[moonie223]

My injectors are ~8 ohms and still High Z. If I were unlucky enough to run batch they'd pop those mosfets for sure.

Plus, I'm just some dummy, but switching times aren't instant and that means your RDSon isn't as fixed as you think it is. The majority of the heat is in switching time, not steady state. The gate charge will prevent the mosfet from switching instantly and during that time it's in a region of exceptionally high resistance.

In other words, when cutting the limit so close you've only half of the calculations there. Next you figure rise times off gate charge and pin current, then you can start estimating how much heat is generated in just the switching time. Then you get to decide if you need a gate driver or not, or just get a bigger fet...

Finally, that 0.9W rating is on 2oz copper in free air with a full square inch devoted solely to dissipation for the mosfet. In other words, it's a useless figure in application.

At the very least I'd run a test or something, might be fine.

[JHolland]

Switching losses won't make any significant difference because injectors run at a relatively low frequency. I am more concerned about the avalanche energy, high-z injectors have a reasonably low energy but the datasheet doesn't specify any rating which is a bad sign.

[moonie223]

So, you're saying you don't think the mosfet will take the freewheel energy? Or, we don't put freewheel/flyback diodes on injectors so that current flow is cut quick as the mosfet shuts off, no slow sloppy shutoff of fuel. That stored energy in the inductor is going to pile up at the mofset D/S and pop it? When I model an inductor and resistor switching as a rudimentary injector in spice I get spikes of 5kv at the drain, makes sense I guess.

[chaoschris]

For all the disbelieving people i made a test with the hardware on two injection channels (inj_a and inj_b).
On Inj_a i had a high impedance injector drawing near one Ampere and on Inj_b I connected a 12V/21W Lamp which was laying around ready to use with some cables on it.
The uploaded sketch was switching the two outputs on for 9 seconds, then switching them off for one second and then switching on again....
The whole afternoon this was running and i cannot say if the Mosfets are really getting warmer or not. They are only somewhat warmer that i couldn't say this is coming from the current of the load or they getting warmer by the pcb. The only annoying thing was the stinking injector which was getting really hot without any cooling liquid...
So i'm pretty sure this will work with all four injector channels as it was intended. The whole thing was drawing 2,7 amperes at 14,4 Volt during on-state.
Best regards,
Chris

IMG_20210114_151826.jpg (4.76 MiB) Viewed 8544 times

[fram]

Good with injectors, at home :around 20°C and low humidity level..
But what about derating temp. when your ECU will be in engine bay with 60-70°C?
Could you also double check with inductive loads, like fuel pump, solenoide valve, others... (keep in mind the peak current when you start this kind of device)

[chaoschris]

with inductive loads, like fuel pump, solenoide valve, others...

Hello fram,
i can simulate the conditions in a car with 80 degrees C, this would be no problem. But I'm pretty sure there will be no problem at all. We are talking about milliwatts on the Mosfet....
What do you mean with fuel pump? Direct without a relay???
Solenoid valve i have never used. do you have any values for me (resistance/inductivity) to get a estimation about this?
Best regards,
Chris

[NickZ]

How do they handle short circuit situations, I have had a few injectors die with a short?

[chaoschris]

No, they must be protected with a 3A or 5A fuse. But this is the common way and nobody should wire the injectors or ignition coils (on the power side) without fuses. It would be also quite impossible on such a small design to provide enough copper on the pcb to handle more than 10 amps for example. The copper lane would be always the fuse...

[JHolland]

What about load dump conditions? you're clamping at 30-odd volts have you tested for that?
The other issue is that the lower injector fly-back voltage the longer the turn-off time, most of the (mainly Japanese) ECUs that I have worked on clamp the injector at 100V.

[PSIG]

Solenoid valve i have never used. do you have any values for me (resistance/inductivity) to get a estimation about this?

If you can use this example (and it is far from the worst), I have a typical Ford/Mazda/Rover PWM IAC sitting on my desk that measures 10-ohms and 50 mH. I have others that are inductively "worse", but they are not handy for measuring ATM.

Ford-Mazda-Rover_PWM_IAC.jpg (43.75 KiB) Viewed 8447 times

However, it is of comparative interest to me as I tested this valve on a v0.4 output, which used a STP75NS04Z (N-channel Clamped - 7mΩ - 80A - TO-220). In testing it would heat the 80A FET rapidly to the point I could not touch it at room temperature. No, the STP FET spec's are not the SOT-23 device you are using, but it tends to surprise people that a 1.5A valve can possibly overheat a large 80A device with 110-watt rating.

A shunt diode solved the overheating, which indicates that flyback energy through the body diode was a substantial source of the heat. Unfortunately or not, a low reverse-V diode slows the valve operation, and the optimal rating will be different and likely higher than the body diode (paradox with that device).

So I am truly not trying to rain on your parade, but offer any reasonable concerns in order to avoid unexpected issues in your final design, slowing development or success. BTW - like others, I am also one that would use paired injectors with your design. I hope it all checks and tests well.