Jump to content

Oscillation and cut out problem with Gotway MSuper V3, ACM, maybe Monster shipped from May 2017 onwards


Recommended Posts

9 minutes ago, meepmeepmayer said:

Mosfets blown at 5 km/h???

May be 5-7 km/h + it was small border (about 3 sm) (i am not sure this is related) + my weight 110 kg. But this wheel had already raised me to long steep slides and there was no problem.

Link to comment
Share on other sites

2 minutes ago, meepmeepmayer said:

This is very interesting. In the other thread something was said about monitoring/preventing too high currents in the latest changes to the wheels. Maybe something produces too high currents? Is that possible?

I don't know. I am just a user. This is my first wheel and experience if only 2 weeks.

When the wheel refused it twitched, it continuously squeaked and smelt of smoke. I could not turn it off. I had to open the case and disconnect the battery.

Link to comment
Share on other sites

MsuperS3plus_MOSFET.thumb.jpg.a8b62605ec6a4dad3b599b53bd016e79.jpg

For anybody more knowledgeable than me about electronics design, here's the data sheet for those MOSFETS: http://www.infineon.com/dgdl/irfb4110pbf.pdf?fileId=5546d462533600a401535615a9571e0b

What makes me wonder:

  • The chips are specified for max. 100V. Isn't using them in an application with >80V cutting it a bit close?
  • According to the date code (P427Y), those MOSFETs have been produced in mid 2014. I wonder, what took them so long to find their way into a GW wheel?
  • If I understand the 12 MOSFET design correctly, multiple MOSFETs are used in parallel to share the load. Just how critical is it, that those devices used in parallel are exactly alike ("matched pairs")?

Any experts here to tell?

  • Upvote 3
Link to comment
Share on other sites

3 hours ago, Tilmann said:

Only very few of us would be willing and able to foot the US$5000+ bill for an ACM built to automotive quality standards

That's true, but an automobile is much much much more complex to manufacture and costs in the order of $10,000 including reasonable quality standards. I can agree that automobiles are amazingly cheap for their complexity, but $5000 for an EUC doesn't seem to be adequate, unless maybe it has a 5kWh battery.

Edited by Mono
  • Upvote 1
Link to comment
Share on other sites

1 hour ago, Sleepy said:

I also have a problem with this series (GotWay MSuper3S+ 1600) (1704).
The wheel was bought 2 weeks ago. The distance is 150 km.
The speed is about 5-7 kilometers per hour. The battery was charged 90%.

Has handed over in repair. I'm waiting.

Fortunately, I was dressed in defense and not injured.

pxz0qn.jpeg

rxgku6.jpeg

ug71d0.jpeg

I believe the MOSFET's should not be blowing like that.  the spec says it should be  able to handle 120 AMPS if properly cooled for a T0220 casing.

Then we just dont have one.  We have two MOSFET per phase on the high side and two  MOSFET per phase in the low side. But for simplicity the power is split in two MOSFET.

So lets put some margins.  Lets design for 100 AMPS Peak per MOSFET.

So technically it should handle 200 AMPS. Also those AMPS are modulated AMPS. Dont know what duty cycle would produce that.

Then lets now say the coil resistance is low enough to allow 200 amps to be induced with a 84 volts battery. That means V=IR so R= 84/200 = 0.45 ohms.

What is the coil resistance for one of the phases?____________.

Then another issue is if you look at watts. 84V*200 = 16,800 watts!  REALLY. No Way.

Do you see how ridiculous this is?  The issues has to be they have defective components or they are damaging the components during assembly. 

Capture.PNG

Edited by Carlos E Rodriguez
  • Upvote 2
Link to comment
Share on other sites

Great post @Tilmann

Scares the hell outta me though.

As a westerner, i expect far more rigorous QA for such an expensive item, especially if it's going to be moving you at fast speeds. It's so odd to me that you can't go to a Gotway website and contact them, or peruse their wheels. Post on their forums, see some videos, discussions, news, etc. I'm glad the fail rate is so low, but yeah, that read like a horror show, lol.

  • Upvote 2
Link to comment
Share on other sites

9 minutes ago, Carlos E Rodriguez said:

I believe the MOSFET's should not be blowing like that.  the spec says it should be  able to handle 120 AMPS if properly cooled for a T0220 casing.

 

Engineers with whom I spoke argue that the problem is most likely in the management of transistors, and not in overload. Forgive me, but my knowledge of English is not enough to discuss this issue in more detail.

  • Upvote 1
Link to comment
Share on other sites

2 minutes ago, Mono said:

That's true, but an automobile is much much much more complex to manufacture and costs in the order of $10,000 including reasonable quality standards. I can agree that automobiles are amazingly cheap for their complexity, but $5000 for an EUC doesn't seem to be adequate, unless it has a 5kWh battery.

Just looking at the ingredients and complexity of the end product, that's certainly true. But please keep in mind, that especially compact cars are produced in huge quantities, while many QA related cost factors are independent of actual manufacturing numbers. Indeed, it may be much more expensive to purchase high quality components in insignificantly small quantities (you just don't have the same pulling power with your suppliers). Also please factor in, that the EUC manufacturers still have to invent all the QA skills, techniques and processes, which automotive developed (and paid for!) millions of cars earlier.

I used my 3D printer analogy to estimate the cost of an ACM produced to highest quality standards:

  • Upvote 2
Link to comment
Share on other sites

Sure, if I would build a single ACM it would cost much much much more than $5000, randomly guessing maybe I could produce one for less than $100,000. It wouldn't be an acceptable price for selling though.

Edited by Mono
Link to comment
Share on other sites

13 hours ago, esaj said:

I don't recall hearing of the legs snapping off or breaking during riding, usually if there's a hall-sensor problem, it's the connector between the mainboard and the motor cabling, on a couple of occasions a sensor has become loose inside the motor after a hard crash or if the wheel was dropped (not lately, I don't think there have been cases in the last year or so, and even then they were generics?). The stator's not turning, so I'd think only major forces affecting the legs/wiring are bumps and drops, and like you said, there's not much mass on these (mainly the wiring, the sensor itself is glued onto the stator and weighs next to nothing by itself anyway).

Not sure if anyone has ever proven it, but it's been mentioned many times that likely the firmware switches over to "sensorless mode" in higher speeds, and just estimates the rotor position and speed, so the hall-sensors would only be used for start up (to know the starting position of the rotor) and at low speeds. Presumably because of latency and possibly noisy output of the sensor with higher rotational speeds? For a single physical rotation of the tire (one turn), there's multiple "electrical rotations" (ie. switching between the phases) as there are so many poles / magnets in the motors. See for example   https://e2e.ti.com/blogs_/b/motordrivecontrol/archive/2014/05/23/mind-spinning-possibilities-hall-effect-sensor-bandwidth-in-bldc-motors   (Do note that they speak of a motor with much lower amount of poles than the usual wheel-motor, and probably their own hall-sensor, to which this appears be marketing material for ;) is likely faster/has higher bandwidth than the cheaper ones)

Same article says its non-sense to say the sensors can not keep up.

"For mechanical systems (i.e. motors), it turns out that it is very difficult to get a magnet spinning fast enough to start to cause Hall effect sensor bandwidth concerns".

"the motor would have to spin at 120,000 RPM ".

That was a 10 pole motor. So lets say we have 100 poles instead.

That would give you 12,000 RPM. Which taking c= 2PI*R. 

So 16 inch wheel. 2*3.15*16/2 = 50.5 inches per Rev.

To go 50 Miles per hour then.( Math checkers welcomed.  I might have made a mistake.

1 mile = 63360 inches.   

(1rev/50.5in)*(63360in/1mi)*(10mi/1hr)*(1hr/60min) = 209 RPM to do 10MPH

to do 50 MPH then 1045 RPM.  

Based on the TI example 12,000 RPM for a 100 pole motor.

We technically only need  1045 RPM so technically speaking the Hall switches are plenty capable of functioning for EUC applications and there is more than 10 times margin.

But maybe I made a math error.

  • Upvote 2
Link to comment
Share on other sites

@Tilmann That was one terrific educational post!  Curious, does GW and KS management have guided tours of their plants or do they allow individual supervised access by appointment?

2 hours ago, Sleepy said:

I also have a problem with this series (GotWay MSuper3S+ 1600) (1704).
The wheel was bought 2 weeks ago. The distance is 150 km.
The speed is about 5-7 kilometers per hour. The battery was charged 90%.

 

Scary stuff! I have never seen evidence of an electrical arc originating from one of the motor wires directly to a Mosfet! The connectors were always the weak link in the chain! Are we now talking inferior wiring and insulation? It appears both the blue and green motor wire took a hit!

Edited by Rehab1
  • Upvote 2
Link to comment
Share on other sites

7 hours ago, Carlos E Rodriguez said:

Same article says its non-sense to say the sensors can not keep up.

"For mechanical systems (i.e. motors), it turns out that it is very difficult to get a magnet spinning fast enough to start to cause Hall effect sensor bandwidth concerns".

"the motor would have to spin at 120,000 RPM ".

That was a 10 pole motor. So lets say we have 100 poles instead.

That would give you 12,000 RPM. Which taking c= 2PI*R. 

So 16 inch wheel. 2*3.15*16/2 = 50.5 inches per Rev.

To go 50 Miles per hour then.( Math checkers welcomed.  I might have made a mistake.

1 mile = 63360 inches.   

(1rev/50.5in)*(63360in/1mi)*(10mi/1hr)*(1hr/60min) = 209 RPM to do 10MPH

to do 50 MPH then 1045 RPM.  

Based on the TI example 12,000 RPM for a 100 pole motor.

We technically only need  1045 RPM so technically speaking the Hall switches are plenty capable of functioning for EUC applications and there is more than 10 times margin.

But maybe I made a math error.

I think your math is right, I tried some values with the equations given in the article earlier, and the margins would seem to be high enough, even though US1881 has half the bandwidth of the TI-sensor, it's still around 10kHz max with the wheels having a few hundred hertz... Don't know if there could be other other issues (high currents in the phase wires running right next to the hall-wires inducing voltage in the hall-wires or such?). Quickly skimming the sensorless vs. hall-sensors -only comments in the Firmware-thread, the end result is... inconclusive :D Lizardmech says it's likely sensorless, as that works well with FOC and also got a piece of code from Firewheel & Gotway -clones that showed sensorless operation. Yet electric_vehicle_lover built a sensored control for the wheel and found some other piece of code that was using hall-sensors only, so go figure ;) 

Lizardmech: I think the position sensors are only used to compliment the sensorless FOC algo, it just feeds the data from them into it along with the park and clark transformations. From my understanding FOC vector control is inherently sensorless and halls or encoder sensors are an additional thing on top of the FOC algo.

 

electric_vehicle_lover:  Running with hall sensors should be no problem for all speeds, as the FOC algorithm will detect the correct angle correction value and apply it continually. I don't think sensorless algorithm is a must have to get the correct angle, as even with an incorrect angle with increase speed, given by hall sensors, the algorithm find the correct angle by looking at Id current that must be zero, other values means the angle should be corrected with proportion. And this works well with my current code.

 

Lizardmech:  It's difficult to troubleshoot hall based FOC because it doesn't seem to be used anywhere else. All the hall based examples I can find are BLDC type, while the FOC are typically encoder or sensorless. I think hall sensor FOC controllers are just sensorless controllers that only use the halls to assist at startup. It's possible there's a reason few FOC controllers operating only using hall sensors exist. Assuming it's theoretically possible for it to work all I can think of is the delay from the hall sensor possibly combined with deadtime being inserted by the gate driver causes issues.

They show part of their FOC code here, it's sensorless with hall sensor startup. Conventional SVM and clark/park transforms.

https://world.taobao.com/item/531928634884.htm?fromSite=main&spm=a312a.7700824.w4004-14846006895.18.36MAJg

Looks like they offer gotway clone and firewheel clone.

 

electric_vehicle_lover: I know I am being telling that EUC original firmware may be using sensorless after some speed and hall sensor just to start and be able to quick change direction. I also implemented only using hall sensors and no sensorless and it seems to be running well, at least for what I can understand, the motor does the same very low noise as original firmware.

Also, as we can see on chinese firmware for EUC: https://eggelectricunicycle.bitbucket.io/FOC--Chinese_controllers_code.html (see first line of FOC_Model()) they seem to use only hall sensor (no sensorless) and just the same way I do, estimate angle between each hall sensor pulse, that's why they integrate the electric angle.

 

Could be either way depending on wheel manufacturer / firmware? :P

  • Upvote 2
Link to comment
Share on other sites

9 hours ago, Carlos E Rodriguez said:

I believe the MOSFET's should not be blowing like that.  the spec says it should be  able to handle 120 AMPS if properly cooled for a T0220 casing.

Then we just dont have one.  We have two MOSFET per phase on the high side and two  MOSFET per phase in the low side. But for simplicity the power is split in two MOSFET.

Do you see how ridiculous this is?  The issues has to be they have defective components or they are damaging the components during assembly. 

There are far more ways to destroy a mosfet than just overcurrent or overvoltage either between gate and source or drain and source (although those are probably the most common), as an example: 

 

  • Upvote 3
Link to comment
Share on other sites

On 6/14/2017 at 6:00 PM, Carlos E Rodriguez said:

I believe the MOSFET's should not be blowing like that.  the spec says it should be  able to handle 120 AMPS if properly cooled for a T0220 casing.

Then we just dont have one.  We have two MOSFET per phase on the high side and two  MOSFET per phase in the low side. But for simplicity the power is split in two MOSFET.

So lets put some margins.  Lets design for 100 AMPS Peak per MOSFET.

So technically it should handle 200 AMPS. Also those AMPS are modulated AMPS. Dont know what duty cycle would produce that.

Then lets now say the coil resistance is low enough to allow 200 amps to be induced with a 84 volts battery. That means V=IR so R= 84/200 = 0.45 ohms.

What is the coil resistance for one of the phases?____________.

Then another issue is if you look at watts. 84V*200 = 16,800 watts!  REALLY. No Way.

Do you see how ridiculous this is?  The issues has to be they have defective components or they are damaging the components during assembly. 

Capture.PNG

@esaj do you agree with my evaluation?

i believe they just need to fix the weak circuit. 

About FOC. Don't toubned additions circuit to trigger the Zero-crossing in Rotherham to implement without the hall sensors?

Edited by Carlos E Rodriguez
  • Upvote 2
Link to comment
Share on other sites

4 hours ago, Carlos E Rodriguez said:

@esaj do you agree with my evaluation?

i believe they just need to fix the weak circuit. 

It's clear that at least one (I think those were low-side on Rehab's board) mosfet on the two phases has been destroyed, can't see if the parallel ones are fried too. There must have a been a lot of heat to burn a hole in the casing, might even be a shoot through? I'd expect that if the mosfet "just" overheats and turns the junctions inside the mosfet into jello, it would just die somewhere around 125-150C junction temperature or whatever, which likely wouldn't be enough to melt the plastic around it, so there'd be no outside signs. Those look more like they exploded, so a very high current spike would seem more likely, but of course I can't say for sure. The large caps in the mainboard can give a very high current spike for a brief moment, depending on capacitor make and model, the internal ESR can be as low as <10 milliohms, if you'd short those at 84V over fully open mosfets with a few milliohms of Rds(on), the momentary spike could easily be several kiloamperes (1000's of amperes), probably the fets would die before they can fully turn on?

 

4 hours ago, Carlos E Rodriguez said:

About FOC. Don't toubned additions circuit to trigger the Zero-crossing in Rotherham to implement without the hall sensors?

Sorry, I'm not sure what you're asking here? Maybe turn off auto-correct? ;)

  • Upvote 1
Link to comment
Share on other sites

On 6/15/2017 at 1:00 AM, Carlos E Rodriguez said:

I believe the MOSFET's should not be blowing like that.  the spec says it should be  able to handle 120 AMPS if properly cooled for a T0220 casing.

Capture.PNG 

(Read that spec sheet wrong.. :wacko: thank you @esaj for pointing that out)!

The mosfet can handle 120A (package limit) but @ 10V not 84V and its also heat limited.

See spec sheet: http://www.infineon.com/dgdl/irfb4110pbf.pdf?fileId=5546d462533600a401535615a9571e0b

Have a look at figure 8, maximum safe operating area, It can only take 10A but for less than 100 usec if you have a voltage of less that 100V.

So its very easy to burn a mosfet with the wrong firmware behaviour.

Edited by Xima Lhotz
  • Upvote 3
Link to comment
Share on other sites

10 minutes ago, Xima Lhotz said:

The mosfet can handle 120A (package limit) but @ 10V not 84V and its also heat limited.

True on the limits otherwise, but the 10V is the Vgs (gate-to-source voltage), not drain-to-source, and the 120A is continuous, not pulsed current.

 

Quote

See spec sheet: http://www.infineon.com/dgdl/irfb4110pbf.pdf?fileId=5546d462533600a401535615a9571e0b

Have a look at figure 8, maximum safe operating area, It can take 100A but for less than 100 usec.

Again depending on the Vds (drain-to-source -voltage, the x-axis in the graph). It does change with the back-EMF (for the low-side) and difference between battery voltage and back-EMF for the high side. The real situation is a bit hard to picture, as the mosfet is "pulsed" by PWM and the (average) back-EMF will change with speed.

 

Quote

So its very easy to burn a mosfet with the wrong firmware behaviour.

True, an extreme situation like a bridge shoot-through (which should not occur if the firmware is correctly made, plus the gate drivers probably have some sort of dead-time insertion) will kill the mosfets in an instant.

Edited by esaj
  • Upvote 1
Link to comment
Share on other sites

10 minutes ago, dalewalker said:

WARNING the replacement boards dont fix the problem. My customer had his board replaced already and today it cut out again sending his acm 1300wh into a deep river. 

He was on a group ride and has about 9 witnesses that seen his wheel get the wabble then cut out. 

I advise anyone thats received a new board to fix the problem to take care because it seems its not over yet.

 

Will update later. Im going to bed. Its quite late here.

I just needed to inform everyone effected before i sleep because if i waited for the morning someone else may get injured and id feel horrible

Is it the new board that gotway said was fixed? My dealer here in norway told me that they wont sent out the new fixed boards before next week.
 

  • Upvote 1
Link to comment
Share on other sites

9 minutes ago, EU GUY said:

Is it the new board that gotway said was fixed? My dealer here in norway told me that they wont sent out the new fixed boards before next week.
 

It was sent before the serious injurys happened but it was the same problem and that was there fix.

He was one of the 1st few that experienced the problem and I got straight onto it because im a good dealer :-p

Its possible all they did was sent another board thinking "hope this one works ok" because before the many and some serious accidents they didnt take it as serious. There was a few around that time and it was a couple weeks before the bad accidents so those bad accidents probably could have been avoided if it was taken seriously 

Link to comment
Share on other sites

  • esaj changed the title to Oscillation and cut out problem with Gotway MSuper V3, ACM, maybe Monster shipped from May 2017 onwards
  • esaj unpinned this topic
  • esaj unfeatured this topic

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...