Puma board for FreeEMS

[nitrousnrg]

You can make a BGA work, but it will have to be on a thin PCB and often requires relief cuts.

Oh, I thought there was quite a steep between a BGA and a normal chip regarding thermal considerations. Ok, I'll have this in mind. Afaik, the thermal expansion of a BGA chip has to be matched with the thermal expansion of the PCB, but I never gave this a deep look.

You may want to look for an online trace calculator.

Yeah, thats perfect, thanks.

I know your spin of this is for $ in the long term.

I wish to, but right now my concern is the cost of the prototype. I've quit my job at the hospital, and I'll be unemployed until january or so. The more it costs, the less I have for this vacations =)
Batch PCB is a nice way to earn a little $, but I'm not really interested now, the PCB should be available at cost. What I want is to sell some boards with freeems here, return a 20-30% or more of profits to the freeems project, and start getting tools to serve this and other projects of mine.

There is a growing movement to make a tax-free zone(zona franca may be equal to free zone) around the harbor next to my city. It is meant to allow exportation of electronic products, no taxes for importation of components, no taxes for exportation of products, pcb manufacturing, pick&place, and its at 5km away of my home (my university is 30+km away!). If that thing really works, I could sell really cheap assembled boards worldwide. There is a big drawback: in the first stage everything must be transported by sea, although 4 years later the airport (only 30km away) will be free zone too.
Anyway, although I'm trying to get involved in the free zone, I'm afraid it could be ruined by politicians or big $-hungry companies.

I'm curious how your doing on the software side of this. I might encourage you to plan for that to take longer than you might expect.

Yes, I know. I think I can make a working system out of this in little time. I still remember that I made a sequential ecu with less than 400 bytes of ram. Also, my car is ready to be plugged to any EMS (I blew a gasket last week, though), so early hardcore testing is available right away.
So, by the time I finish this prototype, and get the mcu/bdm/bootloader to work, the firmware should have a little progress at least. And if it doesn't then I'll hack it until vanilla gets ready to run my engine.
ECUmanager is waiting for this board too, I'm not developing that software until I have something to communicate with.

A million things can go horribly wrong -or not-, but since I want my freeems-powered car the next year, it worths the effort. Plus, I won't be an engineer until the project is running nicely :-)

[nitrousnrg]

Just for future reference:
Today someone gave me a good explanation about power plane vs ground plane.

The doubt I had was what to do with the front layer planes: A 5v plane would add capacitance (which is good), and a ground plane would provide a low impedance path for RF to go out of the board.

Well, that capacitance that stabilizes the 5v line, is also a bad thing, because its a distributed capacitance. The electric field between the planes provides a medium for noise and signals to propagate, it is coupling every part of the board. So, ground planes should be preferred.

Thats all, a basic design tip that could help someone.

[Fred]

I'm curious how your doing on the software side of this. I might encourage you to plan for that to take longer than you might expect.

I'd second that, however I feel the need to point out that the pin-out confirmation and research is something that anyone with some hardware could do, and that 20 or so people have TA cards and that none of them have risen to the occasion and actually helped out where they could! The other part of it, the xgate bit bang code and scheduler, etc, will be getting done in the mean time. I'd love to see you hacking at it to make it work for you, though. That'd be cool.

Fred.

[jharvey]

I'd second that, however I feel the need to point out that the pin-out confirmation and research is something that anyone with some hardware could do.

I don't have the hardware, so I can't do it. If I had other potential uses of the TA card I might consider purchasing it, but I don't so I'll just wait until things progress a bit more. I'm not sure why you haven't gotten buy-in from those that do have the hardware. I might guess there is some skepticism about the need to tune nano seconds on these signals. However I can't speak for others, so it's just a guess. Perhaps they don't know what you mean or what's involved in doing this pin out selection.

[Fred]

I know you don't, and that's cool. No idea what you mean by "tune nano seconds", this is about burning out coils and igniting mixtures at bad crank angles, and flooding cylinders, and wiring fires and nothing more. As for knowing what to do, it's documented in the thread and comments thread for this issue. And that thread is linked from this thread, from yours, from mine, from various other places, too.

[nitrousnrg]

I'll test many pins, but not all. For example, I'm not putting pads to access port A, it should behave just like port B (which I have routed to the 8 injector drivers).

Ports to be tested (in the first working board)
* Port P (PWM outputs)
* Port B (injectors)
* Port T (ignition+RPM)
* ADC Port (I wouldn't care, though)

Port H & J can be interrupt triggers, so some of those pins must be routed to the signal connector, and the left ones should have some pads to see their state with an oscilloscope.


Now an important comment:
The stepper motor driver should be routed to port P, if there is any plan of supporting microstepping (it uses PWM).

[Fred]

I'll test many pins, but not all. For example, I'm not putting pads to access port A, it should behave just like port B (which I have routed to the 8 injector drivers).

No, no, no, Marcos!!! These are the two that REALLY MATTER, along with the fuel pump control pin, whatever that ends up on. A and B do NOT behave the same... and we need to understand why and work around it or fix it so they do. Por favor, bring those ones out too, in place of some others, if you have to, if it's meant for testing. If it's not meant for that, then... I don't know, but the behaviour of those pins is important, perhaps bring out half of A as 12 is the max number I plan to support anyway, at least for now, at least without staging. (and the other four will behave the same way)

[nitrousnrg]

ok, no problem.
I don't get the behavior of this mcu. Is the serial monitor code available?

[Fred]

Yes, it is, but it's not the issue. The issue is due to the modes it operates in, or perhaps just normal behaviour for it. I spent some time digging, but haven't had test hardware really. Maybe I'll get something done in the coming few weeks. I may have a temporary desk while I do a MegaJizz install for a friend.

[nitrousnrg]

ok, but it doesn't make the situation less weird.
It will work, don't worry. Keep it up with docs, site and code ;-)