OBD-II vs FI+VSS - Request for Comment

[peterj]

I have just recently begun investigating the specifics of hypermiling and the use of gauges to help develop good driving habits. I would like to summarize what I have learned (or at least what I think I have learned) in the hopes that others can comment and/or correct me where I have mis-learned.

It seems as if reading the fuel injector (FI) cycle and vehicle speed (VSS) is the only way to accurately determine instantaneous fuel usage. This method will work on just about all gasoline-based fuel injected vehicles, is not overly complicated, expensive, or invasive, but *does* require some gear-head knowledge for installation. There are some inexpensive homebrew solutions, but all commercial equipment is rather expensive.

Reading information from the OBD-II port is limited to post-1996 cars and does not (generally) provide specific information for fuel consumption. The amount of information available through OBD-II varies greatly, and the only information available on all vehicles relating to fuel consumption is mass airflow (MAF). Based on a 14:1 air:fuel ratio, MAF can be used to obtain a rough estimate of fuel consumption. The level of accuracy will vary depending on vehicle type, and will always require some degree of calibration. On the upside, commercial equipment is relatively inexpensive and installation is as simple as locating the OBD port and plugging in a cable.

It seems as if some implementations of OBD provide FI information, so in theory, cars with this data should be able to provide much more accurate information.

The most popular OBD device is the ScanGuage II, but I have been unable to determine which OBD parameters it uses to calculate instantaneous mileage. It seems clear that it must use MAF, but it could bias that based on other values. It is also not clear if it makes use of optional parameters such as FI if they are available. (Any clarification here would be helpful.)

Clearly the FI+VSS method is superior for tracking exact fuel consumption and instantaneous mileage, and will provide the best feedback on driving style. What I am unsure about is how close OBD data will come to the FI+VSS method. Are they close enough such that most drivers in most situations will be able to obtain most of the information they need with OBD in order to maximize mileage?

Many thanks for any addition and/or clarification on these issues.

[cfg83]

peterj -

Sounds pretty accurate to me. I think the SG is relying on the "lowest common denominator" set of information that it can rely on across a variety of car platforms.

Originally Posted by peterj

...

The most popular OBD device is the ScanGuage II, but I have been unable to determine which OBD parameters it uses to calculate instantaneous mileage. It seems clear that it must use MAF, but it could bias that based on other values. It is also not clear if it makes use of optional parameters such as FI if they are available. (Any clarification here would be helpful.)

...

Many thanks for any addition and/or clarification on these issues.

Your post just made me realize that an "OBDII Output Simulator" would be a good way to reverse engineer the SG algorithms.

CarloSW2

[dcb]

OBD, while being as plug and play as you can get from a user perspective, is also a moving target from a development perspective. The standards are still evolving and the hodgepodge of protocols is going to grow or morph into a new (and incompatable with everything) protocol, which can also vary by region. Whereas injectors signals are a pretty stable thing by comparison.

Anyone who can install (or find someone to install) a tachometer in their car can install an injector/vss device. Also, there may be injector leads in the cabin on some cars, I dunno.

MAF is not universally available. More like 2000 something and up, not 96 and up. MPG computation is TRIVIAL with a MAF. Like MPG = (VSS * 710.7) / MAF according to this vpw pic mpg gauge.

The obd mpg question comes up frequently, here's my notes on it without maf, what it might look like (I'm probably being too exacting, some factors won't make a diff):
http://forum.ecomodder.com/showthrea...1.html?p=18665 but only needed for 96 to 2000 something vehicles...


Still there is a lot of obd mpg gauge interest in the public domain, preferably without the expense of an elm chip. Someone should teach an arduino to talk to any sort of obd protocol, but that is not a small order, so it is no suprise that no-one has done that out of the goodness of their heart yet.

[peterj]

dcb: Many thanks for the MAF clarification. I read somewhere that MAF was a part of the baseline requirements of OBD-II, but perhaps it has only been a requirement since year 2000.

Following your link, I eventually wound up at Bruce Lightners's post on the mp3car forums, where he explains how to calculate MPG on cars without MAF. He also indicates that MAF is much more accurate than I thought, and MPG calculations from MAF do not require calibration.

Although I do not have a ScanGuage II, I have read the owners manual as well as many of the posts about it, and now I think I've got a pretty good idea of what's going on inside.

I'm also aware of the mess of OBD, but what I'm trying to do is to try to get some kind of handle on just how useful the data is once you actually have it.

There is no question that for the gearhead, FI+VSS is the way to go. However, the average Joe is looking at several hours of shop labor (at $50 or more an hour) to get a working system installed. An OBD device can be easily shared among family and friends.

[JohnnyGrey]

MAF may be easier to calculate, but it's no more accurate than MAP+RPM. The SG is still unaware of fuel cutoff, cold start / open loop enrichment or decel lean-out. Also, on my car (which has no MAF sensor), it tends to overestimate the effect that intake air temperature has on fuel volume. When the weather changes rapidly, I notice it's off by more when I go fill up. Until the OBD standard has an injector pulse width parameter that is widely supported, FI + VSS will always have the advantage in accuracy.

[dcb]

Peterj, you mean this?

...
As for other ways of doing this, especially if you don't have a MAF sensor, by knowing the displacement of the engine, and after a simple "calibration" using fuel tank "fill-up" data to find the only unknown, namely the "volumetric efficiency" (VE) of the engine, MAF can be calculated from RPM, MAP and IAT. With VE, one can use the following formulas to calculate a synthetic "mass air-flow" (MAF) in grams per second, all without a MAF sensor, using the "Ideal Gas Law", as follows:

IMAP = RPM * MAP / IAT
MAF = (IMAP/120)*(VE/100)*(ED)*(MM)/(R)

where manifold absolute pressure (MAP) is in kPa, intake air temp (IAT) is in degrees Kelvin, R is 8.314 J/°K/mole and the average molecular mass of air (MM) is 28.97 g/mole. Note that, in the above formula, the volumetric efficiency of the (4-cycle!) engine is measured in percent and the engine displacement (ED) is in liters.

You might be on to something Johnny, not that you need me to tell you that But I wonder if the short term and long term fuel trim can be used more accurately predict the fuel consumption in changing conditions.

[JohnnyGrey]

You might be on to something Johnny, not that you need me to tell you that But I wonder if the short term and long term fuel trim can be used more accurately predict the fuel consumption in changing conditions.

Good luck implenting that on a widespread basis. Beyond the basic fields that the SG can extract, pretty much everything else is proprietary. You might be able to do it for the Toyota guys for example, or maybe even just the Corolla guys, but the rest of us would be SOL. I can imagine what a pain it was for the Scangauge programmers just to get all the different protocols to work right, let alone making sense of tables (if they're even available through the port).

[tasdrouille]

Originally Posted by dcb

You might be on to something Johnny, not that you need me to tell you that But I wonder if the short term and long term fuel trim can be used more accurately predict the fuel consumption in changing conditions.

On almost any OBD-II vehicle, SG can get the short term and long term fuel trims.

[peterj]

Going back and re-reading my initial post, I realized I wasn't too clear about my final question. I believe we're all in agreement about the limitations of OBD-II, and that from a gear-head perspective, FI+VSS is in almost all ways a superior approach.

What I am trying to ascertain is that if you are only concerned with the relative accuracy of readings, as opposed to absolute accuracy, is OBD really that inferior to FI+VSS?

Clearly, if you are interested in accurately calculating the number of miles remaining from the fuel remaining in the tank, FI+VSS is the way to go.

However, if you are only looking to display the *relative* efficiency over the previous 5 seconds, 30 seconds and 60 seconds -- the type of information required to develop proper pedal technique -- is OBD really that much inferior to FI+VSS? My hypothesis is that this could be the case. I am curious to to know if anyone has the data (i.e. access to both systems across a spread of vehicles) to prove or refute this hypothesis.

Obviously, there is a very good chance that I may need to actually perform the experiment myself to get an answer to this question. ;-)

[dcb]

Well, you don't have to be *that* much of a gearhead to splice and route a few wires! And it shouldn't take "hours" if there are decent instructions for the vss thingie.

But for the user who is too "lazy" (for lack of a better term) to hookup a vss/injector monitor, they probably aren't going to want to use the exactness of such a setup. Average/relative readings are far better than none, but more exact readings can let you fine tune, but you have to not be lazy to fine tune in the first place. So consider your audience.