THE Overkill - 4WD 36KW 20S13P 100+MPH build from scratch

This build is brilliant. @FullMetal_Machinist is really trying his best, to come up with good solutions. If you guys read bak, you will see, that he never ever said, he will go full trottle on it… but he said, he will climb up the ladder…

Come on guys, just think on this project on the way: What is possible, what isn’t? If you don’t try new ideas, you will never know where an esk8 can be in some years from now. I guess when someone built the very first esk8 everybody around were laughing, that it won’t work…

I myself really like it, that there are some people, who can think out of the box. I even do understand people have their concerns, but come on guys, be positive, or just let @FullMetal_Machinist do his project… It’s his own money he is willing to spend on something he believes can work. Thumbs up man :slight_smile:

23 Likes

Finally someone who knows what i feel. Truthfully i would love to go to a drag strip with my board and outrun 600cc stock bike but it is not something that can be done immediately. A lot of trials, tweeking and fixing stuff that was not good enough is on the way. And it needs to be done, just like any other good build out there.Peole like Kug3lis https://www.electric-skateboard.builders/t/monster-board-trampa-holypro-35-80100-180kv-8kw-fatboy-ss-gear-drive-12s-3p-18ah-dual-focbox/36560 or Pedrodemio also understand as they had to go this road as well. Those are the people that make all the interesting, technical comments instead of typical lols and rotfl. Also seen several attempts of designing ESC on this forum, albeit none of them succeeded for now. Hope mine goes better. Cheers!

11 Likes

I guess you speak about escs with a higher S rating as usual, because if no than you not right with this point. The ESCapes by @stewii have been a great success and still outstanding for a vesc 6 based esc. Please also keep in mind that there is not only this forum where people work on vesc based escs. I just dump this here for example https://vesc-project.com/node/339

3 Likes

@FullMetal_Machinist I love your attitude!

Skating has always been about pushing boundaries - your pushing so many :slight_smile:

Also I think ‘ive built somethingso fast I will have to practice to stay on’ is one of the BEST problems to have ever!

1 Like

Looked over his design and compared to mine. Pity it already is copied by chinese. Read his thread carefully. He used VESC6 as a base so 3 current shunts on high voltage side, i wanted to avoid that at all cost. His board uses single drivers per group of mosfets so he gets 4a driving current over 45nf of gate capacity, I use double drivers per group and different mosfets with less capacity and faster switching so i get 12a driving over 16nf of gate capacity (numbers combined for both). This leads to my design switching almost 10x faster, lower dead times (his are 330ns, mine TBD but 20-30 ns projected). Another beast is that he does not have step-down on his board and it needs separate voltage supply for logic, mine features power section and is perfectly fine with only main voltage. Designs of power stage in both cases look almost identical (component placement, current paths) only difference is component size and count. My design will have round wires 4mm thick soldered onto PCB as it is much cheaper and does not overheat entire PCB during soldering (he mentioned it takes him 5 mins of heating), as he made water-cut busbar from 3mm copper sheet for 50$.

My design is slightly cheaper in comparison, i hope to get to 20S voltage - his is 16S, 18 max.

His design is tested and works after sorting some shenigans, mine is waiting for first PCB to arrive. We will see in a week or two, depends how much free time i will have. Thanks for posting the link. Did not know someone got this far.

8 Likes

I was referring to your last post only. And in this post you was not clear enough formulated. It’s just not right to say that all this forum totally sucked in designing vescs. That’s definitely not right.

Read my answer please carefully! I ask you just to specify your statement. Your idea to create a esc like you doing is outstanding and can’t be compaired with something, somebody tried here before, so you also can’t say we failed in the attempt in doing it!

Sorry for misspelling. Not succeding and failing are different and i wouldnt consider changing 1 chip designing, only minor tweak. Will try my best not to badmouth anyone, but english is not my first language.

3 Likes

All good :wink: waiting 2019 to be the year of 18-20s escs :pray: and maybe some Esk8 Guiness Records…

2 Likes

Start on a steep downhill :joy:

Did u do magnetic modeling to decide on those huge 5mm thick magnets?

@FullMetal_Machinist, build looks awesome, mad but awesome

How are you going to passively cool the motors without airflow and with a current draw that large?

@FullMetal_Machinist Have you considered high temperature super conductors for the stater core? If you add a pressure vent at the ends of the rotor housings and pot the magnets with epoxy (anything to make them smooth all the way around) you can flood the motor with liquid nitrogen and drop the resistance of the motors and pull far more power into them, as long as the magnets don’t crack the torque you get is based only on the internal resistance of the battery and the magnets wouldn’t demagnetize under load. Also the wheels would ‘smoke’ as you go faster and that is amazing

You could also use liquid helium since copper should super conduct at 4K but its kinda expensive

2 Likes

Idea is that motors are so oversized that during normal riding at 50 kph they will be under 10% nominal current and heat will be so small that passive cooling is enough and during high power runs actual working time is 10 seconds and it is not enough time to overheat the motor even at max power, as motor thermal capacity is not small. If the motors start to overheat after all i will be forced to switch the magnets to higher temperature grade, which was yet unavailable when i first bought them. However this is not expensive or time-consuming considering rest of the build at 350$ and 1 day of work. Tires will smoke with actual design, as design was made to use full grip of the tires (supersofts on tarmac). If treaded dirt tires are used it will be impossible to apply over 50% of throttle without smoking them, on gravel or grass grip limit is under 25% of the throttle setting.

3 Likes

Okay makes a little more sense if your not going to use it on 70 mile treks

Any idea what your quarter mile time will be?

0.02 seconds or DNF cause he almost died doing mach 5 on an Esk8.

7 Likes

first post< title says it all THE OVER KILL boldly go where no man has gone before theirs Klingons on the starboard bow

Switching those MOSFETs hard.

Great for minimising power loss across the FETs but man, that’s going to create some nasty voltage spike when switching those high currents. Are you using 100V rated FETs? I may suggest that your controllers will need some very low ESR capacitors across the DC bus to soak up those voltage spikes, and very close proximity to the FETs. I believe such voltage spikes are probably the reason that other fellow’s controllers can only run 18s max even though they are 100V MOSFETs. You’ve probably already considered all this, but just in case, thought I’d mention something. May I ask why wanted to avoid the 3 shut design?

I would love to have 3 shunts, but not on high voltage side. This is what causes 50% of spike problems. Someone https://www.electric-skateboard.builders/t/sesc-a-vesc-6-derivative/58106 has done this by modifying VESC6 firmware to use shunts on ground side. I cannot modify firmware (not my field of work), so i was limited to using 4.12 as a base. For technical side it would add roughly 20 components to the design, no big deal for me. Other spike generating issue is power routing by itself. As described here https://vesc-project.com/node/655 there is problem with ringing at radio frequencies, common to all high-current ESC, however some people do not get to the point of identifying it and vieving spikes on oscilloscope, as most digital oscilloscopes do not “see” frequencies this high - it is over their bandwidth. It is caused by high current path creating a circular “loop” on the PCB itself. It can only be remedied by redesigning with different current paths, ground planes, multiple current layers etc. Problem is more pronounced with high-speed switching as in my design. Spikes can be as high as 50% of voltage supply. Decoupling capacitors (my design has over 30) and their proper placement help somehow but do not get rid of the problem source itself, which is the physical size of mosfets. Low-ESR electrolytic or polymer capacitors are too far (physically) to even count. Getting them closer makes the loop get actually bigger. This is the reason for me to use smaller powerSO-8 package instead of D2Pak-7. Another thing that can be done is isolating external inductance (means motor) by means of RC filters and high-speed diodes, capable of “catching” those frequencies. Mosfet internal diodes are too slow by an order of magnitude to do the job. My design can accomodate 100V or 120V fets. I start with 100v ones, as they have lower internal resistance. First tests will be done at lower voltage, so i have room for spikes, then when filters are working i will up voltage to 20s, as supply voltage has no effect on the presence of ringing, only on spike magnitude. What is important is to run tests at high current and with low external inductance (motor with few turns on the stator), as low inductance promotes spikes. I encountered this problem multiple times with other designs and was able to subdue it somehow. In worst case i myself blew 2000 Amp 1200volt IGBT at 460 volt power supply. Ringing was cauced by long motor leads (industrial application, very big DC motor). Testing in-house had no effect bencause problem was at the installation site. Decoupling had to be applied exactly in the middle of motor leads(sic!). Other time friend from Warsaw University of Technology called me bencause 100KW 30khz transformer i made for them was getting too hot. When i looked at oscilloscope the top horizontal line of square waveform was of 400volt thickness. They didnt even realize it was continous line of 2Mhz spikes. Ended up swearing fucks to a real colonel who was supervising the project. Then i sent them a bill for on-site consulting. So in the end i hope that my previous experience will help me somehow to make this a reality.

I

7 Likes

Excellent,I hope your testing goes well and look forward to hearing about your results when your PCBs arrive shortly :sunglasses: