BESC - beefed up VESC motor controller for high power ebikes

[peters]

It was a double pulse test of a half bridge with 2 parallel FETs on an inductor load, but not perfectly balanced. The end of the first pulse is on the left (turn off). I tried only this hard switching, was not interested in soft switching.
This is the Vds of the same test, measured also on the legs at the package:

DC bus voltage was 64V, overshoot is 13V, and there is some ringing at 42MHz, 13Vpp. The total current was about 240A (just out of the range of a 200A ACS758 which tops at 220..230A).
The Vds edges are sharp (maybe too sharp), but the Id rise time is above 400ns - it appears on the Vds waveform as the chopped corner (L*di/dt) before the falling edge, and it is this long due to the high Qrr.

From where do you see it is EMI sensitive, by the gate ringing?
By your 475 and 575ns which transitions do you mean exactly, can you show on a waveform?

 

[zombiess]

From where do you see it is EMI sensitive, by the gate ringing?
By your 475 and 575ns which transitions do you mean exactly, can you show on a waveform?

From what you said, I thought you had an EMI compliance issue.
The 475ns refers to the time it takes the D-S to switch from 10% to 90% of it's range.


So for some wave form info and what it means, I've typed up a post about ringing and overshoot here
https://endless-sphere.com/forums/viewtopic.php?f=30&t=101577

 

[HighHopes]

TD350E sadly yes is going to be, if not already, obsolete. ST got lucky with that one back in the day when they made the gate driver cause it actually worked remarkably well. this is going back some 15 years. for the low to mid level power, this gate drive even with a boost stage was the goto. now-a-days there are more options that are viable. at higher power levels, say >50kW, you really need to be selective because they don't all work the same quality. the switching IGBT/mosfet creates such noise that it crawls over the plastic which means CMRR in the datasheet is irrelevant. there is just so much to think about in that one circuit.. the gate driver. its the most critical part to a properly working motor drive. for <10kW avago, TI, infineon .. all good options. >50kW you need to be more selective.. and yes, a separate power supply .. an unregulated DC/DC with galvanic isolation preferably with <5pf of input-to-output capacitance is desired. power supply datasheet doesn't list in/out capacitance? probably indicates they either don't know its important or they know its important but their part is so bad in this category they don't want to list it. you found a gate driver power supply and like the price but it doesn't list input/output so maybe use it anyway? that type of thinking is ok <10kW, but it will lead to disaster >50kw. i have designed so many of my own power supplies because i couldn't get the mfg. datasheet to show me the info i needed to see. its a lot bigger on the board to design your own but not that much more expensive. maybe sounds like a waste of time.. but when you play >50kW long enough you learn where your time is best spent.

$0.02

 

[HighHopes]

Peters.. before you get bent out of shape about ringing on a double pulse test. put both diff probes on ONE location, say IGBT gate. then repeat the double pulse. notice you still get ringing? that's basically your probe leads.. it represents the portion of the double pulse scope test results that are not real. there are methods to reduce this so that your scope shows you mostly real results.

 

[peters]

Thanks HighHopes but I know the tricks of these measurements. Only a small portion of the amplitude came from the limited common mode rejection of the probe and another small portion was the coupled inductive noise from di/dt.

 

[jlcortex]

Hello Galp,

I am interested in a couple of this PCBs
Do you still have PCBs availables?

 

[apple2]

I would like a PCB as well if available.
Really like the project and I would be happy to build one for my bike.

 

[fredrikhems]

Sent you a PM galp

 

[ElectricGod]

Any progress on this controller?
will they be for sale soon?

 

[galp]

I tested the controller with a small motor on 10S. The current sensors are reversed by accident but it is solved with a "-" in software.

Results look promising.
[youtube]Vak7w4DLClE[/youtube]

 

[shaman]

Nice work @galp! I actually made the same mistake with my current sensors on my most recent OSHW prototype. Where in the code did you add the "-"?

 

[galp]

The amp gain setting. I think this is the way to do it but I'll ask someone to be sure.

 

[shaman]

Might also work by changing the "GET_CURRENTx()" values hw.h . Only 2 out of the 3 of my current sense amps are swapped by mistake. I'll give this a try.

EDIT:
Actually just noticed the "INVERTED_SHUNT_POLARITY" define. Seems like one could define it and it gets take care of.

 

[galp]

Cool, thanks shaman!

 

[galp]

After a few problems with software and few burnt fets the controller finally works. I tested it @300A phase current and it works flawlessly. My ebike now pops wheelies at any speed. I'm sooo happy!

Ialso designed and 3d printed an enclosure to fit the controller in my ebike. It also holds BMS and cables in place and mounts two 80mm fans to cool the controller.

Now I have to waterproof my ebike and go for a longer ride to see how the controller holds.

 

[shaman]

Congrats! Nice to hear about the 300A phase capability. That's some serious torque! Could you share a bit about what you dealt with in the software/firmware?

 

[galp]

I had to invert shunt polarity and gate disable/enable. I also had few bugs in HW due to bad soldering.

 

[Hackey]

How about replacing ucc21520 with si8233 for half the price and it is also part to part compatible.


Sent from my POCO F1 using Tapatalk

 

[Kalzifa]

First I want to say, wow the controller looks very good.
Sencond I want to ask if it is possible to mount TO-247 Mosfets? I would need around 300 to 400 amps peak and figured a margin would be good. The IRF200P222 is a 200V and 182A max Mosfet with 6.6 mΩ. It even has a lower gate charge but at the same time the turn on delay is 25ns...
Whats your take on these. That would take the controller to next level with (maybe) just a redesign of the Mosfet mounting. Am I right or completly wrong

 

[intra]

Hi, great project!
I'm really impressed in general with the amount of development work by the community that has gone into the VESC Project, without expectation of a return on investment - very refreshing

I am developing a high efficiency & high torque density motor for lightweight electric cars. Basic design spec is max torque = 80Nm torque (will have reduction gearing so torque at the wheel will be higher), max power= 30kW, max speed = 7000rpm, 18 Slots/Coils.
To reduce copper losses in the motor, I'd like to run a battery voltage in the 300-500V range.
There are some off the shelf EV controllers available, but usually current and voltage increase together, eg I can find controllers rated to 200A @ 144V, or 600A @ 400V... but what I ideally want (I think, please correct me if I'm wrong) is something like 250A @400V. I don't want a large, expensive controller rated for much higher current than I need.

So...
Does anyone know of anyone that's already developed a VESC which might be suitable for this application (also it must run in FOC mode, have regen braking, and be reliable, design life of >1000hrs).
If not, what would the challenges be in developing a higher voltage version?
Mark

 

[stan.distortion]

Hi, great project!
I'm really impressed in general with the amount of development work by the community that has gone into the VESC Project, without expectation of a return on investment - very refreshing

I am developing a high efficiency & high torque density motor for lightweight electric cars. Basic design spec is max torque = 80Nm torque (will have reduction gearing so torque at the wheel will be higher), max power= 30kW, max speed = 7000rpm, 18 Slots/Coils.
To reduce copper losses in the motor, I'd like to run a battery voltage in the 300-500V range.
There are some off the shelf EV controllers available, but usually current and voltage increase together, eg I can find controllers rated to 200A @ 144V, or 600A @ 400V... but what I ideally want (I think, please correct me if I'm wrong) is something like 250A @400V. I don't want a large, expensive controller rated for much higher current than I need.

So...
Does anyone know of anyone that's already developed a VESC which might be suitable for this application (also it must run in FOC mode, have regen braking, and be reliable, design life of >1000hrs).
If not, what would the challenges be in developing a higher voltage version?
Mark

https://endless-sphere.com/forums/viewtopic.php?f=30&t=89056

 

[intra]

Great, thanks Stan!

 

[galp]

Hello everyone!

Controller is now extensively tested with more than 2000 km on my 15kW Giant Glory. I have decided it is ready for general DIY-ers and started writting a blog page with details how to build the controller yourself.

http://pavlin.si/besc/besc-g2/

The controller has be runing stable for 2000 km with 300A phase current setting on 20S pack. It is actively air cooled with two computer fans on a small heatsink and mounted inside a mildly ventilated enclosure together with battery pack. It can overheat QS-205 with hubsink easily and doesn't even reach 50°C mosfet temp.

 

[galp]

[youtube]MQYNLKpVmGU[/youtube]

 

[bmurai]

Hey Galp, love the work you've put into this. I'm looking at ordering a few PCB's myself and I wanted to confirm the layer order. I'm thinking the 'IN1' layer is adjacent to the top 'f' layer and the 'IN2' is adjacent to the bottom 'b' layer.

(PS: that's some serious speed going uphill on your bike. Can't wait to try this out.)

Edit: I found a diagram for the naming conventions that confirmed it. I'll be ordering some boards from PCBWAY and hopefully testing it out in the coming couple months. Thanks so much for sharing your design variant.