legacycentral bbs

This conversation started in PM, but it may be of interest to others so I’ll post it here.

Do you know the voltage output range of the stock MAF? I understand it tops out at the 5V reference, but what would it be at idle? I’ve seen your graph, can we correlate output voltage to it? I also under stand that max flow is 200g/m at 5V, true? Do you have any idea what would be a typical flow at idle?

Do you know the injector pulse width at idle (typical range)? And is injector flow fairly linier with changes in pulse width?

What all this is leading to is I’m hoping to use the EJ22T’s MAF and injectors in the earlier EA82T, this would give an additional 50% flow and eliminate the fuel cut issues that occur with any attempt to increase flow (additional boost or otherwise). This would require both ECUs to have similar architecture and input/output ranges, the ECUs are one generation apart so it might be possible. I think I can check the range of the EA82T MAF with a voltmeter, is there any simple way to read the injector pulse width? Maybe some type of RMS voltage, or is an oscilloscope the only actuate way?

Lastly, are the N/A EJ22 MAF and injectors the same as the turbo, if not, what sort of flow are the capable of?

Thanks for your time
Gary

Here is vrg3's responce

I've been searching the site for some info on the stock EJ22T system and you seem to be in the know.

Quote:
Do you know the voltage output range of the stock MAF?


I don't understand what exactly you mean by this question... The signal can range from somewhere around 0 volts up to somewhere around 5 volts, and the meaning of the signal is given by the transfer function graph that I've posted and you've seen.

Quote:
I understand it tops out at the 5V reference, but what would it be at idle?


At idle the signal usually sits somewhere around 1 volt.

Quote:
I’ve seen your graph, can we correlate output voltage to it?


Yes; the horizontal axis is the voltage (or the voltage times 32 or times 8 depending on what units I chose).

Quote:
I also under stand that max flow is 200g/m at 5V, true?


At 5 volts the airflow is around 200 grams per second (not per minute).

Quote:
Do you have any idea what would be a typical flow at idle?


I dunno, somewhere around 10 grams a second maybe?

Quote:
Do you know the injector pulse width at idle (typical range)?


Around 2 or 3 milliseconds.

Quote:
And is injector flow fairly linier with changes in pulse width?


Very roughly, yes. To do it a little more accurately you have to model the time it takes the injector to move from the closed position to the open position. This time is roughly proportional to system voltage.

Quote:
What all this is leading to is I’m hoping to use the EJ22T’s MAF and injectors in the earlier EA82T, this would give an additional 50% flow and eliminate the fuel cut issues that occur with any attempt to increase flow (additional boost or otherwise).


This seems like a cool idea in principle.

How does the EA82T's stock engine management control ignition timing?

Quote:
This would require both ECUs to have similar architecture and input/output ranges, the ECUs are one generation apart so it might be possible.


Weellll... I don't think that's true (the first part, I mean). If you change out the injectors and MAF sensor but keep the ECU, you're basically throwing the architecture out the window. You're going to mangle the input to the system (the airflow signal) and the output of the system (the amount of fuel injected) and hope you can make the two cancel out.

One difficulty you'll encounter is that hot-wire and hot-film MAF sensors typically have quadratic transfer functions, while injector flow is approximately linear. So, you can't just change out the injectors and sensor and get "correct" behavior all the time.

Also, doesn't the EA82T ECU have control over ignition timing? I would imagine it would compute load based on airflow and then determine spark advance using this load computation. If you make the ECU think there's less airflow than there really is, you may get undesirably advanced spark.

You might be able to make this stuff work out in the end though. You'll want good instrumentation, I think.

Another thing you might want to consider... I'll preface by saying I have no idea how the EA82T does its camshaft and/or crankshaft position sensing, but maybe you could look into the possibility of using an EJ22T ECU as well. The hard part would just be to get the right kind of crankshaft and camshaft position signals, assuming the EA82T does it very differently. But maybe you could hack up your crank pulley and a cam sprocket to give the right signals.

Quote:
I think I can check the range of the EA82T MAF with a voltmeter,


What do you mean by the range? Wouldn't you need some kind of flow bench to determine the transfer function?

Quote:
is there any simple way to read the injector pulse width? Maybe some type of RMS voltage, or is an oscilloscope the only actuate way?


Many quality multimeters (especially those designed for automotive use, of course) have the ability to display duty cycle and/or pulse width.

They do also make some very crude "injector duty cycle" meters that basically use a capacitor to integrate a square wave representing the injector drive waveform, and in that way display the average value of the waveform.

Quote:
Lastly, are the N/A EJ22 MAF and injectors the same as the turbo, if not, what sort of flow are the capable of?


90-91 5-speed N/A EJ22s use the same MAF sensor as the EJ22T.

90-91 automatic N/A EJ22s, and all 92-94 N/A EJ22s, use a MAF sensor made by JECS with a maximum flow reading of around 300 grams per second.

As far as I know, all N/A EJ22s use injectors that flow about 280 cc/min, roughly 75% as much as the turbo injectors.

Quote:
Thanks for your time


You're welcome. Good luck!

-Vikash

...why'd the N/A guys get a larger range on the MAF sensor than the turbo guys? Bad Subaru, bad!

I’ve been busy lately, but here are a few more thoughts…

IF the EA82T MAF’s signal ranges from 0 to 5V (I believe it does, still need to test).

IF the EA82T MAF’s transfer function follows a curve “similar” to the EJ22T (big assumption, and much harder to test).

Based on the 180cc rating for the EA injector the EA MAF should flow around 130g/s @ 5V.

MAF ratios: 200/130 = 1.54
Injector ratio: 280/180 = 1.56
Looks like the parts from the ’90-’91 N/A 5MT EJ22 are a good match.

The EA has a lower static fuel pressure 36psi vs. 43psi for the EJ, so I’m assuming I’ll need to raise the fuel pressure to the EJ 43 or so for the scaling to hold true.

I scaled the MAF’s transfer function of the EJ to get a similar function for the EA over the lower flow to see the trend in voltage differential.



The voltage differential increases with higher flow, I believe this will cause a leaner A/F relative to the balance set at idle. But the higher flow will most likely be due to higher boost pressures, which will raise fuel pressure and hopefully offset this.

The ignition timing on the EA82T is controlled by the ECU, but the crank/cam position is sensed by a module in the distributor, and because is adjustable I should be able to shift the whole timing map down. This will mean less total advance, but I should be able to find a balance that mitigates the over-advance due to the lower flow/load sensed. The EA82T also has a knock sensor, which provides up to 5deg of retard.

Thoughts
Gary

Chris - The JECS MAF sensor just has a larger range than the Hitachi one. No Legacy/Impreza/etc turbos used Hitachi engine management aside from ours; all the overseas ones used JECS.

Gary - It's almost certain that the EA82T's MAF sensor has a quadratic transfer function similar to the EJ22T's. But there's no reason to think that the EA MAF sensor tops out right at the EA engine's peak power production. Consider that a 94 EJ22 non-turbo has a MAF sensor that tops out at almost three times the stock output. You need to test this.

You might want to try just hooking up both the EA and the EJ MAF sensors to your engine (in series) and using two voltmeters to collect some ordered pairs under steady state conditions. Then do it again with the sensors in the opposite order to try to reduce the effect of one sensor masking the other or something.

The EJ22T has a static fuel pressure of 36.3 psi.

A smaller "error" at lower flow will actually mean that you'll run richer
than you should at low loads, right? Since the input is about the same but the output (injector size) is higher?

Don't think about fuel pressure. Higher boost pressures do create higher fuel pressures but they don't increase flow.

Yes, I need to test this, was on my "to do" list for last weekend but I got tied up fixing the brakes on my Z and putting the cruise control in my Subie RX. Its on the list for this week.



Great idea! I wonder if I could do that on the bench and back plot against the know EJ MAF data, Hmmm... how much does my shop vac flow.



Good, one less factor to scale.



Yes, but I was assuming I'd make an adjustment to the static fuel pressure to get better idle and cruise, which would translate to leaner on the high end.



Good point, 1:1 RR regs only keep the pressure differential even.

Gary

Bad news… I checked the signal voltage on the EA82T last night and it doesn’t use the same voltage range as the EJ22. Its 3V at idle and 6V during full boost/throttle at redline (2ed gear). So the EA ECU isn’t compatible with the EJ MAF.

Oh well, it was a fun idea. I guess you could try and resolve the crank/cam angle input and use the EJ ECU along with the MAF and injectors.

Gary

6 volts? That's weird... Were you measuring between the ground and signal pins on the sensor itself?

yeah, from signal to ground on the sensor, the manual say the "key on, engine not running" voltage should be between 1 to 2 volts and the voltage on the power side should be 10V or more.

Gary

Weird. Very few analog automotive sensors have ranges beyond 5 volts.

The "10V or more" really means you should see full battery voltage at the power supply.

Any new developments with this AllTalk?

If I'm reading this right, the sensor voltage range is between 1-2V. The power being supplied to the Maf sensor itself should be 10V minimum. Right!

The easiest thing I found was to just get the 1999 MAF element for the Impreza and adjust its range according to the size housing you put it in. I've done this successfully on Mazdas replacing their crappy VAF meter.

-Michael