Ninebot One P...same control board as the E+

[Cranium]

I received my Ninebot One P today and excitedly opened it.  It is sharper looking and I really like the black pads.  I'm glad I've done most of my learning so I won't be scratching up this EU right away.

I took off the pad to expose the control board.  It is version 1.3.1 which happens to be the same as the control board that came on my Ninebot One E+ that I received a few weeks ago.  My E+ was on firmware version 1.2.7 which had no reported issues when the board fried and dumped me to the ground.  I was lucky I didn't get hurt and was able to get a replacement control board in only a week shipped from Ninebot with the help of Forward California.    The replacement board was a version 1.3 board.  I felt I had been downgraded but after looking at the Mosfet ratings, I was ok with it.  It has worked fine so far.

The P uses the EXACT same control board as the latest E+ EUs.  The ONLY thing different electronically one the P from the E+ is the motor which is specified to produce 60% more power and the battery which is 12.5% higher capacity.  Everything else different is in the cosmetics (pads and black rings).  All for a 36% increase in price.

I verified that the MOSFETs are the same number: 100N8F6 (see attached data sheet and picture below).  These MOSFETs are expected to deliver up to 60% more power with the exact same heat sinks and cooling as in the E+.  This means that the P is capable at overheating much quicker than the E+ if you are using the additional power the motor allows.  

The E+ V1.3 board uses a 75NF75 MOSFET and I've attached the data sheet for this as well.  I would include a picture but it was hard as hell to get the number from the 1.3 board and had to use a stereoscopic microscope to read it.  I have a USB camera for it but my computer isn't recognizing it for some reason.

I did some back and forth balancing in my house to see how fast it would heat up.  I loved the additional power which allowed it to be much snappier and for me to feel more confident.  After a bit of time, I got the beeping from the unit overheating.  I checked the app and it showed 158°F.  It took more work than this to get an E+ to overheat doing exactly the same thing (but with less confidence).

So either the 1.3.1 board is either completely underutilized and over engineered OR the E+ or it is over-utilized and under engineered for the P.  My experience with the blown board on the E+ and the overheating on the P leads me to believe that it is the later.  But time will really tell for sure.

Ninebot One v1.3.1 control board E+ & P MOSFETs

MOSFET Differences

V1.3.1 board on the left (E+ & P) and V1.3 board on the right (E+)

 

100N8F6Mosfet.pdf

STP75NF75 Mosfet (3.1 board).pdf

Edited December 2, 2015 by Cranium
Added Mosfet difference table

[Citi Wheel]

The battery is basically the same too from what I saw on speedy feets video, 320 vs. 340 wh 

[playdad]

Wow, I don't know what to say. That is a little worrisome! Please take care when pushing it in the Summer.

There's enough news about exploding/burning "Hoverboards".

[SlowMo]

27 minutes ago, playdad said:

Wow, I don't know what to say. That is a little worrisome! Please take care when pushing it in the Summer.

There's enough news about exploding/burning "Hoverboards".

Well, the P should be safer for it will explode out in the open as compared to hoverboards which explodes inside the houses.  But mind you, my E+ roasted in front of our door.

Edited December 2, 2015 by SlowMo

[cbgti]

It was sure for the mobo... If you put a "P" serial number in your "E" model, your mobo will give you an extra power...

But I'm not sure taht the motor can dissipate more power... Certainly a bit more, but not all the extra power for a long time (like in Uphill per exemple).

[Chriull]

On 2.12.2015 at 5:57 AM, Cranium said:

...

I did some back and forth balancing in my house to see how fast it would heat up.  I loved the additional power which allowed it to be much snappier and for me to feel more confident.  After a bit of time, I got the beeping from the unit overheating.  I checked the app and it showed 158°F.  It took more work than this to get an E+ to overheat doing exactly the same thing (but with less confidence).

So either the 1.3.1 board is either completely underutilized and over engineered OR the E+ or it is over-utilized and under engineered for the P.  My experience with the blown board on the E+ and the overheating on the P leads me to believe that it is the later.  But time will really tell for sure.

 

V1.3.1 board on the left (E+ & P) and V1.3 board on the right (E+)

 

With the P board (1.3.1) shown in the picture on the left side the heatsinks are not screwed togeter? Or was that just from your disassembly? It also seems (from different fotos and videos) that between this two parts of the heatsink no heat paste is used?

Since plastic is a nice thermal insulator the only real thermal exchange can happen through the hole below the metal plate below the motherboard leading to the wheel. There i have seen two different versions in the vides - one hole about 2/3 as big as the board and one very small (maybe from a one C?). 

For a rough approximation i assumed PWM Frequency 25khz, Mosfet switching time 200ns, 1500 W (was the E+ peak motor power imho?) and the thermal resistance of a 80x80x2mm aluminium plate, which should be around 10 K/W.

This should lead to a total power dissipation of the Mosfets of around 9W. This heatsink so could keep the Mosfet Junction temperature below 175°C up to an ambient temperature(heat sink temperature)  of ~75°C. (If the heat sink plate was outside with free convection.)

But the plate is inside the wheel and maybe the venting of the turning wheel is not sufficient to bring the warm air outside? From the tear down videos it seems, that there should be enough place inside to put some kind of "bar" onto the wheel to create a real airflow? There are also low profile heatsinks which one could use to replace/improve the plate?

Or paint/anodize the plate black (for IR) to improve the irradiation.

Could also be, that just the heatsink parts/mosfets are not connected properly and by this increase the thermal resistance too much. Additionaly the contruction with the heatsink for the "upper" 6 Mosfets screwed to the heatsink of the lower 6 mosfets which again is screwed on the plate which finally conducts away the heat seems less then ideal...

 

[Cranium]

4 hours ago, Chriull said:

For a rough approximation i assumed PWM Frequency 25khz, Mosfet switching time 200ns, 1500 W (was the E+ peak motor power imho?) and the thermal resistance of a 80x80x2mm aluminium plate, which should be around 10 K/W.

This should lead to a total power dissipation of the Mosfets of around 9W. This heatsink so could keep the Mosfet Junction temperature below 175°C up to an ambient temperature(heat sink temperature)  of ~75°C. (If the heat sink plate was outside with free convection.)

To answer your questions first:  

1. Yes, the screws missing were from my disassembly.  I was trying to remove the heatsink completely to better expose the MOSFETs underneath and to replace them.  
2. There is thermal paste between the lower heatsink and the bottom plate that the board attaches to but there is not any thermal paste between the upper and lower heatsinks.  I have measured the thermal temperatures directly on the board and the heatsinks and the temperature drops rapidly across the upper heatsink and has little heat dissipation to the lower heatsink.    I have not measured the heat on the back of the plate.  
3. The hole that helps dissipate heat from the plate under the board is about 2/3 the size of the plate.  It does not appear to help a great deal while moving.
4. I would be very hesitant to place anything on the wheel to increase airflow that would be rigid as this would increase the likelihood of a rock, stick or other item to get stuck in the wheel.  And before even considering this, I would want to see what temperature the back plate is reaching to see if this would even be worth considering.

Now for my questions to you:

1. What is the 25kHz PWM assumption based on?  I have wanted to measure the true value on my Oscilloscope but have not yet.  
2. How does the rated MOSFET switching time affect heat dissipation calculations?
3. What are the equations you used to calculate the heat dissipation of 10 K/W?  I've never done this so am quite curious.  What assumptions does this calculation make?

I would like to look at improving the heat dissipation once I understand where the heat is being trapped.  

[Chriull]

55 minutes ago, Cranium said:

 

1. There is thermal paste between the lower heatsink and the bottom plate that the board attaches to but there is not any thermal paste between the upper and lower heatsinks.

A litte bit of paste should help - thats imho quite a design failure...

Quote

1.  I have measured the thermal temperatures directly on the board and the heatsinks and the temperature drops rapidly across the upper heatsink and has little heat dissipation to the lower heatsink.    I have not measured the heat on the back of the plate.  

So the upper heatsinks is just heating the closed compartment and not contucting the heat to the outside... ;(

Quote

1. The hole that helps dissipate heat from the plate under the board is about 2/3 the size of the plate.  It does not appear to help a great deal while moving.

... but this should be the main heat sink - cooled down with fresh cold moving air ...

Quote

1. I would be very hesitant to place anything on the wheel to increase airflow that would be rigid as this would increase the likelihood of a rock, stick or other item to get stuck in the wheel.

Something soft would be more than enough to just move the air - like the soft foam stripes to protect the wheel from scratches...

Quote

1.  And before even considering this, I would want to see what temperature the back plate is reaching to see if this would even be worth considering.

Yes, first the temperature flow has to be possible from the mosfets to the plate, so it makes sense to cool the plate. Like using heat paste and good connections between the different parts...

Quote

Now for my questions to you:

1. What is the 25kHz PWM assumption based on?  I have wanted to measure the true value on my Oscilloscope but have not yet.  

The PWM frequency should not be audible (something >20kHz)

Riding 20 km/h with a 16" the wheel makes 4,3 revolutions a second. Times three for the three motor phases the PWM should be significantly higher than this 12,9 Hz

It should be high enough that the controller can change the power output by changing the duty cycle in small enough steps. So with 25kHz the wheel with 20km/h drives during one cycle  0,28 mm wide. So this should be more than enough for a "fine" control.

And the frequency should be as low as possible, because with every switching the mosfets burn some power.

Quote

1. How does the rated MOSFET switching time affect heat dissipation calculations?

While the mosfet is switching the Voltage from Drain to source changes from almost 0V to the full ~60V. Under full load (the assumed 1500W) a current of 25A is flowing (when the mosfet conducts). So as rough estimation while switching the mosfet you have a power dissipation in average of 25/2*60/2 W. And this 2 times every PWM cycle.

Quote

1. What are the equations you used to calculate the heat dissipation of 10 K/W?  I've never done this so am quite curious.  What assumptions does this calculation make?

10K/W is the thermal resistance of an quadratic aluminium plate 80x80x2mm - thats from the graph at the bottom of this website: http://www.vias.org/mikroelektronik/coolpacks.html. 

(Edit: the heat source has to be in the middle of the aluminium plate for the 10 K/W to come true...)

And this 10K/W is only valid for vertical aligned aluminium plates with unlimited airspace around the heatsink and free convection of the air... ;(

Also you have to add 0,5 K/W thermal resistance between the mosfet substrate and the case of the TO220 (RthJC) and the thermal resistance of the insulation between Mosfet and heatsink (for good materials something in the same magnitude). 

Then you see with the formula Delta T=Rth * P with Rth=11 K/W and P=9W a Delta T from 99K. The max temp at the junction of the mosfet is 175°C, so the ambient temperature for this heatsink has to be below 175-99°C=76°C.

Quote

I would like to look at improving the heat dissipation once I understand where the heat is being trapped.  

The heat is trapped within the compartment - there is no airflow and the plastic around insulats quite good. So the heatsink transports the heat away from the mosfets and heats up the little air in the compartment until the ambient termperature in the compartment exceeds the 175°C-Delta T. Then the mosfet is out of limits and hopefully the overtemp alarm started before

...So imho the best solution would be to mount the mosfets directly on the aluminium plate like seen on this picture for m-ten http://forum.electricunicycle.org/topic/1109-firmware/?do=findComment&comment=11561, put the plate over the hole to the wheel and put something (soft) on the wheel, so you have an steady airflow.

The efficiency of the heatsink gets better, the more surface is exposed to the airflow - so if there is enough place between the hole and the wheel one could use some low profile heatsinks too... And according to this website above the thermal resistance could be reduced by the factor 0,7 if the surface is black (regarding IR radiation)

 

 

Edited December 16, 2015 by Chriull

[Cranium]

@Chriull While you bring up some very good design points, they are general engineering principles that you are applying.  The specific limitations and challenges of this board being mounted to the side of a motorized wheel means that you have to make adjustments to the limitations of your environment.  This is what makes designing these boards challenging.   

Adding fins to the cooling surface between the case and the wheel would be problematic due to the nature of debris being caught up in the wheel that would potentially damage the fins, wheel or the case.  Worse case, this would actually cause overheating since mud, dirt and debris would clog up fins or break off the fins and reduce heat transfer efficiency.  

Adding a soft fin to the wheel may work and I've thought of doing that before with a piece of wiper blade but until I know that there is heat on the plate that is not being dissipated efficiently I won't try it.  But this area is certainly high on my list for where improvements might be able to be made.

The MOSFETs are mounted directly on Aluminum heatsinks (6 on one and 6 on another with both heatsinks attached together).  Just because it is mounted to a plate as you showed does not make it better than a block of aluminum with heatsinks.  There are several other factors which have to be taken into account before a determination can be made on which is more effective.

Painting the surface black better for capturing IR heat (and thus thermal resistance) works if you have an IR emitter that is not in direct contact with heat sink which isn't the case here.  A black painted surface in this application would actually hinder heat dissipation with no other factors changing.  

I'm not trying to pound you and if it comes off that way, I apologize.  But I am pointing out the difference in our approaches.  I am more than willing to assume that there were some smart engineers that have tried many different designs to come up with the one that they currently have in the Ninebots.  If I feel there may be an issue, I gather data to demonstrate empirical evidence that the issue is real and only then do I try to resolve it.  I am not the type that will try things just to see what happens because then if anything changes I would naturally want to credit it to what was done which may lead to inaccurate conclusions on the true cause (whether good or bad).  

But I enjoy seeing others try things.  If there are enough people that provide good data from their own experiments, then it starts to gain credibility that it is actually effective.  

So for me, I am just posting observations and concerns.  Overheating is becoming much more of a real issue for me due to being able to measure and replicate the differences between the E+ & the P.  But I haven't yet nailed down what the exact cause is.  It could be that the design of the Ninebot One will not allow any more cooling and they are trying to do too much with the heavier loading on the same board.  Or it could be that maybe my board doesn't have thermal paste between the board and plate and if discovered is easily remedied and resolves the heat issues I'm experiencing.  

Edited December 16, 2015 by Cranium

[Chriull]

Yes, fins, wipers and all that stuff could maybe help to repair a design issue, but could easily cause negative impact as you pointed out.

The heatsink, the mosfets are mounted to directly, mainly heats up the closed compartment. And the heat capacity of this compartment is very limited. The only real way to conduct away the heat is to the outside, which happens by the plate covering the hole. My idea was that directly mounting the mosfets on the plate (with the other side beeing in fresh air flow) avoids the thermal resistances resulting from the connections between the different heat sink parts.

Imho thats the reason why for the E+ the hole to the outside is much bigger then with ?earlier/the C? modell. Maybe thats the way ninebot also will choose for P v2 or P+.

Or maybe you really just got a model from a bad charge. Ninebot should have done enough tests before releasing the P modell to detect thermal problems and solve them. But on the other side, they are also releasing some firmware trials and destroy their products. And this not only once.. ;(

Unfortionately it seems there are not enough One P Users around to share their experience.

[Jason McNeil]

@Cranium, thank you for sharing this informative comparison of the Ninebot types. As the output power increases, this issue of control-board heat build up is going to become an ever more frequent & relevant engineering problem.  

In the case of the King Song, the battery pack is over double the capacity & number of parallels over the 9B1 P, with the result that overheating can be even more problematic if ridden hard (with a high proportion of acceleration & braking). Even though the aperture to the Wheel housing has more surface area than the Ninebot, it is still less than optimal. 

On the new 16", I suggested that they integrate an internal cooling fan that is activated when the board temperature rises above a defined set-point. It's an enclosed space & of course air is not great heat-conductor, but at least it's better than passive cooling... Pleased to say they've taken up my suggestion & are intending to implement on the production product. 

What's your opinion about contoured or curved fins like the below? Surely Ninebot have the engineering expertise & resources where they can design something clever that minimizes the risk of foreign objects striking during use. At least there is physical space in which they can add the appendages to the inside of the board, but will it be enough to have any effect?

 

A couple months back while doing some limit testing with a number of different Wheels & a datalogger attached, revealed that the average load, over say a minute, is not anywhere near the max rated output, & that the Wheels (in the example below is an 9B1 E) are not very resilient when stressed.

Also power output can experience enormous swings over the course of even a single second.   

Edited December 18, 2015 by Jason McNeil

[Cranium]

8 minutes ago, Jason McNeil said:

On the new 16", I suggested that they integrate an internal cooling fan that is activated when the board temperature rises above a defined set-point. It's an enclosed space & of course air is not great heat-conductor, but at least it's better than passive cooling... Pleased to say they've taken up my suggestion & are intending to implement on the production product. 

What's your opinion about contoured or curved fins like the below? Surely Ninebot have the engineering expertise & resources where they can design something clever that minimizes the risk of foreign objects striking during use. At least there is physical space in which they can add the appendages to the inside of the board, but will it be enough to have any effect?

Setting up convection heat transfer is certainly better than conduction through stagnant air.  Good idea.  I think some sort of heat pipe to transfer the heat from the MOSFETs to the outside air would be great but I can't think of a way to do this without making it more vulnerable in a crash.    

I think the contoured fins would be excellent if they could be integrated to a heat pipe or put on the inside by the wheel if they could somehow be protected from debris buildup and damage.  Damage from rocks would be a concern but a bigger concern would be mud buildup clogging them on the inside which is why I think a better solution may be having this type of heatsink on the outside.  

If they wanted to have them on the inside and protect them, they could create something that would draw air from the outside over the fins that are protected on the inside of the wheel.  They could set up a Venturi nozzle to create the natural air flow.  But I'm not sure what speed you would need to get effective air flow for cooling.  

hmmmm.....How about a heat pipe that transfers the heat to a heat sink with fins in another protected compartment which has vents and allows airflow to the outside?  This would keep the circuit board isolated but allow good air flow for heat dissipation.  But it would require re-design of the circuit board and compartment.  Not an easy do it yourself fix.

 

[Splinter4]

Cranium, 

Where did you get your NB1 P from?  Forward CA says they are still awaiting them to be released in the U.S. and they don't have a solid date.  Do you have an early prototype?

[Cranium]

@Splinter4 I received mine from Forward CA.  

[Splinter4]

That's wierd, I just hit them up on Facebook a couple of days ago and they said still waiting....  I must have misunderstood.  Thanks.

[Paulandjacquelyn]

4 hours ago, Splinter4 said:

Cranium, 

Where did you get your NB1 P from?  Forward CA says they are still awaiting them to be released in the U.S. and they don't have a solid date.  Do you have an early prototype?

They hit US n sold out. I got mine couple weeks ago from ninebot US.

[Jabirish]

On 2 December 2015 at 4:57 AM, Cranium said:

I received my Ninebot One P today and excitedly opened it.  It is sharper looking and I really like the black pads.  I'm glad I've done most of my learning so I won't be scratching up this EU right away.

I took off the pad to expose the control board.  It is version 1.3.1 which happens to be the same as the control board that came on my Ninebot One E+ that I received a few weeks ago.  My E+ was on firmware version 1.2.7 which had no reported issues when the board fried and dumped me to the ground.  I was lucky I didn't get hurt and was able to get a replacement control board in only a week shipped from Ninebot with the help of Forward California.    The replacement board was a version 1.3 board.  I felt I had been downgraded but after looking at the Mosfet ratings, I was ok with it.  It has worked fine so far.

The P uses the EXACT same control board as the latest E+ EUs.  The ONLY thing different electronically one the P from the E+ is the motor which is specified to produce 60% more power and the battery which is 12.5% higher capacity.  Everything else different is in the cosmetics (pads and black rings).  All for a 36% increase in price.

I verified that the MOSFETs are the same number: 100N8F6 (see attached data sheet and picture below).  These MOSFETs are expected to deliver up to 60% more power with the exact same heat sinks and cooling as in the E+.  This means that the P is capable at overheating much quicker than the E+ if you are using the additional power the motor allows.  

The E+ V1.3 board uses a 75NF75 MOSFET and I've attached the data sheet for this as well.  I would include a picture but it was hard as hell to get the number from the 1.3 board and had to use a stereoscopic microscope to read it.  I have a USB camera for it but my computer isn't recognizing it for some reason.

I did some back and forth balancing in my house to see how fast it would heat up.  I loved the additional power which allowed it to be much snappier and for me to feel more confident.  After a bit of time, I got the beeping from the unit overheating.  I checked the app and it showed 158°F.  It took more work than this to get an E+ to overheat doing exactly the same thing (but with less confidence).

So either the 1.3.1 board is either completely underutilized and over engineered OR the E+ or it is over-utilized and under engineered for the P.  My experience with the blown board on the E+ and the overheating on the P leads me to believe that it is the later.  But time will really tell for sure.

Ninebot One v1.3.1 control board E+ & P MOSFETs

MOSFET Differences

V1.3.1 board on the left (E+ & P) and V1.3 board on the right (E+)

 

100N8F6Mosfet.pdf

STP75NF75 Mosfet (3.1 board).pdf

Hi, 

my ninebot dead and I'm not sure what's the problem, where did you get the control board from please? 

My ninebot only 3 months old, and I bought it from eBay but stall under warranty, do you think if I contact ninebot service by email they could help me and send me new board or I have to buy one online? 

Please advice.

[Vik's]

@Jabirish You can always contact Ninebot directly or through your seller. Many users got help from Ninebot despite that theirs 9b1 were bought from non-authorized sellers.

[Benphysics]

On 12/1/2015 at 8:57 PM, Cranium said:

I received my Ninebot One P today and excitedly opened it.  It is sharper looking and I really like the black pads.  I'm glad I've done most of my learning so I won't be scratching up this EU right away.

I took off the pad to expose the control board.  It is version 1.3.1 which happens to be the same as the control board that came on my Ninebot One E+ that I received a few weeks ago.  My E+ was on firmware version 1.2.7 which had no reported issues when the board fried and dumped me to the ground.  I was lucky I didn't get hurt and was able to get a replacement control board in only a week shipped from Ninebot with the help of Forward California.    The replacement board was a version 1.3 board.  I felt I had been downgraded but after looking at the Mosfet ratings, I was ok with it.  It has worked fine so far.

The P uses the EXACT same control board as the latest E+ EUs.  The ONLY thing different electronically one the P from the E+ is the motor which is specified to produce 60% more power and the battery which is 12.5% higher capacity.  Everything else different is in the cosmetics (pads and black rings).  All for a 36% increase in price.

I verified that the MOSFETs are the same number: 100N8F6 (see attached data sheet and picture below).  These MOSFETs are expected to deliver up to 60% more power with the exact same heat sinks and cooling as in the E+.  This means that the P is capable at overheating much quicker than the E+ if you are using the additional power the motor allows.  

The E+ V1.3 board uses a 75NF75 MOSFET and I've attached the data sheet for this as well.  I would include a picture but it was hard as hell to get the number from the 1.3 board and had to use a stereoscopic microscope to read it.  I have a USB camera for it but my computer isn't recognizing it for some reason.

I did some back and forth balancing in my house to see how fast it would heat up.  I loved the additional power which allowed it to be much snappier and for me to feel more confident.  After a bit of time, I got the beeping from the unit overheating.  I checked the app and it showed 158°F.  It took more work than this to get an E+ to overheat doing exactly the same thing (but with less confidence).

So either the 1.3.1 board is either completely underutilized and over engineered OR the E+ or it is over-utilized and under engineered for the P.  My experience with the blown board on the E+ and the overheating on the P leads me to believe that it is the later.  But time will really tell for sure.

Ninebot One v1.3.1 control board E+ & P MOSFETs

MOSFET Differences

V1.3.1 board on the left (E+ & P) and V1.3 board on the right (E+)

 

100N8F6Mosfet.pdf

STP75NF75 Mosfet (3.1 board).pdf

Since 100A and 80A is relatively close considering  that each pair of transistors add up to at least 16A when the wheel would require a maximum of 100A at extreme conditions, and because both wheels are significantly below 75 V, the only real difference is the 176w vs the 300w. This said, I really don't know if the 300w seems safer... On the other hand, the axle on my Ninebot One E+ broke on the PCB side, and the wires that went through the axle also got sheered at the same time. this fried two mosfets 100N8F6 the two first transistors starting on the left of the board as seen in your pics. Obviously, their parallels remain intact; only one of the pairs fries at a time. Now, I'm looking for replacements, but they all seem to come from China or Hong |Kong which will take at least 2 months to arrive. Do you know of anywhere in the US or Canada where I would be able to get them sooner?

Mind you, I bought a KS 18XL a couple of weeks ago and enjoy it for its speed. On the other hand, I did like my Ninebot One E+ for its agility and to help other learn. I also enjoyed the Ninebot  One for practicing freestyle; its a lot easier than my heavy KS.

Ben