DIY Electric Car Forums banner

61 - 80 of 118 Posts

·
Registered
Joined
·
730 Posts
Discussion Starter #62 (Edited)
Schematics for RS232, digital outputs (relay drivers) featuring fully protected smartmos Infineon BTS3110N HITFETS, and version 2 of the digital inputs.

EDIT: forget isolated inputs, not enough pins on connector. Missed that bit in Damiens video.
Digital inputs V3 has transient suppression, because the STM32 has +/- 5mA functional EMI susceptibility on most GPIO pins.
Some even less: 0 mA on PA4,PA5 and PC13..PC15. No current injection allowed on these pins for reliable operation.
 

·
Registered
Joined
·
730 Posts
Discussion Starter #63
CONTROLLER BOARD DESIGN FROM SCRATCH

Well, it has become clear from Damiens video and posts in his DIY Tesla Controller thread that a Tesla drivetrain is a CANBUS system.
The highly integrated drivetrain is a great starting point for a SiC build, but there just aren't enough pins on the external connector for a lot of parallel I/O. Only brake and throtlle.
I'm going to start from scratch with the design of a CAN and SPI based controller board with xxxuino compatible Atmel MCUs.
Probably a 32 pin ATMEGA64M1-AZ for the main board and another AVR on an external CAN satellite board for I/O: throttle, shift stick input etc.
 

·
Registered
Joined
·
37 Posts
Oh yeah, IBGTS for now, and once again found confirmation that an ARM 32bit has lots of spare resources like computing power in a motor controller.
That microcontroller is a monster, and most important, the FOC code and graphical user interface (CAN-enabled) is already developed =)

And for the powerstage I guess it could be a matter of swapping the TO247 IGBTs for TO247 SiCs when the day comes.

I should make a thread in this forum someday about these
 

·
Registered
Joined
·
2,170 Posts

·
Registered
Joined
·
730 Posts
Discussion Starter #67 (Edited)
Phew, Damien (jackbauer) had me worried a litlle bit when he posted that the current signal from the Tesla drivetrain is digital (as in: not analog). But luckily, it looks like it is a sigma-delta bitstream. Phew again.

nitrousng said:
And for the powerstage I guess it could be a matter of swapping the TO247 IGBTs for TO247 SiCs when the day comes.
Well, it isn't that easy. I tried two TO247 SiC in parallel (in my charger design thread) and messed it up pretty bad.
But I'll try it again when prices come down and if I can find the free time for it.

nitrousng said:
That microcontroller is a monster, and most important, the FOC code and graphical user interface (CAN-enabled) is already developed =)
and
dcb said:
what is wrong with a code compatible board with more pins?!?

http://www.mouser.com/ProductDetail/...LYMmnI2g%3d%3d

with all kinds of other features on it (built in easy to use programmer and etc, even handles arduino shields, and an online compiler if you like)
Yeah, that's it. For both. Very nice boards, but with too many features. I want to reduce it to the bare essentials for a Tesla drivetrain. So less pins is better (also for DIY). Probably even no FOC. Well, maybe.

And I am not sure anymore about porting software. It is not that difficult to implement basic ACIM control and CAN connectivity. On a 16 MHz ATMEGA SVM VFD results in a CPU load of about 10%.

I'm not going to use development boards / shields in my build.
 

·
Registered
Joined
·
730 Posts
Discussion Starter #68 (Edited)
Here are the schematics of the ATMEGA64M1 uC basic ACIM controller.
Overcurrent protection is in SW. IGBTs are protected by DESAT.
DESAT still results in a shutdown of the power stage controller section of the uC.
Other parts can survive an overcurrent condition for a longer period of time.
Response time approximately 500usec (group and phase delay of current sensor filter), which is adequate for motor windings etc.
 

·
Registered
Joined
·
2,229 Posts
Could you give a description of the current sensor filter circuit? Might just steal it:D
 

·
Registered
Joined
·
730 Posts
Discussion Starter #70
Chebyshev LPF, characteristic frequency 1000Hz, 0.2 dB pass band ripple.

REV 2 has already been designed :D (for FOC): 900 Hz, 0.05dB ripple.

Feel free to copy it. Especially for educational purposes. :D

No guarantees. Certainly isn't going to be fast enough for HW detection of overcurrent.
 

·
Registered
Joined
·
730 Posts
Discussion Starter #71
The topology in the sigma-delta filter schematic can be used for a great number of filter types.
Change the following parts

R7,R8 910
R5,R6 6800
R3,R4 5100
C6,C7 1.5n
C3,C4 150p
C2,C5 330p

Opamp MCP6292

and you get a Bessel 3rd order Low Pass Filter with a characteristic frequency of 48KHz, low signal distortion in the pass band and high attenuation (60dB+) above 1MHz. Step response 0 to 100% is 7 usec.
 

·
Registered
Joined
·
730 Posts
Discussion Starter #73
Textbook classic. The topology is widely used and appreciated for its simplicity and versatility. It has been around since 1955.

Don't hesitate to make good use of it if you think it's ideal for your application.
 

·
Registered
Joined
·
2,229 Posts
Thanks Tony. I'm going to give it a shot in the next rev of the Tesla board.
 

·
Registered
Joined
·
2,229 Posts
What clock speed would that equate to for a 16 bit conversion Tony?
 

·
Registered
Joined
·
2,229 Posts
:D What R value does a LTC1799HS at the set pin need to run at 10 MHz?
I have been trying to work that out myself. I can't make the equation in the datasheet work. Not much of an engineer:eek: So I just put a 1 Meg pot on the PCB:D
 

·
Registered
Joined
·
730 Posts
Discussion Starter #79
I hope my filter design can approach 9 to 10 bit resolution (the ADC of the ATMEGA64M1 is 10 bit).
BUT: I'm going to use a 900 Hz 3rd order Chebyshev and the 16 MHz CPU clock.
That is, if you find that the P85D front drive unit uses a similar circuit as the RWD.

Might give you some indication of what can be expected from a 48kHz Bessel @ 10 Mhz.
 
61 - 80 of 118 Posts
Top