I suppose, the two options for the toggling would be

(1) The scangauge doesn't sample the injectors quickly enough, so, while the
actual inject pulse moves between e.g. 1 to 1.2 to 1.4 to 1.6 to 1.8 to 2.0 units, the scangauge can only get an 'accuracy' of 1, or 2
(2) The ECU is actually at the lower limits of accuracy, meaning that it can only make the injector pulse move in steps of '1' unit. If this is the case, driving with the MPG showing the higher value will actually save that much petrol!.

If people have been hypermiling and getting within 1-2% of the figures, it seems that (2) may be the explanation!.

I've wondered if it was this too. However, when the display toggles between the two different values (e.g. in top gear @ steady speed 70 km/h) I can't perceive any change in engine note / accel / deceleration. Not saying it makes it impossible that it could be the case. Just saying.

If it was (2), then, thinking about it further... If an engine is on low load, then it can actually run lean, without detonating, and will only begin running unevenly once it becomes very lean (maybe > 16:1), and, modified turbo cars on full boost can go to 8:1 (and run fine)... So, if there were large differences in the air-fuel ratio, then you still might not notice.

The Scanguage gets its data direct from the ECU, so, it might be that (2) is indeed the case - and that, keeping the MPG figures on the better of the two options might be very good for the economy.

One other possibility is that the ScanGauge can only calculate the instant MPG from the data from the ECU and maybe that is not providing the information to the SG to come up with anything more accurate that what we see. As far as the injector pulses being measured by the SG that may not be the case. Imagine if the injector pulses were tracked and counted but only the gas burned was saved in 0.1 gallon increments. Then in a tank with 50 stops and startups might loose the fractional parts of the tenths of a gallon and that can add up to quite a bit of gas.

As far as the readings you are getting that don't seem smooth - it is caused by the resolution of the numbers it calculates with. I also have seen the numbers not change as I moved the throttle and I don't know if the ECU is not measuring the movement of the throttle butterfly or the SG is not seeing a big enough change to increase or decrease to the next 0.1GPH digit.

Now, to throw another possibility into the mix. Another of the ways the SG calculates mileage is by monitoring the MAP sensor. At one time, I had spliced a potentiometer into the MAP circuit and was getting some fantastically high instant mpg readings; on the order of 70-90 mpg at 50-55 mph cruise. I was then told by Fran at hydrogen-boost.com about the SG/MAP sensor link. Something else to consider.

Not in my case - I don't have a MAP sensor - I did in the Geo but not in the xB - they have a mass air flow sensor instead.

While the Scangage is incredibly useful as a monitoring and feedback device, it overstates my F350's mileage by about 25%. I get 19-20 highway mpg in my diesel at fill up, but the Scangage trip meter says 25 mpg or so.

BTW, I spoke with a servive rep at Granatelli motorports, and they are going to send me a replacement module to see if my the truck stops misfiring and gets the promised mileage.

I don't know how a scangage works, but this quote seems accurate.
I've watched the injector pulsewidth and O2 volts on my dataloggers under various conditions. During normal cruise I can see the O2 volts cycle above+below stoich like they should. This is the ecu trying to keep the A/F ratio at 14.7:1. The ecu can add/subtract fuel accurately enough to consistently cycle the O2 volts around .15 to .75v. At the same time the injector pulsewidth shown on the logger remained the same at around 1.79ms for example. The ecu uses a two byte variable to control fuel, while the logger only sees the hi byte. I can also see the airflow reading from the MAF vary slightly at the same time, so I know the ecu is tracking the airflow and adjusting fuel to match it.

My car is OBD1. I can't imagine OBDII having LESS accuracy.

DRW -

I think the ScanGauge is great because it takes advantage of the OBDx specifications, but all our engines and ECUs are different, so the SG can't be considered a perfect measure across all car platforms. It is a jack-of-all-trades because of the specification, but it is up to the car manufacturers to implement the specifications. The SG is on the outside of the black box of the ECUs, so it is vulnerable to the tolerances/deviations in the manufacturer's implementation of the OBDx protocols.

And now there will soon be a CAN protocol, isn't that right?

That's my two bytes,

CarloSW2

Whatever happened to the "Space Kom" MPG meter? I had one of these jewels on my earlier test vehicle(s), but I let it go with my last sale! Bad mood, I guess....Very bad decision on my part!
This MPG computer used a speed transducer...in series with the speedo cable, and a (gas)flow transducer...light activated, no moving parts!...spliced into the pre(I think)-fuel pump line. These two voltages generated were fed to a calc-circuit in the meter housing...to give a meter deflection. No averaging function...just simple MPG, 5 or 6 times a second!
I didn't agree with the calibration procedure...but it gave beautiful deltas if you were changing...hills, tires, loads, traffic, etc.! They're out-of-business, now. The co. that bought all the rights to this product / concept are also out-of-business! Too bad! (sniff!)