Solar car races

Post and discuss your VESC projects here.
rew
Posts: 943
Joined: 25 Mar 2016, 12:29
Location: Delft, Netherlands.

Re: Solar car races

Postby rew » 05 Feb 2017, 08:38

lizardmech wrote: but motor control looks easy enough.
There are a lot of these things that "look easy enough" until you dive in. An example is a datasheet of a CPU. You might glance through it and think: Looks usable. Then when you start using it, precisely that section that you need is absent, badly written or inconsistent.

For C2000: It took me a long time to get the compiler to work. One of the things was that gcc-for-C2000 had a bug. It was noticed years ago, but nobody had taken the trouble to fix it for the distributed gcc. So I had to go in and patch it myself.

Oh anotherthing that looks good on the surface but is in fact "bad": They give you the code to drive the motors. However the code that you need to call to get the motor to spin is so complicated that they have to give you that code too. They end up writing a code generator for you. So you go clicketyclick, and it generates your source code for you. But now... when something changes, you get several not-pleasant choices: rerun the code generator and re-merge your code into the generated code. Rerun the code generator, try to figure out what changed and merge that into your code.

Similarly, they GIVE you the startup source code. YOU are meant to change it if you want to use an external clock or something. So when you get a new revision of the compiler suite, you get to review if they've fixed anything in one of the startup files that they have GIVEN you. Much better would be to have a default startup file. The startup file could call different hooks so that if necessary you can change different parts of the startup sequence. And if really necessary, you could override the global startup file.

You end up with a mess of system files and user-source files in your working directory. It's a mess.

Chibios suffers from some similar problems, but it is not that bad. You are supposed to COPY the ch.conf and hal.conf files from an example, and modify those to your liking. This breaks down when chibios changes something and requires a change to all the .conf files.

tecnologic
Posts: 15
Joined: 26 Feb 2016, 08:23
Location: GER / Hamburg

Re: Solar car races

Postby tecnologic » 06 Feb 2017, 09:22

@rev: full ack, i had the same issues squared. dual core c2000 :x

lizardmech
Posts: 171
Joined: 19 Jan 2016, 10:54
Location: Australia

Re: Solar car races

Postby lizardmech » 07 Feb 2017, 15:49

The analogue devices cm40x mcus have M4F + DSDF +high res PWM, downside is expensive dev boards, I couldn't find exact details for PWM res.
I was looking at the STM32H7 specs but I can't see any specialized PWM just the boost for higher clock speed.

What is the most demanding part of the VESC firmware? The transforms for FOC or the sensorless observer?

tecnologic
Posts: 15
Joined: 26 Feb 2016, 08:23
Location: GER / Hamburg

Re: Solar car races

Postby tecnologic » 07 Feb 2017, 20:32

Both are not very expencive in computing time. What pwm/ control frequency are U targeting?

STM32H743 has hrpwm as the cm40x.

lizardmech
Posts: 171
Joined: 19 Jan 2016, 10:54
Location: Australia

Re: Solar car races

Postby lizardmech » 08 Feb 2017, 03:18

You need 100khz-200khz for high efficiency ironless motors. ERPM is typically extreme as well. The ST page lists the H7 as having 2ns PWM resolution, better than the F4 but still a long way from the c2000 and KV5x, those are 150ps and 250ps unless I'm reading something wrong.

High res PWM has some other possible uses as well, I have seen some ideas such as using a multiphase bidirectional buck-boost running around 1mhz to perform voltage and current regulation instead of the motor. With GANfets it's possible to use such extreme switching speeds meaning only a small inductor is needed, it would look something like the VRMs on high end computer graphics cards. The same thing is useful for supercapacitor banks, they require bidirectional buck-boost to work well with electric vehicles, with slower silicon solutions inductor sizes became an issue.

tecnologic
Posts: 15
Joined: 26 Feb 2016, 08:23
Location: GER / Hamburg

Re: Solar car races

Postby tecnologic » 08 Feb 2017, 08:20

OK, just a quick calc. 200kHz -> 5µs / 2ns = 2500 -> >11Bit. So this is just the same Resolution as VESC has now at 40kHz. KV5x is also great only disadvantage is that Freescale designed this and the doku will be crap, but thats no show stopper. The missing interfaces to connect better adcs is the only problem with this cpu.

Both the Buck and the FOC do not need such high control freq. 50kHz is enough, when u have not more than 5kHz electrical field revolutions. Meaning 10 control steps per rev is a minimum to maintain a stable observer. So 200kHz PWM with every 4th cycle being a control cycle will suffice.

When u talk about ironless motors are this machines PMSM or sync reluctance machines? With SyncRel u are lost with current VESC, than u have to dive in the mathmatics and read some dissertations.

regards

Tec

lizardmech
Posts: 171
Joined: 19 Jan 2016, 10:54
Location: Australia

Re: Solar car races

Postby lizardmech » 08 Feb 2017, 08:43

They're still PMSM just lack any sort of iron stator, there's many variation on the design, CSIRO motor and thin gap. You can reach 96% efficiency but they have two limitations, stator temps are difficult to control, 100khz+ switching frequency is beyond most controllers, you need high speed high accuracy current sensing to go along with it. There's only one company selling controllers for them they charge $5000 just because no one else sells any and solar car teams will usually pay it.
Image

tecnologic
Posts: 15
Joined: 26 Feb 2016, 08:23
Location: GER / Hamburg

Re: Solar car races

Postby tecnologic » 08 Feb 2017, 13:54

These are cool motors. Do u have a datasheet of this? Whats the mean coil inductance? Are the holes on the composit part for hall sensors? Then the software design is quite easy.

How many poles do these motors have? Whats the peak rpm? U sad some thing about 10kW of power, rated? The CSIRO motors i found had 1.8kW rated power. What voltage has the battery pack?

lizardmech
Posts: 171
Joined: 19 Jan 2016, 10:54
Location: Australia

Re: Solar car races

Postby lizardmech » 08 Feb 2017, 16:06

Can't find specs for any of the coreless motors, the CSIRO one has a huge spec sheet but they didn't bother to test inductance and just say to add three 100uh inductors to it because their motor controller wont work. They run it at about 130V supposedly they can generate about 50nm and 10kw of power but overheat in a few minutes, 1.8kw is max sustained power. The specs list around 100mOhm for resistance.

The interesting thing about the motors is because they have no metal stator they are pretty easy for hobbyists to build. The main reason you don't see many of them as most people don't have access to a suitable controller. You really need FOC with sine wave current outputs. Most hobby ESC are just BLDC while most FOC controllers including VESC can't handle the switching requirements.

tecnologic
Posts: 15
Joined: 26 Feb 2016, 08:23
Location: GER / Hamburg

Re: Solar car races

Postby tecnologic » 09 Feb 2017, 08:38

Ok lets recap the motor hast 100mR Phase resistance with approx 5-10µH inductance, meaning worst case 5µH/100mR = 50µs electrical time constant.
so in 50µs the current reaches 63% of the short circuit current of 130V/100mR = 1300A this leeds to 820A/50µs --> 16.4A/µs current slope. At 100kHz this leeds to 82A current ripple at 50% duty. With 10kW max power at 130V this means the current reaches zero in every pwm cycle even at near max load!!! But this is only the case at standstill with rpm the emf lowers this effect.

So the 100µH extra are needed. Then u are at around 8A ripple at 100kHz. To eliminate the 100µH u need around 1MHz switching freq. Or a very good decoupling of the FETs to feed the 16.5A/µs. But this kills the most controllers, because every nH in the conducting path counts here.



may be u have a look at : http://www.sinusleistungssteller.de/en_index.html.


Return to “VESC Projects”

Who is online

Users browsing this forum: No registered users and 1 guest