A locost EFI
Page 8 of 10

Author:  JPS Europa [ October 8, 2015, 9:14 pm ]
Post subject:  Re: A locost EFI

HI Marcus,
I dont worry too much about unused pins on a microcontroller. I do set them as outputs though. Doesnt hurt to ground them though.
I dont know what the idle air solenoid draws, but I do know that Brand MS uses a pair of IRF540's to PWM drive the 2 coils. Those things can handle a ton of current and I use them on my induction heating project. I am sure they are overkill, but they do the job.
I know that there is a resistor mod description out there where they tie one side of the valve just closed and just drive the second coil with PWM to open...sort of an electric spring. I dont know what the value of the resistor is offhand...but that would give you a good idea of max current needed. You could probably find on the forum what the value is with a bit of searching.
Started to work out what my engine mounts look like and welded up some mounts that I can bolt onto the motor for welding up a little frame.

Author:  horizenjob [ October 9, 2015, 11:04 am ]
Post subject:  Re: A locost EFI

Thanks Paul. I will probably compromise then and not fix all of them, just most of them.

I was thinking the interfacing with this unit would mostly be on CAN. Now I see that the little converters from USB to CAN are sort of expensive, much more than a converter from serial rs-232 to USB. So I'll route the serial to the CAN connector too and figure a way to make it selectable. My head has been in the tiny details so long I need to reacquaint myself with the bigger picture.

I was avoiding lights on the board because people won't be able to see them usually, but I think I may put a few back in to help debugging.

Author:  300D50 [ October 9, 2015, 7:58 pm ]
Post subject:  Re: A locost EFI

Might want to look into using a bidirectional switch, like this one.



Looking around a bit more, the TS5A23159 may be a better option.



Author:  horizenjob [ October 12, 2015, 12:46 pm ]
Post subject:  Re: A locost EFI

Hi Walt thanks for that pointer, I'll keep those chips in my bag of tricks. I was sort of staying away from analog mux style chips because the ones we used at my last job were very sensitive to static electricity. People would pick up a board to look at it and it then it would be dead and we'd have to replace the chips. These have an ESD rating though so they might be a lot better.

I'm trying to rethink thru my connection options for the user. I had a DB9 connector for CAN because I thought that would be good for connecting two of these units together for expansion. The expansion was to cover cases with more complicated motors and also possibilities like collecting data on suspension and other performance items. In this early phase though it turns out that adapters between USB and CAN for example are not as cheap as I would like for more casual customers, they run from $150 to $400.

Then I thought that a cable adapter for USB to serial would make sense. These are available cheaply and once you get up to $25 they come in wide variety, including versions that go straight from USB to 5V TTL serial which is what the processor actually speaks. So that saves on serial support chips on the board and also helps allow faster data rate. This solution allows use of a simple little 1/8" phono plug like people use with their cell phones to listen to music with. The phone plugs are small, take little board space for the socket, are economical and also come in a sealed enclosure.

The next option is to just take the USB to serial converter parts and out them on the CPU board. Then a person would connect with a normal USB cable, nothing special required. This is also the most economical solution.. It costs maybe $5 to do this and everyone probably has USB cables. It would provide the highest speed connection which would be nice for debugging when you are loading new code.

They make heavy duty USB connectors for automotive use, which is good even though they look a bit clunky. I don't know yet how well they are sealed, but they are wrapped in a shrouded plastic case so maybe they are.

I was asking advice here, but probably I've talked myself into using the USB connector...

Author:  horizenjob [ January 5, 2016, 3:40 pm ]
Post subject:  Re: A locost EFI

Long time, no update. Unlike building a real thing on a table, like a car, my struggles don't make for very interesting pictures. I stopped posting pictures of the board in the design tool, because despite my struggles the pictures looked basically the same.

Here is a much better picture of the board as it sits now - because it is now sitting in front of me! I am pretty excited about that. So excited that I feel like using a whole bunch of smilies, but will restrict myself to this one which I've never used before.

Initial inspection shows the board to be very well made by Advanced Circuits. I will try to get some close up pictures, but this is about as good as my laptop can do. I can't find my 10x magnifier right now and don't think the laptop will do that well thru the microscope but may try that too.

I am in the midst of nailing down the bill of materials and getting my orders for parts out. Then the whole pile o'stuff gets put in a box and shipped back to Advanced Circuits for assembly. These boards will be assembled with a robot. Their big machine can put 40,000 parts per hour on a board. It takes about a day of work to set up the machine with my parts and teach it where they go. By my calculation it will take about 18 seconds per board, so I think they will use a slower machine for me :rofl:, at least until I order a big pile.

ECU_CPU8b_0p2-top.jpg [ 248.2 KiB | Viewed 4019 times ]

Author:  russian2 [ January 5, 2016, 4:20 pm ]
Post subject:  Re: A locost EFI

1... 2... 3... 4... 17... 18...

Show us an assembled board already! :)

You've got yourself a sweet New Year present!

Author:  300D50 [ January 6, 2016, 2:01 am ]
Post subject:  Re: A locost EFI

One of my friends bought a Madell DP2006-2 pick-n-place in late 2014, and has been itching for projects...

Let me know how many unique parts are on the board, and the dimensions of the board, and I can see if his machine can support it. 8)

Author:  horizenjob [ January 7, 2016, 2:21 am ]
Post subject:  Re: A locost EFI

300D50, that might be interesting. Off the top of my head it's probably about 30 different parts and the board is just a bit over 3"x5". Does your friend make the solder stencils and have a re-flow oven?

Author:  300D50 [ January 7, 2016, 2:26 am ]
Post subject:  Re: A locost EFI

The machine has a paste dispensing system, so it automatically deposits according to the cream layer.

He has an IR reflow oven rated for lead free and leaded solder. The plan is to get a second one for true ROHS compliance, since this one is mixed usage.

Size wise the board fits, not sure if we have enough feeders for 30 unique parts in one sitting though.

Author:  horizenjob [ January 7, 2016, 2:37 am ]
Post subject:  Re: A locost EFI

Forgot to mention the board has parts on both sides. I am going thru a bunch of double checking and working on the BOM now so I'll have a real number pretty soon.

Author:  300D50 [ January 7, 2016, 2:44 am ]
Post subject:  Re: A locost EFI

Ah, so far we've only done single sided, and our current oven isn't set up for double right now, no easy wat to do standoffs for the second side parts after the first side is reflowed. No adhesive dispenser either... :(
It would probably be best to do the professional placement for testing, and after confirmation of the design we could look into a tool up. It's going to need to be adressed (double sided capabilities) anyways, so I'll check his roadmap.

Any plans for Wankel engine support? :twisted:

Author:  horizenjob [ January 7, 2016, 4:43 am ]
Post subject:  Re: A locost EFI

Any plans for Wankel engine support?

You would have to explain what's special about them. There is an issue with unusual engine support because this CPU has a special module that does some of the timing used int he ECU. There is code supplied by the chip maker, but it is sort of generic. Writing you own code for this special purpose part of the chip requires it's own compiler and probably some education. The main processor is a conventional PowerPC and is handled with normal tools.

Is your question mostly about the ignition or do you have unusual requirements for injectors too? Does your ignition look like a 4 cylinder with an odd shape crank? Like you delay one set of plugs on a rotor a bit or do both plugs on a rotor fire at once?

Author:  300D50 [ January 7, 2016, 6:09 am ]
Post subject:  Re: A locost EFI

Working from memory, you have essentially a combustion event every 180 degrees, alternating between the rotors, and advancing one face of the rotor each time.
So 0° = rotor 1 face 1, 180° = rotor 2 face 1, 360° = rotor 1 face 2, 540° = R2F2, etc.

So instead of 1-3-4-2, it's R1F1-R2F1-R1F2-R2F2-R1F3-R2F3 for the order.

One "cylinder" takes 1080° instead of 720° to complete an ICPE cycle.

Because of this, you can pretty much fuel it like a regular four cylinder engine, since you still have an intake event each 180° of shaft rotation. Just set it as twice the displacement as far as fueling is concerned.

There are primary and secondary runners with their own injectors, and primary/secondary throttle plates that correspond to those injectors.

Only the primaries fire till the secondary plates start to open, then both fire. Batch firing seems to work OK, sequential also works. Depends how flexible the injection can be.

At 3800 RPM, the auxiliary 5th and 6th intake ports open to allow more charge volume into the engine. Fueling is adjusted in the maps based on the known actuation point.

At 5200 RPM the VDI (variable dynamic intake) kicks in, and shortens the runner length for top end torque.

Both of the intake port systems are solenoid controlled.

Ignition wise, it's rotor 1 leading, rotor 1 trailing, rotor 2 leading, rotor 2 trailing, ad nauseum.

You have 180 degrees between leading plugs, and the trailing fires between 0 and 15ish degrees after the leading plug, depending on how you dial things in, load, phase of moon, and if Elvis has been sighted on the drag strip. (Sorry, joke had to go someplace.)

On the stock system, there is one leading coil that fires both leading plugs at the same time in waste spark, and a trailing coil that has two inputs, select and fire. Select chooses the rotor, fire fires the selected rotors trailing plug.

The RX-8 uses 4 individual coils, but they are quite a crappy design compared to the old setup. The change from 2 coils was made for cost reasons and packaging concerns.

A four coil system can be made with GM D585 coil packs or RX-8 packs.

Idle timing is around 20°ATDC for face 1 rotor 1, advancing up to around 30°BTDC at max rpm.

I ended up consulting the megamanual, rx7club, and another site found by google search for ignition timing information, since that part was hidden behind a missing synapse in my brain. mega manual
coil and additional timing diagrams

Author:  russian2 [ January 7, 2016, 8:32 am ]
Post subject:  Re: A locost EFI

horizenjob wrote:
Any plans for Wankel engine support?

You would have to explain what's special about them.


Author:  rx7locost [ January 7, 2016, 10:12 am ]
Post subject:  Re: A locost EFI

In the FC 13B engines, there are 2 coil-packs. One has a single coil, the other has two coils. The leading plugs get fired in waste spark mode, hence the single coil (2 plug leads). The trailing plugs must be delayed at variable angles depending on demand conditions, alternated due to the port locations. The trailing plugs reportedly are there to help a complete burn. It is not absolutely necessary unless you are running a cat converter and have to meet emissions.

in some situation, the plugs can fire after tdc which was a bug that I found in the MS-II firmware which has now been fixed.

It is somewhat difficult to describe.

Page 8 of 10 All times are UTC - 5 hours [ DST ]
Powered by phpBB® Forum Software © phpBB Group