Aftermarket Injection, Why no MAF? Options

[McMark]

I was just thinking about how all the aftermarket injection systems all use vacuum signal which can cause problems with wild cams. How come nobody is using a Mass Air Flow sensor? It can't be that complicated. Famous last words.

[DBCooper]

All the systems I can think of can be configured to use either a MAP sensor or MAF. Most discussion is about MAP sensors because they're simple, easy to fit, and will work on any car that has a reasonable vacuum signal. That means they're the better option for most cars, and especially for cars being upgraded from carburetors. But if you have a MAF system in place you should still be able to use it with a progammable EMS, as long as you don't change the engine's VE much to take it out of the MAF's flow range.

[Dr Evil]

I thought that MAFs were a pain with wild cams. Something about the pulsing air intake causing havoc with the meter. I could be wrong.

[McMark]

So what sensor technology works with wild cams? Is there anything?

[Mueller]

alpha-n works best with wild camshafts....it is almost like an electronic version of the 911 MFI or a kugelfischer fuel injection system....

it's RPM and throttle position based (I think, it's been a while)

[ClayPerrine]

A Vane Air Meter... (the L-Jet AFM) is lousy with a radical cam. The reversion pulses foul up the readings back to the computer. Same goes for the sensor plate on CIS.

D-Jet is a MAP sensor system, and it gets fouled up by the low vacuum.

You can use a Hot Wire MAF. It doesn't get screwed up by the reversion pulses (no moving parts). Hot wire MAF can be retrofitted to a Motronic system.

Or use the Alpha-n system.

[Dr Evil]

Thats what I was reffering to, Clay. You nailed it. Ya, a hot wire set up would have no moving mass to negotiate, only temperature variances. What used such a system?

[ClayPerrine]

Bosch LH-Jetronic... (Volvo) Nissan.. most later models. GM .... Ford... Also available conversions for the 911 Motronic setup on the 3.0 and 3.2 SC.

[gaz914]

QUOTE (ClayPerrine @ Dec 20 2004, 12:34 PM)

You can use a Hot Wire MAF. It doesn't get screwed up by the reversion pulses (no moving parts). Hot wire MAF can be retrofitted to a Motronic system.

I don't agree.

If a cam is "wild" enough to generate recursion pulses of enough significance to affect "flap type" metering, then the hot wire will also be affected.

Granted the "problem" is different but there must be an effect.

If you can imagine a "block" of air being sucked past the MAF, stopping and then pushed back out again (reversion) - what does the MAF "see".

What it "sees" is 1 block of air moving past, then another block of air moving past - it cannot dscriminate which direction the air is travelling.

The effect? The air is metered TWICE, so at low speed (where the reversion occurs) the mixture will be very rich.

I think that most "serious" engines use TPS mapped (alpha-n) fuel injection for this reason.

[Series9]

3.6 has a MAF hot wire (it's really a hot spike, but same thing).

That's it right behind the air filter.

[lapuwali]

MAP based systems get confused by the rapid fluctutations in manifold pressure caused by wild cams run at lower revs. Some smoothing of this in hardware (small restriction) or software (averaging) help, but also reduce accuracy. At higher revs, these pulses naturally smooth out.

MAF based systems get confused by the extreme reversion in the intake caused by lots of overlap (exhaust pulses can flow "backwards" all the way back out the intake valve and up the whole intake tract. However, even carbs get confused by this. This causes double carburation, with the backwards wave adding fuel just like the forwards wave. If you run carbs open, you can even see this as "standoff", with a fog of fuel/air hovering just outside the carb mouth during these reversion periods. These come in harmonics of some set frequency based on the length of the intake tract and the cam timing. The torque curve ends up looking like a series of hills, with the peaks where the timing all works together and troughs where they all add up to badness. A great many motorcycle engines have such torque curves, and the last two-valve bike engines were the worst, as they were pushed to compete with the four-valve designs. My '86 FZ600 had torque peaks at 3000, 5000, and 7000rpm, with the torque at 6000rpm lower than the torque at 4000rpm. Weird curve, hard to ride.

Alpha-N is, as Mike says, just RPM and throttle angle (maybe some correction for temp). The injection is tuned for a particular atmospheric condition and state of engine tune. Change anything and you had to remap the injection. Fine for a full-on racing engine, a pain in the ass for a road car, esp. if you lived near any mountains. Some have included a barometric sensor to handle this to some degree, but you're still only guessing at loads (uphill at the same RPM/throttle setting requires more fuel than downhill).

Some systems blend these systems together. They'll rely on the MAP signal at higher revs, and alpha-N at lower revs, for example. The system needs to know how to do the blending, however. Some of the more sophisticated systems (like Motec and Electromotive) do this blending based on yet another map.

MAF also hasn't been widely used in the aftermarket as new MAF parts are very expensive (or impossible to get at all) in small quantities, and basing products on junkyard parts isn't a good business strategy if you're charging $1000 or more for the ECU. In the DIY ECU world (pre-Megasquirt), many people used junkyard hot-wire MAF. MS itself is MAP based, but there are people working on MAF code for it.

Theoretically, MAF is really the way to go for a street engine, as just about everything is automagically accounted for. A modern hot-wire MAF with a long enough intake tract will be unaffected by the cam until it's very wild indeed. If someone sold a MAF in single unit quantities for a reasonable price, I'm sure we'd see it in the aftermarket more often.

[McMark]

Well KitCarlson is a MAP/Alpha-N hybrid I believe. So that's good.

I just thought someone would have picked up a hotwire setup from another car at the parts store and built and injection system around it.

What if there was a system that had MAP/Alpha-N/Hot Wire? If you biased the info so at low RPMS MAP barely counts, Hot Wire counts a little more and Alpha-N counts a lot, etc it might work well. Of course that also might make the system prohibitively expensive. (IMG:http://www.914world.com/bbs2/html/emoticons/confused24.gif)

[lapuwali]

Again, they have, in the DIY world. There's an old list, perhaps defunct now, called DIY_EFI, which consisted of a group that reverse engineered GM ECUs, and attempted to build what turned out to be an overly complex 32-bit ECU. It was this group that the Megasquirt group came from. A number of people there used junkyard MAFs in their applications successfully.

Using both MAF and MAP isn't really useful. What you really want to know is airflow (MAF), and MAP plus revs will allow you to estimate airflow. The great thing about MAF is that there is no estimation. You know exactly how much air in coming in, so you theoretically don't even need to know the throttle position. You don't need to know how many cylinders, or how big those cylinders are, or how fast the engine is spinning. In practice, having a throttle sensor helps with idle and WOT control (that's why early L-Jet TPSes only had two microswitches for idle and WOT), when knowing airflow isn't adequate for really fine-grained control at idle, or for an extra rich mixture. You also don't really get much acceleration enrichment that way (L-Jet and K-Jet fudged this with flapper overswing).

Since there are problems using MAF with REALLY wild cams, and because many of the early MAF sensors were overly restrictive (think L-Jet flappers), and because measuring air pressure is relatively easy compared to MAF, MAP was used first. I strongly suspect this is why D-Jet was developed before L-Jet. Even K-Jet was invented after D-Jet, and it's entirely mechanical.

IMHO, for street engines with reasonable cams, a MAF can be used without any problems. You'll note that OEM EFI systems on cars are now pretty much universally driven by MAF sensors. MAP/revs systems haven't been used in OEM applications pretty much since the D-Jet days.

[airsix]

Great post James!

I did a poor-man's blending of MAP and alpha-n once just for kicks. How'd I do that with a budget digital ECU? I just set up my MAP source slightly upstream of the butterfly valve so that at idle the MAP source was shut off from the intake and read atmospheric. At part throttle the port was exposed to vacuum as the butterfly moved past the port. Then I programmed my injection maps like it was an alpha-n system below 2,500 rpm and a MAP system above that. It worked pretty well for injection but I switched back to a "normal" mode because the MAP signal is also used for ignition advance and I couldn't set up the vacuum advance curve properly with no vacuum signal at low throttle possitions (when vacuum is high and you need more advance angle).

Another point I don't think was mentioned is that a MAF sensor presents an intake restriction that is not present in with a MAP system.

-Ben M.

[McMark]

Yeah, reading your post, I realize having MAP and MAF is redundant. What was I thinking?

Thanks for the info. (IMG:http://www.914world.com/bbs2/html/emoticons/biggrin.gif)

[ClayPerrine]

QUOTE (gaz914 @ Dec 20 2004, 04:32 PM)

[I don't agree. If a cam is "wild" enough to generate recursion pulses of enough significance to affect "flap type" metering, then the hot wire will also be affected. Granted the "problem" is different but there must be an effect. If you can imagine a "block" of air being sucked past the MAF, stopping and then pushed back out again (reversion) - what does the MAF "see". What it "sees" is 1 block of air moving past, then another block of air moving past - it cannot dscriminate which direction the air is travelling. The effect? The air is metered TWICE, so at low speed (where the reversion occurs) the mixture will be very rich. I think that most "serious" engines use TPS mapped (alpha-n) fuel injection for this reason.

The problem with a Vane Air meter is not it's inablility to measure the air coming in correctly, its problem is that the reversion causes the vane to flutter, changing the reading sent back to the computer.

The Hot wire mass air flow sensor does not have a mechanical flap. It heats a wire, and measures the amount of current necessary to keep the wire at a precise temprature. It doesn't care if the air is flowing in or out. It does not matter if the air gets measured twice. The software can be written to take into account the reversion pulses at low speed.

[rick 918-S]

QUOTE (lapuwali @ Dec 20 2004, 02:51 PM)

MAF based systems get confused by the extreme reversion in the intake caused by lots of overlap (exhaust pulses can flow "backwards" all the way back out the intake valve and up the whole intake tract. However, even carbs get confused by this. This causes double carburation, with the backwards wave adding fuel just like the forwards wave. If you run carbs open, you can even see this as "standoff A modern hot-wire MAF with a long enough intake tract will be unaffected by the cam until it's very wild indeed.

It's my understanding that the "fog" is really fuel vapor that is blown back out of the plenium through the venturis by the reverse pulse. Carbs only work on a venturi principal and only in one direction. The problem comes from the cancelling of inward air drawing fuel vapor to the cylinders. That's why carbs with big cams are set up so rich. It's an attempt to cancel the harmonics until until the revs pass that point.

I agree with the intake plemium size and runner length theory. Hence the shoe box tunnel ram on my old BB Chevy. I had a 580 lift with a 490 duration and lots of overlap. The long runners and shoe box helped straighten out the vapor. Same theory should work with injection. In the factory publication "Chevrolet Power" They promoted the use of the Crower MFI. It was a crazy thing with long stacks. It had a big mechanical pump that was driven off the front of the cam. You adjusted the charge with a "pill" like meter. Crazy stuff in the day... (IMG:http://www.914world.com/bbs2/html/emoticons/aktion035.gif)

[Bleyseng]

QUOTE (ClayPerrine @ Dec 21 2004, 05:48 AM)

The Hot wire mass air flow sensor does not have a mechanical flap. It heats a wire, and measures the amount of current necessary to keep the wire at a precise temprature. It doesn't care if the air is flowing in or out. It does not matter if the air gets measured twice. The software can be written to take into account the reversion pulses at low speed.

Or like on my Ranger truck, you move the MAF 3feet away from the TB connected by a tube. I just changed the MAF (60mm) and it was only $70 so they are not pricey. I would think this is the way to go cuz its pretty simple EFI which is good.

Geoff

[lapuwali]

QUOTE

It's my understanding that the "fog" is really fuel vapor that is blown back out of the plenium through the venturis by the reverse pulse. Carbs only work on a venturi principal and only in one direction.

Most carbs don't work all that well "backwards", but many will still supply fuel to the reverse pulse. The venturi works either way. An open throat carb like a Weber will certainly do so. The mixture will just be weaker in the reverse direction. However, if the mixture has already been metered some fuel once, any more will still make it rich. If it gets even more on the trip back through, it will be very rich.

Some time ago, there was a failed attempt by the guy who ran Dynojet, the bike jet and rolling dyno company, to make an all analog ECU for bikes. At the time, carbs were still almost universal on bikes. He cited the reverse flow problem specifically, quoting numbers he obtained on a dyno with a high-precision flowmeter. Run the bike through the rev range on carbs and the flowmeter numbers matched the torque curve, except the "troughs" in the curve showed even more fuel being used than on the peaks.