SGII Not quite right

[thebrad]

The estimations aren't quite right on my Honda Civic, upon installation of the device they were far too generous, then it sensed a disconnection so I had an asterisk next all stats for the remainder of the tank. Upon refill it just underestimated everything for the next two tanks. I just reset it on my most recent fill up and it is back to grossly overestimating, but I hope upon refill it will attempt to recalibrate and become more accurate.

In regards to figures, over estimation: I hit 80 mpg current today.
Under estimation: I was lucky to hit 30mpg.
For actual fill numbers I average around 55 mpg, check profile for tank specifics.

[dremd]

If it freezes up, do a reset, refuel, do not adjust fill.

your issue is likely due to SGII not registering a significant portion of the tank.

[cfg83]

trebuchet03 -

Originally Posted by trebuchet03

...

But I'm a gasser... and here's the twist! I have NO CLUE how load is calculated... It's a MKIV Jetta 2.0 - but load isn't what I expect. If I keep constant throttle while accelerating, load goes up. If I floor it at low speeds, load doesn't change much, and goes up with speed.... Just goofy... Not exactly related, but just throwing it out there that calculations seem to vary from car computer to car computer....

IMO this makes sense because OBDII is a specification that VW implemented. Each manufacturer has to meet the specification, but their implementation could lead to variations in output. In a VW service department, the custom VW software that interacts with their implementation could account for the variation, so probably no biggee. When the SG interacts with it, it may be "exposing" the variation.

If I was a manufacturer, I would emphasize the accuracy of sensor output that effects emissions requirements. Other than MPH and maybe throttle position, I don't know what that is because I'm not in the smog test business.

CarloSW2

[diesel_john]

dremd
I noticed on the scanguage.com site that now the TDI's are not all covered by scanguage. I wonder of VW figured out a way to fullfill the OBDII obligation, while at the same time forcing people to go to their dealers. The whole reason for OBDII in the first place.

[dremd]

Originally Posted by diesel_john

dremd
I noticed on the scanguage.com site that now the TDI's are not all covered by scanguage. I wonder of VW figured out a way to fullfill the OBDII obligation, while at the same time forcing people to go to their dealers. The whole reason for OBDII in the first place.

Hmmmm Very Interesting. I actually purchased one through a group buy on TDI club direct from Liner Logic. I wonder if they are working on the issue/ are concerned about the devices performance on the TDI . . . ..

[dremd]

I drove the 2000 TDI NB with SGII yesterday, it appears as though the SGII is accurate(based on fillups on a regular commute) and some quick quasi experimentation makes it appear as though it is off in the opposite direction than mine (that is lower indicated consumption on high flow rates and higher indicated consumption at low flow rates). Economy stays rather near 30mpg from 55mph to 82mph, we all know that is incorrect.


Differances in her car from mine.

1) Auto
2) 11 mm pump head (autos have 11mm 5 speeds had 10 mm)
3) lower mileage she is around 120,000 miles I'm around 175,000
4) More worn engine, hers had the wrong oil for the majority of its life, cam is worn down, and I suspect the rest of the engine is the same
5) I THINK more Aero drag (bug Vs Golf), (also missing suspension dams, under tray,an added spoiler, and wind deflectors on windows)
6) Larger, non Low Rolling resistance tires
7) Timing belt done by stab and pray (unknown pump timing)


Any thoughts?
Encourages me that some cars read one way and others read the other . . . . .


This makes me finally realize why she keeps telling me that she might as well drive 80; because her milage is basically the same, her scangauge has been giving her false information!!

[monroe74]

Excuse me for jumping in here.

Scangauge is great, but it has some inherent limitations. SG is obviously relying on the OBD2 protocol. As far as I can tell, OBD2 simply does not report injector behavior (pulse width/duty cycle). Why not? Who knows. I guess the folks who designed OBD2 in the early 90s were not thinking about what a bunch of people are thinking about now: accurate, realtime reporting of fuel consumption. I think monitoring injector activity is probably the best way to do that, but OBD2 doesn't get us there.

Since the injector information is not available, SG calculates fuel flow indirectly, by relying on something else that OBD2 does report: air flow. I guess that method works well enough on most vehicles, but it seems to have problems (at least for some people) on two types of vehicles: diesels, and vehicles with lean burn (Honda VX, HX and Insight).

Lean burn is a challenge for SG because it's relying on the assumption that mixture is usually close to stoich. OBD2 does provide O2 sensor data, so I think theoretically SG could make corrections, using this data, but I think it doesn't do that. Anyway, that's still indirect and prone to error, compared with watching the injectors directly. Trouble is, OBD2 doesn't allow that.

I think something similar might be happening with diesels. I have a feeling that SG will have an accuracy problem with any vehicle that is not running mostly at stoich, for whatever reason.

A digital DMM with a dwell feature ($33) can monitor the injectors directly. That might be a good way to go if you want to watch instantaneous fuel consumption in a more direct manner.

Anyway, maybe this answers some of the questions that were raised. Someone please speak up and correct me if I'm missing something.

[monroe74]

Another problem with assuming stoich is that even 'normal' cars don't run stoich when cold. My hunch is that this is another factor that introduces errors, for all cars, into the data SG displays.

Another problem with calculating consumption indirectly (instead of looking directly at the injectors) has to do with DFCO (deceleration fuel cutoff). I think SG only knows about this indirectly, because the user enters a constant representing the known DFCO threshold in rpm. This probably works well in most cases, but there are things that can go wrong, so it's not as good as watching the injectors directly.

A good discussion about mixture issues on diesels is here:

http://www.eng-tips.com/viewthread.c...=130502&page=1

[cfg83]

monroe74 -

Thanks for this analysis. I have had similar misgivings, but you describe it a *lot* better than me.

This is one reason why I am skeptical of what the SG reports when someone applies the "IAT Resistor Mod" to fake HAI temperatures to the ECU/PCM. I think the SG relies on the IAT temp, but the ECU/PCM mostly ignores the HAI during closed-loop operation. Garbage in, garbage out.

CarloSW2

[monroe74]

cfg: "you describe it a *lot* better than me"

Thanks for the kind words.

"IAT Resistor Mod"

Interesting, I hadn't heard about that before. Yes, another good example of how the SG can get fooled because it's not watching the injectors directly.

It's probably worth mentioning that the SuperMID and the Opengauge take the approach of watching the injectors directly. In other words, they don't rely on OBD2 the way SG does.

I don't intend to be slamming SG, because it's a great tool, and very popular. It's just a good idea to understand its strengths and limitations.