Puma board for FreeEMS

[nitrousnrg]

The guy (aka Molak) didn't tell me something he saw, he made me realize about things that I've seen with my own eyes.

I had a lunch with the biggest tuners of buenos aires a while ago, and they shown me their 4cyls 500+hp cars. 3 rails are in a 4wd volskwagen saveiro, 4cyl and 700hp (yep, kind of a funny car), and they were talking about 6000cm³ injectors. All the cars I saw were tuned for drag racing, so long term reliability, idling, etc aren't the primary goals for these guys. The thing is, most of the tunable EMS users are this crazy (here). Almost all of them frequents the dragstrips, its not common to see a programmable ecu in the street. Really, a big share of the EMS market has 2 rails, to feed their ethanol/methanol powered engines. I totally agree with you, I prefer one big injector, and one precise injector, but this is not how it is done here :(

The good thing is that it has the *potential* to drive 8 low Z injectors, but for those who don't need the 8 drivers, I can leave some of them unpopulated (easy because its thru hole!:), saving a few bucks.
I'm kinda worried about removing 2 ignition drivers, but since I never saw a 6cyl cop setup its probable that it'll end with 4 drivers.

We could test lots of circuits/behaviors in this first version, and then make the "worldwide" one (the latter is quite ahead of my immediate expectations, but why not). I'd like to help with the firmware once I have the hardware. I'm not a PC programmer.

Hope that all makes good sense :-)

Yep, as always. The problem are the argentinian engine tuners! :P

[jharvey]

I just checked out the design, looks like it's coming along nicely. I don't see any mounting holes. With out them, the board would likely be held in place via the leads of the drive silicone, and would likely have mechanical stress issues. The 1.0 version used the TA card's mounting holes for this feature. I think it would also be good if one of these mounting holes is connected to GND and is located close to where the GND connection leaves the board. I'd let the mounting holes float. This would make your enclosure grounded and would help keep other signals out.

[nitrousnrg]

True, thanks for taking a look at it. I have to search for cases, maybe I can find a nice aluminum enclosure. So far, the board is 10x13cm.
My thoughts:
1. It depends on what I can find.
2. The case must be metallic, and grounded. Maybe with the power ground.
3. 15 TO220 is quite a lot to screw to the case, but the weight is unbalanced. With 4 or 5 mounting holes in the board should suffice, but again, it depends on which case I'll have.

I'll let you know if I find a case (I'm aware of the case thread)

Pushed up the last changes. Nothing exiting, it has a BDM connector and 8 injector drivers.

[tpsretard]

i have tuned many systems on many different setups.
from stock street cars to wrc engines, to rally cross.

There are times where staged injection with 2 or more fuel rails is the best way to go.
The WRC escort ran 2 440 cc injectors on each cylinder, once the cross over points were done was simple to tune.

a methanol burning 4 banger i worked on has 850cc primary and 1600cc secondary's, the car is run in a type of autocross, so drivability is key, it would not be possible for 1 big injectors to do it, at 40od psi of boost all the injectors are maxed right out.

the moral of the story is it needs to be flexible, most of the time it ends up being pectel as #1 choice or motec.

I also like my ignitors external and a lot of people do also, it gives more flexibility. as you can then use oem drivers, or cop with them built in, switch to cdi or any other fun and fancy combo you want.

[Fred]

Dual rail, sure, low z on both, maybe, triple rail, no, only for show, not for go. I don't know about you, mate, but I couldn't live with myself creating something to support faggots that make cars for people to look at, not drive ;-)

As for the grounded case, you must NOT connect that to the internal grounds. No way. You'd be violating all good grounding principals by doing that as it will end up screwed to a firewall 2cm from the ground point for the heater fan and injecting noise all over your internal grounds. Metal, yes, grounded, yes, connected to functional grounds, no way.

Fred.

[nitrousnrg]

Damn, I was writing the reply to tpsretard and a sudden logout made me loose all I've written.

the moral of the story is it needs to be flexible, most of the time it ends up being pectel as #1 choice or motec.

I want it capable. Able to drive 8 injectors or 4 coils. If the user doesn't have low Z injectors, then I won't put the drivers in it. The same applies for the ignition, if he has an external driver, then excellent, I replace the internal driver for a mosfet :-). A high end version, a low buck version, and probably more can be made from the same pcb.

The low end market is ruled by MS (it cost uSs400 tough), and I don't want to compete in the same niche, so I'm going for the differentiation. Megasquirters fail when a sequential or high end in some way control is required, so those people need to buy a much more expensive haltech or motec EMS, at least doubling the cost of the setup. Besides, by fabricating it here I'm avoiding the assassin taxes applied to importation, making a quite important advantage to get more people to use it.

BTW, nice input!

Dual rail, sure, low z on both, maybe, triple rail, no, only for show, not for go. I don't know about you, mate, but I couldn't live with myself creating something to support faggots that make cars for people to look at, not drive ;-)

Hell no! LOL I'm not going to put 12 injector drivers. afaik, there are 2 or 3 cars with that setup.

As for the grounded case, you must NOT connect that to the internal grounds. No way. You'd be violating all good grounding principals by doing that as it will end up screwed to a firewall 2cm from the ground point for the heater fan and injecting noise all over your internal grounds. Metal, yes, grounded, yes, connected to functional grounds, no way.

Agreed. But... the power ground is completely isolated from cpu/analog circuits. Those grounds (power/cpu) joins in the battery so I don't see any problem with that. In fact, at this moment of agonizing sleep deprivation, I'm thinking about leave floating the case to ground it at the firewall, not really to preserve noise from entering to the board, but to don't put more noise in the chassis ground.

[nitrousnrg]

Almost routed :-)

Still rough in the edges, but it passed the DRC, and I have time to improve it. The only unconnected pads are the power outputs from the cpu (undefined yet), crank/cam signal tracks (not sure if they follow the official pinout), and the connection between the switched 12v and the divisor to measure the battery voltage. This last one is the only I find hard to route.

One more thing is the front plane: it needs a plane in the front too, but I'm not sure about making a ground plane, or a power plane. IMO, ground planes provides a lower impedance path to the battery. In the other side, the capacitance formed from the stacking of a ground plane and a power plane could be beneficial, who knows.

If someone see something wrong or improvable, please let me know (Jared, I'm still looking for cases :-)

[jharvey]

About the case, you have seen the WS-4517 from here?

http://www.xtech-outside.com/products/e ... S4517.html

About the schematic, looks good to me. I think there will be some empirical tunning required, I believe the 10K resistors on digi and an protection circuits are a bit high and could be lowered. But eh, they are probably good as they are. Some empirical data will likely help confirm/deny different theories about this. Perhaps a quick-ish test would help verify this before parts are ordered and populated.

There is a small issue with C83 overlapping with R157, ect. Functionally OK, but a bit hard to read. Same thing for R177, ect.

Also I have a tendency to make my pins "bidirectional" this removes annoying drc complaints.

Looks like pin 14 on U19 should be ind_fuel_pump not ind_fuel_pmp. PT6,7 should probably have NC's attached.

I'm significantly ignorant about the BDM connector. I believe that's required and I'm not seeing it. Is that what BKGD is about?

Perhaps a decoupling cap on U19 would help prevent mild RF leakage on the power and gnd planes.

About the PCB layout, silk screens are a bit large and they appear to be in the same locations as the defaults put them. I bring it up, because I have a tendency to forget about them, then I'm reminded about them when I get the PCB. Which is followed by series of short words and increased blood pressure.

Something that's been brought up in the past was a non conducting ring around the perimeter of the board, connected to gnd when gnd leave the board. Theory being that it captures stray RF ect and channels to to GND before it couples with the board. Your GNDed case should do this, so it's semi already taken care of. I don't see harm in doing it for the PCB as well. Figured I'd bring it up.

About soldering this up, have you seen the TechFX reflow module?

http://www.thesiliconhorizon.com/cms/in ... &Itemid=57

I made a re-flow oven for about $200 usd. Works well, and much nicer to use than an iron. I think I may have already mentioned that, so it might be redundant.

Keep it up, it's looking good.

[nitrousnrg]

Just corrected the schematic's DRC. The DRC I did was for the pcb, I've been overlooking the schematic lately.
Now I only have two errors, due to the default FT232RL component.

About the protection circuits... I'll try something closer to 1k than 10k. Like you said, its subject to further testing.

There is a small issue with C83 overlapping with R157, ect. Functionally OK, but a bit hard to read. Same thing for R177, ect.

True, tough to read. Corrected now.

Perhaps a decoupling cap on U19 would help prevent mild RF leakage on the power and gnd planes.

Yep, added in the cpu side. I have quite a dilemma here: I don't want to add another regulator to put a 5v reference to the power side. I'll see, maybe a zener could be enough.

About the PCB layout, silk screens are a bit large and they appear to be in the same locations as the defaults put them. I bring it up, because I have a tendency to forget about them, then I'm reminded about them when I get the PCB. Which is followed by series of short words and increased blood pressure.

ha! I totally understand the last part, but not the first. What locations/defaults are you talking about?
I've been asking to my colleagues about the distance between the smd resistors, I don't know... maybe they are too close.

Something that's been brought up in the past was a non conducting ring around the perimeter of the board, connected to gnd when gnd leave the board. Theory being that it captures stray RF ect and channels to to GND before it couples with the board. Your GNDed case should do this, so it's semi already taken care of. I don't see harm in doing it for the PCB as well. Figured I'd bring it up.

Yes, I recall that. The bottom GND is just huge, I should probably do the same with the front; maybe with just a plane instead of a ring. Good question for my RF professor.

And... WOW at both links!
I love the case, I was wanting something like that. The coolest-looking cases are the audio power amplifiers, and I'm sure I can find who the hell makes them here. Its excellent to give me ideas, but its a shame that I'd pay so much in taxes and shipment for it.

The oven controller is amazing too (didn't know about it), I'll share the links with a few mates here :-)

Your thoughts are highly appreciated, keep in touch :-)

[jharvey]

About the PCB layout, silk screens are a bit large and they appear to be in the same locations as the defaults put them. I bring it up, because I have a tendency to forget about them, then I'm reminded about them when I get the PCB. Which is followed by series of short words and increased blood pressure.

ha! I totally understand the last part, but not the first. What locations/defaults are you talking about?
I've been asking to my colleagues about the distance between the smd resistors, I don't know... maybe they are too close.

The text is between the pads, and overlaps the pads, such that much of the silk screen will be blanked out. I believe this is how it was positioned in the library module file that created the working copy of the module. That's what I meant by default, it was simply how it was on the template. If you rotate the text 90 degrees, it should fit just fine. Also the font is fairly large on say R47, where it can be smaller like on D31. However, larger text is easier to read.

And... WOW at both links!
I love the case, I was wanting something like that. The coolest-looking cases are the audio power amplifiers, and I'm sure I can find who the hell makes them here. Its excellent to give me ideas, but its a shame that I'd pay so much in taxes and shipment for it.

I know that many of these companies simply use a foundry like Almag.

http://www.almag.com/

A China or Taiwan foundry would likely be lower cost. The problem with going direct to a foundry is that you have to buy bulk. The extrusion dies often cost several thousand dollars. If you ignore the die costs, the extrusions are very low cost, typically it's around $8usd/lbs. Most cases are well under a pound. So your often left paying the taxes and fees of companies that will sell it to you with out charging the tooling costs.

The oven controller is amazing too (didn't know about it), I'll share the links with a few mates here

I've also seen folks use a toaster, on it's side with the spring mechanism disable, for small boards. When done this way, you can make a reflow oven for under $100. I used some $18 SSR's from futurlec. It took them some months for the parts to arrive, but for the price, you can't beat these SSR's.

http://www.futurlec.com/Relays/SSR40A.shtml

I went with the infrawave oven because it combines infrared and conduction, and it's large enough to do a typical PCB. The conduction creates air patterns that help smooth out hot spots often found in pure infrared ovens.

Your thoughts are highly appreciated, keep in touch

No pat on the back required, a much better way to say thanks is to complete this project and make it run the vehicle. Keep it up and I'll assist in what ways I can.

You may also want to keep an eye out for tenting as noted in sparkfuns tutorial found here.

http://www.sparkfun.com/commerce/tutori ... als_id=115

At this point, nothing else is really jumping out at me. There are some more traces to route, ect. I'm sure you know those. Looks like you are approaching rapidly on a working board layout. Let me know when you think it's nearly done and I'll look it over again.