Injector Control Options

[jharvey]

Any chance there are other vendors out there for this product? I see it can be obtained at future.

http://www.futureelectronics.com/en/Sea ... e,Nea:True

However, you have to plan to purchase 1k of these guys, so an upfront cost of $4.5K.

[jharvey]

I think I might be a bit confused here. I don't fully understand the rise and fall times noted. On pg 10 of the datasheet, I see "Output ON Current Limit Fault Filter Timer (Short to Battery Fault)" at 60uS up to 90uS. The rise time below it specifies mS not uS. Then just below that it notes "Output Slew Rate (No faster than 1.5 μs from off to on and on to off)" at 5V/uS. So I would expect that for up to 14V, that's nearly 3uS. Seems to me that those values are all over the place. I would expect my tolerance to range from 3uS and up to a 7mS. Could I be reading those wrong? EssEss, is this something you can measure to confirm or deny?

Aren't these injector rise and fall times much slower than the 1uS we ball parked a couple posts back? Perhaps a proper injector design shouldn't have a 1mS pulse, or perhaps these are intended for low power applications, where it's easy to get one injector to do it all. I like some of the features, and digi drive stuff, but I'm a bit concerned about it's drive capability. Perhaps these can be used as general purpose drives instead of injector drivers? I'm also a bit concerned about that .2ohm Rds. 4X of that is about 1 watt to dissipate when driven saturated, and more when in the transition states. That looks like a lot of heat to deal with.

I see the two external resistors for the ignition sensing, do they have some recommendations about what those resistors should be? I see the 4 channels of ignition in one chip as handy. Rise and fall times around .2uS ect. Those still require external drive silicone, so your ultimate drive delays are determined by the external silicone. I like the ignition sensing feature. I also seem to recall DaveBMW's BMW used a 300 ohm resistor for this type of sensing.

[tpsretard]

I took some pics of the motec M400 board this morning and a few part numbers along with it..
a quick run down of it is

Outputs
* 4 x Injector outputs—high or low ohm
* 4 x Ignition outputs
* 8 x Auxiliary outputs—for functions such as camshaft control, drive by wire throttle, boost control, nitrous injection,
idle speed stepper motor and many more

Inputs
* 8 x Analogue voltage inputs—fully configurable including custom calibrations
* 6 x Analogue temperature inputs—fully configurable including custom calibrations
* 1 x Wideband Lambda input—for Lambda measurement and control
* 4 x Digital/speed inputs—for wheel speeds and function activation

The board has 3 things you notice right away.
1, the 18 TO220 devices on the board.
2, the fact that there are 3 MC33385DH low side drivers on the board.
3, what seems to be an insane amount of flash.

Now i am by no means saying we copy this ecu, but if you are building a car there is no sense in reinventing the tiers for it..

The motec ecu is one of the best, so we can take some ideas from them i do not see any harm in that, most of the components are commonly available anyway..

8 of the TO220's are IRFZ34N's, these seem to be fuel and ignition, these seem to be GPO's
2 are LM2937ET 8 volt regulators.

8 DPAC transistors are IRF647N's

the rest i cant see what they are as there is an epoxy coating on everything.

I like the sound of the low side drivers though.. These seem to be what give the Motec the ability to do diagnostics on there fuel and ignition channels.

http://www.datasheetcatalog.com/datashe ... 85DH.shtml

now a pic or 2

[EssEss]

I think I might be a bit confused here.

that just says:
1) an injector shorted to batt (when transisitioning off->on) will nominally be detected in ~60us
2) an open detect (when transitiong off->on) will happen nominally ~7.5ms
2) an open detect (when transitiong on->off) will happen within 100 to 400us

once again, the slew rates and propagation delays in the datasheet are a non-issue - they're known constants. you should have seen fet specs 20yrs ago, you'd think this project would have been impossible if those are your only driver criteria

[AbeFM]

That's what I got out of it too - it's length of time to detect various faults.

The heat.. Sounds like a lot, but isn't. 1 watt will happily radiate away. And it's not one 100% of the time. If you're that worried, thermal-glue a small heatsink on the chip.

It's sold as PWM capable, even without looking at the numbers, unless Freescale is straight up misrepresenting itself, it's likely fine.

My question - is the PWM scaled to be proportional to some smart input (i.e. you program this chip (voltage compenstaion, p&H current, etc) then send it digital signals), or do you drive the PWM from the CPU directly, just letting it act as a dumb current switch.

Also, somewhat unrelated, at least on the miatas, the variable valve phasing takes careful control of the current to the valve - it sounds like maybe this chip would be helpful for doing that easily?

[jharvey]

I'm trying to find an article/web page I once found, and I'm having a heck of time finding it now, perhaps it went the way of bit rot. It was relative to injectors. I thought I'd see if someone else read it, perhaps they can point me to it.

The article was from a fellow that did a bunch of testing with both high and low impedance injectors. He built a bench test circuit and measured the voltage of the injector(s) as it turned on. He measured things like the rise and fall times of the injector. I also recall he could see the needle bounce. The rise was as expected with an inductor, except for some little ripples at about the time when he expected the injector to be opened. He measured the tau at various points of the rise time.

I seem to recall he was doing this as part of a project while in school. Perhaps it was a senior project or what have you. Does that sound familiar to anyone? I'd like to find it and add it to the articles section. Or perhaps someone with a setup might be able to take some readings for us? I know EssEss has a nice scope Sorry for putting you on the spot. Would you be interested in looking at the rise/fall times of an injector? I'd like to learn more about both. Mostly the fall times. I'd like to compare the fall of a snubber-ed (or flyback diode) circuit vs, the fall of the protected MOSFET's. I believe the snubber is a logarithmic decay, and the protected MOSFET is a linear decay. I'd find it interesting to see how they compare side by side.

[tpsretard]

I do not think it is what you read, but it dos have a little bit of what you talked about.

http://sonic.net/~mikebr/ecm_555/injector_drvr.html

[jharvey]

There we go, that's exactly the place I was looking for. Specifically this page is what I was thinking about.

http://sonic.net/~mikebr/ecm_555/inj_inductance.html

[EssEss]

thats just a simple characteristic of a solenoid .. and yes, clamp does have some influence on close times. I've been told by a factory oem tuner for ford that they purposefully avoid driving the injector hard to reduce wear on the seat .. if you were to drive it hard enough you'll get bounce on the seat and wear it out a lot faster.

I've done all the experiments already, I don't have the desire to do a writeup. my goal was to figure out a way I can map back some physical open/close times (static/dry) to dynamic deadtimes ... I learned a lot in the process. I started w/a simple inj driver, pulse generator, a knock sensor, a scope and excel. After I got familiar w/that I got the SAE inj flow spec and learned a whole lot more; mainly that my original idea was horribly flawed. so now, I simply flow against my selected driver and move on. The only thing I do dry (off the flow bench) is the vbatt vs deadtime curve to adjust my static flow calcs .. thats about the only thing that has translated btwn the two domains decently (but not perfect).

after you do your comparison and learn some stuff, whats the next step ? what I'm trying to avoid here is to sound 'preachy' -- I dont' want to throw info out there that sounds like I know it all .. I don't .. but I've learned a lot on my own along the way because I could never find the answer until I just did it myself. If you do it yourself and want to compare notes, I'm all for that.

[AbeFM]

http://abefm.smugmug.com/FreeEMS/Build-10

This is the stuff I did for the project, wow, many months ago. Specifically in the second test:



It's hard to pick out, but I remember being pretty convinced that I could see the pintle moving, and then stopping, hence a turn in the curve - once energy isn't going to accelerating the pintle, and just into building fields. Maybe I was just thinking it.

I could probably do another, sometime. Someday I might get work to pop for a better scope - but one cheap enough I can still bring home.