DIYEFI.org Forum

MAX9926 chip can do VR or Hall. The circuit board screw pads on the blank plate can be cut off - if you're doing enough carving at that faceplate to get a large connector in there, you'll be cutting them off anyway!

Karri,

Check out this site:
http://www.k-data.org/kdfi-v1-4-with-enclosure.html
Yes these are MS clones.
Running an early version on a motorcycle.

I don't have time to explain. Either trust me, or don't. Your call.

Fred wrote:I don't have time to explain. Either trust me, or don't. Your call.

If that was addressed to me, my comment wasn't an argument; I know rather little about CAN and that side of things, and I'm looking to learn.

No, it was to Karri. Maybe he still doesn't trust me after the aerodynamics discussion I learned a few things in 5 years ago? :-)

No seriously, I want(need!) to know the problem with injectors and IGBT's. Datasheets and example projects don't seem to indicate such issues, in all it's pretty much a MOSFET input, BJT output type of deal. Flyback diode arrangement is different since there none inherently in the device, but external diode is a simple, already existing component on the board so adding more isn't an issue. There is also a tail current present in the IGBR, but it's only a problem with high speed switching power sources, as the turn-off delay is only some microseconds (MOSFETs are a hundred time faster than this but this is still hundreds of times faster than the quickest injector). Also there is a chance of IGBT latching if the turn-on pulse is too brief, but again, with injector timing it won't become an issue.

I'll settle with a good MOSFET, but I do like to know why I'm using it. With IGBT the reason was it's damn near indestructible.

BTW if that aerodynamics discussion was about the hood risers, then no, I didn't even remember that one until now

I've put more inputs and outputs to the circuit than currently available on the CPU, but what I had in mind was shift registers for the outputs and multiplexing for the inputs. I2C is not implemented yet in the code, but it's saving grace is you don't need to know the topology of the accessory chips like you do with SPI. Right now not having code does not limit from what there already is, but it's more of a future proofing idea than acute need.

The VR question is a tricky one. The MAX9926 does a great job, I did look into it while planning the features, but the number of required components is greater than with a schmitt-trigger buffer,an consumes pins from the header. The only loss of benefit with an external unit that I can see is missing the direction sensing on the crank wheel, but since that's only required in certain stop & go solutions (where the combined alternator/starter turns the engine so the cylinder at the beginning of the work cycle is just past TDC to allow spark start on a direct injection engine), the need for it is a *long* way into the future. Configuration between hall/opto and VR at the currently available tech requires opening the enclosure, which I want to avoid. Of course, it would be possible to configure the input with programmable flip-flops, but the code isn't there yet and I wouldn't hold my breath waiting for it (and you'd still lose two more pins.)

About the flexible injector/igniter configuration, I don't think there's any new code required. Looking at the Jaguar IO specs, there's eight high impedance injector drivers and four igniter channels available. Now, it seems a good solution is to simply hardwire the first two injector driver channels and the last two igniter driver channels since those are (nearly) always used, and for the two dual purpose drivers use a simple OR logic gate to allow either the ignition or the injection channels to control the drivers.

Figuring out the high current drives. So yeah, higher resistance internal resistance with IGBTs is bad for the injector response time with 12(14) volt system. Going for MOSFETs this time. The VNx14NV04 would indeed seem like the ideal choice, the TO-220 package version is VNP14NV04. The injector circuit is the same old, no reinventing the wheel there. Troubleshooting LEDs are in place too. The same circuit should be OK for the PWM task too. Would there be a future use for the in-built fault indication? It would make troubleshooting easier with code to support it..

I reconsidered the idea of having two dual-purpose transistors shared for injection/ignition. I plan on focusing on the Bosch ignition modules, and since those only draw 22mA at most, there's really not a great need for extra heat sinking. Surface mount transistors should do fine. A datasheet from Bosch had a nice reference design for 3rd party EMS design. It does seem like inverted setup compared to usual FreeEMS behaviour (MCU pin high, output to ignition module low) so it needs to be inverted twice... other than that, doesn't seem too complicated.

I learned the hard way that mazda handles fuel pump/ignition and injector power relays in very different manner. The fuel pump and ignition power are switched as they should be, but for some reason the injector relay solenoid is on constant power. It would be a quick fix to have an extra relay to isolate the coils, but there's probably something why it was designed that way. Any case, since the fuel pump/ignition relay is going to be in a switched line, that's the candidate for switched power detection. I just need to figure out the exact circuit for it.

What I was thinking of powering the unit with is Freescale MC33730. It has a SMPS part for power supply and two linear regulators for sensor reference voltage. There'a also a sleep mode and quite a lot of other interesting features too.

KW1252 wrote:It does seem like inverted setup compared to usual FreeEMS behaviour (MSU pin high, output to ignition module low) so it needs to be inverted twice... other than that, doesn't seem too complicated.

What? That's not true. The vast majority of ignition drivers require a high input to dwell. This is the FreeEMS default behaviour, and is in hardware, not software.

Use FET-drivers for ignition outputs, and keep the voltage supply to them configurable in hardware, too.

how can the mcu pin be high and the output be low if the signal is inverted twice?

Am i missing something or wouldn't you only need one inversion?

To invert something twice is the same as not inverting the signal at all isn't it?

Yeah, I got a little confused there for a minute. Of course the system isn't inverted; I guess I'm not at my sharpest at midnight anymore.

Anyway, the actual transistor for the ignition module driver is going to be a FET and not some odd throughhole BJT. The provided schematic was a nice gesture from Bosch, it eliminates a bit of guessing.

Any thoughts about the malfunction detection, worth the effort and board estate?