Making Begode wheels safer: PWM tiltback with custom firmware

[alcatraz]

The better stock pack for the MCM5 has got to be the NCR18650GA 800Wh. That's 3P. Those cells are 15-20A cells. They almost pass the overheat test at 20A. So 3x20A ~ 60A.

The Samsung 40T overheats at 30A but passes 20A. Let's call them 25A cells. At 2P you have 50A discharge. 

In any case. Not a whole world of difference depending on what you compare the cells to.

That is really good that they're that close at 3.6v. Below 3.3v you're going to see voltages separate a lot.

[Leskont]

Let me see this in practice)))

Moreover, there are no permanent currents of 50-60А in MSM. Only short-term peaks. And 40T copes with them better.

Edited July 22, 2022 by Leskont

[alcatraz]

Forgot to say why they separate more at lower voltage. It's because the bulk of the charge is around 3.7v (the nominal voltage). When you near depletion the voltage changes rapidly. So small differences in capacity will display themselves as voltage differences. 

[alcatraz]

If you want to build a high current monster look for Sony VTC cells. They're 18650 but still match the 21700 currents, so you can get maybe 50% higher currents because of the higher number of cells.

[Leskont]

2 minutes ago, alcatraz said:

Forgot to say why they separate more at lower voltage. It's because the bulk of the charge is around 3.7v (the nominal voltage). When you near depletion the voltage changes rapidly. So small differences in capacity will display themselves as voltage differences. 

Let me see this in practice))). Okay?
I promise to report on the tests conducted here...

[Leskont]

Just now, alcatraz said:

If you want to build a high current monster look for Sony VTC cells. They're 18650 but still match the 21700 currents, so you can get maybe 50% higher currents because of the higher number of cells.

I've already done all, thank you)
In my MCM already 40T

[alcatraz]

3 minutes ago, Leskont said:

Let me see this in practice)))

Moreover, there are no direct currents of 50-60А in MSM. Only short-term peaks. And 40T copes with them better.

Yeah, it's possible that 40T has less sag. I just compared the heat side of things. But I think they are connected, even if not equally proportional. Heat + sag + capacity.

Having a smartBMS is a fantastic upgrade. Was it hard to find space for it inside the small shell?

[Leskont]

3 minutes ago, alcatraz said:

Was it hard to find space for it inside the small shell?

Pretty simple.
There's a lot of space there.

[Leskont]

Moreover, there is a place for 3p 21700, but I don't need it anymore.
UP:  https://t.me/gotwaymcm5/49868 Here is a little photo of my MCM

In this form, he has already driven 1000+ km

Edited July 22, 2022 by Leskont

[alcatraz]

Yeah, I think I posted the first picture of a 1100Wh MCM5 here on the forum. I found it for sale in the chinese 2nd hand pages. Can you believe it? (not a news article or something like that) 

[Freestyler]

On 7/21/2022 at 2:36 PM, Leskont said:

Hi, now I'm ready to try this firmware on the Tesla T3. Model: T3, Model code: 1601001

I also want to try it on MSM5 later. There I have a custom battery on Samsung 40T high-current cells. And I was interested in the battery percentage scale in euc-dash. Is it possible to correct it in your firmware? The fact is that 40T can be discharged deeper than stock Begode

I have prepared a firmware for the MCM5 with lower voltage limits for you to try. Instructions in your PMs.

 

Low voltage alarm from 68v -> 62v

Empty battery tiltback from 66v -> 60v

[Ranulf]

Hello @Freestyler, after a crash due to a low-batt mcm5, I'm really interested by your custom firmware. Is it possible to test it on my 84V mcm5? (I can't PM you, I'm a new user :-)) 

[jshreder]

On 7/12/2022 at 3:11 PM, Freestyler said:

Previous, but I can port them to the newer ones if desired.

Hey there, love this project! I have an EXN HS (C30) that I've been riding with the newer firmware (72v cutoff with broken tail light), just hit 1k miles. Would it be possible to get a version of that firmware with your PWM cutoff? 

I mean, my ideal would be a working taillight with 72v cutoff and your PWM tiltback haha, but I'll take what I can get!

I'm pretty familiar with flashing boards so I'd love to take a crack at it myself if you are up for sharing the firmware. Regardless, thanks for this awesome contribution to the community!

[Vince.Fab]

@Freestyler I would also love to try your dynamic tiltback on my new RS Molicel c30.  Coming from a Kingsong s18 that was pushed to the limit way too much, it would be great to not just rely on beeps on this wheel.  Thanks!

Model RSC30

Firmware - 1902002

[Bizra6ot]

@jshreder I don't know if you were able to contact him and install his FW, but there was a new update on the begode app early this week that fixes the taillight 

[Magman116]

15 hours ago, Vince.Fab said:

@Freestyler I would also love to try your dynamic tiltback on my new RS Molicel c30.  Coming from a Kingsong s18 that was pushed to the limit way too much, it would be great to not just rely on beeps on this wheel.  Thanks!

Model RSC30

Firmware - 1902002

I have the same RS as you.  You will love the Dynamic PWM Tiltback with the accelerated stability.  I have been discussing with @Freestyler about getting a version that takes advantage of the high current Molicel batteries.  The firmware is supposed to have a lower voltage cut-off and an increased phase current.  I am excited to try that version whenever he gets it done.  I think we can get some extra range with the low voltage cut-off adjustment and a slight increase in initial acceleration with the phase current adjustment. Definitely looking forward to trying it out.

The dynamic tiltback FW will give you an added layer of security that you will not over-power the wheel regardless of the battery charge level or speed.  It will automatically adjust based on PWM feed back.

I have been putting this FW to the test and it does function at many different speeds depending on battery level.   I ran the wheel down to about 40% charge and it was triggering the PWM tiltback at about 34mph.. On a full charge, the tiltback kicks in at about 48mph, but have pushed this wheel to 54.3mph once; was trying to find the limits of the wheel and don't plan on ever hitting that speed on this wheel again. The light will come on too to let you know encase you don't feel the tiltback.

[Vince.Fab]

@Magman116that sounds great!  Not sure I'll ever reach those speeds cause big boned but that's pretty wild.   Hit 40 on it today and couldn't even imagine that much faster.   Euc World safety margin alerts seem to be pretty laggy, any idea if they are accurate?

[Magman116]

@Vince.FabI didn't trust the safety margins of EUC World much until I loaded @Freestyler's PWM Dynamic Tiltback firmware. I tested the tiltback doing lift tests then set out to test it riding and various speeds seeing when the tiltback would kick in.

With the alarms in EUC World, I never really trusted them and have watched videos a cutouts without triggering alarms. Now it could be the alarms were disabled by the user.

I was very surprised I was able to hit the speeds I did before the PWB tiltback kicked in.  I really don't like the feeling of going over 45mph unless I had a smooth tail wind. 40-45mph seems to be a better or more comfortable range for me.  Wind is a big issue at those speeds especially gusty or cross winds that can really mess up your day.  When I analyze my tours in EUC World, most of my riding is in the 35-40mph range on the road and much less off road.

Now if we can get updated firmware with the low voltage cut-off and the phase current increase, these RS19 C30 wheels with the Molicel P42A cells will be all that much better preforming.  Phase current increase should also help in braking.

 

[Chriull]

50 minutes ago, Magman116 said:

@Vince.FabI didn't trust the safety margins of EUC World much until I loaded @Freestyler's PWM Dynamic Tiltback firmware. I tested the tiltback doing lift tests then set out to test it riding and various speeds seeing when the tiltback would kick in.

With the alarms in EUC World, I never really trusted them and have watched videos a cutouts without triggering alarms. Now it could be the alarms were disabled by the user

Euc world pwm alarms for begode wheels are calculated/estimated by the app and not meant to help going anywhere near the limits.

Freewheelers firmware uses the build in pwm value without any delays.

This should make a huge difference!

[Freestyler]

On 8/26/2022 at 3:37 PM, Chriull said:

Freewheelers firmware uses the build in pwm value without any delays.

I'm regularly been mistaken for Freeforester, but first time as Freewheeler!

 

Apologies everyone for being absent. In August no one works in Greece (too busy island hopping ) and then I got covid and was off for a few more days.

I will get slowly get back to everyone that messaged me, thanks for being patient!

Edited September 1, 2022 by Freestyler

[Magman116]

@FreestylerI'm glad you are well and still with us.  The safety of Begode / GW owners depends on you.  

Seriously though, I am glad you are ok and back at it.

[meriwald]

@Freestyler

I'm impressed with your work, how did you get ahold of firmware, isnt it encrypted stored in the nand?

Do you think it's possible to apply old settings from boards before v14 for rs ht to make it as strong as MSP was? New boards v14 and higher are so sluggish

[Chriull]

On 7/15/2022 at 2:18 PM, Freestyler said:

It's something that has been bothering me as well, but I simply lack the knowledge to understand it.

We now have a very good warning system to prevent against overspeed cutoffs, but we still don't know the overtorque / acceleration limits.

For example: How fast I can accelerate to 50km/h on a wheel that can do 60km/h (even at 10% battery)?

That questions are all covered by the pwm %.

As found out by @jancellor in 

Somehow summarized - without load by speed one uses up pwm % from 0% at standstill up to 100% at lift cut off speed.

So at any given speed (actual speed / lift cut off speed) pwm % are already "used up". So for burden only ( 100% - this used up pwm %) are availabe.

Once one reaches 100% one overleans.

So the qestion how fast can i accelerate reduces to how fast will the pwm % reach 100%. With actual speed/max speed and battery charge already regarded.

If one monitors pwm % change rate one can "predict" how long it will take until one overleans, if one continues with constant acceleration/burden/...

On 7/15/2022 at 2:18 PM, Freestyler said:

Let's say a wheel has 250a phase current limit.

As reported in the other topic, for example the EX20S firmware cuts off at 200A motor current - so for this limit some seperate alarms/actions have to be implemented. This is not covered by pwm %.

On 7/15/2022 at 2:18 PM, Freestyler said:

We now know how much headroom we have when we do crazy leans at low speeds, but as the speed increases so does the back EMF from the motor. (which cancels the battery supplied voltage)

As speed increase pwm % does too, so this case is covered, too.

On 7/15/2022 at 2:18 PM, Freestyler said:

You can't push 250a when going 40km/h. So how much current can you push at each speed? What is the formula?

You need information such as the motor kv and winding resistance. Maybe PWM information is useful here. Can you help @Chriull?

The formula would include too many unkowns like you mentioned and (by burden, over time and temperature changing) internal battery resistance etc...

And yes, pwm % covers it all!

On 7/15/2022 at 2:18 PM, Freestyler said:

If we know the max current for each speed (or PWM), then we can set an alarm in a fixed percentage.

I understood that your firmware triggers tiltback at some fixed/?choosable? pwm % threshold?

So this gives some fixed safety margin for each situation - regarding battery voltage, current, resistances, temperature, ...

Just not the firmware cut off for high currents - but imho there possible prewarn times will be always way below human reaction time so this is something people have to learn to avoid.

On 7/15/2022 at 2:18 PM, Freestyler said:

Btw the latency of the app is not higher if any at all, because to sound the beep a timer is used internally and you might be in an unfavorable time window relative to the interval.

That's why sometimes you see the alarm state in packets faster than hearing the beeps. Euc alarm has very good latency as well.

I think the very nature of extreme leans is that even with alarms it's not guaranteed to save you. If you are in an extreme lean position and hear the beeps at 80% of max current, you might not have enough time to react.

Still it would be immensely useful to slowly learn the acceleration limits of your wheel, because as it is right now we are completely blind.

Your customized firmware tiltback trigger is activated by the generated packets?

That would be "bad" news - as the next "improvement" would imho be "dynamic" warnings/tiltback. So not at a fixed pwm percentage - as this could create a huge prewarn time before overlean if one accelerates slowly and some inhuman short prewarn times if one accelerates very fast!

By calculating the pwm % change rate one could trigger the alarm some 0,x seconds before reaching overlean (- y%). But for this change rate to be usefull one has to "denoise" it/get rid of the "spikes" by doing some avaeraging/low pass filtering the original pwm % from the control loop, which should be available some 100 times a second... The external available 2-5 pwm % samples a second are too late and too inacurate for such measures...

Edited September 2, 2022 by Chriull

[Freestyler]

Exactly. I'm interested in the knowledge not provided by the PWM.

For example I like to do some accelerations on my wheel where I hit around 150 phase current (wheel has 250a limit).

I stop at around 45km/h, so the PWM does not exceed 50%. (free spin of the wheel is around 94 km/h at full charge)

But I'm afraid of accelerating harder, because I might hit the current limit. Not the "stall" phase current limit of 250a imposed by the firmware that can only be hit from an almost stand still.

I want to have a graph like the one you created in your anatomy thread and the one in the ex20s thread in order to know how much phase current can I reach while going at 30km/h for example. Is it 200a? Is it 150a?

I know it depends on many parameters that we don't know the exact value and we have to guess, but I don't even know the ballpark

 

45 minutes ago, Chriull said:

Your customized firmware tiltback trigger is activated by the generated packets?

That would be "bad" news - as the next "improvement" would imho be "dynamic" warnings/tiltback. So not at a fixed pwm percentage - as this could create a huge prewarn time before overlean if one accelerates slowly and some inhuman short prewarn times if one accelerates very fast!

No! I was only referring to the triggering of the beeper. It's not instantaneous even if you trigger it from the firmware, so sometimes you might see the alarm in the bluetooth packets before the buzzer activates. The loop runs a lot more often as you said.

The good thing is that it already works somehow like that. The more aggressive you are, the lower the tiltback speed.

Observe the first 45 seconds of the following video: 

You can see that at that battery percent, with normal driving the tiltback engages at 27 km/h.

at 40 seconds I do a quicker acceleration so the tiltback speed is activated at 23 km/h. The PWM only spikes to 53%.

So you'll be much sooner at full tiltback and it will be almost impossible to push further.

 

 

If you go extreme like this guy here (timestamp 4:12), then no warning system will save you.

You are too far gone  The spike from the tiltback will probably drop you anyway.

 

[Chriull]

2 hours ago, Freestyler said:

Exactly. I'm interested in the knowledge not provided by the PWM.

Why stick with phase current and make it more complicated if one can have all of it simply with pwm %?

2 hours ago, Freestyler said:

For example I like to do some accelerations on my wheel where I hit around 150 phase current (wheel has 250a limit).

I stop at around 45km/h, so the PWM does not exceed 50%. (free spin of the wheel is around 94 km/h at full charge)

Lets imaging free spin speed is 100km/h at the current battery charge state for easier math:

With no load/burden (==no acceleration, friction or air drag) pwm would be at ~45% at 45 km/h.

For acceleration, friction and air drag one has still 55% pwm % reserve.

If one wants to accelerate up to 70 km/h one has 70% pwm % used up already and only 30% for acceleration, air drag and friction.

So maximum possible phase currents between at 45 km/h to possible at 70 km/h are 55 to 30.

The real actual value depends on your battery charge state, internal battery resistance, coil resistance, temperature and battery 'burden state".

All of this is already included and regarded in the pwm %...

If you really want such a figure for one stateof your wheel (battery, coil,...) log some lift spin cout outs. Draw it in a phase current vs speed diagram.

Then try to spin up the wheel to some higher speed below cut off and then brake it again as hard as possible.

Draw the log data in your phase current vs speed diagram. If you see that it hit the limit line you now can connect the limit line. Voila!

Proceed with the following with extreme caution (or just don't do it!) - these new wheels are heavy and extreme powerfull!

If not, you have to break the wheel harder (let the tire touch something ?slippery?) at higher spinning speeds until succesfull.

Or you have by chance some log of an already happened overlean? Use this log!

Another, safer way is to extrapolate some current/speed plot by the recorded pwm% and draw the limit.

If you happen to get some nice diagramms you'll have to shift the limits with battery charge state for future use.

Maybe adopt for increasing battery resistance too over time.

2 hours ago, Freestyler said:

I want to have a graph like the one you created in your anatomy thread and the one in the ex20s thread in order to know how much phase current can I reach while going at 30km/h for example. Is it 200a? Is it 150a?

I know it depends on many parameters that we don't know the exact value and we have to guess, but I don't even know the ballpark

Read through the anatomy and the there linked previous topic and try to redo the calculations and graphs. That should give you the picture