Memory dump of ECU

[TomsSVX]

dood... Ur in puerto rico??? No wonder I cannot get a hold of u Give me a buzz after you get back and settled

Tom

[Dessertrunner]

I don't know if this is the right place to post this so tell me if its not. Tom asked about Knock Sensor and as a result I looked at the table then following typing the reply I had a look at the other tables and realized what Subaru have done. If you look at the overall data for timing and Knock sensor and to a lessor degree the Air/Fuel they are all constructed in a way to start pulling the potential power of the engine. Seeing as we were looking at Toms I will talk about the the USA ECU. I may get some of this totally wrong but it sure should start debut.
Raw timing data if I am correct in assuming that the USA cars are rev limited at 6,000 you can see it in the raw timing data map at the line 6400 load 44 to 64 the numbers are total different to the numbers at 6,000 reason is somebody wrote new numbers when the decision to rev limit the engine was made. Compare it to the JDM timing map and you will see what I mean.
Knock sensor the only time they allow you to get maxium power is between 2800 to 4400 then they pull it back because they don't want the engine generating full power at the time the rev limiter comes in. When you think this throught it appears that the higher the revs the higher the allowiable knock reading but Subaru have backed it back.
Air fuel my only comment would be that If the converted numbers are Air/Fuel then I would search the net to find out what the best air/fuel number is to get maxium power and load that into the table at all the high revs & Load areas.
So there it is for what its worth hope to get my tables this week and see what we got.
Tony

[b3lha]

Following last week's discussion about engine load, I decided to try and find out how the SVX ecu derives it. I haven't figured it out yet but I've found the first piece of the puzzle.

As we know, the MAF uses a voltage between 0 and 5v to tell the ECU how much air is being sucked into the engine. The more air, the higher the voltage.

In the ECU, the MAF signal wire is connected to an analogue to digital converter. The ECU program reads the voltage as a number between 0 and 1023. 0 being 0 volts, 1023 being 5 volts, and each step in between being about 0.00489 volts.

But the ECU doesn't want to know the MAF voltage, it wants to know the Airflow. It finds the airflow by looking up the MAF signal on a graph. The data for the graph is 64 sixteen-bit numbers starting from address 0x8080. This is what it looks like



For simplicity I have converted the units of the X-axis to MAF voltage. The Y-axis is a sixteen bit number (ie. between 0 and 65535), which will need some conversion before you can express it in litres per second.

I'm sure that I have read somewhere that the MAF voltage is supposed to be proportional to the airflow, but the graph seems to indicate that at low flow rates, it is not. I assume that the graph was calibrated experimentally using a real intake.

Looking at the top of the graph, you can see that the graph maxes out around 4.7 volts. As far as the ECU software is concerned, that is the max airflow possible, even if your MAF can send a higher voltage. I guess that anyone who wants to shovel extra air into their engine will need to change the MAF for one that can read a higher airflow, and then modify this graph to match the performance of their new MAF.

[svxistentialist]

Originally Posted by b3lha

I guess that anyone who wants to shovel extra air into their engine will need to change the MAF for one that can read a higher airflow, and then modify this graph to match the performance of their new MAF.

That's an interesting deduction, Phil.

The forced air people who discovered the problem of the MAF maxing out don't do this exactly. They use a Nissan 300 Turbo MAF that is calibrated to yield 0-5v for a much higher airflow.

So far as I know there is no modification to the graph used to compensate, like you are suggesting.

Of course Lan's chip might actually have modified that graph, I don't know. Does anyone?

From what I have seen I don't think that everyone who has force fed [SC or turbo] their engines has used Lan's chip, but I could be wrong.

Joe

[b3lha]

Originally Posted by svxistentialist

That's an interesting deduction, Phil.

The forced air people who discovered the problem of the MAF maxing out don't do this exactly. They use a Nissan 300 Turbo MAF that is calibrated to yield 0-5v for a much higher airflow.

So far as I know there is no modification to the graph used to compensate, like you are suggesting.

Of course Lan's chip might actually have modified that graph, I don't know. Does anyone?

From what I have seen I don't think that everyone who has force fed [SC or turbo] their engines has used Lan's chip, but I could be wrong.

Joe

I'm not saying that using a different MAF won't work. I'm saying that this graph is calibrated to the stock sensor. A different MAF will probably have a different voltage/airflow curve. The MAF readings seen by the ECU will be then sub-optimal in areas where the new curve deviates from the stock curve.

I can't say how that would translate into an engine performance difference, particularly since the fuel and timing maps would most likely be modified to compensate for the inaccuracy in the airflow measurement.

[Calum]

Yea, the Nissan ecu works the same way. When you change maf sensor you change the VQ table (i.e. look-up table for maf voltage to some unitless mass flow number). Whats handy is you can use the stock VQ table for whatever maf sensor your switching to, assuming you've got the rom for the ecu that originally worked with that maf. Alternatively you pay for a lab to test the maf and produce the same chart for you. VQ tables are pretty easy to come by for most common Nissan mafs (300z, etc). Also, for different cars that use the same maf sensor but have different intake piping you'll see slight differences in the VQ table. I assume this is because the VQ table has been custom measured for a given intake piping. These small changes really have no effect on the drivability of the car, and aren't worth worrying about.

The VQ table is unitless. There is probably some second variable that converts it to meaningful numbers. In the Nissan ecu this is the 'k value', and how it works is basically K x Q/rpm is your base injection time (where base injection time is the injection time needed for 14.7:1 for a given airflow and rpm). Q is simply the look-up value from the VQ table, and K is a ratio of the maximum mass airflow the sensor is capable of measuring and the size of your fuel injectors.

Back out the calculation the ecus is doing with the VQ table, dollars to donuts you'll find something similar to the k-value. These ecus seem to do just about everything else the same.

[svxistentialist]

Originally Posted by b3lha

I'm not saying that using a different MAF won't work. I'm saying that this graph is calibrated to the stock sensor. A different MAF will probably have a different voltage/airflow curve. The MAF readings seen by the ECU will be then sub-optimal in areas where the new curve deviates from the stock curve.

I can't say how that would translate into an engine performance difference, particularly since the fuel and timing maps would most likely be modified to compensate for the inaccuracy in the airflow measurement.

Yes, I'm not doubting it will work, just pointing out that I'm not aware of any of our maxed MAF people having adjusted the map in the software.

Just as an observation the fact that the graph is not linear at low airflow may well be a function of the physical measurement of the cooling effect on the element when fed air through a cylinder of a fixed size and volume, possibly even fixed cross-sectional area and volume.

Or put another way, using a different MAF the observed cooling effect may have the same graph shape [or as near as dammit, empirically ] and the said different MAF is merely calibrated to deliver the expected voltage over a higher range of airflow conditions.

You are probably correct that the accuracy at least will be sub-optimal at lower flow and voltage regions. Higher up it will be in the linear and accurate range of the graph/system.

Joe

[svxistentialist]

Originally Posted by Calum

Yea, the Nissan ecu works the same way. When you change maf sensor you change the VQ table (i.e. look-up table for maf voltage to some unitless mass flow number). Whats handy is you can use the stock VQ table for whatever maf sensor your switching to, assuming you've got the rom for the ecu that originally worked with that maf. Alternatively you pay for a lab to test the maf and produce the same chart for you. VQ tables are pretty easy to come by for most common Nissan mafs (300z, etc). Also, for different cars that use the same maf sensor but have different intake piping you'll see slight differences in the VQ table. I assume this is because the VQ table has been custom measured for a given intake piping. These small changes really have no effect on the drivability of the car, and aren't worth worrying about.

The VQ table is unitless. There is probably some second variable that converts it to meaningful numbers. In the Nissan ecu this is the 'k value', and how it works is basically K x Q/rpm is your base injection time (where base injection time is the injection time needed for 14.7:1 for a given airflow and rpm). Q is simply the look-up value from the VQ table, and K is a ratio of the maximum mass airflow the sensor is capable of measuring and the size of your fuel injectors.

Back out the calculation the ecus is doing with the VQ table, dollars to donuts you'll find something similar to the k-value. These ecus seem to do just about everything else the same.

Thanks Calum. That's good information.

Do you know what the term VQ means?

Since it is a measure of the volume of air cooling the element as it is ingested by the engine, could it mean Volume Quotient?

Or does it mean Voltage Quotient?

Or neither?

Joe

[Nomake Wan]

Oh wow, so the 300ZX values are available online? I wonder if the code would just drop right into our ECUs (via a ROM of course) or if they'd have to be modified to match our code...

This is getting interesting.

[b3lha]

I was looking at the code again today and it looks to me that if the MAF is not working (ie. Check Engine code 23), then the ECU will attempt to guess the airflow based on the RPM and the TPS signal. It uses the following map from location 8B30.

The X axis is RPM. I'm not 100% sure of the scaling factor. If it is the same as the other maps then it will be 50, meaning that the chart only goes up to 4800rpm.
The Y axis is the TPS signal. Again, not sure of the scaling factor. Possibly y*5/255 volts, meaning the chart only goes up to 16% throttle.
The Z multiplied by 128 equates to the airflow from my previous post.
The UK Model uses the same graph as the USDM.

[mbtoloczko]

Nice job. ......

[Dessertrunner]

Phil I have a question? In section 2.7 page 16 of the Subaru manuel it talks about the fact that the Air/Fuel ratio is set to suit the ideal ratio for the Cat's and to meet emmision standards. If this is the case then we need to run two different maps and us a switch between the two.
My question Phil is do you know which memory location it tells the ECU that the car is a Catalyst model. As I remember there are cars out in the world that don't have O2 sensors. If we are going to get maxium power we need to tell the car to look at a different map not the O2 one. On our rally car they don't run O2 sensors because they need a richer mixture to get maxium power from the engine.
Second thought is if the O2's are disconnect does the ECu look at a new or alternative fuel map?

[Nomake Wan]

As far as I know, the car runs crappier if you unplug the O2 sensors.

[Dessertrunner]

I had a bad O2 and car was running rough untill I unplugged it it ran better. I am suggesting if we find the writee write in the ECU to change it from running on O2, same way you stop the rev limiter.
Tony

[b3lha]

Originally Posted by Dessertrunner

Phil I have a question? In section 2.7 page 16 of the Subaru manuel it talks about the fact that the Air/Fuel ratio is set to suit the ideal ratio for the Cat's and to meet emmision standards. If this is the case then we need to run two different maps and us a switch between the two.
My question Phil is do you know which memory location it tells the ECU that the car is a Catalyst model. As I remember there are cars out in the world that don't have O2 sensors. If we are going to get maxium power we need to tell the car to look at a different map not the O2 one. On our rally car they don't run O2 sensors because they need a richer mixture to get maxium power from the engine.
Second thought is if the O2's are disconnect does the ECu look at a new or alternative fuel map?

Hi Tony,
Looking at my JDM code, I can't see any sort of flag in the memory to tell the ECU whether it has a catalyst or not. The O2 sensor processing can't be turned off except by modifying the software. I suspect that the non-catalyst models like Danny's Middle-Eastern version have different software with the O2 sensor processing removed. It would be interesting if we could get a dump of Danny's ECU to compare.

With regard to your second question. With the O2 sensors unplugged the ECU will choose a default value and use that in place of the correct O2 sensor reading. I don't know whether the default value remains constant, or whether it varies according to other sensor inputs, like rpm for example. The O2 sensor code is fiendishly complex and I haven't managed to make much sense out of it so far. When I figure it out I'll post it here.