batch-fire on SFI engine?

[KLattig]

Wondering if anyone can give some guidance on running a SFI engine (LX9 - 3.5l from 2006 Malibu) in batch-fire mode. I've got this engine in an MGB, running & driving on MegaSquirt II/extra, but can't pass emissions - I have a minor mis-fire at idle causing high HC in the State emissions test. I think I've tried everything - AFRs from 12ish to 15, ignition advance from 6-25BTDC, various idle speeds; and nothing seems to have any affect on the misfire. It's a 40k mile engine with new plugs, coil, & wires.
I'd heard that batch fire would result in lower idle quality, but didn't think it would impact emissions much (I'm testing at about 1500ppm HC, vs 500ppm max). The engine runs well at speed, and passes the Loaded test with flying colors (~300ppm HC). But at idle, I can hear an occasional miss, and the idle is clearly wandering around.
So, am I chasing my tail? Maybe there's something about the manifold / cam that requires SFI to idle well? (Hoping not to have to upgrade to MS3 with SFI!)
Thanks for your thoughts!

[caffeine]

batch-fire on SFI engine?

What injectors are you using? Have you set the injector lag times table correctly? If that table is off it can cause some stumbling/misfiring at low pulsewidths i.e. idle and low load conditions.

If you are using 3400 injectors you should be able to extrapolate the correct values from a stock 3.1 binary. Otherwise for 3500 injectors you may need some trial and error.

Are you using a wideband? What are the AFRs at idle and do the AFRs stay stable?

[KLattig]

What injectors are you using? Have you set the injector lag times table correctly? If that table is off it can cause some stumbling/misfiring at low pulsewidths i.e. idle and low load conditions.

Are you using a wideband? What are the AFRs at idle and do the AFRs stay stable?

Interesting idea... I'm using the stock injectors that came with the engine - not sure exactly of their specs. I have not specified injector timing - not sure if I can with MSII. Given that I fire 3 injectors at once, what should my target timing be?

I am using a wideband (Innovate MTX-L). AFRs are reasonably stable (+/- 0.3). I can of course adjust the AFR based on the VE table; I've tried several settings ranging from 12ish to 15ish, with no affect on the occasional mis-fire.

[caffeine]

batch-fire on SFI engine?

There should be a table for injector lag time vs voltage. It may have a different name and I can't imagine any standalone not having such a table.

Different types of injectors have different lag times and also running the injectors at different fuel pressures will also have an effect on lag times. It is basically to tell the computer how long it takes for the injector to open so it knows how much time to add to the pulsewidth.

Having incorrect values can cause erratic behaviour in low load situations where the lag times take up more of the total pulse width.

I would start with the settings in a stock 3.1 tune and then adjust higher/lower as necessary. You will mostly be focusing on the settings around 12-14V.

The number of injectors firing at once doesn't really have anything to do with it.

[caffeine]

I went through a similar issue but WAY more pronounced with the USCAR style 550cc injectors in my turbo engine. It would always have trouble maintaining a smooth idle. After adjusting that table and getting the values correct I immediately had a perfectly smooth idle.

[Superdave]

I think Caffeine is on the right track, I've had the same issues with MPFI. I have gotten them to idle as smooth as glass but it takes a lot of fine tuning.

15-20 degrees at idle is about where my MPFI tunes sit with a stock cam. Idle AFR's are usually mid 14's.

I had a customer that tried to run my tune in his car with those 3500 injectors instead of the recommended 3400's, It kept having idle issues and I'm pretty sure it was the injector offset tables. Never got the chance to fix it, i just refunded his money and he went and got it tuned at a shop. Never heard if they had any success.

[KLattig]

There should be a table for injector lag time vs voltage. It may have a different name and I can't imagine any standalone not having such a table.

Different types of injectors have different lag times and also running the injectors at different fuel pressures will also have an effect on lag times. It is basically to tell the computer how long it takes for the injector to open so it knows how much time to add to the pulsewidth.

Having incorrect values can cause erratic behaviour in low load situations where the lag times take up more of the total pulse width.

Yes, I see where I can adjust injector 'dead time', as well as a voltage compensation figure. I currently have 1.0ms & 0.2ms/V in there (but no basis for these figures). My idle pulsewidths are roughly 2.5ms.
I had given the matter some thought when I was designing the system, and concluded that I should be able to adjust for this using the VE table. This still seems logical to me, but having read much more about it this morning, it seems that I've gotta figure out the actual dead times.
Looks like I have a test bench to build - stay tuned for results. Thanks for the feedback, caffeine!

[KLattig]

I had a customer that tried to run my tune in his car with those 3500 injectors instead of the recommended 3400's, It kept having idle issues and I'm pretty sure it was the injector offset tables. Never got the chance to fix it, i just refunded his money and he went and got it tuned at a shop. Never heard if they had any success.

Superdave, can you elaborate on the 'recommended' 3400 injectors? Is there a particular reason to use 3400's instead of the ones that came stock with my 3.5l?

[caffeine]

I think it's mostly because the 3500 injectors/fuel rail use a different style (returnless) fuel system and as a result for most swappers it simplifies things to use a standard return style rail and injectors. 3400 injectors will not fit in the 3500 fuel rail because they are too long.

If you don't have the lag times dialled in I don't believe it is possible to tune for them with the VE tables. I would start playing with that table and I bet you will be able to get rid of that misfire.

[Superdave]

Superdave, can you elaborate on the 'recommended' 3400 injectors? Is there a particular reason to use 3400's instead of the ones that came stock with my 3.5l?

On most of the swaps guys are using the standard return type fuel systems (a majority of the chips I sell are for J and L body cars upgrading from 2.8/3.1 MPFI engines). It's easier to use a 3400 fuel rail and injectors. They have a good spray pattern and are good for 250 HP or a little more.

Attached is the BPW vs Battery Voltage offset table from one of my 3500 tunes. I think it's pretty much the stock 3.1 values, they always seemed to work well enough in most cases.

[KLattig]

On most of the swaps guys are using the standard return type fuel systems (a majority of the chips I sell are for J and L body cars upgrading from 2.8/3.1 MPFI engines). It's easier to use a 3400 fuel rail and injectors. They have a good spray pattern and are good for 250 HP or a little more.

Attached is the BPW vs Battery Voltage offset table from one of my 3500 tunes. I think it's pretty much the stock 3.1 values, they always seemed to work well enough in most cases

Ah, that makes sense. Since I was starting from scratch with the MGB, returnless using the stock injectors & rails seemed easiest.
Thanks for sending that "BPW" chart over - very interesting. Can you confirm the P/N on the injectors you're referencing? Mine say "12586557" & I've been unable to find any data on the web about them.
In Megasquirt (at least the version I'm using), I can only enter a 'base' dead time value for 13.2V, and a delta V value. Your data is pretty linear in the range of 11-16V, so this should work OK (assuming we're actually talking about the same injectors).

[caffeine]

batch-fire on SFI engine?

The table he sent you is what he uses for 3400 injectors. I don't think you'll be able to find lag times specifically for the 3500 injectors so your best bet is to start with that data and then start adjusting to see what makes a difference. With my 550cc injectors which are the same style as 3500 injectors (USCAR style), I found that the lag times were longer than earlier multec style injectors.

So to start off I would try adjusting the value to a bit longer than what's in the table above. Keep adjusting until you notice a difference. If it makes the problem worse, adjust to shorter times instead.

Also if you're running a different fuel pressure from stock that will affect lag times as well.

[KLattig]

So to start off I would try adjusting the value to a bit longer than what's in the table above. Keep adjusting until you notice a difference. If it makes the problem worse, adjust to shorter times instead.

Also if you're running a different fuel pressure from stock that will affect lag times as well.

Thanks for the clarification, caffeine. I spent a little time today working out how I'd make a test bench to get exact dead time numbers, and eventually concluded that it may be worth simply winging it with some educated guesses at the right numbers. I was thinking of this procedure:
- set an arbitrary dead time
- drive & tune (MS has an "VE Analyser" function that works with a datalog to help you update the VE table and achieve your target AFR across the various bins)
- see if the idle is better & make an educated guess at at a new dead time
- repeat

But your comment about "Keep adjusting until you notice a difference" has me wondering if there's another way to tell if a particular dead time setting is better/worse than another. Can you elaborate?

Thanks for your help on this!

[caffeine]

When I increased the dead time for my 550cc injectors, I immediately noticed the difference in idle.

Either the idle will get choppier or it should smooth out. If you're adjusting the wrong way I imagine it will become apparent fairly quickly.


Also, you should test the adjustments with the engine fully warmed up. I noticed with my 550cc injectors that before I adjusted the table the idle would be much smoother on a cold engine (likely because a colder engine will run a bit richer).

[js35]

Ah, that makes sense. Since I was starting from scratch with the MGB, returnless using the stock injectors & rails seemed easiest.
Thanks for sending that "BPW" chart over - very interesting. Can you confirm the P/N on the injectors you're referencing? Mine say "12586557" & I've been unable to find any data on the web about them.
In Megasquirt (at least the version I'm using), I can only enter a 'base' dead time value for 13.2V, and a delta V value. Your data is pretty linear in the range of 11-16V, so this should work OK (assuming we're actually talking about the same injectors).

If you have swept the AFR from 12-15 and this was confirmed with the wideband, then I don't really see a fuel issue here. Whats your pulsewidth millisec at idle? What are you using for a fuel pump and external regulator? you do have an external FPR right? The stock pump has regulator in the tank so it just returns right into the fuel module, so technically it is not a "returnless", if you have some external pump hooked straight shot into the fuel rail you are dead heading the pump and probably seeing 70-90 psi, this can hold injectors closed and increase the dead time so much that less fuel will actually be injected.

unless its a dead hole missfires can be hard to diagnose without obd2 or an oscope, pulling plug wires one at time usually doesnt cut it, here's something to try: every missfire you get a puff of cool air and raw fuel(<-high HC) out of the exhaust valve, pull any heat shielding off the manifolds if installed and let it idle and miss for afew min. hit every runner with a non contact temp gun, write down the numbers. temps will vary slightly but the missing cylinder will stand out cooler, if they all are about the same or vary wildly then it may be in the settings or mechanical issue. If you find a cold one swap the plug with another cylinder and recheck afterwards, do a compression test and if you have it a leakdown as well. low compression missfires can show up at idle only if its not too low.
sometimes a little whif of propane from an unlit torch into the intake can smooth out and idle and tell you if one or more cylinders are lean