www.tdiclub.com

Economy - Longevity - Performance
The #1 Source of TDI Information on the Web!
Forums Articles Links Meets
Orders TDI Club Cards TDIFest 2016 Gone, but not forgotten VAG-Com List Unit Conversions TDIClub Chat Thank You




Go Back   TDIClub Forums > VW TDI Discussion Areas > Upgrades (non TDI Engine related)

Upgrades (non TDI Engine related) The place of handling, lighting and other upgrades that do not relate to the performance or economy of the TDI engine. In other words upgrades to your TDI that don't fit into TDI Fuel Economy & TDI Engine Enhancements.Please note the Performance Disclaimer

Closed Thread
 
Thread Tools
Old July 24th, 2007, 06:05   #16
Occams_Razor
Veteran Member
 
Occams_Razor's Avatar
 
Join Date: Jun 2003
Location: Dayton, Ohio
Default Cluster Mods - Fuel Consumption Fudge Factor #4

Tried to kill two birds with one stone today.

My odometer has been reading about 0.8% short, according to my GPS. In addition, my MFA economy readings have been reading high by about 4%.

So I changed my K-value down 1% (from 3528 to 3493...fewer impulses to the mile will raise the number of miles odo'd per mile traveled), and changed the MFA fudge factor from 100 to 105. The K-value change will set the odometer correct, but will *also* (I guess) raise the error in the MFA by another 1%, so the 105 should bring it down by 5%.

I'll report back in a few tanks and let you all know how it worked.

Interesting note: Now that I've changed the K-value, my cluster coding has changed from 05232 to 05230. The final "2" used to indicate a fixed K-value of 3528. Since there is no coding value for 3493, it's reverted to "0" on its own. "0" usually means that the cluster receives the impulse readings through the CAN-bus.

I have to remember not to recode the cluster with the trailing "2".

--Chris

Nihilator
Veteran Member
__________________
"When you have two competing trading theories which make exactly the same predictions, the one that is simpler is the better & more profitable one."
Occams_Razor is offline  
Old July 24th, 2007, 06:07   #17
Occams_Razor
Veteran Member
 
Occams_Razor's Avatar
 
Join Date: Jun 2003
Location: Dayton, Ohio
Default Cluster Mods - Odometer Correction

Quote:
Quote:
Originally Posted by Nihilator
My odometer has been reading about 0.8% short, according to my GPS. ...

So I changed my K-value down 1% (from 3528 to 3493...fewer impulses to the mile will raise the number of miles odo'd per mile traveled),...

--Chris

After about 100 miles, the error in the odometer had swung to 0.3% in the opposite direction...my car's odometer was reading about 0.3% too high, in comparison to my GPS. So I went back in and changed my K-value to 3503. Another hundred miles and it was within 0.1% of the GPS.

Unfortunately, every time I changed the K-value, it resets the cluster's trip information, so I had to wait until today to start the data collection for the fuel consumption test.

Again, I'll check back in to let you know how it went.

--Chris

Nihilator
Veteran Member
__________________
"When you have two competing trading theories which make exactly the same predictions, the one that is simpler is the better & more profitable one."
Occams_Razor is offline  
Old July 24th, 2007, 06:09   #18
Occams_Razor
Veteran Member
 
Occams_Razor's Avatar
 
Join Date: Jun 2003
Location: Dayton, Ohio
Default Cluster Mods - Speedometer Error Correction

Okay, so we figured out how to adjust the speedometer scale, too. It's really involved at the moment, and I'm sure there's an easier way, but I can now get rid of that infernal 3.5mph speedo error!

http://www.gti-tdi.de/board/thread.php?threadid=2158

--Chris

Nihilator
Veteran Member
__________________
"When you have two competing trading theories which make exactly the same predictions, the one that is simpler is the better & more profitable one."
Occams_Razor is offline  
Old July 24th, 2007, 06:10   #19
Occams_Razor
Veteran Member
 
Occams_Razor's Avatar
 
Join Date: Jun 2003
Location: Dayton, Ohio
Default Cluster Mods - Speedometer Error Correction #2

It's going to be a pain, though. After determining what the new values are supposed to be for the scale's eight two-byte words (not tough), they have to be written to the cluster, as there's no adaptation value for this. Well, maybe there is, but no one's pointed it out yet.

Then I have to recode the cluster to remove the checksum error. But recoding the cluster will wipe out the changes I've made to my K-value, so I'll have to go back and make *that* change again!

Then there's there's 2,000 miles worth of long-term fuel economy accuracy checks.

In the end, though, I'll have a cluster that accumulates miles *accurately* ( <0.05% ), calculates fuel economy *accurately* ( <1% ), and displays speed *accurately* ( <0.5% ). Why VW couldn't do this in the first place is beyond me, but I'll be pleased as punch to get it working the that way. After only 49,500 miles.

Worth noting: I've only filled once since changing the fuel consumption fudge factor, but my trip odometer + distance till empty figure has been consistent at about 430. Hasn't varied much beyond 10 miles from that figure at any point. Noice!

--Chris

Nihilator
Veteran Member
__________________
"When you have two competing trading theories which make exactly the same predictions, the one that is simpler is the better & more profitable one."
Occams_Razor is offline  
Old July 24th, 2007, 06:12   #20
Occams_Razor
Veteran Member
 
Occams_Razor's Avatar
 
Join Date: Jun 2003
Location: Dayton, Ohio
Default Cluster Mods - Speedometer Error Correction - Results!

My speedometer is DEAD ON ACCURATE now. I love it. It was worth the hassle.

--Chris

Nihilator
Veteran Member
__________________
"When you have two competing trading theories which make exactly the same predictions, the one that is simpler is the better & more profitable one."
Occams_Razor is offline  
Old July 24th, 2007, 06:14   #21
Occams_Razor
Veteran Member
 
Occams_Razor's Avatar
 
Join Date: Jun 2003
Location: Dayton, Ohio
Default Cluster Mods - Speedometer Correction - Results! #2

Quote:
Quote:
Originally Posted by Nihilator
Then there's there's 2,000 miles worth of long-term fuel economy accuracy checks.

In the end, though, I'll have a cluster that accumulates miles *accurately* ( <0.05% ), calculates fuel economy *accurately* ( <1% ), and displays speed *accurately* ( <0.5% ). Why VW couldn't do this in the first place is beyond me, but I'll be pleased as punch to get it working the that way. After only 49,500 miles.



--Chris

Update:



After over 1800 miles, everything (odometer, speedometer, fuel economy calcs) is as close as I could possibly hope it to be.

I kinda wish my economy were a bit better (all you diesels, just zip it!), but oxygenated fuel and cold temps are the bane of gassers. It'll be back up to 31 in the spring.

--Chris

Nihilator
Veteran Member
__________________
"When you have two competing trading theories which make exactly the same predictions, the one that is simpler is the better & more profitable one."
Occams_Razor is offline  
Old July 24th, 2007, 06:16   #22
Occams_Razor
Veteran Member
 
Occams_Razor's Avatar
 
Join Date: Jun 2003
Location: Dayton, Ohio
Default Cluster Mods - Linear Coolant Temp Gauge

By the way folks, I now have a linear coolant temp gauge in my MKIV instrument cluster. The gauge as it comes from the factory is programmed to sit at dead center, on 190F, for all coolant temperatures between (about) 182F and 207F. This is certainly reassuring, especially to non-enthusiasts, but I wanted my gauge to show *actual* temperature, dammit!

I found the line-and-a-half of code that specifies the response curve, altered it to be completely linear, and reinserted it. It still needs some tweaking (because at the same time, I went from 1J0 gauge faces to sport edition 1J5 faces), but now my coolant temp gauge will move in response to every change in coolant temperature! Rockin'!

--Chris

Nihilator
Veteran Member
__________________
"When you have two competing trading theories which make exactly the same predictions, the one that is simpler is the better & more profitable one."
Occams_Razor is offline  
Old July 24th, 2007, 06:20   #23
Occams_Razor
Veteran Member
 
Occams_Razor's Avatar
 
Join Date: Jun 2003
Location: Dayton, Ohio
Default Cluster Mods - Linear Coolant Temp Gauge #2

Quote:
Yeah, I want more info on the coolant temp mod. It's always bugged me that the needle doesn't vary a bit once it's pointing staight up, even though I KNOW there's gotta be some variances happening.

Let us know the secrets, GrandMaster Chris!

Nuje
Member
<Sigh...> Fiiiine.

In my cluster (a 1J0 920 926 F), there are two sets of six two-byte words. The first set is located at 0x02D4 to 0x02DF. The second set is at 0x02E0 to 0x02EB. Translated to decimal and plotted against each other (first set of six as x-values, second set as y-values), you get this:



Change the two middle y-values so they're in a straight line with the point before and after them, and you get this:



Translate the decimal values back to hexadecimal, and reinsert them in the cluster code. Done.

It's not perfect. I think the coolant temp sensor has a non-linear response curve itself, because event though I now have a linear curve in the EEPROM, a plot of coolant temp (as measured with VAG-COM) vs. gauge needle location (i.e. lowest end, 1st tic, 2nd tic, etc.) isn't linear. Still have to work on making it *exactly* right, but for now, it's great.

--Chris

Nihilator
Veteran Member
__________________
"When you have two competing trading theories which make exactly the same predictions, the one that is simpler is the better & more profitable one."
Occams_Razor is offline  
Old July 24th, 2007, 06:21   #24
Occams_Razor
Veteran Member
 
Occams_Razor's Avatar
 
Join Date: Jun 2003
Location: Dayton, Ohio
Default Cluster Mods - Warning! Warning!

I hate to say this, because I so want to do it to mine , but these eeprom mods really should only be done by someone who knows what they're doing. If you have to ask a lot of the basic questions about how it's done, that should be a hint that you should find someone to do it for you.

I've screwed around w/my cell phone and there are warnings all over the web sites saying "you're responsible if you hose your phone".

[/soapbox]
Now on with the modding!

SilverGhost
Veteran Member

[IMG]image.php?u=30596&dateline=1133893405[/IMG]
__________________
"When you have two competing trading theories which make exactly the same predictions, the one that is simpler is the better & more profitable one."
Occams_Razor is offline  
Old July 24th, 2007, 06:23   #25
Occams_Razor
Veteran Member
 
Occams_Razor's Avatar
 
Join Date: Jun 2003
Location: Dayton, Ohio
Default Cluster Mods - Linear Coolant Temp Gauge - Explanation

Quote:
Quote:
Originally Posted by Nuje
Ummm....thanks Chris. But, uh...like....umm...what does that mean?

I did a eeprom dump of my cluster (1J5 920 846 C), but how do you know which six "two-byte words" to look for? How do you decide what does what looking at the hex?

From 0x002D0 to 0x0030F, my "words" look like this:
E0 03 10 04 F0 00 90 01 50 02
60 03 E0 03 10 04 10 00 10 00
8C 02 8C 02 8D 04 09 05 87 8A
77 00 78 1E 00 27 88 2F 10 38
98 40 20 49 A8 51 30 5A B8 62
40 6B 0A 00 2A 00 46 00 60 00
78 00 8E 00

Dazed and confused in hex-land....
It looks like the 2-byte words are big-endian format... The bytes have to be swapped then converted to decimal before graphing. Then use the slope-line formula (remember highschool algebra?) to calculate the new points for a linear fit. Convert back to hex and swap the bytes to get the new word values.

Here is a raw EEPROM dump from my 1J5 920 946C cluster starting at address 0x02D4:

F0 00 90 01 50 02 60 03 E0 03 10 04 38 00 38 00 AE 02 AE 02 AA 04 25 05

Here is the processed data showing the before and after mapping:


In other words I'll have to edit the EEPROM dump and change the byte values as follows:

Address Before/After
0x02E4 AE/A9
0x02E5 02/01
0x02E6 AE/B3
0x02E7 02/03

I'll load these values into my cluster in a few days and let you know how it works.

It looks like the mapping may be different on almost all clusters. The one thing that is common is that they all seem to have a flat-spot in the midrange values.

TD

TeleDawg
Veteran Member
__________________
"When you have two competing trading theories which make exactly the same predictions, the one that is simpler is the better & more profitable one."
Occams_Razor is offline  
Old July 24th, 2007, 06:33   #26
Occams_Razor
Veteran Member
 
Occams_Razor's Avatar
 
Join Date: Jun 2003
Location: Dayton, Ohio
Default

Quote:
Quote:
Originally Posted by Nuje
Two questions:
1) How did Chris find out/discern that the 0x02D4 and 0x02E0 areas were the ones that affected the temp. gauge display?
It was an educated guess. I knew roughly what to look for, and where, from previous work with the speedometer and fuel scales. And I have hundreds of EEPROM dumps to compare against each other, so that helped.


Quote:
Quote:
Originally Posted by Nuje
2) Regarding messing with the eeprom: if one were to do something "bad" (write some bogus values), would it fry the cluster? Would it be possible to write back the original values to the changes by simply loading a saved copy of the original eeprom?
It likely would not "fry" the cluster, although it's very slightly possible you could lock yourself out of the cluster and not be able to re-write the original dump to it. Highly unlikely, though...Once I received a cluster that was essentially completely wiped clean of all data from the EEPROM, and after a try or two I was able to write the dump from a similar cluster onto it, making it useful again.

--Chris

Nihilator
Veteran Member
__________________
"When you have two competing trading theories which make exactly the same predictions, the one that is simpler is the better & more profitable one."
Occams_Razor is offline  
Closed Thread


Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 22:48.


Powered by vBulletin® Version 3.8.5
Copyright ©2000 - 2018, Jelsoft Enterprises Ltd.
Copyright - TDIClub Online LTD - 2017
Contact Us | Privacy Statement | Forum Rules | Disclaimer
TDIClub Online Ltd (TDIClub.com) is not affiliated with the VWoA or VWAG and is supported by contributions from viewers like you.
1996 - 2017, All Rights Reserved
Page generated in 0.16601 seconds with 9 queries
[Output: 109.74 Kb. compressed to 93.40 Kb. by saving 16.34 Kb. (14.89%)]