20s Little FOCers on stooge SRB v5#1

3 Likes

While I was at it I notice how exposed the underside of this Vesc is and am worried that the excess of dust / debris may short something or do some other not so awesome stuff to it so I am think of taking the pcb out of the aluminum casing and throwing it in a ultrasonic cleaner filled with isopropyl the only thing that am worried about is somehow the caps or capacitors still having remnants of power… I haven’t turned this Vesc on since esk8con but I’m not sure how ballsy I wanna be is there a way to test if it still has voltage or power lying in it or am I just being over cautious!? Ideally after this if the cleaning is needed I’m gonna try and figure out how to seal this Vesc up so dust won’t immediately come back in. Thoughts suggestions are much appreciated… thanks guys :slight_smile:

1 Like

:+1:

a poor connection would be a higher resistance connection which would lead to heat, which might lead to desolder.

given it’s 20s, the currents are lower for the same power as a 12s board. which reduces the concern.
but it’s a fine upgrade.

2 Likes

This is the heatsink. there will be thermal pads or thermal paste joining the fets to the heatsink. you’ll need to replace it with fresh stuff if you touch it. and be cautious about separating it to not damage the fets.

i personally wouldn’t touch it unless it was broke.

maybe.

it can’t hurt to seal things up better. but i’d skip taking it apart for the extra super clean unless you can tell there’s a problem in there.

2 Likes

What kv motors were you having issues with heat at 20s?

135 7490’s which we thought would be better because of the hollow core but wasn’t. Too many losses in play. I’ve had better luck with the 138kv 6485 reachers and hopefully will be going 4wd soon which will help even more with heat dissipation.

1 Like

The 7490s have 10 pole pairs, which makes it very prone to iron losses. If you want to spin large motors at high rpms you are better off with the 6385 or 6395, you can even up the KV with those

4 Likes

Just to chime in, high kv 7490s is why my v5 is so inefficient at >100wh/mile. At full speed, the iron losses alone consume like 5kW of power. Using them was the worst decision I made with the design. (there’s more detail in my build thread)

8 Likes

I Will seal it up and leave it alone thanks fess!

hmm Just thinking what if…
20s6p using 21700 cells,
potentially Utilizing P50B Cells when they become Available


3 Likes

4 Likes



Playing around with some prototype concave designs
Super rough idea of what it’s supposed to look like but it should help with toe side turns and with heel side turns you’ll pivot your foot off the the concave and rest it on the flatter portion.

I’ll keep you guys updated once I start 3D printing real deal and whether it actually helps.

4 Likes

Concave makes a huge difference! What’re you modeling in, is that blender(?)

Haven’t tried adding it to the rear of my v5 (yet), but I’ve got a mbs camber pad on the front. You probably don’t need as much as you think, just 5 degrees made a huge difference.

5 Likes

Sorry for the late response man I just had a sailboat to mexico and I got back a couple hours ago… it’s embarrassing to say I’ve used solidworks, fusion360 and also some other programs but since I’m currently not in school I dont have the licensing for it anymore but I use tinkercad for quick ideas like this hence why it isn’t as detailed. But it honestly works great, I find it much faster for basic things you are modeling. But I wish I had the cracked version of solidworks Lolol I’ve been trying to download it off torrent sites but it’s like awfully complex to unpack as a desktop application. Maybe it’s not that complicated but idk I’m having trouble anyway.

Graduated two years ago, so I lost access to student version of solidworks and fusion 360 too.

I’ve switched to onshape for all my stuff, highly recommend.

1 Like

Ill definitely try it out thanks man

final thoughts

power

TL;DR; castle more power.

v5#1(vesc) never quite hit the same power is v5#3 (castle).
i’ve measured max 16kw on v5#3 and max 12.5kw on v5#1.

Specs wise, the lilfocers should have been able to pull 8kw/pc.
It’s quite possible that the difference came from v5#1(vesc) having one less lipo (5 instead of 6) worth of power.

subjectively, v5#3(castle) felt more powerful to myself and others who rode both. Gialong at AVS in particular put them through their paces on that course, and chose v5#3 to race.
IMO part of that subjective feel is the the throttle difference.

Relatedly if you look at performance of the single hobbywing single motor setup @MarioChacon has been running. And note that everyone in vesc land is going to 4wd to get the same power, it really paints a picture that the RC car ESCs are well ahead of a majority of VESCs in terms of power right now.

IMO tho littlefocer’s were very close in this setup just not quite there. maybe gofoc-300s could do it ? maybe it was the batteries? i’d still like to find such a setup. but i’ve lost my test platform.

throttle, duty_cycle control(castle) vs current control(vesc)

throttle responsiveness

TL;DR; castle more control vesc more smooth

Current control lacks the responsiveness that you notice when running at the edge of traction and balance around a turn on the track. It also lacks the punch of duty_cycle control. I suspect both of these are due to input lag and that PID tuning may help. Sadly, I never got around to testing PID turning. Punch wise, a litte throttle expo might help current control feel more jumpy. but it won’t help the lag.

Note Vesc throttle ramping needs to be set very very low (off), as it’s an artificial lag in response that would skew all these tests in duty_cycle’s favor.

throttle range

TL;DR; mixed, vesc more power fidelity, castle more speed fidelity

if you’re at mid speed with duty_cycle control and let off the throttle. you have to come all the way back to mid throttle to match your speed again, and then only the remaining range of the throttle is available to add more power. you could see this as less range to control remaining power or more range to carefully match current speed which matters balancing through turns.

with current control if you’re at mid speed and you let off the throttle, you only have to travel a small distance before you’ve matched enough current to maintain current speed, and the rest of the throttle is available to control the remaining amount of power. this more range to control remaining power less range to carefully match current speed. which matters if you wan to carefully add more power and maintain balance in a turn. enjoy the symmetry.

IMO the matching the speed bit feels like more fidelity. the more range to control the power feels ‘smoother’

These differences in control require vastly different muscle memory and the reason it’s so troubling to switch between the two. (on high power boards)

smooth starts

TL;DR; vesc more smooth.

castle background

In the castle controllers, there are two kinds of motor current limiting: punch control, which we run at 100%, and torque limiting, which we have disabled. Torque limiting is motor current limits, and thus we’re running “none” all the time. I’m not sure how punch control is implemented, it’s intended to prevent burnouts. I suspect motor current limits only when rpm is very low.

duty_cycle control simplified model

With duty_cycle control if you command 50% duty_cycle, the motor will instantly try to turn at 50% top speed (x max power %) If it can’t, the motor instead demands current which results in torque until it reaches that speed. The amount of current demanded is proportional to the difference in actual rpm and target rpm. This can be a lot. [up to infinite if the motor is held stopped. ]

castle smooth starts

If you give 5% throttle the esc targets 5% duty_cycle and will use any amount of current to get there. To get a smooth start you have to smoothly control the throttle so that the delta between current speed and commanded speed (duty_cycle) ramps up at your control. if your muscle memory is vesc you will usually push more throttle then you need and it will jump. in fact for anyone you’ll usually push to fast into it. it requires fine control at the starting end.

you can learn this, but it’s the harder of the two types.

current control simplified model

With current control, if you give it 50% throttle, the ESC tries to target 50% of motor current max, in order to do this it begins watching the current sensors and using a PID loop tries to adjusts duty_cycle until it hits the desired current. this PID loop provides a ramping that contributes to “smoothness” and also a delay that contributes to “lag”

vesc smooth starts

From start on a vesc if you give 5% throttle, you’re only asking for 5% current. The vesc will begin ramping duty_cycle at whatever the PID control’s response rate is, until it sees 5% current. this is likely going to be less than 5% top speed so the rider will give more throttle before getting to the same 5% speed in the castle exampel above.

So, vesc gets to use more range of the throttle on start,
and gets the built in ramping of the PID loop.

both contribute to more smoothness over castle / duty_cycle control.

Adapting between the two

it’s not easy. familiarity bias comes into play a ton.

going from current_control → duty_cycle

you will notice that starts are jumpy, less power fidelity.

you will notice that after reengaging throttle you need to throttle in further than you’re used to to to match your current speed.

going from duty_cycle → current_control

you will notice you over throttle when reengaging throttle. you need less throttle to hit the matching speed point before you add power.

you will notice notice that power seems to engage later on, you’ll hear people say “more mid range power”. I think this is just PID lag.

you will notice power feels like it runs on as you attempt level off speed. this is because in duty_cycle you hold throttle to maintain speed, with current control you need to learn to back off throttle and match current to speed.

summary opinions

As someone is used to the GT2B and duty_cycle control, and used to hoyt pucks on current_control on less powerful boards. and now GT2B on vesc. as well as OSRR on vesc. I will say that you can get used to both. switching requires conciously switching your muscle memory.

However, I did try to run duty_cycle on hobby wing using a hoyt puck, IE using my thumb for the first time on duty_cycle. This was the hardest thing for me to adapt to. In particular I had trouble finding matching speed point exiting turns and it was throwing off my balance.

So my vote goes to learning duty_cycle control slightly more difficult than learning current control.

Therefore, I concede that @jack.luis is right. ( by small margins) vesc current control is smoother and easier to deal with in a way that makes it better for a street board.

I still maintain you can learn both and it’s not that big a deal. but I admit it’s there even when you overcome familiarity bias. I do maintain the fidelity is better on duty_cycle due to lack of PID lag, if you’re pushing your board that hard like track riding does.

fin.

Thanks for reading this far. I hope someone found this brain dump useful. I worked hard to learn something from it all i figured someone might find it interesting.

sry it’s so long, I spent time to make it shorter. I’m out of time to make it actually short. :smiley:

19 Likes

related to the duty_cycle vs current control talk:

I personally think it’s input lag, but this thought takes it further, suggesting that it’s directly speed control. (duty_cycle) which helps manage your grip around turns.

i should probably break this control mode topic out of this thread.

1 Like

I thought VESC can also run duty cycle control? Or is that something different from the duty cycle control of the castle / hobbywing?

1 Like

funny. i was just making this subtopic it’s own thread.
and I opened with that difference.

let’s continue there.

leave this thread to be the journey of V5#1

4 Likes