Injector Control Options

[Fred]

Not even a single lol? :-(

[EssEss]

lol

when I think of p&h, I think PWM'ing .. which automatically rules out high-side under any condition when the terms are (injector + p&h).

otherwise switching time of the driver is miniscule relative to the physical open/close mechanical variance. I don't want to poo-poo anyones ideas .. this is targeted for hardcore diy efi (which is why I'm here), but (for me) injector high-side driving for traditional injectors is automatically ruled out for experimentation. There are simply too many other tweakable things that I will play with. So feel free to do whatever you want on the board, after all you're the only one thats even taking those steps for our benefit. Anything high-side on that board will either not be loaded or used for dash indicators on my personal board.

Thank you for your efforts! I appreciate it.

If we plan for one injector pulse per revolution, we have a min period of 5mS. If we plan for sequential injection, the max time we have to inject is around half of that.

.. is the only statement I *slightly* disagree with .. if you have a 10ms window for 720* then I see no reason why you can't squirt for all 720* ? in my experiments (personal and professional) injector phasing relative to valves at > ~rpm/3 has not been crutial. I'm no expert, but I've stumbled many times along the way and have built up my own personal mental dictionary of which things are low-hanging fruit and get the biggest bang for the buck -- this experience/intuition has helped me converge on the important things much faster and earlier in designs.

[Fred]

Other than the fact that it is a terrible way to determine accuracy needs. Idle is where it counts for injection, and angle accuracy for ignition. The full load stuff is almost irrelevant.

Fred.

[EssEss]

yes, maybe I should have said it like Fred. The scheduling variance which is based off of a sloppy drivetrain, by far, swamps anything related to a switching time (especially when widths are highly controlled). So that makes it a bad metric to choose drivers on ... my metric of PWM'ing is where it became important. I just wanted to be clear through verbosity

[jharvey]

So at idle, let's do that math again.

If we figure the RPM around 600RPM, it would correlate to 10 rotations per second. If we plan for one injector pulse per revolution, we have a period of 100mS. If we plan for sequential injection, the max time we have to inject is around half of that. So a max potential pulse width of 50mS for 50% duty, more likely to be around 5mS for idle, perhaps less if the injector will allow it. If we want that pulse to be accurate to say 1%, we need the rise and fall to be less than 50uS. If we want that pulse to be accurate to say .1%, we need the rise and fall to be less than 5uS.

[edit] if 4 stroke, the period is 200mS with max potential window of 50mS at up to 25% duty. The min pulse is still around 5mS.[/edit]

Hmmm, can we get a good idle with an injection pulse of less than 5mS?

Other delays will of course have to be calculated out, but should be fairly constant. So I'll ignore them for this approximation.

[EssEss]

driver switching time (rise/fall) is pretty close to a constant over all conditions - one driver vs. another takes a little longer to switch ? +/-xx ticks to compensate. it will certainly vary less than the open/close rate of an injector over temp/pressure.

Assuming you know the deadtime, compensate for battv and have a timer w/enough resolution your mass injected will be highly repeatable. it's a non-issue to me. My little rinky dink diy flowbench has proven that to me time and again.

After looking at Fred's code I see where the magic 50ms number comes in .. I was getting confused by your wording. My interpretation of your wording describes the limitation against crank angle when inj time is actually wrt tmr freq.

[BenFenner]

So at idle, let's do that math again.

If we figure the RPM around 600RPM, it would correlate to 10 rotations per second. If we plan for one injector pulse per revolution, we have a period of 100mS.

We are dealing with 4-stroke engines here, yes?
(Pardon the sarcasm. I don't know where it comes from.)

[jharvey]

I believe we are planning for both 2 cycle, or 4 cycle. In either case the window for injecting is approximately half a cycle. As noted above, at some currently unspecified RPM, it's OK to rely on the engines inertia to allow for injecting outside of the window. Gut feel tells me that above 3K RPM you don't really need to inject in that narrow window.

I agree, that with a 4 cycle, it would change the overall period by about twice, so when I claimed an over all period, then 50% duty, I should double the period, and the note 25% duty for the normal expectation. I think I'll edit the above to include that.

The key item I'm looking at is how fast of an on time and how fast of a drop time do we need, those values don't change with the above notes.

I believe the injector pulses could go down to about 5mS, and up to full on, although typically not more than 80% on. The min 5mS is where we need the fastest on and off times, so that's been where we are looking to establish the specs for what's required.

[jharvey]

So based on the above, It sounds like the general preference is low side drive, with on and off transition times of less than about 10uS to 5uS. So far we have one that meets these requirements, the VPN MOSFET.

I know IGBT's are known as OEM inductive drive silicone, mostly because of their increased isolation voltage, but where typically considered slow because their duel stage design had an increased capacitance. I decided to take a look for protected IGBT's, so far I found this one that is a close match to the above noted specifications.

http://octopart.com/parts/search?q=ISL9V5036&js=on

I see the rise time less than 7uS and is typically less than 1uS, and the fall time is less than 15uS, typically around 2.8uS. It's Vclamp is around 300V, and can drive 31 amps at 110C. I don't see Rce, so I don't know how much energy it would dissipate when saturated. Cost under $2 from Newark. Comes in D2Pak and TO220 packaging.

I think this chip shows that gen 2 IGBT's have broken down some of those boundaries once associated with IGBT's.

[jharvey]

I also see this one, appears it was release in 2009, so it' s new silicone intended for automotive ignition.

http://octopart.com/parts/search?q=STGB10NB40LZT4&js=on

Can also be obtained from Mouser

http://www.mouser.com/ProductDetail/STM ... 8Ymg%3d%3d

I don't see a posted rise time but that is typically either equal to or less than the fall time, and the fall time is less than 10uS at 150C, and is 6uS normally. It's Vclamp is around 350V, and can drive 30 amps at 100C. I don't see Rce, so I don't know how much energy it would dissipate when saturated. Cost under $3 from Mouser. Comes in D2Pak and I2Pak (TO220 with out a hole) packaging.