Forgive me, I'm trying to thread a needle. The important part of the post is the principles, but it will sound like a sales pitch and that is not my intent. There are lots of ways to get the job done, I just happen to be familiar with our stuff.
First, it is important to understand why OBD-II MPG is inaccurate. I cover a bit of it here:
http://ecomodder.com/forum/156981-post21.html
The fundemental problem is that the unit doesn't typically know what the lambda, or AFR, of combustion is. This is 14.7:1 (really lambda 1.0, which is not quite the same thing), a lot, but not always, and it can depend on even the route being driven.
So, when you are calibrating such a unit, you are not just adjusting for sensors, but also for driving style and typical routes and terrain. If you are trying different driving styles, then your calibration becomes moot.
So, from a strict instrumentation standpoint, measuring injector duty cycle is much more accurate for comparative purposes. It also has to be calibrated, because of the mechanical properties of the injector, but even without calibration, comparisons of readings have merit.
Second, instantaneous readings are not a very good way to make meaningful comparisons, at least not when gauging small changes.
I'm not saying that such devices aren't useful, but the instant reading and the actual results are two different things. For example, and instant reading will reward any decrease in TP, but if you are going up an incline, this is not the most fuel efficient way to operate.
I think that when you are trying to make comparisons, logging and post operation analysis is the way to go. That's how we do it in performance operations. You are really interested in the total result, not instant by instant readings. Also, logging lets you focus on executing the techniques.
If I were going to try to collect data for quantitive comparisons, I'd start by logging OBD-II data (like I did in the post above). This is nice because it lets me also review the consistancy of execution of the technique as well as measure the fuel results.
If my need for fuel consumption accuracy was fairly precise, I'd add a wideband O2 setup. In the case above, I'd put the wideband in the pre-cat bung for the regular O2 sensor, then have the wideband controller simulate narrow band output for the ECU. This would save me installing a new bung. Then I'd chain that through the OBD-II module so that the wideband data is logged in sync with OBD-II data.
Wideband + MAF is pretty accurate for MPG. Plus, it let's me have a glimpse into how much economy is really about the technique, and how much is ECU specific behavior.
If that precision is not enough (limited by ECU response rate), I'd wire up injector duty cycle as well to a little analog/clock input module and chain it in as well.
For 'real time' monitoring, I might wire up a gauge (there is a serial output as well as USB/Wi-Fi), but you get the idea...
Now, again, none of this meant to sell gear. I bet it would be easy to apply the same principles to something like the MPGGuino. My points are ultimately just:
1. Understand that OBD-II is generally only air side monitoring, so different technique throws off precision
2. Meaningful evaluation is a lot easier (and I think a lot more accurate) if you log the data and analyze it after the fact.
Regards,
-jjf