G1 ECU Thread

[Cryser]

So I just wanted to throw this out there:

I have a theory on why the ECU has 4 fuel maps and 3 ignition maps.

There are bypass control solenoids that control when the butterflies in the intake manifold open. There are 2 of these solenoids, one to open them part way in mid range and other fully open them in the high end. Now on VTEC motors and VTC(i-VTEC) motors, such as the H and K series, Every time the cam is changed to a new position the ecu switches to a new fuel and timing map, in order optimize the new flow into the engine.

Seeing as how an egt based fuel map wouldn't need anymore timing changes since it's just changing the amount of fresh air/fuel mixture able to enter I could see the maps being for when these solenoids are engaged changing the pathways in the intake manifold.

Now going by the ignition maps there is a change in timing between the maps at noticeable locations where these solenoids would be activating, this timing changes are mirrored in the fuel maps but to a much lesser extent, which makes sense. As the amount of air isn't really changing its the velocity at which the air is entering that is.

If the above is indeed the case then there should be 2 variables to when these solenoids are activated somewhere in the ecu coding. If these variables were to be found then through the ecu we would be able to control when these pathways open and be able to tune them for optimal power.

[sr5guy]

Originally Posted by Cryser

So I just wanted to throw this out there:

I have a theory on why the ECU has 4 fuel maps and 3 ignition maps.

There are bypass control solenoids that control when the butterflies in the intake manifold open. There are 2 of these solenoids, one to open them part way in mid range and other fully open them in the high end. Now on VTEC motors and VTC(i-VTEC) motors, such as the H and K series, Every time the cam is changed to a new position the ecu switches to a new fuel and timing map, in order optimize the new flow into the engine.

Seeing as how an egt based fuel map wouldn't need anymore timing changes since it's just changing the amount of fresh air/fuel mixture able to enter I could see the maps being for when these solenoids are engaged changing the pathways in the intake manifold.

Now going by the ignition maps there is a change in timing between the maps at noticeable locations where these solenoids would be activating, this timing changes are mirrored in the fuel maps but to a much lesser extent, which makes sense. As the amount of air isn't really changing its the velocity at which the air is entering that is.

If the above is indeed the case then there should be 2 variables to when these solenoids are activated somewhere in the ecu coding. If these variables were to be found then through the ecu we would be able to control when these pathways open and be able to tune them for optimal power.

You made me take another look at the ECU coding tonight. I have been working off of a (37820)-PL2-A62 bin file which is a 1990 ECU. I had some trouble decoding it because some of the code is internal to the processor and there were basically jumps within the coding to areas that I couldn't "see" without pulling the MCU off of the board and making a special reader for it (not happening). Well, I had an '88 M/T ECU here from l4zy415 and decided to socket it and get a read from the chip.I compared the chip to the PL2-A62 bin and noticed some serious differences. It turns out that the 88 ECU doesn't use the internal MCU ROM and all of those "programs" I couldn't see before were included right on the chip! Also, the offset and memory map of the '88 ROM was different as well (+8000 offset for the ROM chip). All of the fuel and ignition maps and basic settings were in the same place on the chip irregardless of the offset.

Anyway I installed my emulator, datalogging header and TTL->Serial cable on the ECU and started fooling around. I started headfirst into the serial communications routines and first found the baudrate. The stock setting is 125000 baud. Yes, 125,000 BAUD!!! That is incredibly fast for an ECU this old! It is a non-standard baud rate but I have a special terminal program that allows me to open up the serial port at any speed. Also, tunerpro supports non-standard baudrates thankfully. As a sidenote, I confirmed that this ECU does not facilitate 2-way communication in any way.

The second battle was to figure out how the stream was laid out. In stock form, the ECU sends a 130 byte repeating stream. I wasn't interested in this much as this mode would only provide updates for each value every 2 seconds or so. Also, the stream was a direct, consecutive memory dump and did not reference an "address table" within the ECU. This stream had two header bytes - $33 $CC (note that they add up to $FF for a external parity check - fucking honda and their perfection) followed by 128 data bytes.

What I was interested in was the (factory provided, but disabled) 16 byte data stream - that also has 2 header bytes that add up to $FF - $11 $EE. This is followed by 14 data bytes that are referenced by a table at $753C on the ROM. Any address that is put in this table will be streamed out of the serial port directly in the order of the table. So if the table included the addresses for RPM ($0081), Speed ($006D), L STFT ($116D) in that order they would be streamed out (and changeable) in that order. This special 16 byte streaming mode was turned on by changing the value at $75A6 from 1 to 0.

Ok, so whats all this mean?!?! Serial communications is working in tunerpro! I still have to come up with some solid conversions to get the raw data into a human readable form, but the hard part is over!!!! Let the fun begin.

-Matt

[Cryser]

So if I understand everything correctly your saying:

1. The bin and xdf file provide for the 90 ecu can be used for at least the 88 ecu and presumably the other years as I doubt honda would completely change the addresses of the ecu for 1 year. So in essence we have the locations for the major parts for basic chip tuning to be possible and we can extend that bin file as more addresses are found for say the solenoid engagement rpms, if they exist.

2. The offset is different presumably because of the use of the MCU in the 90 ecu, what year was this introduced 89 or 90?

3. The ecu has an internal datalogger that can be turned on by changing the value at $75A6 from 1 to 0, this is awesome! This would allow remote tuning to be available to people using this datastream as long as the appropriate addresses were logged!

As for conversion for the values that fill the fueling tables I would suggest just using the raw number. All the honda tuning platforms I've used all seem to just spit out a 4 digit number with the general consensus that higher = more fuel.

For the Timing values the tunerpro XDF file uses the following formula to get a degree output on the map: ( X - 24 ) / 4

Not sure if those help.

I have noticed in regards to the ignition maps the following to be true on the bin

For all 3:

the first 4 columns for all rows are the same

So at idle and very high vacuum the timing is static between the tables for all rpm ranges

The first 4 rows for all columns as well are all the same

idle and low rpm ranges regardless of vacuum timing is static

The last 3 columns for all rows are the same

This would be the above atmospherically or "boosted" hence why timing is soo retarded in these areas at high rpm.

For the differences in the maps From ign table 1 to 3 seems to be slight advancements in timing in the mid rpm and vacuum ranges while table 2 is a huge advancement in timing in these ranges.

My preliminary guess would be:

Ignition table 1 would be fully closed
Ignition table 3 would be partial open
Ignition table 2 would be fully open

Since these are arbitrary labels provided by the creator of the .xdf file it wouldn't be safe to match up the same labeled fuel maps to these ignition maps, the fuel map differences are much more subtle sometimes the variances are very small.

I beginning to wonder if both rpm AND manifold pressure are taken into account in the ecu's decision to open the butterflies or if the engine calibrators just deemed it unnecessary to tune portions of the switch over maps that would never be touched. I am leaning towards the second scenario simply because the pressure variable as a trigger to engage VTEC wasn't introduced until after the K series motor ecu was made.

On the P28 ecu(most common ecu to be used with Hondata S300 and chrome) the Variables for VTEC are, engine oil pressure, engine temperature, vehicle speed, engine speed and throttle position. It wasn't until K-pro that you are able to use a VTEC window to engage the system, or the VTC system, in regards to manifold pressure. I don't think honda would have ditched that control variable if they found it important in '89

[cars-all-mighty]

Originally Posted by sr5guy

You made me take another look at the ECU coding tonight. I have been working off of a (37820)-PL2-A62 bin file which is a 1990 ECU. I had some trouble decoding it because some of the code is internal to the processor and there were basically jumps within the coding to areas that I couldn't "see" without pulling the MCU off of the board and making a special reader for it (not happening). Well, I had an '88 M/T ECU here from l4zy415 and decided to socket it and get a read from the chip.I compared the chip to the PL2-A62 bin and noticed some serious differences. It turns out that the 88 ECU doesn't use the internal MCU ROM and all of those "programs" I couldn't see before were included right on the chip! Also, the offset and memory map of the '88 ROM was different as well (+8000 offset for the ROM chip). All of the fuel and ignition maps and basic settings were in the same place on the chip irregardless of the offset.

Anyway I installed my emulator, datalogging header and TTL->Serial cable on the ECU and started fooling around. I started headfirst into the serial communications routines and first found the baudrate. The stock setting is 125000 baud. Yes, 125,000 BAUD!!! That is incredibly fast for an ECU this old! It is a non-standard baud rate but I have a special terminal program that allows me to open up the serial port at any speed. Also, tunerpro supports non-standard baudrates thankfully. As a sidenote, I confirmed that this ECU does not facilitate 2-way communication in any way.

The second battle was to figure out how the stream was laid out. In stock form, the ECU sends a 130 byte repeating stream. I wasn't interested in this much as this mode would only provide updates for each value every 2 seconds or so. Also, the stream was a direct, consecutive memory dump and did not reference an "address table" within the ECU. This stream had two header bytes - $33 $CC (note that they add up to $FF for a external parity check - fucking honda and their perfection) followed by 128 data bytes.

What I was interested in was the (factory provided, but disabled) 16 byte data stream - that also has 2 header bytes that add up to $FF - $11 $EE. This is followed by 14 data bytes that are referenced by a table at $753C on the ROM. Any address that is put in this table will be streamed out of the serial port directly in the order of the table. So if the table included the addresses for RPM ($0081), Speed ($006D), L STFT ($116D) in that order they would be streamed out (and changeable) in that order. This special 16 byte streaming mode was turned on by changing the value at $75A6 from 1 to 0.

Ok, so whats all this mean?!?! Serial communications is working in tunerpro! I still have to come up with some solid conversions to get the raw data into a human readable form, but the hard part is over!!!! Let the fun begin.

-Matt

[sr5guy]

Originally Posted by Cryser

So if I understand everything correctly your saying:

1. The bin and xdf file provide for the 90 ecu can be used for at least the 88 ecu and presumably the other years as I doubt honda would completely change the addresses of the ecu for 1 year. So in essence we have the locations for the major parts for basic chip tuning to be possible and we can extend that bin file as more addresses are found for say the solenoid engagement rpms, if they exist.

2. The offset is different presumably because of the use of the MCU in the 90 ecu, what year was this introduced 89 or 90?

3. The ecu has an internal datalogger that can be turned on by changing the value at $75A6 from 1 to 0, this is awesome! This would allow remote tuning to be available to people using this datastream as long as the appropriate addresses were logged!

As for conversion for the values that fill the fueling tables I would suggest just using the raw number. All the honda tuning platforms I've used all seem to just spit out a 4 digit number with the general consensus that higher = more fuel.

For the Timing values the tunerpro XDF file uses the following formula to get a degree output on the map: ( X - 24 ) / 4

Not sure if those help.

I have noticed in regards to the ignition maps the following to be true on the bin

For all 3:

the first 4 columns for all rows are the same

So at idle and very high vacuum the timing is static between the tables for all rpm ranges

The first 4 rows for all columns as well are all the same

idle and low rpm ranges regardless of vacuum timing is static

The last 3 columns for all rows are the same

This would be the above atmospherically or "boosted" hence why timing is soo retarded in these areas at high rpm.

For the differences in the maps From ign table 1 to 3 seems to be slight advancements in timing in the mid rpm and vacuum ranges while table 2 is a huge advancement in timing in these ranges.

My preliminary guess would be:

Ignition table 1 would be fully closed
Ignition table 3 would be partial open
Ignition table 2 would be fully open

Since these are arbitrary labels provided by the creator of the .xdf file it wouldn't be safe to match up the same labeled fuel maps to these ignition maps, the fuel map differences are much more subtle sometimes the variances are very small.

I beginning to wonder if both rpm AND manifold pressure are taken into account in the ecu's decision to open the butterflies or if the engine calibrators just deemed it unnecessary to tune portions of the switch over maps that would never be touched. I am leaning towards the second scenario simply because the pressure variable as a trigger to engage VTEC wasn't introduced until after the K series motor ecu was made.

On the P28 ecu(most common ecu to be used with Hondata S300 and chrome) the Variables for VTEC are, engine oil pressure, engine temperature, vehicle speed, engine speed and throttle position. It wasn't until K-pro that you are able to use a VTEC window to engage the system, or the VTC system, in regards to manifold pressure. I don't think honda would have ditched that control variable if they found it important in '89

1. Yes the definition can remain the same. How the processor sees the ROM is different, as the offset is different. For a laymen's explanation of offset, the first fuel map is located at $604E in the ROM. In the disassembly, the processor references the fuel map based on an offset. In the '88 ECU the offset is +$8000, so $604E+$8000 =$E04E. The 90 ECU is a +$4000 so $604E+$4000=$A04E. So, even though the processor "sees" a different address, the map is still in the same place on the ROM - $604E.

2. I don't know when the change occurred, all I have are samples from the '88 and from the '90.

3. Yes, I have most of the critical addresses identified for datalogging, now the challenge is to come up with the conversions to real world values.

4.The reason for the x and y scalar issues is because the definition that was posted earlier in this thread was thrown together quickly and the scalars were carried over from the G2 rom and not corrected. I found the RPM scalars are defined in the ROM. Unfortunately, the MAP scalars are not scaled and are referenced directly from the filtered signal from the MAP sensor. So instead of referencing a direct millibar reading I am going to set them up with manual percentages of load. Yes the fuel reading should and will be kept raw and unconverted.

5. The different fuel maps are most definitely not VIS flap based. The VIS flaps don't really have the effect you think they may have, they may bump up torque a few foot lbs in the mid range, but i can almost guarantee that you wouldn't notice the difference if they were working or not. I haven't dug deep enough into the disassembly to tell exactly what their reason is, but an educated guess based on the G2 ROM is that they are high and low load maps (mostly TPS based) and EGR active maps.

-Matt

[Cryser]

wish I was closer, I would drive down and let ya read my 89 ecu =)

If I find anything useful for the conversions I'll post up

[sr5guy]

You could help by taking a picture of the inside of your ECU. Mostly need the part number off the processor.

-Matt

[Cryser]

sure, I'll try and grab a picture of it tomorrow

[3GCVC]

Originally Posted by sr5guy

You could help by taking a picture of the inside of your ECU. Mostly need the part number off the processor.

-Matt

Would you want anything from me?

I have the below ECUs

1989 2.7 EDM ECU (from no cat converter model)
1989 2.7 JDM ECU
1990 2.7 JDM ECU with traction control

If you need / would like to see ECUs from JDM and or EDM 2.0 (non-turbo) and 2.5 let me know and I will keep a lookout for them.

[Cryser]

had to tear down and rebuild an engine for the race car so I didn't get a chance to snap a photo. Ordering my rebuild stuff for the legend weds, when I get my pay check so I'll get it as I tear into the motor hopefully this weekend/next week.