1981TTA

Well, given the recent discussions regarding running at non-14.7 AFR, it's about time we come up with some code to automatically run at a calibrated AFR and close the loop with the WBO2 sensor.

Here's what I have so far (please feel free to comment/add/delete as necessary)

1) Modifications will be made to a 7749 ECM with $58 software. (Why, because that's what I have to test with... )

2) Calibration table to convert WB volts to AFR.

3) Calibration table of desired AFR vs. RPM and MAP. (This would be dimensioned just like the main spark table.)

4) Existing AFR modifications (AE/PE) would be left intact. (However, I'd imagine the need for portions of the existing PE algorithm could be comprehended with the table in #3...)

5) Modify existing INT logic:
5a) Allow closed loop operation during PE.
5b) Remove "slow trim/fast trim" logic/tables.
5c) Add proportional gain calibration (KPROPGAIN)
5d) Add integral gain calibration (KINTGAIN)
5e) Utilize existing transport delay algorithms for INT update
5f) INT calculation becomes : INT(k+1) = (INT(k)+(KINTGAIN*AFRERR)) + KPROPGAIN*AFRERR

6) Existing BLM operation remains basically unchanged.

After completing the above, we'll have to take a look at how things like DE and DFCO affect the above. We'll also need some volunteers to try out some of the software changes. I can bench test and test P/N operation in-vehicle. But, I don't have the capability to drive down the road. (Well, I do. My Trans Am doesn't.....)

So..... what do you think.......?

gta324

This sounds great, too bad I dont have the knowledge to help out, I can try it out later... but there are some experts here that I think could help.

2) Calibration table to convert WB volts to AFR.

I most WBO2, atleast innovate, you already could do this. It has 1 prog. output...

/N.

Z69

Little soon in my learning curve for me, but
I can provide a little help.
E-mail me.

TRAXION

After talking to Mike Davis with regard to how he configured the stock AUJP binary to read the wideband, I was surprised to find that there isn't just one simple lookup table used to convert wideband volts to an AFR. He used 2 lookup tables with different slopes due to the DIY-WB V vs. AFR not being linear.

Tim

1981TTA

I just created a table that mirrors the conversion table that was supplied with the Zeitronix sensor. I guess I've never done the math to see how much of a difference there is between the piecewise linear approach and some sort of smooth fitted curve....

Is there anything documented on how the AUJP curve was implemented and/or what sensor was used? I don't understand why two tables would have to be used, either.....

RBob

iTrader: (1)

The use of one or two tables is probably dependent upon whether an indexed lookup table or an interpolated lookup table is used. With an indexed lookup only one table is required. That table will also be larger then an interpolated table. However, the benefit would be faster code.

RBob.

{edit: a single interpolated table can also be used as long as the stoich point is set to a hard data point in the table. By a hard data point I mean where a table row is at stoich. This will break the table between the two curves at that point}

TRAXION

I remember Mike saying that he used the built-in STOCK code routines for the lookups and that interpolation is taking place. He said the DIY WB had 32 points and the GM tables allowed for 16 points to it was a match made in heaven to use 2 tables.

If what I said was wrong ... don't blame Mike ... blame me because that just means that I misunderstood what he said. But, it was something like the above.

Tim

1981TTA

Interesting...... I created a 21 point table for the Zeitronix sensor. While this also uses the stock table lookup routines, it isn't using one of the most efficient ones. (Extra ROM needed for number of table entries plus an "extra" INC X operator.) I'd imagine I could have found a way to reduce throughput by working with fewer entries and/or other processing. At the time, I just went with the first thing that came to mind that got the job done.

Grumpy

Do you ever check your PM's?.

RBob

iTrader: (1)

This is the beauty of software. The development of code allows the creativity of the programmer to be unharnessed. There is an unlimited number of methods and techniques available to reach the same goal.

RBob.

1981TTA

I hadn't until 5 minutes ago....

3.8TransAM

If you are to the point where you can write/modify code heavily(one day i will get that far i hope), you might want to check inot bruce's $60 which is a new improved version of $58 that he wrote(created) and has added to quite extensively.

It also has much more available space for additions and modifications.

Not sure if this is what he ahs sent you already or not.

If u need hammerhead to help you out with simple stuff or grunt work, let me know, I'm not alot of use in the whole rewriting code or creating new stuff, but I'll throw in any way I can.

Always been fascinated by widebands since I knew what they were.

later
Jeremy

1981TTA

Yep. I just got a copy of his code. I was in the middle of a similar effort to generate source code on a $58 mask. Although I got to the point I could make changes and reassemble the code successfully, my "source code" isn't as well commented and organized as the $60. Both to make life easier for me *and* to try to reduce the number of versions of software floating around, I'll use the $60 code for development from this point forward.

It occurs to me based on some of the comments in this thread that it might be a better approach to make these changes one step at a time. As I've mentioned, I'm using a Zeitronix WB sensor. Maybe this should start with simply implementing the WB Volts-to-AFR table and outputting the result in the ALDL. This way, anyone who has a WB could try this with their favorite sensor to verify functionality across a wide audience. After this has been proven to work with enough people, we could move on to something like implementing the WB for closed loop at stoich only. Again, this should be of interest to a number of people who would like to replace their NB with the WB without having to install an extra bung in the exhaust system.....? Next, closed loop during PE with existing routines. Finally, separate commanded AFR tables for non-14.7 operation! I'm sure there will be things to consider and work out at each step of the way. But, this should give us a high level idea of how things might proceed.

I have to thank everyone for their offers of help and suggestions with this effort. In addition to the responses in this thread, there have been quite a few PMs and e-mails offering support, too.

So, if I were to make a version of $60 available with the WB volts-to-AFR conversion available, how many people would be interested? I'd also include a *.ECU file for Tunerpro that could be used to modify the table. (Maybe someone could make a Tunercat compatible file from this??) And, what sensors are being used? I know this works with the Zeitronix hardware. I don't know about LM-1, DIY-WB or others.....

As always, open to suggestion!

3.8TransAM

Well,

when Bruce and I get a chance to play, I'm assuming the TTA will switchover to $60 code and my Formula was built with a turbo in mind, so sometime in the next year when I undergo the transformation it to will run on $60 code.

Theres 2 for me lol

Also I have the Innovate wideband, I know the conversion are also out here for the Innovate sensor as well.

The Z and I wideband seem to be the 2 most commonly bought as well as the obligatory DIY widebands of which I have two of.

Everything would be able to remain the same with the code and we would just have to change the AFR vs voltage, correct?
later
Jeremy

1981TTA

That's how I'd like to do it. At this point, I'm curious whether or not all the WB sensors could use the same table. I *think* they will. But, I'm not sure......

skwayb

Turbodig (A fellow SyTyer) already has the WBo2 outputing into the ALDL for code $58. He puts it in the Target AFR field and dumps the RAW value to one of the EGR fields so you can make sure you are getting data from the F14 pin on the ecm.

He has patches for serveral different WBo2 brands. Some are linear so he uses a formula to calculate the output numbers in the stream. For the Non-linear units he has a look up table. I helped test them all and they work really well.

He has the patches up on his site. Take a look at them, they may help you get a jump on what you are trying to do.

http://digware.lunarpages.com/digware/index.html

All the instructions for which pin on the ECM to hook the WBo2 to are there also with pictures.

1981TTA

Do you know if the voltage curves are publicly available for the various "patches" on that page? I'd like to know whether or not a 22 point table will suffice for a generic WB voltage-to-AFR table. Of course, I expect the truely "linear" output versions of sensor to not be a problem. But, the non-linear ones....

My Zeitronix sensor was supplied with a table of 22 voltage/AFR pairs. If possible, it would be good to see the same for the other sensors.

RBob

iTrader: (1)

For the DIY-WB I use an 85 byte table that extends from 10.0 to 24.0 AFR. Can eliminate 9 bytes by taking the table to a max of 20:1 AFR, which is about all that is required. By creating a table of AFR's based on ADC counts it is an easy indexed lookup. Neither the 85 byte or an 76 byte (85 - 9) byte table is all that large.

Could even put a series of tables in for some of the more popular WB's (DIY-WB, LM1, Zeitronics) and have an option word for the user to select which one to use.

There tends to be a lot of EPROM space available in the ECMs we use.

RBob.

1981TTA

Hey.... Now that I like!! After "expanding" the $58 code to use 32k space, there's all sorts of room for code and calibrations. (I wish the same could be said for RAM space.)

This would eliminate the need for implementing the "largest" table and/or rescaling a particular sensor's transfer function into that table...... Have I mentioned I like this idea......? RBob, I like the way you think!

skwayb

Yah I know the voltage curves for most of the patches. The linear ones are straight forward. Usually something like
AFR = Volts*2 + 10 or AFR = Volts*2 + 9 ,etc

The ones that use a table I made all the graphs and data points for Dig to get the values needed. I can send you the excel spreadsheet I made for them if you need.

3.8TransAM

I know the WB voltage to AFR charts ahve been posted here for the Innovate LM1 and the DIY WB. Having trouble finding the posts that contained it however.

If u want, later when I find my Innovate box I can post the volt vs. AFR for it. Raucher should have the DIY chart.

later
Jeremy

1981TTA

3.8TransAM, if you have that information, I'd appreciate it if you'd post it.

skwayb, sent you an e-mail.

Thanks!!

skwayb

Here is the spreadsheet with the AFR to Volts values for various WB02 controllers.

http://www.nwstp.com/files/paul/AFR_Voltages_Master.xls

skwayb

I also found some notes I had made:

Notes:

• FJO - Uses Table, has correction for voltage readings. Not an ideal unit. Data stops at 15 afr at 2.54v. Half of the usable range is above 15 afr.
• AEM UEGO - Uses Table
• TE 2.0 SvOut, DIY, TE 1.1/1.5 Vout - Uses Table, also for TE has correction for resistor factor.
• Innovative - Linear Output. Default setting for Analog 2 output is 1v - 2v (10 afr to 20afr) so Formula is Volts*10. Used default setting to come up with formula.
• TE 2.0 WBlin output - Linear output Formula is Volts*2+9 (Verified Our ECM Patch works and the data is correct in the AFR Target field of Datamaster)
• PLX - Linear output. Formula is Volts*2+10
• AEM (one of their units has a WBlinear output) - Linear output. Formula is Volts*2+10

vernw

iTrader: (2)

The spreadsheet link isn't working.....

skwayb

Should be working now... Sorry about the bad link.

1981TTA

I'm still getting the "not found" error.....? Anyone else have any luck?

3.8TransAM

http://digware.lunarpages.com/digware/wb.html

Original post with wideband patch in it for the $58, has almost any wideband I can think of listed as compatible with it. So all theconversion charts may well be listed here.

Still trying to find my Innovate box to verify the AFR to volts conversion.

later
Jeremy

skwayb

Ok the spreadsheet link is really working this time.. LOL...

Yah all the wb02 patches on Digs page were developed off my spreadsheet. Dig had me do all the charts, tables and find all the info out for him. Then I helped test them.

Well the innovative is programmable so you can change the default formula they had for their WBLin output. Initially it was set to 1v = 10 afr and 2v = 20afr. So the forumla is AFR= V*10. If you recalibrate the output to something else... that patch on Dig's page will not work.

I need to talk to Dig and see if we can put what the forumla was we used for the patch or the table of values. After looking at it, it is hard to tell what the patches used.

Synapsis

Will the end result be available to the needy masses? I'm actually holding off buying a WB02 for awhile to see if this will become available because it might influence my decision of what features I'd get in a unit.

1981TTA

My intention from the start has been to make this public. As per the previous messages, we should have support for quite a few sensors. If this works out, you should be able to get whatever sensor you'd like and still be able to use it with $58/$60!

3.8TransAM

Let me know if those links from skwayb and myself gove you enough information. I can still dig up the original posts if need be..

let me know
Jeremy

1981TTA

After what seems like a *very* long time making changes and turning ***** on the ECM bench, I believe the software is ready for the next level of testing. I've uploaded the binaries for the $58 mask (not the $60, yet) to Moates fileman site. The file is named WB_Patch_$58.zip in the Custom Modified Binaries folder. (However, it appears the name is truncated to "WB_Patch_.zip".) Briefly, this software will allow a user to select from 8 pre-configured WB sensor transfer functions and output the AFR value in the ALDL stream. There's a text file in the .zip file with the binary that describes, in detail, the changes.

Please let me know if anyone tries this and has success (or failure). If this appears to work reasonably well, we can start working on the next steps implementing closed loop WB control...

gta324

Does this put out data in the EGR field as "digs" version does when you scan the car?

/N.

1981TTA

I wasn't aware of any changes to the EGR ALDL information. The only change is overwriting the "Desired/Target AFR" with the WBO2 reading rather than the ECM's target value. (It seems like this was what he did, too..???)

That being said, we can do anything we want with the ALDL and the information sent over the link. If there's a different field that would suit this better, we can do that!

RBob

iTrader: (1)

My preference is to not overwrite the desired AFR in the ALDL stream. But to use a more worthless location. Byte 31 is a candidate: NUMBER OF MINOR LOOPS BETWEEN REFERENCE PULSES, can't say that would be of much use.

The reason is that using a WB and tuning while in open loop the WB AFR can be compared to the desired AFR. From this VE table corrections can be made.

RBob.

1981TTA

RBob -- I agree. During development, I would overwrite the boost parameters with whatever I wanted to see. This works nicely with a datalogger like Tunerpro where the variables can be rescaled. However, for users of Datamaster, I'm not sure they have the same capability. There aren't many variables scaled to 0.1 to choose from..... My intent is to make this as "open" as possible so an individual isn't forced into using a particular piece of hardware/software. However, if this becomes too restrictive, we can always change that view.

As always, I'm open to suggestion. If a consensus can be formed on where to put the reading, I can make the change.

vernw

iTrader: (2)

Wish someone would do the same thing for $8D on a 730..... I'd be willing to field test, but can't write the code.....

scooter

iTrader: (11)

Yeah I second that, I was going to order an Innovate WB at the end of the week

gta324

Looking through the code changes now and what have you done to make the EGR working? its not the same as Rbob posted before...

I have started to copy my settings in the old bin over to "yours" and give it a try.....

/N.

1981TTA

I don't know how you're translating the calibrations between bins. But, if you aren't already, you should be able to simply copy everything from 0x0000 to 0x076F directly over. There are various binary editors that can make this as easy as a copy-and-paste operation.

I don't recall the implementation specifics of RBob's EGR changes for V8. However, I think I've implemented software that will do the same thing. It basically boils down to replacing the cylinder select check with #$FF when the cylinder select calibration is set to $00. I know I see the EGR "function" as far as an LED on an ECM bench with this change. I will admit I haven't checked non-V8 operation to make sure it still works for other cylinder values....

Let me know how this works for you. I'm excited to see some "real world" results in-vehicle.

1981TTA

I'm willing to give this a try. The changes I made to the $58 software should largely translate to $8D directly. Of course, this is with the assumption there's enough room in the EPROM and RAM.

What broadcast code(s) should I try? I know there's a super "something" BCC that's been worked on by members of this board. You might be able to tell I haven't kept up with it given my immediate interest in $58..... I guess I'm going to get started re-reading all the "how to convert 730<->749" threads to figure out how to convert my ECM bench!

skwayb

Nice job. I need to upload it to my Ostrich and give it a shot on a V6. Wish there was a 3 bar version... be nice if you could put that in the code too. Toggle between 2 bar and 3 bar would be a bonus...

Too bad you didn't have any bits left to put in a RAW patch for the WBo2. That way if a sensor comes out that you don't have in the bin you could just use the RAW value and then in TunerPro you could put in a table or formula for it. I know it kind of screws over the Datamaster users but it would be a nice option.

Just a thought...

1981TTA

There's LOTS of ROM left to play with. I can add another WB Options word that will allow another 8 selections......

I'll need some additional information. First, what's the transfer function for volts-to-MAP of the sensor? Next, what parameters/tables get modified, if any? (I.e. is it sufficient to lose resolution in the main spark table with the addition of a 3 bar sensor? Or, is the table size increased?)

Sounds like I need to start working on version 2!!

Z69

Good job Don.

$58 FAQ

I can look over the 8D stuff and one of us can bench it unless someone else is already at it.

gta324

here are the 2 metods Rbob posted:

1.
Addr is: to:

$2E8F $B6 $86
$2E90 $C0 $FF
$2E91 $09 $01


2. 100% version
Addr is: to:

$2E8F $B6 $CC
$2E90 $C0 $C8
$2E91 $09 $00

$2E92 $C6 $01
$2E93 $C8 $01
$2E94 $3D $01


I'm using tunerpro/tunercat and compare my old 16k bin with your WB bin and copies all the changes

It feels safer than copying everything in "hex format" I dont want to miss anything hardcoded in "your" bin

/N.

RBob

iTrader: (1)

What I had done was a simple proof of concept change. It also allowed for an easy patch to be made so that the bin could be used on 8 cyl engines and allow EGR. The down side is that it would only be correct for 8 cyl engines. Not a problem for people that want to patch their own bin and use it.

What '81TTA did was to change the code in such a way that it will work on 4, 6, and 8 cylinder engines. This required 2 additional instructions to the code. For code that is being released to the general public it is a much better method of implementing the change.

RBob.

skwayb

For the 3 Bar information I would look at this post Grumpy made on the 3 bar patch for code 60.
https://www.thirdgen.org/techbb2/sho...ht=3+Bar+Patch

Look here for the information I have on Brian Green's changes he made to all the tables for 3bar vs 2bar. The table comparisons are a little ways down.

http://www.nwstp.com/forum/chiptable.asp

Not sure if you lose resolution in the tables since you rescale the table. Grumpy would be a better person to ask on that.

Great work btw.. If you get the AFR vs Map Vs RPM table working that is going to be killer. How are you going to handle the Boost Multipler table?

vernw

iTrader: (2)

If there's anything I can do on the $8D version, like in car testing, etc., please be sure to let me know. I'm not well versed enough to handle doing coding changes, but would still like to help out if possible so that I'm not just being a "sponge" on this effort.....

If you like, I can even email you a 730 $8D BIN to start working from.... just let me know.

ty1295

I am anxiously awaiting your developments.

Downloaded your current version last night, looked good on tunerpro.

Truck will be running hopefully this weekend, and we can transfer over my fuel tables and give it a shot soon.

I have several other trucks I can test on also. This will be a big help.

I don't know that much assembly language, but be glad to help anyway I can.