Injector Control Options

[jharvey]

A MOSFET's like VNP20N07 noted in the beginning of this thread has an Rds of .05 ohms, and only dissipates .05W via resistance. The VNP5N07 has an Rds of .2 ohms, and dissipates .2 watts via resistance heating. Those are fractions of the .4 watts being dissipated via 70V OVP period. Heating via Rds is certainly important, but the OVP range is where it really heats up, both in peak heat, and average heat.

I based the above off the VNS14NV04 in the SO8 packaged. That has a Rds of .035 ohms, for .035W Rds heat. The key is that a 70V OVP will cause more heat dissipated in the ECU, and the faster decay of the 70V is about .5mS where the 40V will decay in about 1mS. So a variation of about .5mS from one to the other.

[Fred]

Any chance of creating a variable OV circuit and using one un-voltage-protected fet and one injector to play with this on a scope?

My gut feeling is that there isn't that much kick back or dissipation time. I could be completely wrong, though.

[jharvey]

There might be a chance of of making it using an OVP MOSFET instead of my switched zener thing. Now that I have a zener, I can try making it run via MOSFET, then try making the MOSFET driven by the zener. I don't think it will change the results much, as the current varies and the voltage stays constant. I buggered the above picture. See these updated pictures.

The 40V, appears to dissipate about 4 watts total, and the delay from when it's turned off, to when it actually off, is about 1 mS.


The 70V, appears to dissipate about 4.2 watts total, and the delay from when it's turned off, to when it actually off, is about 1/2 mS.


I was measuring the current in the wrong area on the first try, as I was measuring the leakage current in that switch, which was modeled with a 1K resistance when open. That's now mega ohms, and current is measured in the correct place. This then shows there is much more heat being used in the ECU from the OVP part of the cycle. Which would get worse if one PWM's the MOSFET.

[Fred]

That seems more like it! :-)

[jharvey]

My gut feeling is that there isn't that much kick back or dissipation time. I could be completely wrong, though.

That seems more like it!

Those two quotes seemed a bit odd to me. At first my small wattage claim was to high, then I come back with watts, and it's about right. Just seemed odd to me.

I'd like some empirical data to back it up. The average of 4 watts seems high to me. Does anyone have an injector and OVP MOSFET top test with? Of you one has an installed system, can they confirm/deny the injector wave form?

My primary concern is caused by seeing the 40 watts peak. I'm concerned that it won't disperse evenly, which will cause rapid changes in temperature on the chip, before it makes it to the board and such. Putting on a heat sink will help it disperse, and reduce thermal stress.

[Fred]

What seems more like it, is the two being very close together dissipation wise, and the higher voltage one turning off a bit faster.

Your math for adding it up looked suss in that email that you sent, but I have been to busy to work through it myself. I wasn't looking at the numbers themselves, just the way you came up with it. It could still be iffy, I don't know. If I get time, I'll look it over. Fact remains, any of the VNP units work admirably with high Z applications. Hence gut feel.

[jharvey]

Yeah, the big place I wonder if it's suspect, is my decay current prediction. I figured it started at 1 amp, and dropped to 0 amps over blah time in a fairly linear fashion. So I figured it's the same as .5 amps for blah time.

Those numbers just seem big to me. Using a 12 ohm injector is about right, correct. I just don't quite trust the results, they don't sit quite right with my gut feel.

[jharvey]

http://wiki.freeems.org/doku.php?id=inj ... ver_theory

[Fred]

Bad math! 9000 RPM with sequential on a four stroke is 75 injections per second. Each electrical pulse has a maximum duration approaching, but never equal to 13.3 recurring milli seconds.

Worst case is likely to be 6 injections per engine cycle, though, with a dizzy input and a 6 cylinder (wouldn't work well enough to use practically on an 8, 10, or 12), and definitely works very nicely on a lower power 4. Thus 2.2 recurring milli seconds per pulse and 450 injections per second, HOWEVER, 450/second is totally unrealistic because a distributor engine with 6 cylinder will simply not rev that far without CDI, which would be more expensive, complex and noisy than a better input signal while being less reliable.

Thus I'd say 9k with wasted and semi seq is a good number, past that there isn't enough dwell time, and as such, 150 injections per second of 6.6 recurring milli seconds each is probably a good place to do your testing.

You got lucky! :-)

Or not. 266.6 recurring is reasonable on a 4 cylinder that revs to 8k with a dizzy. Each pulse is approaching 3.75 milli seconds in this case.

Do as you wish, now :-)

Fred.

[jharvey]

The page is titled "injector_driver_theory" not injector selection or ignition. You're right my initial draft did mention things like sequential injection, which has several implications. I changed that to instead read per rev injection. My rough calculations were not to show how to select an injector, but to show the ball part pulse range of the injector, and to show how different approaches can effect the decay.

I'm a bit confused why you brought up ignition relative to that page, and in this thread. How's that effect the injection driver theory? Perhaps because it could be nearly the same circuit for devices that are sharing the PT port.

However I do enjoy getting lucky

I have updated it and gotten through at least on proof reading. It's a wiki, so feel free to update and fix it there instead of pointing it out to me here. Check it out at this link.

http://wiki.freeems.org/doku.php?id=inj ... ver_theory