Accurate instrumentation possible on a diesel?

[TheTestPilot]

Hi Oldbeaver

I haven't given up ,life just gets in the way.

What pressure is in the injector feed line(s) of your engine?
Are the injectors bank inject or sequential?
Common rail fed injectors?

Regards
Tp

[oldbeaver]

Hello TheTestPilot,

Tks for coming. Well, the injectors are not common rail. The engine uses indirect injection. Each injector has over 20:1 of pressure I asume, as they must defeat compression in the cylinder.

I recall I have a compression test to the engine, I can search it if you want.

What are you thinking about?

Looking forward to hear about you and yr thoughts,

Oldbeaver

[TheTestPilot]

This is just theory , fit a high pressure sensor (0 -5 v output) just before the injector ,this should produce a pulse with a width that will vary with the amount of diesel being injected ,I am basing this idea on the Nissan 4.2 turbo diesel ,as I have a vague idea how this mechanical injection system works.
Depending on the residual pressure in the injector line ,after the flow stops,there may be the need to do some electronic conditioning so the MpGuino can see the signal .

[oldbeaver]

Dear TheTestPilot,

Clever idea of yours! I think this should work. Sure will need some electronics to convert pressure properly to a variable that represents volume injected.

Thinking loud on the same idea, we can count the injection pulses per minute, just counting the rpm pulses, right? Injection pulses per cylinder should be rpm/2 in a 4 stroke engine. As my engine has 6 cylinders, number of total strokes are 3 per revolution of the engine.

Based on the injection pump data, I can know an approximate figure of injection volume per stroke either. Further on, I can make a family of convex curves that represent the volume injected per stroke as a function of rpm. This function is a parabola (quadratic) function.

So total volume is K*rpm^3 (approximately).

What I don't know is how does throttle position determine to the amount of fuel injected on each stroke. There are models for that.

But this is just an approximation to the real injected volume.

Your approach, allow us to have a measure of real volume injected, whatever the surrounding conditions. I like it very much.

We need a pressure sensor that gives a continuous measure of pressure, so we can generate a continuous flow value as well. Definition of minimum and maximum pressure (this has a name I cannot recall now) to determine pulse duration is crucial too. Having pulse duration and pulse flow (stroke duration and stroke flow) will get a very good approximation to real volume. Then comes callibration.

Maybe we can make:

Volume per stroke = Stroke duration x mean pressure

Volume per stroke = Stroke duration x (Max pressure - Min pressure)/2

Futher we must count the number of stroke per unit of time, to get:

Volume/hour = volume per stroke x number of strokes/hour

I don´t know what kind of signal is MPGuino expecting, but this kind of signal should suit.

My problem is to build an extra circuit to make all the necessary processing previous to input the signal to MPGuino.

Therefore, in your approach the pressure sensor is key.

This is my comment. Looking forward to yours.

Oldbeaver

[TheTestPilot]

Hi Oldbeaver

This idea came from my ongoing development of a flow sensor for a carburettor & the MPGuino.

I was originally just going to fit a sensor to the injection line & feed it to the MpGuino injector Input ,but a little research later ,I discovered the crack pressure for the potential test vehicle's injector is approx. 100 bar ,as yet I have been unable to find what the actual injection maximum pressure is.

My though train is to fit a properly rated sensor( DIY GUYS,DANGEROUS PRESSURES HERE KNOWLEDGE REQUIRED) ,clean the static part of the signal out , detect the upper & lower dynamic signal points & turn what is left into pwm output compatible with the MPGuino injector input with some analog electronic parts(the design has been floating around in my head for some time , easy )

I just need to see the test vehicle in more detail tomorrow to make sure I am on the right track.

I like some of your ideas ,although I need to keep this really simple to prove the concept as other life priorities are calling.

Tp

[AndrzejM]

Don't have access to old diesel engine with standard injectors right now. But I'm still thinking of "listening" the injector with the microphone. With an amplifier and some signal cutting I think that it will be possible to "hear" the difference between injection pulses for different loads.

[P-hack]

I think the pressure is fairly predictable since it is a spring loaded injector, so that if you just time the event (i.e. with a piezo clamped to the fuel line) you can get a good approximation using inj delay somehow (and rpm).




The noisy signal from the adapter (diesel timing light) needs processing power, but they are simple devices (and you just clamp it on a fuel line):

[oldbeaver]

Dear dieseler's,

To me all sensors mentioned are ok: pressure, sound or piezo. However, the less invasive and the simpler the better. The piezo electric seems to me to have some advantage as piezo adapters and timing lights for diesels already exist. This should ease the transducing part of the process with piezo sensors, in comparison to the other two.

But, some working with the signal may be necessary anyway. The graph published by P-hack probably represents the situation. If vibration is like the graph, we should be able to convert that into digital pulses representing injected volume.

To make it simple: we may think that the fuel injected is proportional to the area under the curve. If that is true, we need to get duration of pulse and mean voltage generated. And we need to count number of pulses per unit of time, of course.

I would study in the oscilloscope what happen with the injection pulses under different loads. My only concern with this signal is: would the graph get that variation? If so, there should be a variation on the voltage generated.

AndrzejM believes that sound on injectors should vary with load, so signal generated on the mike devoted to capture it. I think he is right. However, working that signal should be a lot harder.

On the other hand, the pressure sensor proposed by TheTestPilot may probably be the most direct and "true" measure of injected volume per stroke. However, this tranducer looks the hardest to manage to me.

Question to P-Hack: what source for that piezo timing light systems do you know?

Oldbeaver

[P-hack]

Originally Posted by oldbeaver

Question to P-Hack: what source for that piezo timing light systems do you know?

Oldbeaver

There's a few but they cost hundreds (google "diesel timing light" or "diesel pressure transducer"), but I think someone with a diesel and a scope and a couple bucks for some radio shack piezos should see what they can come up with (note, the piezos can be wired to cancel out noise). Maybe take two, and take them apart and cut them down to a strip (with the leads intact) and clamp them on opposite sides of the fuel line with a hose clamp (and a bit of hard plastic) and see what kind of signals you get.

[P-hack]

also, there is the tiny-tach, has some sort of diesel transducer listed for $58 (don't know anything else about it)
Design Technology, Inc., Home of the Tiny-Tach - More Tachometers
part number 28032