite immediately after i started my acceleration. So by an soft tilt-back (~1-2°) i would have had to lean even more forward to keep up the acceleration and so noticed that the wheel warns me. Keeping the "original" lean forward would have resulted in a slower acceleration and a safe and nice end-speed with absolutly no surprise or danger...

So since the wheel constantly measures the speed and by this can easily compute the acceleration it's an very easy calculation if one would reach the tilt-back speed limit within 2-3 seconds and it's time to start tilt-back nicely!

Another fast and dangerous "fast tilt-back" situation i presumably encountered on my 9Bot E+ was once i drove and the battery charge went under a certain threshold which lead to a lower speed limit for tilt-back. So since i was actually driving above this speed limit tilt-back kicked in immediately. This situation could also happen with every wheel that has battery charge depended speed limits. Or also with (over)power, temperature, etc triggered tilt-backs. If in this case the tilt-back comes as hard and powerfull as the wheel can manage it's just again an irresponsible and not acceptable firmware issue - that's easily to program to perform the tilt-back nicely so that it's again a warning the driver can recognize and respond to it without being surprised!

The "rear-up" condition reported from @LanghamPi fortunately did not encounter till now, but i remember that there where some posts already about this happening. Could this be some kind of warning triggered by to fast deceleration (regenerative "overpowering", overvoltage warning)? Did you confront Inmotion or your reseller with this strange behaviour?

Are there any other "fast and/or dangerous" tilt-back conditions already identified?

Would be great if they would be resolved with new firmware updates! Would be imho a reason to get the first  2.0 firmwares!

Maybe @Jason McNeil @tinawong @Diana@szkingsong.com and other reseller/manufacturer representatives could push the designers to resolve this issues?

(Sorry to adress mainly kingsong representatives since i just know their names - and i have no idea how their firmware development state/implementations is in regard to tilt-back's)

Quote

Is there any chance to get these data in some ascii format to play around with them?

Attached you find the zip file with the original data in csv format for Current, Speed, Altidute, Pitch, Voltage and Altitude. The graph started at 26 Minutes and 26 Seconds, so the "interesting" data is starting at 26*60+26=1586.00 sec (first column). You have to divide the data for (Current,Speed,Altidute,Pitch,Voltage,Temperature) by (100,1000,100,100,100,10) to get the Values in (A,m/s,m,°,V,C).

The 9Bot E+ sends Voltage values unfortunately at much lower sampling intervalls as the other values, so load dependend voltage drops are not really "visible".

*1) while reaching/hitting the max torque speed limit one could feel the "softness" of the wheel, brake and balance it to come to a stable state again. But under the above assumption that one accelerated really hard to the limit, so the the generated momememtum takes one up until the cut-off speed that recovery is imho just theory.

*2) If cut-off speed is reached before the max torque speed limit of the motor it would be really a very bad and irresponsible controller design!

9B_20160415_105325.zip

Link to comment
Share on other sites

Thanks @Chriull for the data, I will have a look. That voltage is not available is a pity.

7 hours ago, Chriull said:

Imho cut-off speed is not (easily) reachable while driving. Cut-off speed is normally reached with very low load (like lifting the wheel).

Agreed, that is also my (rather limited) experience and that is how it should be. 

Quote

Maybe with extrem acceleration one could hit cut-off speed by the build up momemtum while already "going along the max-torque speed limit"

But under the above assumption that one accelerated really hard to the limit, so the the generated momememtum takes one up until the cut-off speed

I don't think that works, because there is no inertia or memory in acceleration, it's all stored in the speed (impulse or momentum is speed times mass and the mass doesn't change). If you stop pushing due to lack of power you stop accelerating immediately

Still, when going downhill with tailwind one could imagine to hit cut off speed despite the lack of motor power :huh:

It would be useful to always make the distinction between insufficient torque and cut off (or cut out), because the first is instantaneously recoverable when the demand is lowered.

Link to comment
Share on other sites

Analysing the data, I found a negative correlation between tilt back angle and current in the above graphs. It remains negative when taken from second 31 or when removing acceleration in a multivariate linear regression. AFAICS it's not likely a direct causal link, as I still have no reasons to believe that tilt angle effects energy demand.

Link to comment
Share on other sites

  • 8 months later...
On 30.4.2017 at 10:20 AM, Mono said:

Analysing the data, I found a negative correlation between tilt back angle and current in the above graphs. It remains negative when taken from second 31 or when removing acceleration in a multivariate linear regression. 

Since i reread this as this topic came up again lately,  i just did some (limited) reading - so negative correlation means that higher tilt back angles go with lower currents and vice versa?

No (low) correlation would be values around zero?

(... took some time to realize this...? )

Quote

AFAICSit's not likely a direct causal link, as I still have no reasons to believe that tilt angle effects energy demand.

When tiltback hit me, my dearest wish was to stop or at least stop acceleration - the "negative correlation" shows that i achieved this goal? And by this could easily be a "spurious correlation" and no causal link.

My explanation would be the following:

One can drive quite normal with tiltback (keep the speed constant by balancing, accelerate if one leans forward or brake by leaning back) - just like without tiltback. Many riders also adjust the pedal tilt to their preferences - establish some "constant (negative) tiltback" for normal driving. So acceleration depends an the leaning and not on the pedal angel. (Or more correct on the acceleration induced by the leaning).

There could be a correlation and causal link between the tilt back angle change (?rate?) and current: to achieve a (positive) tilt back angle change the wheel has to accelerate (increase the current). But this could be just a small peak in the ongoing "self balancing noise". The main correlation/causal link should be again induced by the drivers behaviour. It one leans forward despite the tiltback the wheel accelerates and uses more current, if one balances the tiltback speed stays (no acceleration and just the self balancing current occurs) or one leans back and decelerates (negative current).

So i'd fully agree to your "I still have no reasons to believe that tilt angle effects energy demand."

Link to comment
Share on other sites

5 hours ago, Chriull said:

Since i reread this as this topic came up again lately,  i just did some (limited) reading - so negative correlation means that higher tilt back angles go with lower currents and vice versa?

yes

Quote

No (low) correlation would be values around zero?

yes

Quote

(... took some time to realize this...? )

When tiltback hit me, my dearest wish was to stop or at least stop acceleration - the "negative correlation" shows that i achieved this goal?

Good point. I guess that's why I computed the correlation also after I removing acceleration. 

Quote

And by this could easily be a "spurious correlation" and no causal link.

My explanation would be the following:

One can drive quite normal with tiltback (keep the speed constant by balancing, accelerate if one leans forward or brake by leaning back) - just like without tiltback. Many riders also adjust the pedal tilt to their preferences - establish some "constant (negative) tiltback" for normal driving. So acceleration depends an the leaning and not on the pedal angel. (Or more correct on the acceleration induced by the leaning).

There could be a correlation and causal link between the tilt back angle change (?rate?) and current: to achieve a (positive) tilt back angle change the wheel has to accelerate (increase the current). But this could be just a small peak in the ongoing "self balancing noise".

RIght, I estimated the additional speed above:

On 4/22/2017 at 12:47 AM, Mono said:

The effect from the change of tilt angle (here I stand corrected): because changing the tilt angle backwards increases the speed of the motor traveling relative to the shell, changing the angle requires energy. The amount however seems to be rather miniscule. If we travel 20km/h=5.6m/s with an 18" EUC and change the tilt angle from 0º to -10º in 1 second (pretty scary, IMHO), the shell position changes over the wheel circumference by 4cm = 10/360 * 18" * π. Hence, the circumferential rotation speed increases for 1 second by 0.72% = 0.04m/5.6m, i.e. by less than one percent. I am actually not sure what the power demand of this mechanism is (between 0 and 1.4% seems a good guess), but to all I can tell it must be negligible. Tilting the wheel also lowers its centre of mass. Lowering 20kg by 1cm in 1s may deliver 2W for 1s at most.

 

Quote

The main correlation/causal link should be again induced by the drivers behaviour. It one leans forward despite the tiltback the wheel accelerates and uses more current, if one balances the tiltback speed stays (no acceleration and just the self balancing current occurs) or one leans back and decelerates (negative current).

Here is another question that bugs me a little: what makes "self-balancing current" different from "keeping-the-same-speed current" if the speed is non-zero? Isn't both of them just a sequence of small accelerations and decelerations around a given speed? It is noteworthy that the wheel doesn't do anything on its own to "self"-balance but only reacts to the drivers input in the pedal positioning (the pedal tilt). AFAICS it is a perfectly valid model to think of the pedals as a (rather sensitive) throttle pedal.

Quote

So i'd fully agree to your "I still have no reasons to believe that tilt angle effects energy demand."

 

Link to comment
Share on other sites

27 minutes ago, Mono said:

Here is another question that bugs me a little: what makes "self-balancing current" different from "keeping-the-same-speed current" if the speed is non-zero? Isn't both of them just a sequence of small accelerations and decelerations around a given speed?

Yes. Also the same while accelerating, if one can differentiate between the lean to accelerate and the leaning changes by the inbalances. This would allow one to get the acceleration current and the same as above "self balancing current". (If rider inbalances are comparable while riding at a steady speed and while acceleration)

Quote

It is noteworthy that the wheel doesn't do anything on its own to "self"-balance but only reacts to the drivers input in the pedal positioning (the pedal tilt).

As the wheel does "rider+itself balancing" this is achieved by reactions to the pedal tilt. The feedback loop (pedal angle to motor drive) is established by a P(roportional)I(ntegral)D(erivative) controller it is not only a direct (proportional) reaction.

Quote

AFAICS it is a perfectly valid model to think of the pedals as a (rather sensitive) throttle pedal.

+1. With the addition, that the wheel tries to work against any pedal angle changes from the driver side.

Link to comment
Share on other sites

25 minutes ago, Chriull said:

As the wheel does "rider+itself balancing" this is achieved by reactions to the pedal tilt. The feedback loop (pedal angle to motor drive) is established by a P(roportional)I(ntegral)D(erivative) controller it is not only a direct (proportional) reaction.

Good point. It would be interesting to know how different that actually feels to a P-only controller using, say, the quadratic deviation to the target angle as input. I guess the I-part alleviates the pedal dipping under constant load, and the D-part prevents a sudden push-through.

Link to comment
Share on other sites

4 minutes ago, Mono said:

Good point. It would be interesting to know how different that actually feels to a P-only controller using, say, the quadratic deviation to the target angle as input. I guess the I-part alleviates the pedal dipping under constant load, and the D-part prevents a sudden push-through.

That could be a nice question for "our" firmware guys. Whats the input and output values of the pid controller and which factors for the p, i and d parts work (respectively how to determine sane values).

Just looked at http://ctms.engin.umich.edu/CTMS/index.php?example=InvertedPendulum&section=ControlPID it seems that only a P controller should lead to overshoot reaction (wobble). The D part should help to eliminate this.

Maybe the I part could "solve" the in the article mentioned "sideward move problem", adjusts gyroscope drift, or whatever? 

Maybe @esaj got into this with his selfbalancing robot?

Link to comment
Share on other sites

2 hours ago, Chriull said:

Just looked at http://ctms.engin.umich.edu/CTMS/index.php?example=InvertedPendulum&section=ControlPID it seems that only a P controller should lead to overshoot reaction (wobble).

That seems immediately obvious if we take the angle theta as input to P, it's not so obvious if we use the squared angle as input.

Link to comment
Share on other sites

4 hours ago, Mono said:

That seems immediately obvious if we take the angle theta as input to P, it's not so obvious if we use the squared angle as input.

The controller normaly works on the "error" (difference of actual value to desired value) - so a square function would not work out. The sign has to "survive". (Or the "error" always has to stay positive..)

An optimal transfer function can be calculated for a given system - as far as i have read again some articles revealing a bit if my fading memory of past education somehow with laplace transformation and step response...

The same can be achieved with a PID controller. Or it could be set up with some trial an error/setup guides and reach (more or less) satisfatcory results... And the developers can reuse readily available and tested source code.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...