This Forum is now CLOSED use the link to get more details viewtopic.php?f=3&t=13924#p102587
By Thack
#60274
Oh, I forgot to mention. My earlier measurements suggested three pulses per cycle for some of the time. This was a mistake. In fact, when an injection pulse spans TDC on the compression stroke, I was counting half of it in one cycle and half in the next, thus registering it as two pulses when in reality it's one pulse that spans two cycles.



Only when I graphed it did it become obvious what was happening. With luck it will become more obvious to you, too, when I've uploaded the graphs!
By Thack
#60307
So here is a picture illustrating what the fuel injection does.



This is taken from a flat-out acceleration run, starting at 3000rpm in top gear, with the throttle held wide open for thirty seconds. The road was flat, there was no noticeable wind, the engine was fully warmed up and the ambient temperature was around 20C.



The picture shows a number of snapshots going down the page; each is numbered with the time at which it was taken (beginning at 0s). For each snapshot you will see a diagram of the pulse timing, the throttle opening and the engine rpm.



The pulse diagram has time along the x axis, and the blue blobs show the start and end of the pulses. The x axes are to scale, which is why they get shorter as the engine rpm gets faster. Above the blobs is a note of the total injection pulse duration.



The x axis lasts for ONE FOUR-STROKE CYCLE (two revolutions of the crank). Unfortunately I couldn't conveniently trigger from exactly TDC; rather I've triggered from the spark event which is perhaps 20 degrees before TDC(??). Therefore the coloured line above the diagram showing the strokes is only approximate. To summarise: each snapshot goes from one spark event to the next spark event.



Unfortunately I didn't realise the engine often injects right across TDC. This makes the diagrams harder to interpret, but I'll explain next.



Image



Please begin by looking at the snapshot labelled "3s". Because of the difficulty I mentioned earlier, the first, longer blob on the left is actually the remainder of the pulse which started in the previous cycle, and the small blob on the right is the start of the next injection pulse, which continues into the next cycle. So in reality we are seeing ONE PULSE PER CYCLE, it's just that it is spread across TDC on the compression stroke. (Triggering off the spark event was a bad idea!) Because the pulse length doesn't change much between consecutive cycles we can still add the blobs together to get a good idea of the total injection duration per cycle, and you can read this from the middle of each pulse diagram.



As you scan down the page you will see the injection pulse duration remains pretty constant, between 12.1ms and 12.5ms, even though the engine speed is increasing. This duration presumably provides the perfect mixture for steady-state wide-open-throttle acceleration.



After 30s I roll off the throttle. At 31.3s the throttle is at 54%, although the engine had barely slowed down. The fuel injection pulse is now down to 10.6ms. As I continue to close the throttle the injection pulse gets shorter, until it is only 3.4ms at 32s. Note how the pulse is shortened by delaying its start - the end point remains the same.



We skipped the first three seconds earlier. At 0s - when I had just snapped open the throttle - the injection time was longer, at 14.8ms. This shows the temporary acceleration enrichment, which allows the engine to respond instantly to rapid throttle demands. Over the next three seconds it gradually falls to 12.6ms, where it goes into the steady-state explained earlier.



Finally, notice how the extra fuel is supplied by an additional pulse, this time around TDC on the exhaust stroke. This second pulse gradually gets shorter, and then disappears, over the three second period.



I've got some more diagrams which I'll upload separately. These show what happens at idle, and a few other places. This use of a second pulse is quite common. The second pulse is always around TDC on the exhaust stroke.
By Thack
#60308
Here is another assortment of snapshots, showing various throttle openings and engine speeds (mostly low throttle and speed).



Image



It isn't at all clear to my under what circumstances the fuel injection decides to use two pulses rather than one. In the acceleration run it was for the temporary acceleration enrichment, but in these examples it often happens at quite low throttle openings and engine speeds. I might look into this a bit more.



Meanwhile, I apologise to the 99.5% of the board members who aren't remotely interested in this! :-)
By jefrs
#60314
Thack, you need to verify from other sources that injecting against a closed valve is normal since it is at least in part the induction, the air being drawn past the injector that allows fuel vaporization. I distinctly remember a set of Gardner injectors that were vacuum operated, the pump being driven by manifold vacuum, i.e. they would only inject on the intake induction stroke with the inlet valve open. In that they are much like a carburettor in that fuel is only drawn from the jet when air passes over it when sucked into the cylinder. The fuel leaves the injector in atomized form, you will appreciate that a (partial) vacuum will vaporize it into the air. Sanity check - this finding is not making sense.
By Thack
#60330
jefrs writes: "Thack, you need to verify from other sources that injecting against a closed valve is normal since it is at least in part the induction, the air being drawn past the injector that allows fuel vaporization."



Actually I don't think I "need" to do anything of the sort - I offer my findings on a take-it-or-leave-it basis. However, the abstract of this paper suggests that closed-valve and open-valve injection timing both have advantages and disadvantages, with neither being obviously better than the other overall.







Also, if you want to learn more about it I suggest you use Google to search for the three main injection regimes: sequential, bank-fire and gang-fire. Sequential is the one where the injection timing occurs separately for each cylinder, normally during the intake stroke. Bank-fire fires the injectors in groups (thus some will be against a closed valve) and gang-fire fires all the injectors simultaneously (ditto). It is normal to switch between regimes depending upon engine speed and load.





By Thack
#60331
Sorry - totally messed that up. Click on the green text for the link to the paper!
By jefrs
#60332
Thack, no ill will here, a "sanity check" is simply what we called it when we passed our findings over to a colleague or two prior to publication to make sure it all made sense. It's not really doing that. I suggest you research other published papers. It is not the done thing to publish findings as-is, take it or leave it, but to make absolutely sure you're right by double double checking everything. I would not like you to hand in your thesis and have them pull it apart. Discuss it with your tutor.
By Thack
#60338
Ah, I see. Yes, of course - for the purposes of the board we post what we like, for informal discussion. It is fine that we challenge each other - and respond likewise.



As for a final year thesis, then I totally agree that it needs to be far more rigorous, referenced, and so on. Interestingly, it was in the second year that our tutor taught us about the various fuel injection pulse timing regimes.



It's worth pointing out that injection which isn't timed to the intake pulses has been quite common historically. The Bosch K-Jetronic system injects continuously. It was fitted to loads of cars back in the 70s.



Personally I am only 90% confident in my findings because the hardware I'm using is barely up to the job, which is why it took so long. Ideally it would need to be backed up by running the bike under load in a lab when it's wired up to an oscilloscope. Perhaps I can do that in the coming months.
By jefrs
#60344
From what I've read from 'reliable' online sources such as Motec, injector systems like Keihin try to deliver fuel before the inlet valve closes, using what is known as End Of Injection (EOI). Some systems such as Toyota use SOI. I do not know the inlet timing for the EFI, lots of stuff for adjusting cams but not degrees when it opens or closes - anyone?



My back of envelope for piston stroke, example 4800rpm (a nice divisible number) is 80 revs per second = 160 strokes (up and down) per second, so a stoke takes (1/160) 6.25ms. An typical injector may take 10ms and so would commence squirting against the closed inlet. There is a problem noted here that wetting the port walls provides poor atomisation, although good for starting.

At low revs the inlet is pulsing, at high revs it may be considered continuous flow. Our injector may be able to deliver sufficient fuel within out inlet valve's opening period on our slow-revving engine.



Motec forum response here but I think he got his crank degrees per 10ms a bit miscalculated http://www.motec.com/forum/viewtopic.php?f=39&t=769
By jefrs
#60345
Bosch injectors are used on BMW, and Bosch won't tell anyone how it works. It is a different system to Keihin.

Shop for accessories at Hitchcocks Motorcycles