Z-man
Senior Member
Posts: 335
Loc: Foxfield, CO
Reg: 12-30-03
|
09-16-07 10:46 PM - Post#1255165
Take it easy...it's not Holley's, so you can calm down. It's mine.
Anyone want to look at what I've been playing with the last couple of weeks? I realize that, in light of the fact that Doug has officially announced that Holley is releasing a free new upgrade product for all of us any day now  , this effort is somewhat in arrears, but hey...maybe some of you might want to give me feedback.
I wrote a little package that would analyze my C950 logs and allow me to quickly apply changes to my fuel map. It's been a real bear since my wife hates doing this type of stuff, and so does anyone else I know. I got real tired of driving around trying to watch the laptop, or trying to analyze things when I got back home. I was thinking that this might be useful to a person new to the C950. It can be used to quickly create a "base map" and get it nailed down without the need to have a co-pilot along with you for the ride.
Here's a snippet of what the application does:
Introduction
C950 CoPilot ("CoPilot") is a tool that is designed to assist in the tuning of an automobile using the Holley Commander C950 Pro hardware and software. CoPilot utilizes two main operational modes:
1) acting as a copilot while the driver operates the automobile, CoPilot can automatically adjust the fuel map to bring the fuel map values within the parameters set for closed loop compensation in the C950 software, and
2) acting as an analysis tool, CoPilot can help analyze a log file that was captured during a drive on a laptop using the C950 Datalogger. CoPilot can help determine where to make changes in the fuel map, and automatically update your fuel map for you if you want.
I'm most interested in feedback from the log analyzer part of the software. You can try the "auto-tune" part too, but it's really "alpha" at this time. (If you want to see how it works, you can run it "manually" and watch it "tune" your map.) With the log analyzer, you basically take a drive and log it. Then run analyzer and apply those changes to your map. Lather, rinse, repeat.  What it does is look at each data point in your log. If the A/F ratio is out of spec for that particular MAP and RPM, it calculates how far off that cell is and remembers it. For instance, if the closed loop compensation is at 90 and you really wanted it to be at 95, then that cell is off 5%. You can tell the analyzer to adjust the value in that cell by 5%. Actually, you tell the analyzer to adjust every cell in the map that is out-of-spec. You can then save the map and take another drive. Within a few drives, you'll have a base map.
If you guys think it's crap, I'll go back to my daytime job, but if you think there's some merit, I'd incorporate your feedback - just drop me an email (see my profile). Keep in mind, I'm trying to create something that would help a "noobie" to quickly get just a "base map" established. Acceleration enrichment and temperature modifiers would be done as usual (or could be added later). I know most of you have already got your base maps, but you'll just have to pretend - or start with an old map.
To download the software, go to my website (http://temp.corvetteforum.net/c3/zman/index.shtml) and pick the second button. There is a link at the bottom of the page. (I know "CoPilot" has been used everywhere, but right now, I really don't care).
Help is really sparse because I just threw it together, but I'd be glad to send you more info if you get stuck or have questions.
NOTE: Doug (or anyone) - this software does not interface with the C950 electrically, does not attempt to read the incoming data stream, and does not attempt to modify any of the code in the C950 Pro software. It just runs simutaneously on the same machine..
|
 DeltaT
"2nd Year" Silver Supporting Member
Posts: 975
Loc: San Jose area, CA, USA
Reg: 06-20-02
|
09-17-07 08:15 AM - Post#1255370
In response to Z-man
I'll try it and review. The closest I got was creating some Excel macros that would lookup cells from my datalog and ID which cells were low to target and which were high.
This could be fun!
Jim
|
CamaroFever
Member
Posts: 19
Loc: Kansas
Reg: 06-28-04
|
09-17-07 07:16 PM - Post#1255907
In response to DeltaT
I would like to try it. I am kind of a newbe to this although I have had my C950 for some time it is still a long way from being tuned. this might help alot.
Mark
| 1985 IROC-Z 383 Stealth Ram MPFI, Holley 950 Commander |
|
Z-man
Senior Member
Posts: 335
Loc: Foxfield, CO
Reg: 12-30-03
|
09-17-07 09:09 PM - Post#1255971
In response to DeltaT
I'll try it and review. The closest I got was creating some Excel macros that would lookup cells from my datalog and ID which cells were low to target and which were high.
This could be fun!
Jim
Thanks guys. Delta - that sounds similar to what this does...
|
 Doug_F
Chevytalk Moderator
Posts: 4224
Loc: Bowling Green, KY
Reg: 08-20-01
|
09-18-07 04:26 AM - Post#1256067
In response to Z-man
The big tricks with self tuning is to be able to blend the cells with the algoritm and not just "beat on one cell". Also knowing when to not self tune.
Hope it works good. I'll try to take a look if I get a spare minute.
Doug
1972 491 EFI Nova
10.50 @ 127.5, 9.61 138 on a small hit |
|
Z-man
Senior Member
Posts: 335
Loc: Foxfield, CO
Reg: 12-30-03
|
09-18-07 07:25 AM - Post#1256173
In response to Doug_F
The big tricks with self tuning is to be able to blend the cells with the algoritm and not just "beat on one cell". Also knowing when to not self tune.
Hope it works good. I'll try to take a look if I get a spare minute.
I noticed that the other "self-tuning" packages picked a range of cells like you mentioned. I could easily do that, but I wasn't sure which cells to group. For instance, would you update all surrounding cells too, or a cluster of 4 adjacent cells just above, below, right and left? All of the surrounding cells is a group of 9 cells - kind of large...but maybe that's the way to go. My goal was just to get a quick base map because most of the Holley maps are just a "good start" since each user usually has a different engine configuration.
Actually, you got me thinking now. Hmmmm....I can just put in a little graphic option that lets the user decide. I'll get right on it... 
|
zwede
Senior Member
Posts: 1233
Reg: 01-10-03
|
09-18-07 08:13 AM - Post#1256202
In response to Z-man
I would think it should interpolate like the C950 does. Example:
You have a cell at 2000 rpm and the next one at 2500 rpm.
You're currently at 2250 rpm and lean so the SW needs to richen the map.
Since it's at 2250 rpm it needs to richen both 2000 and 2500 equally.
If the current rpm is 2400 rpm it would richen the 2500 rpm cell 80% and the 2000 rpm cell 20%.
Now add in the load scale the same way (sliding percentages).
Edited by zwede on 09-18-07 08:13 AM. Reason for edit: No reason given.
|
Z-man
Senior Member
Posts: 335
Loc: Foxfield, CO
Reg: 12-30-03
|
09-18-07 08:22 AM - Post#1256209
In response to zwede
I would think it should interpolate like the C950 does. Example:
You have a cell at 2000 rpm and the next one at 2500 rpm.
You're currently at 2250 rpm and lean so the SW needs to richen the map.
Since it's at 2250 rpm it needs to richen both 2000 and 2500 equally.
If the current rpm is 2400 rpm it would richen the 2500 rpm cell 80% and the 2000 rpm cell 20%.
Now add in the load scale the same way (sliding percentages).
So, a "weighted" interpolation adjustment - good idea. I'll toss that in to the auto tune part. On the log analyzer, each cell already has a particular calculated variance based on the driving conditions, so I would think that is even better. Maybe not?
I've found that, on lower value cells, since the number needs to round to the nearest integer, so sometimes the number doesn't change. Example, if you are off 3% on a cell that has a value of 12, nothing happens since the nearest integer value for that cell is 12. Not as large a problem on the higher value cells.
|
zwede
Senior Member
Posts: 1233
Reg: 01-10-03
|
09-18-07 09:19 AM - Post#1256268
In response to Z-man
Yeah, the rounding error is a problem. Especially with weighted cells as you would probably never change them (change would always be less than 1).
Only perfect fix I can think of would be to have a private fuel map with floats that you make your corrections to and whenever a cell in this private map changes numbers you update the C950 map accordingly.
|
Mkelcy
"4th Year" Gold Supporting Member
Posts: 53
Reg: 03-30-04
|
09-21-07 06:43 PM - Post#1258721
In response to zwede
I've had my 950 running now for 2 days - this being the second. I will confess I was pretty discouraged last evening. The maps offered by Holley that seem to work with my engine are a narrow band map for a 383 with 36 pound injectors - which I have - or a wideband map for a 383 that seems to use 30 lb. injectors. Last evening I entered the fuel map values from the narrow band map into the wideband map and loaded the wideband map - as so modified - for tuning today.
I also downloaded the program with the intent to try it today. So far, it seems to have been very helpful. I had some issues getting it to load on my Win XP laptop, comdlg32.ocx didn't seem to want to get registered, but I found a fix on the web. I use a USB to serial converter, so I had to do the data logging with all other windows in the Holley software closed, but it seemed to work. I did several iterations - lather, rinse, repeat - and believe the basic fuel map is significantly improved from where I started. Thank you.
I have some small throttle movement issues - it seems to hesitate starting from a stop and on slight acceleration, but all in all, the degree of O2 correction is much lower than when I started, with only 3-4 iterations. I noticed some aberational proposed changes - adjacent cells at approximately the same values with one being extremely rich and the other extremely lean. It may be that my 5-10 minute data logs simply didn't compile enough data - or the serial to USB connection was corrupting the data. Is there a recommeded datalog size? Other than that, I'm very pleased with what I was able to accomplish, with a big assist from your program - today.
Unfortunately as i was doing my data logging, the wheel came off - literally - as I was heading home from my mountain road run, so I ended up waiting for the tow truck for about an hour with a very nice CHP officer. Given my attempts to record data at higher loads and higher TPS ranges on a mountain road, it was the most amazingly successful accident I've ever had. The car (a '68 Camaro) is undamaged - it rode on the driver's side rear rotor once the wheel left, the wheel looks okay, I'm putting up a big "Check Lugnut Torque" sign in the shop, and planning on continuing to tune tomorrow.
Two questions - what's a safe yet economical A/F ratio for long distance cruising. I'm planning a long trip back east in a week or so, and would really like to see 20+ mpg if I can.
Also, how "raw" in the Co-Pilot function? I'd love real time tuning, but not at the risk of cells being changed to zero or 700.
Thanks for doing this program.
| Mike - '68 Camaro with some stuff done to it |
|
zwede
Senior Member
Posts: 1233
Reg: 01-10-03
|
09-21-07 07:36 PM - Post#1258763
In response to Mkelcy
If the motor will take it you can lean it out into the mid 15's. Once you get to about 16:1 fuel economy tends to start dropping even if it's not misfiring.
Whether or not you'll be able to get it that lean depends mostly on the cam. The more cam, the more likely it will start complaining (surging/missing) as you lean it out. Start at 14:1 and note how the engine runs and then gradually take it leaner.
Remember: You can't damage an engine by running lean as long as it's at LOW LOAD.
Edited by zwede on 09-21-07 07:37 PM. Reason for edit: No reason given.
|
Z-man
Senior Member
Posts: 335
Loc: Foxfield, CO
Reg: 12-30-03
|
09-21-07 11:02 PM - Post#1258834
In response to zwede
I've always wondered how lean you should be setting it for cruise. The Holley manual is very conservative in this regard - they point out that it's best to be slightly rich. However, my engine seems to accept leaner settings. I'll probably aim more for 14-15 instead of 13-14.
By the way, I've modified the CoPilot for those of you playing with it. Please download the newer version. Based on feedback, I changed the way the log analyzer saves the data (directly to the file itself). I also took a shot at Doug's suggestion regarding inclusion of the surrounding cells. Right now, the CoPilot part is nebulous because I still haven't figured out the best way to implement the changes to surrounding cells. Right now, it will change all the user selected cells in the same direction as the central focus cell, and by only a value of 1 at a time. But it occured to me that some of the surrounding cells might be "off" in a different direction, so I need to test that area better.
Mkelcy - Wow - close call on the wheel! I've had one get loose, but never fall off. Sounds like you were lucky - did you get a lottery ticket just in case?
|
DeltaT
"2nd Year" Silver Supporting Member
Posts: 975
Loc: San Jose area, CA, USA
Reg: 06-20-02
|
09-22-07 04:08 PM - Post#1259177
In response to Z-man
I'm playing successfully up in the 15.5 - 15.75 AFR range for low-load cruise. The suggestions I had for increased advance (higher than I would normally consider) I got here were helpful, and at the highest AFR's I'm now running 37 and slowly increasing it. I'll probably stop at 38.
My mileage has definitely gone up. At ~70 now I am getting almost 20mpg. It's just hard to stay there...
Jim
|
Mkelcy
"4th Year" Gold Supporting Member
Posts: 53
Reg: 03-30-04
|
09-22-07 05:55 PM - Post#1259244
In response to DeltaT
At ~70 now I am getting almost 20mpg. It's just hard to stay there...
Jim
You can't go that fast or you can't go that slow?
| Mike - '68 Camaro with some stuff done to it |
|
Doug_F
Chevytalk Moderator
Posts: 4224
Loc: Bowling Green, KY
Reg: 08-20-01
|
09-22-07 07:30 PM - Post#1259296
In response to Mkelcy
At ~70 now I am getting almost 20mpg. It's just hard to stay there...
Jim
You can't go that fast or you can't go that slow?
With 500 RWHP probably that slow!
Doug
1972 491 EFI Nova
10.50 @ 127.5, 9.61 138 on a small hit |
|
jleeheat
Forum Newbie
Posts: 45
Reg: 01-24-07
|
 11-15-07 07:19 AM - Post#1296267
In response to Z-man
Z-man,
I'm looking forward to using this impressive software you've developed. I'd like to thank you for offering it to us.
My question is, should I use CoPilot in open loop mode or closed loop. C950 software as you know is adjusting for target mixture and I didn't know if it would be counter-productive to operate CoPilot in closed loop mode. Also, wouldn't it beneficial to let CoPilot alter ALL surrounding cells rather than the "X" pattern. I was wondering why you wouldn't want all cells manipulated.
Thanks,
Jim
Edited by jleeheat on 11-15-07 07:19 AM. Reason for edit: No reason given.
|
Z-man
Senior Member
Posts: 335
Loc: Foxfield, CO
Reg: 12-30-03
|
11-15-07 05:55 PM - Post#1296753
In response to jleeheat
My question is, should I use CoPilot in open loop mode or closed loop. C950 software as you know is adjusting for target mixture and I didn't know if it would be counter-productive to operate CoPilot in closed loop mode. Also, wouldn't it beneficial to let CoPilot alter ALL surrounding cells rather than the "X" pattern. I was wondering why you wouldn't want all cells manipulated.
Thanks,
Jim
Thanks - It's designed to operate only in closed loop mode and will stop tracking when the system goes open loop. Even though the C950 is adjusting the mixture, often times the selected "starter" map is farther off than the adjustment range. (Otherwise we could just all use the same map regardless of our engines!) The purpose is to try to get a custom base map specifically for your setup in a rapid manner. Also, you really should try to get the map tuned specifically for your engine. If a cell on the map is too lean, there will be a certain lag time before the C950 can figure out that is needs to add fuel. If the cell was adjusted correctly, the C950 will immediately be injecting the correct amount of fuel when the engine hits that cell.
You can tell CoPilot to adjust all the surrounding cells - not just an X pattern. My main challenge right now is coming up with an algorithm that best handles odd circumstances. For instance, if an adjacent cell(s) is higher than it should be, and the target cell is lower than it should be, all cells are lowered - so that drags the adjacent cell farther away from what it should be. That's not what should happen, so most users are reporting that they are letting it focus on only one cell.
By request, I'm putting some finishing touches on an "EFI dashboard" and some other enhancements to the Log Analysis portions of the package. Keep checking back...
Edited by Z-man on 11-15-07 05:56 PM. Reason for edit: No reason given.
|
artic-trans-am
Member
Posts: 3
Reg: 02-05-08
|
02-06-08 01:43 AM - Post#1359713
In response to Z-man
hi
I will try this program when I mount my engine again. Do this work on all the 950 software, wide and narow band O2-sensor?
is there anyone else that have working on programs like this for the comander 950?
|
Danny Cabral
Member
Posts: 1335
Age: 39
Loc: Ellington, CT
Reg: 11-03-04
|
02-06-08 04:24 AM - Post#1359743
In response to artic-trans-am
Funny you bring this up. Just the other day I was just wondering if there was any news or updates with Z-man's CoPilot program.
May God's Grace Bless You.
'78 BRONCO: 508" stroker, TFS heads, C950 EFI, A/C, Lentech Strip Terminator wide-ratio AOD, 3:1 Atlas II, max built Dana 44/60-lockers-4.10s, hydroboost/4-disc brakes, ram-assist/heim-flip steering, 4" lift, 35" tires |
|
artic-trans-am
Member
Posts: 3
Reg: 02-05-08
|
02-06-08 04:54 AM - Post#1359758
In response to Danny Cabral
well, I hope someone will create a software that can autotune the C950.. It is a problem to tune that thing. I just don't understand why Holley did not do this by them self..
that would be a good feature!
IB
|
Twilightoptics
Senior Member
Posts: 739
Loc: Vancouver, WA
Reg: 06-26-04
|
02-06-08 07:19 AM - Post#1359853
In response to artic-trans-am
I'm anxiously waiting too
'87 IROC-Z 355 HSR C950 Lunati Voodoo Solid 241/249@.050 .555/.576 lift(1.6) 110LSA, Pro Topline Iron 200cc/64cc, T56
'04 Jeep Wrangler "Rocky Mountain" 4.0L/5spd
'82 RX-7 GSL |
|
 2Loose
"6th Year" Platinum Supporting Member
Posts: 2947
Age: 67
Loc: 3000 miles west of Guadal...
Reg: 03-17-03
|
02-06-08 01:29 PM - Post#1360106
In response to Twilightoptics
Well, after watching here and other web sites for awhile, I just gave in and coughed up the cash for the FAST XFI setup for my LT motor. The LT is far from stock, and I wanted to be able to log it and easily tune it, and this seemed like the best path. It will be awhile before I get it all running, but I will let you all know how well it really does work.
Meanwhile I will keep watching here to see how the C950 progresses/evolves....
Aloha,
Willy
Slow Old Man = Fast Old Car
Picture Trail
2Loose
Chevys: 55Hardtop, 554drgasser, 554x4truck,
58prostreettruck, 57Olds, 70BuickGS
Avatar: Bill Stinson |
|
Z-man
Senior Member
Posts: 335
Loc: Foxfield, CO
Reg: 12-30-03
|
02-06-08 05:23 PM - Post#1360299
In response to artic-trans-am
hi
I will try this program when I mount my engine again. Do this work on all the 950 software, wide and narow band O2-sensor?
is there anyone else that have working on programs like this for the comander 950?
Wide band / narrow band - it shouldn't matter, but you'll need the pro version. No one has mentioned using narrow band yet.
I've got most of the gauges created and working. But to do this, I had to basically upgrade my whole package because the older system didn't show graphics well enough in my opinion. I also started down the path of creating a backward compatible version for older laptops with Win 95/98 but stopped that effort.
Another problem is that someone else is already using the term "CoPilot", so I'll have to dream up something else... 
Soon...
Edited by Z-man on 02-06-08 05:25 PM. Reason for edit: No reason given.
|
Twilightoptics
Senior Member
Posts: 739
Loc: Vancouver, WA
Reg: 06-26-04
|
02-06-08 05:36 PM - Post#1360315
In response to Z-man
lol 950-Passenger
'87 IROC-Z 355 HSR C950 Lunati Voodoo Solid 241/249@.050 .555/.576 lift(1.6) 110LSA, Pro Topline Iron 200cc/64cc, T56
'04 Jeep Wrangler "Rocky Mountain" 4.0L/5spd
'82 RX-7 GSL |
|
artic-trans-am
Member
Posts: 3
Reg: 02-05-08
|
02-06-08 11:16 PM - Post#1360521
In response to Twilightoptics
you do a great thing you create!
I look realy forward to test this, I didn't know anyone did this programing for C950.
There is a friend of me that did build a ECU at school. he made software that tid the tuning it self (self learning). I did a copy off that ECU, but I am a idiot with "programing language" so I send it to him for programing. But it look's like he never find time to do it. This was a ECU kit you by from VEMS.
http://www.vems.hu/wiki/index.php?page=GenBoard%2F...
|
DeltaT
"2nd Year" Silver Supporting Member
Posts: 975
Loc: San Jose area, CA, USA
Reg: 06-20-02
|
02-07-08 05:51 PM - Post#1361073
In response to artic-trans-am
GoPilot. Co-Mander. 950 Lieutenant. 949. Gofer. Twilight's Last Gleaming.
There's a lot more where they came from. Send $20 for more, or $50 for no more!
Jim
|
laiky
Member
Posts: 79
Loc: NYC
Reg: 09-25-02
|
02-08-08 06:05 AM - Post#1361397
In response to DeltaT
how does one get a copy of this autotuner software? is it for sale? Laiky@aol.com if it's free. otherwise please explain.
| 1984 Camaro Z-28
383/TKO/HSR/C950
AFR 190's/Comp Roller/SLP 1 3/4 to full SLP System
|
|
Danny Cabral
Member
Posts: 1335
Age: 39
Loc: Ellington, CT
Reg: 11-03-04
|
02-08-08 06:36 AM - Post#1361420
In response to laiky
how does one get a copy of this autotuner software?
It's on Z-man's website: http://temp.corvetteforum.net/c3/zman//CoPilot.htm
May God's Grace Bless You.
'78 BRONCO: 508" stroker, TFS heads, C950 EFI, A/C, Lentech Strip Terminator wide-ratio AOD, 3:1 Atlas II, max built Dana 44/60-lockers-4.10s, hydroboost/4-disc brakes, ram-assist/heim-flip steering, 4" lift, 35" tires |
|
Twilightoptics
Senior Member
Posts: 739
Loc: Vancouver, WA
Reg: 06-26-04
|
02-08-08 07:30 AM - Post#1361457
In response to Danny Cabral
But I thought it wasn't finished?
'87 IROC-Z 355 HSR C950 Lunati Voodoo Solid 241/249@.050 .555/.576 lift(1.6) 110LSA, Pro Topline Iron 200cc/64cc, T56
'04 Jeep Wrangler "Rocky Mountain" 4.0L/5spd
'82 RX-7 GSL |
|
Kingtal0n
Member
Posts: 6
Reg: 04-23-04
|
02-10-08 12:36 PM - Post#1362940
In response to Twilightoptics
I Wrote something similar in the past when holley first released its original C950. without intercepting the datastream the only thing one can really do is watch colors, and change text, which is all you really need to do to begin with. Using a single function I was able to add data to a list, using (itemdata) to capture af values. the array would widen are more af values are logged to (item data) with "," commands between each AF. The line would look something like this:
12,15 (item data = "15.5,15.3,15.7<"
Use (mid) to break them up
for i = 1 to len(itemdata)
x = mid(itemdata, i, 1)
if x = "," then
x = mid(itemdata, 1, i - 1)
total = (total + x) / number of air fuel values variable
this will give you an average A/F for that cell..
Then just interpollate by handing it to a function that grabs the average item data from adjacent cells
cell = (coord.x - i, coord.y - i) for instance in a loop to get surrounding cells, with coord.y being defined as a TYPE.
The result was a 3 hours program that stored and compared a/f values to cell values and adjusted them accordingly. I had max limits set to prevent overadjustment, as well as #adjustments made per session and a function to delete stored array values after a change was made, and several text boxs with witch to adjust the rate at which this all happens for slower computers.
|
November Members
"Become a Supporting Member" 
johnhem (5)
