Injector Control Options

[baldur]

My design does all of this and only requires one output pin - did you read through the whole thread???

I did not read through the whole thread before, Fred just asked that I posted my opinion on this after we had some discussion on flyback suppression, which I did.
No, your design doesn't do all of this. It would dissipate more power during the current limiting period than the design I described earlier but like I've said, there are compromises to be made.
Care to run another simulation? I'd be interested in seeing the gate, source and drain voltages of M2 related to ground, also the voltage at C1.
You state that a comparator cannot operate at part-voltage and only latched to a rail. I must say I have observed otherwise, the comparator is really just a high gain amplifier and there is a sweet spot you can hold it at. However the positive feedback should help it oscillate.
Now, it's a clever approach to current limiting but I'm not convinced the flyback current won't burn it up.

[Delta]

My design does all of this and only requires one output pin - did you read through the whole thread???

I did not read through the whole thread before, Fred just asked that I posted my opinion on this after we had some discussion on flyback suppression, which I did.
No, your design doesn't do all of this. It would dissipate more power during the current limiting period than the design I described earlier but like I've said, there are compromises to be made.
Care to run another simulation? I'd be interested in seeing the gate, source and drain voltages of M2 related to ground, also the voltage at C1.
You state that a comparator cannot operate at part-voltage and only latched to a rail. I must say I have observed otherwise, the comparator is really just a high gain amplifier and there is a sweet spot you can hold it at. However the positive feedback should help it oscillate.
Now, it's a clever approach to current limiting but I'm not convinced the flyback current won't burn it up.

I said its always trying to drive to a rail - Yes a comparator can be held at a certain output voltage - by quickly changing the inputs about each other. In this case that does in fact happen - keeping the output just below the threshold voltage of the mosfet - which turns it off - then obviously the current falls and the voltage falls at the negative input, which pushes the output above threshold voltage just long enough to raise the voltage at the input of the negative terminal so it turns off again. But the point was more that having a comparator there means that something must always be in flux or, as you said, due to positive feedback it must drive towards a rail. In this case the feedback is fast enough to hold the mosfet near threshold voltage and push above or below quickly to turn on and off - which is what we want.

I find it difficult to imagine that a transistor based circuit doing limiting would EVER use less power than a switched mosfet circuit - I'm failing to understand how it could be - I'll stick your circuit in LTspice and have a gander. In the mean time - enlighten me.

I'll run those sims for you of gate source drain C1 etc, brb.

[Delta]

Top pic - gate voltage - you can see it sits near the threshold voltage and pulses up to turn on when its limiting current.

2nd pic - Pink is C1 voltage. Red is Vdrain you can see it turns on and off while limiting.

3rd pic - Vsource - voltage value closely mirrors injector current.

4th Pic Injector current


I'm not an automotive engineer - but I'm a reasonable electronic one. I've used a very similar circuit in real life already to control hydraulic solenoids. While this ISN'T my field of expertise - I'm copping a lot of flack over this circuit - perhaps rightly so I'm not sure. Its always good to have more than one set of eyes on something and some critique - but even so, everyone is having a go, I'm pretty sure I've covered everything in the simulator - I think it would be more contructive if someone would test it in real life with an injector, and come back with potential REAL LIFE problems, rather than theoretical ones. (I'd test in real life myself but I don't have an injector setup to do so) I'm sure everyone has a circuit that does this, so if everyone thinks this circuit just isn't going to work and thinks everything is going to burn then choose another. I'm happy to defend my work but it seems no one wants this.

[jbelanger]

I don't think people are attacking your design. I think most people are interested and since this is new they just want to understand it and make sure it works as intended and that the usual issues with low-Z drivers are considered namely heat, noise and turn off time.

And I agree that it would be nice to have this tested on a real injector to validate it. Unfortunately, I'm not setup to perform the test either so hopefully someone else will be able to do some tests.

Jean

[baldur]

Correct, I'm not attacking anyone or anything. I'm just saying what I see when I look at the circuit diagram. I've found that spice apps are generally not all that good at estimating component temperature. In my experience it's good to get a second view on whatever problem I'm solving, even if the observations may not be 100% correct it does drive me to prove or disprove them.

For example, how high does the C1 voltage rise after a minute (or as long as it takes to stabilise) at 80% duty cycle with a 10ms pulse length?

[Delta]

So I suppose the pertinent question is - do you think it works or not.

[Fred]

I don't think people are attacking your design. I think most people are interested and since this is new they just want to understand it and make sure it works as intended and that the usual issues with low-Z drivers are considered namely heat, noise and turn off time.

Exactly!

I'm impressed and very interested in seeing this built and used at some point. Whether it's in the main board or not doesn't phase me, I still want to see it tested because it's interesting.

Heat, I trust that you know exactly what will be happening there.
Noise, it will have some, but PCB design and the snubber stuff should take care of that.
Switch off, I don't understand your explanation, though I do want to. Conventional wisdom says "let it spike as high as possible" = fastest switch off. Conventional wisdom says "hold it close to 12v rail during pwm" = least noise and heat. I just don't understand the circuit enough to know what is happening under these conditions. Any chance of taking another crack at explaining it to the inferior kiwi ;-) {just pulling your leg}

Definitely don't be discouraged though your input and feedback is very much valued by all of us!

Fred.

[davebmw]

Delta, this will work well, we just need to do some real world trials and maybe tweak the values for the snubbers to kill EMI. as different PCB layouts will have quite different characteristics.
Even though I am not running LoZ injectors I would still want this option for future upgrades.
Don't get disheartened by others that don't understand, I didn't at first and was too quick to judge until i looked again and you confirmed how it worked.
Remember most people think this piece of precision timing should be controlled by the CPU what they don't realise is that before Microcontrollers were around this is what was used and still is in many fields. Especially military products they are still skeptical about software stability.
If its good enough for them to use it, we should be OK

[Delta]

Correct, I'm not attacking anyone or anything. I'm just saying what I see when I look at the circuit diagram. I've found that spice apps are generally not all that good at estimating component temperature. In my experience it's good to get a second view on whatever problem I'm solving, even if the observations may not be 100% correct it does drive me to prove or disprove them.

For example, how high does the C1 voltage rise after a minute (or as long as it takes to stabilise) at 80% duty cycle with a 10ms pulse length?

C1 voltage stabilises after about 100ms. Voltage is highly dependant on injector inductance and resistance - the highest i've seen it stabilise at is 35V if it turns off during the hold section. The highest values are actually seen if the duty cycle is low enough to switch off at the peak of the 'peak' part of the cycle, then the voltage on the cap can stabilise as high as 45V. So the caps would have to be 63V rated for safety.

[jharvey]

A while back someone mentioned I should toss out a draft of an idea I had for P&H. I'm just getting to it now, here's a quick snap shot of what I was thinking.



I drew some of the expected wave forms across the top.

Features I like about this approach include, dissipated heat is outside ECU box, small variations in capacitance and inductance won't change your signals, noise floor isn't raised. If you don't want the P&H because you're running high Z, you simply don't put in the max chip and 2nd FET.

Down sides, it doesn't self compensate, so you have to crimp your injector wires correctly.

Other notes, I used a resistor pack to show the max chip because it's an 8 pin package, so ignore the insides of that symbol. Also this watch dog timer isn't quite right, but it's close, and I'm sure the remaining small stuff can be found in another chip.