progress on the banshee...

[pmbrunelle]

most EFI applications I have used in the past (granted, use was limited) performed calculations similar to what was described by not including target AFR, having the VE table control fuel, so I stuck with that, or at least I think that was my mindset at the time. include target AFR sounds simpler by the way you describe it.

I haven't seen a VE table that looks like your's before, but most were using different algorithms for fuel control, so they would look differently. I'd imagine enabling incorporate AFRTarget makes the majority of that difference.

Right, it's a matter of preference, though includeAFRtarget seems simpler to me.

If I weren't running includeAFRtarget, then my "VE" (well, not strictly volumetric efficiency) table values would be greater in the parts of the table where I had a desired mixture richer than stoichiometric.

I've actually been using the VE analyze feature, in MLV, but not the autotune as much.

Whatever way works for you. I like the autotune in TS to rapidly iterate in the beginning, and to use MLV later to fine-tune using a bunch of data. For the last VE fine-tuning iteration I did, I drove around for about a month, datalogging on the SD card (no laptop in the car). I then ran MLV on the month's worth of datalogs.

currently MS3 doesn't support my crank and cam position sensor arrangement for actual sequential injection.

Hmm, so it looks like your injectors are firing twice for every 4-stroke cycle.

If you could fire the injectors once for every 4-stroke cycle, then the same fuel could be delivered with a single, longer pulse. You would only have one dead time error per cycle, not two.

Even without the full sequential support in MS3, you can turn off sequential and run batch fire (2 squirts alternating).

With 2 squirts alternating (one bank squirted, then the other bank is squirted), you'll have the same pulse width as with a full sequential setup.

What OE application were they used in? You may be able to find OE injector table data for them... that you know is backed up by a lot of test time.

That test data will only be applicable if you duplicate their test condition in your car.

As much as dead time depends on the injector, it also depends on fuel pressure, and the flyback voltage (likely to vary between MS and OE test setup/ECU) of your ECU.

I'm not aware of them being used in an OE application, I have thought about getting another set and having them flowed and getting as much data as I can from a fuel injector service, I'm not sure how necessary that is though.

I think that the in-car method could be quite sufficient.

With the car running, toggle between Batch fire (injector firing once per cycle) and Semi-sequential (injector firing twice per cycle) :

Sequential test.jpg (155.18 KiB) Viewed 1053 times

If the ECU is exactly compensating the dead time, then total fuel delivery will be the same regardless of whether it is delivered in one or two shots per cycle. Therefore, measured AFR won't change as you toggle the setting.

Once you have found the dead time that keeps AFR constant, confirm by commanding a 10% richer mixture (either by the VE table, or AFR table if you run includeAFRtarget). If the measured AFR is 10% richer, then you're golden. After all, the goal of getting the correct dead times is that the ECU can adjust the mixture as desired, and obtain the desired results.

You'll want to repeat this test with the alternator disconnected, so that the dead time is calibrated at two voltage points (running voltage and battery voltage).

Since you have a pretty long duration camshaft, you may want to perform the dead time testing/tuning while revving the engine to 2k-3k, so you don't get into fuel bypassing the combustion chamber during the overlap period at idle speed.

This method has worked for me (at idle), but using a milder camshaft than yours.

[ericjon262]

I'll mess with the dead times next tuning session, thanks for the advice!

I haven't played with the car much, but the MS3 is having problems with my crankshaft position sensor signal. I initially thought the problem was noise from the ignition system, I took a composite log of the crank sensor signal, and didn't see noise, I also ran the ignition in test mode at various levels, and saw nothing on the log. One of the admins on the MS3 forum mentioned that the waveform was distorted, confusing the ECU, I'm going to try and put a new crankshaft position sensor on it and see how that helps.

[ericjon262]

Sync loss fixed, there was some stuff on the end of the crank position sensor, it appeared to be metallic/ferrous, for now, I'm going to tell myself it was wear products from the original cam, lifters, rockers, and timing set. hopefully I'm not wrong. whatever it is/was, it looks like it had been there for a while, now the engine runs to 6600+ RPM on a free rev with no sync loss.

[Shaun41178(2)]

Progress

[The Dark Side of Will]

You running a filter magnet now that you've seen that?

[ericjon262]

You running a filter magnet now that you've seen that?

I still need to order one, but yes.

=====================================

I just finished restoring the VE and ignition tables to single tables for the entire KPA range, a few weeks ago when I was chasing idle issues due to the VSS, I split them into 20-100 KPA, and 100-300 KPA, thinking part of my problem may have been table resolution, since it didn't solve my problem, I figured it would be best to make them individual tables to simplify tuning, once the tune is more dialed, I may split them again, but trying to keep track of 4 tables takes more effort than 2.

[ericjon262]

With the car running, toggle between Batch fire (injector firing once per cycle) and Semi-sequential (injector firing twice per cycle) :
Sequential test.jpg

I didn't plan on getting into this today, but here I am...

I idled the engine up to about 1400 RPM, and toggled the test,



there was a fairly clear line between semi sequential(red line above), and batch fire(red line below), I take that to mean my commanded dead times are longer than actual, and should need to be shortened a little bit. The difference in AFR currently appears to be about .8

here's the same chart, with PW added.



tomorrow I'll play with it more, and actually adjust dead times some, I'm not sure how much adjustment I'll need to make, so I'll just play with it, a bit.

[Shaun41178(2)]

Is that 1.6 Milliseconds batch or sequential?

[ericjon262]

Batch, semi sequential is about 1.17 or so.

[Shaun41178(2)]

And you normally run in sequential? 1.17 is into the non linear zone for those injectors in all likelihood. Also I think you are running cheap china injectors? This can result in diff amounts of fuel being sprayed for each pulse width. Obviously this would cause idle problems.

1.6 is much friendlier to the injector

Hell I'll have to check but even my 80s are around 1.6 to 1.8 at idle if I am not mistaken and I think you said yours are 60s

Just an observation

[pmbrunelle]

Sync loss fixed, there was some stuff on the end of the crank position sensor, it appeared to be metallic/ferrous

I can definitely believe that the stuff would cause a sensor malfunction.

I have seen a situation where someone drilled a steel plate above an open box of bare position sensor magnets. The chips fell down and stuck onto the magnets.

A whole bunch of assembled position sensors failed their end-of-line performance test... good times

[ericjon262]

And you normally run in sequential? 1.17 is into the non linear zone for those injectors in all likelihood. Also I think you are running cheap china injectors? This can result in diff amounts of fuel being sprayed for each pulse width. Obviously this would cause idle problems.

1.6 is much friendlier to the injector

Hell I'll have to check but even my 80s are around 1.6 to 1.8 at idle if I am not mistaken and I think you said yours are 60s

Just an observation

it's idling fine down there right now, unfortunately, I'm stuck with semi sequential or batch fire until the next rev of the MS3 firmware. full sequential should operate at about the same PW as batch fire, so once the firmware comes out (if it comes) I'll swap over to that. that's odd that you have so much more PW, with that much larger of an injector, that being said, your micro is batch fire, in batch, mine is about 1.6-1.7.

My injectors are genuine Deka 60's, unless Lingenfelter was selling knockoffs when I bought them. thankfully, most of my idle problems seem to have gone away now that I have the new VSS installed.

Sync loss fixed, there was some stuff on the end of the crank position sensor, it appeared to be metallic/ferrous

I can definitely believe that the stuff would cause a sensor malfunction.

I have seen a situation where someone drilled a steel plate above an open box of bare position sensor magnets. The chips fell down and stuck onto the magnets.

A whole bunch of assembled position sensors failed their end-of-line performance test... good times

oh geez! what a mess that must have been! as soon as I saw it, I was both disturbed, and a bit relieved, as it was kinda a smoking gun. I'm not super worried about it yet, because the engine itself is about 18 years old, with an untold number of miles, that could have been accumulating the entire time.

[ericjon262]

I played around with dead times a bit, it seemed as though I could make the difference bigger, but not much smaller, I suspect this is because, as Shaun alluded to, my injectors are running on the ragged edge in semi sequential. I settled out at .697, down .003 from my starting value of .700. in this process, I found the engine ran a bit smoother in batch fire than in semi, so I decided to just run it in batch for now.

for a while, I had been using the Mega log viewer VE analyze feature, but I've noticed that it doesn't seem to actually correct the tables as it should, I'm not sure if this has to do with my setup specifically, or something else. using the "Tune for you" function seems to do a much better job, I just like being able to see stepped changes. but I guess that's what the heat maps are for.

The car scoots nicely, I'm really curious what it will feel like when I enable 100% TPS, it comes on hard,

I ordered a filter magnet, who knows when it will be here, but when it does I'll run the engine a bit, swap the filter, cut it open and see what I find. I have an 1800 lb pull magnet in the shop, but I thought that may crush the filter... lol.

[Shaun41178(2)]

Don't remember if you mentioned your fuel pressure but you could lower it to 42ish if you are running it higher.

That will bring your pw up.

[Shaun41178(2)]

Just checked a datalog of my injector pw.

2.2 Ms at idle. 177 coolant temp

IMG_20220310_210433616_MP~3.jpg (3.49 MiB) Viewed 960 times

[pmbrunelle]

I played around with dead times a bit, it seemed as though I could make the difference bigger, but not much smaller, I suspect this is because, as Shaun alluded to, my injectors are running on the ragged edge in semi sequential. I settled out at .697, down .003 from my starting value of .700.

For this to work, you might need to be driving/cruising with a load while a passenger twiddles around with a laptop.

Constant-speed dyno would be ideal, but only if tuning other things as well. I wouldn't rent a dyno just to mess with dead times!

[Honest Don]

I ordered a filter magnet, who knows when it will be here, but when it does I'll run the engine a bit, swap the filter, cut it open and see what I find. I have an 1800 lb pull magnet in the shop, but I thought that may crush the filter... lol.

Magnetic oil plug from an LS car is a nice addition as well

[ericjon262]

I've heard one report (from a guy on facebook soo....) that ferrous material on the CPS is fairly common I'm still going to run the magenet, and continue to cut my oil filters for inspection.

[ericjon262]

got the filter magnet in today, I think tomorrow, or maybe tonight I'll get a new filter and change the oil, I think I can get it done without even jacking the car.

I've been slowly stepping the throttle up, and the VE table is beginning to smooth out quite a bit, and have a shape more like Patrick's it still has some abnormalities that need to be sorted, and everything above or to the right of the red lines are parts of the table I have no data for yet, although I have done some extrapolation so the table doesn't have a straight up cliff on it.



The switch to batch fire definitely improved the run quality of the engine, I really need to start working on startup characteristics now, warmup enrichment, after start enrichment, ect. as the car doesn't like to stay running after starting. once it's run a 15-45 seconds, it's ok, but the immediate after start isn't there.

The car is getting better almost every time I drive it, even without idle control enabled, it rarely misses idle once it's running, first gear goes by in a blink, babying it in first yesterday, with the throttle limited to 60%, and bumping the rev limiter, it went 0-60 in less than 6.5 seconds. I'm wondering how it will respond when it starts actually getting force fed...

[pmbrunelle]

I really need to start working on startup characteristics now, warmup enrichment, after start enrichment, ect. as the car doesn't like to stay running after starting. once it's run a 15-45 seconds, it's ok, but the immediate after start isn't there.

MegaSquirt afterstart enrichment is often described as being used to keep the engine alive for the first 2-3 seconds after starting, but on my own car the afterstart enrichment at 0°C has a 45-second taper time.

I could not tune the car using only warmup enrichment (depending on coolant temp only). There really was an apparent lean condition that tapered off over a period of 20-30 seconds or so (visible in the wideband datalog).

I think that the intake valve is warming up following startup, and that as the intake valve warms up, fuel that hits the valve vaporises better and becomes more combustible.

The time constant of the intake valve temperature may be in the tens of seconds, so my afterstart enrichment taper time reflects this.

Intake valve temperature doesn't rise at the same rate as coolant temperature, so it makes sense that there should be a different enrichment scheme to deal with that phenomenon (though MegaSquirt doesn't specifically address intake valve temperature).

High Value V6 may differ in the details from my Gen 1, but the general idea of fuel spraying onto a valve is still there.