Archives

After my second tank of running my IAT resistor mod, I noticed that my SGII was a 1.7gals short on fillup. The scangauge went to 11.7% over and the next tank sucked. It took some under-hood tinkering for me to figure this out.

I made a crude lowpass filter connected to my #1 injector and measured the DC voltage with my DMM to give relative fuel consumption readings. The average voltage with the IAT resistor was 0.32v. I removed the resistor and plugged the harness back into the sensor. The voltage jumped up to 0.50v (running rich) and over the next three seconds went back down to 0.32v only now, the SGII reported 0.4gph instead of the 0.3gph it claimed when the resistor was in place. The SG seems to watch what the ECU would go by if it were in open loop mode. The IAT mod was not leaning out my cruise, only the wide open throttle condition where I actually wanted the power (the loss in power was quite noticeable). In short, the IAT mod did not fool the ECU, only the SG.

The ScanGauge is very good at measuring how much AIR goes into the engine, but not so much with fuel. The SG doesn't seem to care about injector pulse widths which is probably why it doesn't account for lean-burn in Hondas or injector cutoff while coasting. I think the SG assumes a fixed air to fuel ratio and reports fuel usage based on how much air the engine is sucking in. This brings me to my last project...

My O2 sensor intercepter. The idea behind this mod was a continuation of a failed mod, a simple battery powered voltage offset I could apply to the O2 sensor in series to make the ECU cut back on fuel. I added 0.3v to the line and the ECU threw a code "O2 sensor voltage high" and resorted to open loop mode. I scrapped this idea shortly after.

This new box is different though. It reports 0v-1.0v to the ECU depending on whether or not the input from the O2 sensor is above or below an adjustable threshold. It uses a 7806 voltage regulator, three 2N2222 and one 2N2907 transistors and a bunch of resistors. It has two LEDs, a green one for power and a red one for O2 sensor activity.

I installed the box and set the threshold to 0.20v, the lowest I could manage without hesitation. I drove my sister to school on a highway trip to give the ECU a chance to learn and the SG reported lower than usual numbers, 34 where I had gotten 42 in the past. At first I thought my box was destroying my FE until I got to the gas station. This time I had burned 12% less fuel than the SG thought I did, probably because it thought I was running a A/F ratio of 14.7 to 1. To my best estimation, I had managed an A/F ratio on that trip of 16.4:1.

The best explaination I can think of is that before, the car would cycle between normal and rich to prevent NOX formation. The guys on jbody.org have mentioned that our cars tend to err on the rich side. Rich is probably a more eco-friendly condition since the cat eats up most of that stuff anyway. So instead of cycling from normal to rich, the car now cycles between lean and normal. Extra air in the engine isn't really giving me any more power, but it's not costing me any more fuel either. Extra fuel in the engine doesn't give me more power, but it does cost me extra fuel.

Later tonight I got a chance to confirm my results. I went on a 40 mile highway trip with my recently calibrated SG and averaged 43.8 mpg, beating my 55mph steady state record of 42.1 mpg. I expected better, but a 4% improvement isn't too bad at all.

Hi Specter:

___I moved your thread to the Technical forum from the FE forum as I thought it would best fit here. I was even tempted to post in the SG-I/II forum as well but it really did not fit there either …

___I have to agree that SG-I/II does not have the ability to read the injector pulses from the OBD-II stream for a SI-ICE right now but looks at air flow for the FE calc’s. We see this whenever our hybrid and non-hybrid automobiles go into Fuel cut at speed yet mpg does not go to 9999.

___The SG-II – Diesel a/b setup might be looking at injector pulses but I am not sure just yet. With the lean A/F ratios however, I can bet that would be a complete mess for us let alone thinking we are running a 1.9L TDI’s injector size spread and if that data is even in the stream?

___Good Luck

___Wayne

Aww crap, now all of 3 people are going to read my thread. :(

Something tells me the diesel fuel injection is just derived from the TPS and/or MAF.

Tracking fuel usage from injector pulse widths alone is tough since every car has different flow rates and fuel pressures, but the SG could easily compare arbitrary values to the rest of the numbers it uses to determine fuel consumption and once trained (and adjusted after a few tanks), derive fuel usage from these pulse widths alone. Seems like lazy programming to me. If they ever fix that issue in a SGIII, I'll buy it and get rid of my current one.

Heh. I tried doing this on my old Trooper, years and years ago.
My friends all believe it led to me burning the exhaust valves,
although I have it on good authority that the heads in those
engines just generically sucked and needed valve jobs on a
fairly regular basis anyways. But it was fun to play with, and
I didn't make it throw a "too lean" code too often. Unfortunately
it didn't seem to do *squat* for MPG.
.
The problem with a typical O2 or even wideband sensor is that
the algorithm is always wandering back and forth *across* stoich,
not parked directly on it. That's what those slow up-and-down
waveforms are all about. So in way it isn't the DC level you
get from the sensor so much as it's about *when* it heads high
or low. So I think that to really fool the system in a way that
makes sense to it, you'd have to delay the rise rather than offset
the whole thing by some applied bias. The DC bias may have
some amount of that effect simply given how the waveform looks,
but where the peaks are is then all wacked. And of course you're
creating your NoXmonster in the process.
.
Still, it's interesting. Wanna post your circuit somewhere?
.
As far as measuring injector duty cycle ... that's sort of what
the "sweet spot meter" does, but the limits I'm playing with
on that are based on scope timing measurements. Once I figure
out the right duty cycle boundaries and plunk the right components
into the proto-board, I can take the scope out of the car and
run around on just the meter for a while. Until I come up with
another half-ass theory and have to get out the scope again.
.
In related news, I have a Vetronix Mastertech MTS3100 to borrow
for a month -- yup, the "genuine article" Toyota Hand-Held Tester,
supposedly able to grok all the seekrit Prius stuff. From the
very same pile of professional scanners, in fact. Once I make
sure its comms aren't broken, I intend to do a lot of testing;
anything y'all want me to try and find out, I'll work on.
.
_H*

We played with a number of profesional scan guages at the Up Your Volts training session.
Only the factory tools could read all the data. the injector pulse with may be available with a different tool.
Because the pulse width and frequency change as RPM change,and the pulse is usually accompanied by a large fixed width inductive voltage spike, a simple low pass filter may not be as meaningful for efficiency determinations as measuring the real pulse width with a scope or a processor based measurement.
Mike

So in way it isn't the DC level you
get from the sensor so much as it's about *when* it heads high
or low.

My first project added an offset. What the box currently installed in my car does is send 1v to the ECU whenever the O2 sensor reads higher than X, where X is some voltage from 0v to 1v. When the voltages is lower than that threshold it sends 0v. The ECU will never throw a code because it always gets its data from the circuit. I saw this "EFIE" available online that uses an offset so I based my first circuit on that and it didn't work.

The 4% or so improvement I got I think comes from not running lean, but not running too rich since extra air is free but extra fuel is not.