torqueboards
1 MW
Lol.. lots of people must live on the top of a hill. I don't ever have an issue over charging a battery pack through re-gen.
10S custom skate ESC: testers wanted!
- Thread starter austindavid
- Start date
austindavid
100 W
- Joined
- Apr 15, 2014
- Messages
- 204
Right. First let's solve the problems we actually have -- namely, a good ESC. If, after that, we find a substantial number of people wrecking batteries because they live on top of a hill, we can talk. But really I'll just tell you to charge to 4.1V/cell and leave room for that initial downhill segment.
Some progress updates:
I have spent some time testing the new PCBs now. One hardware problem I found was that the current measurement on one shunt was a bit noisy. This was because the path for one of the sense wires was kind of bad. I fixed this on the prototype by adding a small orange wire and fixed the CAD layout of the PCB. Here is a picture (the big red wire is not a layout problem fix, I just happened to short the pcb with a screw driver and had to patch that):
Since it is constantly raining here, I had to do some tests outdoors in the rain (with my kind-of sealed box):
To stress the ESC I set the current limit to 70A and aggressively accelerated and braked for 2km. When going quite steep uphill with full throttle, I noticed how the temperature limit started to kick in towards the end of the hill, meaning that the MOSFET temperature was over 90 degC. Since I was using the 60V FETs and a sealed box without air circulation and no heat sink at 70A current, this is not bad at all. To run well on 70A, I also had to change the commutation algorithm a bit, adjust some hardware filters and implement runtime current shunt offset estimation. Right now I feel quite confident about the robustness of the ESC.
The next time it isn't raining and I have some time, I will do the video logging again including temperature overlay.
The next step for the ESC is to order new PCBs with the current shunt path fix and test them. If that seems to work, we can go for the bigger order with assembled PCBs.
I have spent some time testing the new PCBs now. One hardware problem I found was that the current measurement on one shunt was a bit noisy. This was because the path for one of the sense wires was kind of bad. I fixed this on the prototype by adding a small orange wire and fixed the CAD layout of the PCB. Here is a picture (the big red wire is not a layout problem fix, I just happened to short the pcb with a screw driver and had to patch that):
Since it is constantly raining here, I had to do some tests outdoors in the rain (with my kind-of sealed box):
To stress the ESC I set the current limit to 70A and aggressively accelerated and braked for 2km. When going quite steep uphill with full throttle, I noticed how the temperature limit started to kick in towards the end of the hill, meaning that the MOSFET temperature was over 90 degC. Since I was using the 60V FETs and a sealed box without air circulation and no heat sink at 70A current, this is not bad at all. To run well on 70A, I also had to change the commutation algorithm a bit, adjust some hardware filters and implement runtime current shunt offset estimation. Right now I feel quite confident about the robustness of the ESC.
The next time it isn't raining and I have some time, I will do the video logging again including temperature overlay.
The next step for the ESC is to order new PCBs with the current shunt path fix and test them. If that seems to work, we can go for the bigger order with assembled PCBs.
Some video overlay logging indoors at chalmers with the new PCBs. I have improved the mAh counters and added the MOSFET temperature that the new PCB can measure. Since the floor was slippery I couldn't reach so high currents without making the wheel spin. I will do video logging outdoors as soon as it stops raining.
[youtube]gPOxwxR8cfk[/youtube]
I also changed the motor pulley to 15t, giving a gear ratio of 1:2.67. This works well on the 168kv 63mm motor and I can climb the steepest hills now. The top speed is about 27 km/h now, so increasing the voltage to 10s or 12s would be perfect for this single motor setup. Smaller motors (50mm) should should probably have 1:3 or 1:4 gear ratio for single motor setups.
Next week I will probably send this ESC to Austin so he can test it as well. I will also assemble another one with 40V FETs for up to 8s setups and test it.
[youtube]gPOxwxR8cfk[/youtube]
I also changed the motor pulley to 15t, giving a gear ratio of 1:2.67. This works well on the 168kv 63mm motor and I can climb the steepest hills now. The top speed is about 27 km/h now, so increasing the voltage to 10s or 12s would be perfect for this single motor setup. Smaller motors (50mm) should should probably have 1:3 or 1:4 gear ratio for single motor setups.
Next week I will probably send this ESC to Austin so he can test it as well. I will also assemble another one with 40V FETs for up to 8s setups and test it.
torqueboards
1 MW
Awesome vedder.. cont current looks about 10-20amps. I've been looking into 13S 8.8ah li ion packs. Should be real nice. Also plan on using 14T/44T for 1:3.14. My new motors 5065 200kv + pulleys will have keyways.
speedfast said:
I haven't implemented that yet, but it could probably be done quite easily. I haven't used hall sensors much yet, so the sensorless part is where I have spent the most effort by far. Is there a specific reason why you want to use hall sensors? Unless you have an application like a servo that switches direction all the time, running sensorless should be better.
beto_pty
1 kW
Hi torque..
a bit surprised by your battery choice... you have always been the light/portable advocate .. what gives?
Dont get me wrong... as the 45 km range, 2 hour ride advocate im thrilled yuo are spending some time on this...just wondering why...
a bit surprised by your battery choice... you have always been the light/portable advocate .. what gives?
Dont get me wrong... as the 45 km range, 2 hour ride advocate im thrilled yuo are spending some time on this...just wondering why...
austindavid
100 W
- Joined
- Apr 15, 2014
- Messages
- 204
Personally I'm willing to carry an extra kilo if it means I can last a whole season on a $200 battery string, and use a charger costing under $150. The trick is that (in addition to being very light) LiPo have a very high discharge rate compared to most other chemistries.
Have a look over at the for sale new section. Schwibsi makes battery packs made to order that dont need a BMS and cost just as much as lifepo4. Supposedly the cells that are in makita tools iirc.
Maybe higher voltage (and so less amps and more efficiency) would be the key, making the battery bigger but might be worth it. Only problem is we would need lower kv motors or an easy way to cusomize RC ones.
Maybe higher voltage (and so less amps and more efficiency) would be the key, making the battery bigger but might be worth it. Only problem is we would need lower kv motors or an easy way to cusomize RC ones.
made_in_the_alps_legacy
10 kW
@ Torque, assuming 2C discharge maximum from a LiIon pack, you will need like 10Ah if you want 20A current draw,
but you could get up to 1100W primary electric power from a 10Ah at 13s
but you could get up to 1100W primary electric power from a 10Ah at 13s
It is my first project with electric motors ... I m beginner
But i read a lot and i like these DLRK .
I was thinking that sensored will be ever better than sensorless mode and that the main interest of sensorless is to reduce hardware issues .
And i was thinking that now, sensorless mode with the latest implements was very very near efficient that sensored mode .
But you seem to said that sensorless mode has better performance than sensored mode . Why ?
You use FOC for sensored too ?
If it works exactly the same, i will use in sensorless mode .
It's just that i have already the motor and the sensors .
torqueboards
1 MW
Might need a 38" board to fit everything. Should be nice though with the added power and Wh.
I've been changing my battery packs more often than I would like but perhaps that it is also my lipo battery packs deteriorating.
The price difference isn't much different either. I'll just have a lot more room for pretty much everything and it will be a bit more well rounded battery pack. Increased power/speed on the low end and high end + added mileage. Sure, I'll sacrifice a smaller board + weight but I'll have a separate board for that. I'm also setting up a pneumatic board so a pack like this which isn't that much different from cost will be used + with Vedder ESC's it's doable.
I'm going to standardize 7S and 13S packs for my stuff. These packs have BMS and one plug chargers. Considering a decent lipo battery charger is $160 (350W + iCharger 208B). I prefer for a 13S 8800mah pack lithium ion pack and a 2A/5A charger.
Which if you add up the costs of lipo + battery pack and don't forget to add the low cycle life of lipo's.. It's a perfect trade
Price difference isn't much and the lithium ion is a better trade off. It's just these packs aren't readily available so we turn to Lipo's.
The packs are 40A Cont and 80A Peak which I probably won't need.
It's a step up from the standard 6S lipo's and 6S Car ESC's that are readily available.
Both Evolve and Boosted are using 36V and 42V and all their boards are 38" + and weigh about 17-22+ lbs.
Don't get me wrong. I still like the smaller boards which I'm still going to have but a 7S 8800mah pack. Hopefully, I can wire it up so I can still swap those packs.
I've been changing my battery packs more often than I would like but perhaps that it is also my lipo battery packs deteriorating.
The price difference isn't much different either. I'll just have a lot more room for pretty much everything and it will be a bit more well rounded battery pack. Increased power/speed on the low end and high end + added mileage. Sure, I'll sacrifice a smaller board + weight but I'll have a separate board for that. I'm also setting up a pneumatic board so a pack like this which isn't that much different from cost will be used + with Vedder ESC's it's doable.
I'm going to standardize 7S and 13S packs for my stuff. These packs have BMS and one plug chargers. Considering a decent lipo battery charger is $160 (350W + iCharger 208B). I prefer for a 13S 8800mah pack lithium ion pack and a 2A/5A charger.
Which if you add up the costs of lipo + battery pack and don't forget to add the low cycle life of lipo's.. It's a perfect trade
Price difference isn't much and the lithium ion is a better trade off. It's just these packs aren't readily available so we turn to Lipo's.The packs are 40A Cont and 80A Peak which I probably won't need.
It's a step up from the standard 6S lipo's and 6S Car ESC's that are readily available.
Both Evolve and Boosted are using 36V and 42V and all their boards are 38" + and weigh about 17-22+ lbs.
Don't get me wrong. I still like the smaller boards which I'm still going to have but a 7S 8800mah pack. Hopefully, I can wire it up so I can still swap those packs.
torqueboards
1 MW
okp said:
Haven't tried it much yet or researched on my options. I do know they are a bit heavier which is why I am leaning more towards Lithium Ion.
speedfast said:
I do not use FOC at all, but hardware-wise it is possible. I might implement it in the future, but as far as I know there are no real benefits for small low inductance/high RPM motors. Most likely, FOC will run slightly less efficient if the ESC switching losses are included on a 60k ERPM motor.
The difference with hall sensors is that startup with load will become better, but as soon as the motor is running sensorless operation will time the commutations better and be more efficient. Sensorless startup and low-speed operation is something I have spent lots of efforts on, so it works quite well on my ESC.
onloop
10 kW
- Joined
- Jun 24, 2013
- Messages
- 600
Is the RPM counter correct? seems really high.
it would be great if you can add some more power use data, such as average Wh use per second. or real time Wh use per second. Or peak Wh per second & avg Wh, or a chart that shows total battery Wh and shows Wh consumption in real time & as a percentage of the total.
this data would make it easier to see how to make the most efficient board. also when using the pneumatic wheels it would be good to know if they are much worse in efficiency when compared to other wheels.
thanks
it would be great if you can add some more power use data, such as average Wh use per second. or real time Wh use per second. Or peak Wh per second & avg Wh, or a chart that shows total battery Wh and shows Wh consumption in real time & as a percentage of the total.
this data would make it easier to see how to make the most efficient board. also when using the pneumatic wheels it would be good to know if they are much worse in efficiency when compared to other wheels.
thanks
Jan Christian
100 W
- Joined
- Oct 26, 2013
- Messages
- 129
Yes very good work! I think its ready for group buy. This is probably more testing than the chinese companies has ever done
RPM in the video is measured in electrical RPM, so you have to divide the value by 7 to get the mechanical RPM since it is a 14-pole motor.
W = joules / second
Wh = (joules / second) * 3600 seconds = joules * 3600.
Wh / second = joules * 3600 / second = W / 3600.
Therefore, Wh / second is the same as the displayed wattage divided by 3600
This also makes sense, because if you, for example, consume 300 W for one hour, you consume 300 Wh.Regarding battery watt hours consumed and charged, it is already displayed right under mAh consumed and charged. When those values reach the capacity of the battery, the battery has to be charged.
I could average data over the last 5 seconds or so and display things like Wh/km if this speed is maintained etc. I think that would be useful.
Thanks!
I have connected the ESC to my laptop with an USB cable. The video overlay is done in real time with a Qt application that handles sound synchronization and overlay and opencv to capture and compress the video in real time. This is done with 4 separate threads to keep it smooth and synchronized. I have uploaded the full source code for that application here:
https://github.com/vedderb/bldc-logger
There are also some instructions if you would like to try it yourself.
Thanks! I would like to order one more version of the PCB though, because I made a small adjustment to one of the current shunt paths to get better current readings. I will also send my prototype to Austin and let him test it.Yes very good work! I think its ready for group buy. This is probably more testing than the chinese companies has ever done
