MPGuino for Audi

[rmccomiskie]

I've been lurking here for a while. Now I have an iDuino and I'm ready to start implementing in my 97 Audi A6 Quattro. Fortunately, Audi provides several connection points for VSS. It's a signal that's output to the other systems by the speedometer. Even better, Audi provides a fuel consumption signal that's intended for the optional trip computer. I'm hoping this signal will avoid the need to measure injector pulses directly.

Here's what the VSS signal looks like at 60 mph:



I've been told the signal from the sender is 8 pulses per tire revolution. Given the stock tire size, I expected about 11 pulses per 100 mS. The oscilloscope proved it.

Now the fuel consumption signal at idle (700 rpm):



And at 60 mph steady:



And accelerating hard 55 to 65 mph:



The only expanation of the fuel consumption signal that I've found is this:

"The fuel consumption signal is a data message which is conditioned by the engine control unit. The sum total of the high levels during a defined period of time corresponds to the injected fuel quantity."

So, it sounds like I just have to count the pulses that occur within, let's say, 100 mS. The problem is, what volume of fuel does that number relate to? I may have to just estimate a ballpark volume and then adjust with every fillup.

I know that I get about 22 mpg at 75 mph. If I use the scope to capture the number of pulses at 75 mph, I should be able to get close to the fuel volume those pulses represent.

Does anyone have a better way to do this?

[dcb]

looks like about 14040 vsspulses per mile?

The "fuel consumption signal" looks basically like an injector signal.

[rmccomiskie]

Now I'm confused. The stock size tire turns 811 times per mile. The sensor is supposed to output 8 pulses per tire rev. 811*8=6488 pulses per mile.

Crosschecking that number, if the scope shows 11 pulses in 100 mS at 60mph, that's 110 pulses per second * 60 sec = 6600 pulses per minute (1 minute = 1 mile at 60 mph). So 1 mile = 6600 pulses. Pretty close.

What am I missing?

[dcb]

I guessed 117 pulses per second (7 pulses in 60ms) per the graph, and the guino will multiply that by 2 (counts rising and falling edge). You can always calibrate it using the cars odometer so close enough for getting started.

[secmil]

I'm currently working on getting mine to work on my '98 A4, and was wondering where you were able to tap into the fuel consumption signal...

I was looking through all of the wiring diagrams in the bentley manual, but I guess I missed that whole section.

[rmccomiskie]

Originally Posted by secmil

I'm currently working on getting mine to work on my '98 A4, and was wondering where you were able to tap into the fuel consumption signal...

I was looking through all of the wiring diagrams in the bentley manual, but I guess I missed that whole section.

Ahh...an Audi owner to commiserate with!

Bentley wasn't much help for me. I found a thread on AudiWorld on retrofitting a trip computer. The fuel consumption signal came from a connector behind the passenger kick panel. I checked my car and it has the proper color wire in that spot. I hooked up the scope and it looks like the right signal.

It's a Blue/Black wire in the red connector here:



I removed the connector bracket, pulled the connectors and pushed a solid wire into the top of the connector to bring the signal out to where I could attach an alligator clip:



Remember, this is a 97 Audi A6. I don't know if the A4 will have this signal in exactly the same place. Audi is, however, pretty consistent. Since the A4 has a trip computer as an option, there must be a consumption signal somewhere. I'll bet that you will find a Blue/Black wire in a red connector in roughly the same place.

For the VSS, I found a schematic that showed the VSS going to the cruise control module. I pulled the module connector and found the signal there on a White/Blue wire. I found the same colored wire behind the kick panel and a continuity check showed them to be connected. So, I pushed a solid wire into the kick panel connector to bring that signal out.



Let me know if have any questions. Good luck.

Bob.

[Kaneda]

good to see others trying to remedy the audi mpg curse ... '99 A4 2.8 here

[rmccomiskie]

Can't cure the curse that Audi wrought. Best to know your enemy well though! Jump right in and join the fray.

[rmccomiskie]

The only expanation of the fuel consumption signal that I've found is this:

"The fuel consumption signal is a data message which is conditioned by the engine control unit. The sum total of the high levels during a defined period of time corresponds to the injected fuel quantity."

So, it sounds like I just have to count the pulses that occur within, let's say, 100 mS. The problem is, what volume of fuel does that number relate to? I may have to just estimate a ballpark volume and then adjust with every fillup.

After logging more fuel consumption signals, I'm coming to the conclusion that the 97 A6 signal is one pulse for each injector firing and the pulse width is the length of time that the injector is open. So, the signal includes the injection time for all 6 injectors not just one. Summing the pulse width time is the total injector open time.

Now I need to monitor the pulse width at 75 mph where I get about 22 mpg. Then I can calculate the approximate injection time per gallon. Soon.

[dcb]

If it actually conditions the signal, then you might need to adjust your "Injector DelayuS" from its default of 500. Or just tap an injector lead

[secmil]

Wow, many thanks for those pics for fuel consumption signal! I'll be tearing my car apart again this week I guess...

Assuming I can actually use the signal, and all.

[rmccomiskie]

I did some logging on the way home last night. At 75 mph, the engine is turning 3400 rpm and burning 22 mpg (my guesstimate). Here's the injector pulse width at that speed:



This one happens to be 550 micro seconds width. I logged a number of these and the average is around 562 uS.

For my 2.8 liter 12 valve V6, I calculate 100,890,240 uS per gallon. Once I get this installed and running we'll see how accurate that is.

BTW: I'm using a sound card oscilloscope program on my laptop to capture the signals. I have an audio cable connected to the car with a couple of resistors to drop the signal voltage below 1V. It's plugged into my laptop microphone input. Works just fine at the frequencies we're dealing with.

[rmccomiskie]

Originally Posted by dcb

If it actually conditions the signal, then you might need to adjust your "Injector DelayuS" from its default of 500. Or just tap an injector lead

Question: Since the Audi fuel consumption signal appears to be one pulse for every injector, do I have to change any settings. In my case, it's as though I am connected to all 6 injectors not just one.

My assumption is that the normal MPGuino hookup is one injector and the pulse width of that injector is being scaled up to represent the rest of the injectors. Right?

[secmil]

I would just assume that our setting for microsec/gallon (as it's called in the workspace) would just be 1/6th as much since there are six injectors reporting instead of just one.

I wish I wasn't having so many hardware issues so I could play with it and see what's what...

[dcb]

Originally Posted by rmccomiskie

My assumption is that the normal MPGuino hookup is one injector and the pulse width of that injector is being scaled up to represent the rest of the injectors. Right?

Basically.

Honestly though, I have no idea if that signal you have found will be accurate. We account for mechanical delay when measuring the injector pulse. I don't know if what you are looking at has accounted for that or not. You will have to carefully measure that pulse alongside the injector pulse and do some figguring if you want to be reasonably certain of the accuracy. I don't know how conditioned that signal has been, you are breaking new ground.

[rmccomiskie]

Originally Posted by dcb

...I have no idea if that signal you have found will be accurate. We account for mechanical delay when measuring the injector pulse. I don't know if what you are looking at has accounted for that or not. You will have to carefully measure that pulse alongside the injector pulse and do some figguring if you want to be reasonably certain of the accuracy. I don't know how conditioned that signal has been, you are breaking new ground.

Good point.

I'll check the pulse width at idle. Since the 'Injector DelayuS' value is 500uS, I'd expect to see a pulse width of 500uS plus whatever injection time is needed to sustain idle. If the pulse width is less than 500uS (assuming that's the correct delay) then it's likely that the fuel consumption signal subtracts settling time from the injector pulse and directly represents the actual injection time.

If I can easily get at an injector wire I can compare that signal as a crosscheck.

[rmccomiskie]

I logged the fuel consumption signal pulse width at different speeds and compared them:



The typical pulse widths are:

Idle = 450 uS
50 mph = 620 uS
60 mph = 740 uS
75 mph = 870 uS

Given that the idle pulse width is relatively large compared to the rest, I suspect that the settling time is contained in the signal. In fact, I'd bet that the fuel consumption signal is just a composite of all 6 injector signals with no other conditioning.

Also, since the Idle PW is less than 500 uS, the delay time should be something less. Maybe 400 uS. I have no idea. I'll have to play around with the settings against actual fuel burn to get an empirical value...in my 'spare' time.

[rmccomiskie]

I found this interesting bit of information at http://www.injectorcleaning.co.uk/flow.htm

Bosch Part Number: 0280150552 (crossref to 1992-98 Audi 12V V6)
156 cc/min @ 3 bar (44.1 psi)
Resistance: 15.9 ohms

Audi typically specs their fuel rate at 43.5 psi so 3 bar is close. I'd expect the installed flow rate to be about 150 cc/min.

[rmccomiskie]

Bump.

[dcb]

Are you sure you want to go through all this to avoid tapping into an injector wire? There are a few unknowns. What is the pulse width of the injector at idle?