Puma board for FreeEMS

[KW1252]

I have to admit I've been somewhat cursory on catching up with the development, but I was thinking about how much emphasis there will be on CAN connectivity. Right now there's CANRx and CANTx pins exposed, but to use them a tranceiver circuit is needed. So, how's the policy on that?

[Fred]

I have to admit I've been somewhat cursory on catching up with the development, but I was thinking about how much emphasis there will be on CAN connectivity. Right now there's CANRx and CANTx pins exposed, but to use them a tranceiver circuit is needed. So, how's the policy on that?

Circuitry and code, I welcome people adding both in a purely experimental off-by-default way.

Fred.

[jharvey]

Connector card should allow for those to be implemented. Would you be interest-able in taking a stab at the first connector card?

[KW1252]

Here's a quick shot at what the CAN connector card would look like:



Of course, DE-9M is not a good idea to put inside the case. A proper setup would have a 2x1 pin header with a FTP cable connected to the ECU connector.

[Spudmn]

There needs to be a place for the termination resistor if it is needed.

Also it's a good idea to run the GND line along with the CAN differential pair. This will keep both ends of the CAN bus at the same potential.

In my set up I am going to run power as well, so that I can power little CAN Modules off my CAN bus wiring. You may want to add 12V to that connector.

[KW1252]

The device is configured as high-speed unit on the connector, I was thinking of cable-terminated bus; with a resistor, it could use a configuration pin for slow speed mode too. The DE-9 footprint was the first idea, but I would much rather have a pin header instead and have the actual CAN connection on the ECU connector, with the power line routed from elsewhere.

[nitrousnrg]

Glad to see these things being discussed :-)

About the 12v in the header, what kind of things would you be hooking there? Those sort of 12v are very crappy.

[lumpensack2003]

The device is configured as high-speed unit on the connector, I was thinking of cable-terminated bus; with a resistor, it could use a configuration pin for slow speed mode too. The DE-9 footprint was the first idea, but I would much rather have a pin header instead and have the actual CAN connection on the ECU connector, with the power line routed from elsewhere.

You mixed up two different resistors used with the xceiver:
- at Pin 8 you'll connect an appropriate resistor for slope control. As you wrote, connecting Pin 8 to GND will select High-Speed mode with steepest slopes.
- nevertheless there should be an option to stuff a 120Ohm resistor between CANh and CANl. Following the CAN specifications, the bus has to be terminated with 120Ohm at both ends. Assuming the FreeEMS is on end of the bus, the termination resistor should be there in most likely every installation.

Furthermore, there should decoupling capacitors (10nF and 220nF) close to Pin 2 and Pin 3.

Yes, it's a good idea to carry +12V and battery GND with your CAN signals in the CAN cable. Due to CAN being a differential bus, there is no need to connect the CAN cable GND to FreeEMS GND. Connecting both GNDs is potentially harmful.

[KW1252]

Okay, caps added to the circuit. Rs pin does more than just control the slope. If the pin is put high, the chip will go into standby mode, still functioning as RX node at reduced power consumption. I made a simple circuit that uses IGN voltage to drive this. I also used a configuration pin to determine whether controlled slope or fast signal rise and fall is used. When there's no voltage on the FET gate, the pull-up resistor puts the Rs pin high. Once voltage is applied, the pin falls to low, and into the set mode whether the path to the ground is shorted or through a resistor.

Here's the quick and dirty schematic; the component designator for the transistor is wrong and needs to be fixed and the resistor values aren't even close.


and the board.



There's 12v running now to the board, but I still wouldn't consider running it to the header, as switched source from outside the ECU box works just as well.

[Fred]

I figured out the reset/serial state stuff. After pulling that resistor off and doing the power-from-usb-bus mods, it, again, wouldn't boot into user code, marcos, please be more careful with your language, you kept saying "the mcu wouldn't start" it does start, code runs, and exhibits specific behaviour of not running FreeEMS code. This diagram shows why, it's from AN2548:



So, we need a pullup on it!

I'll take a crack at it now, as I have to use it, but maybe there is a better way than what I do.

Fred.