DDFI-3 tuning with stock o2 sensor

[lowkey]

I reas somewhere that the midpoint on the o2 sensor is something very relevant. The stock calibration puts the mid point at 0.4885v and the tuner lowered it to 0.47v, which is the average of rich and lean voltages. But the point on doing this I dont know yet.

I was wrong before, the goal he was after was in fact to change the target AFR value. I do not yet understand what would be the difference of a short spread VS a wide spread if the midpoint is the same number.

It seems that changing values to shift mid point will change the sensors target AFR. With the modified midpoint value of .47 the target AFR is 15.1, remember the lower the midpoint value the leaner it will read.

Have a look at this chart
http://cherrypicker.tripod.com/id12.html

Using that chart, if we take the stock mid point of 0.4885v the target AFR is reading is 14.8.


You can see the richest value possible to be read by the narrow band is only 14.4 AFR, much to lean to tune WOT.

 

[d_adams]

Dang, I have 4.2 sensors

http://www.innovatemotorsports.com/support/49.php

The new dual wideband from innovate the dlg-1 uses 4.9 sensors. If I didn't already have what I have, I would get that

My question on this particular setup is, does it have only ONE narrow band simulated output or TWO? I almost bought this the other day but decided to do a little more research on them first. 10 phone calls to Innovate went completely unanswered on Friday at 3PM (one hour prior to their advertised closing) so I don't know what the output is. I need 2, otherwise it's wasting money.

 

[AZmidget91]

My question on this particular setup is, does it have only ONE narrow band simulated output or TWO? I almost bought this the other day but decided to do a little more research on them first. 10 phone calls to Innovate went completely unanswered on Friday at 3PM (one hour prior to their advertised closing) so I don't know what the output is. I need 2, otherwise it's wasting money.

Jacob has one. Apparently it is 3 outputs, a digital output for the wide bands, and 2 analog outputs which can simulate the narrowband.

 

[mcaff]

Jacob has one. Apparently it is 3 outputs, a digital output for the wide bands, and 2 analog outputs which can simulate the narrowband.

AZ from this can I assume I can buy one Innovate LC-2 kit plus one spare bosch 4.9 and make everything work together?

 

[mcaff]

I was wrong before, the goal he was after was in fact to change the target AFR value. I do not yet understand what would be the difference of a short spread VS a wide spread if the midpoint is the same number.

It seems that changing values to shift mid point will change the sensors target AFR. With the modified midpoint value of .47 the target AFR is 15.1, remember the lower the midpoint value the leaner it will read.

Have a look at this chart
http://cherrypicker.tripod.com/id12.html

Using that chart, if we take the stock mid point of 0.4885v the target AFR is reading is 14.8.


You can see the richest value possible to be read by the narrow band is only 14.4 AFR, much to lean to tune WOT.

I think I'm too stupid to understand the rational behind this. It doesnt make sense to me changing the midpoint voltage of the sensor. As per the graphic you posted, that would make mixture even leaner as 0.50v corresponds to 1 lambda factor.

 

[lowkey]

I think I'm too stupid to understand the rational behind this. It doesnt make sense to me changing the midpoint voltage of the sensor. As per the graphic you posted, that would make mixture even leaner as 0.50v corresponds to 1 lambda factor.

You are exactly right in your thinking! That is what I'm saying, if you want to bump the fuel up a bit go to a mid point of .5v or higher, this will also effect the open loop fuel bumping it richer across the map as well.

I still don't understand the thinking of what the tuner was trying to accomplish you referenced who changed the mid point to .47v

 

[mcaff]

Mcaff: Thanks My controller accepts 2 sensors and the single gauge has 2 displays.

AZ: I saw that thread a while ago, but didn't worry when Theoctopus replied:

"Looks like it [Innovate] would work fine.

The LC-1 [4.2 sensors] gets close enough for normal use, but it's up to you if you want to spend the extra money on a 4.9 controller."

Now I'm curious which O2 sensors I have! The kit is pretty new, I wonder if they changed what O2 sensors they supply since that thread is from 2011.

Would you mind sharing with us what controller you have? Actually I'd like to know everything you have in your setup as I am about to purchase one for my Uly. If possible I'd install two wideband sensors connected to one controller and from it to the ECM. If this setup gets too complicated I can just leave the stock NB in there to feed the ECM and use the WBs only for tuning.

 

[mcaff]

Found a good discussion around o2 voltage in here:

http://badweatherbikers.com/buell/messages/32777/383786.html?1218319578

 

[mcaff]

Posted by Gunter, one of the guys behind ECMSpy, in another forum:

QUOTE
From the Screamin Eagle manual "Closed Loop Bias adjusts the AFR control point from nominal. Note: Lower values (O2 voltage bias) will cause a leaner AFR. Higher values will cause a richer AFR."
QUOTE

That's pretty much plain bull****. Because it's impossible.

NB-O2 sensors are not made to and do not - I repeat: DO NOT - signal a specific mixture except lean (= O2 partial pressure higher than stoich) or rich (= O2 partial pressure lower than stoich). They won't as they can't. Especially not if it comes to a rich mixture. A NB-O2 sensor will also never ever signal constantly 0.50 volts or some other midpoint voltage , unless it's broken. Period, over and out. We have thousands of miles of wideband logging showing clearly, that there's no relation between NB-O2 voltage and mixture except the qualitative determination of being above or below stoich. Run away from people telling you something else, as they prove only having no clue of what they are talking about.

NB-O2 signal varies with O2 sensor temperature and O2 sensor age. A lean voltage of 0.12 volts (typ.) increases to about 0.20 volts, a rich voltage of 0.75 volts (typ.) drops to 0.65 volt. So it seems only reasonable to use an algorithm which is capable to cope with this situation. Every ECU I know of, evaluates only the voltage jump to trigger a EGO correction: if voltage is above rich threshold (~ 0.58 volts) fuel will be reduced, if voltage is below lean threshold (~ 0.42 volts), fuel will be increased. If voltage is between the rich and lean threshold, the signal will be evaluated as erraneous and be ignored. The impact of moving the "error range" on mixture is neglectable, because
a) NB-O2 sensors do not signal a voltage near the thresholds (unless defective) and
b) the integral factor of the PID regulator is chosen that way, that a voltage jump/mixture change is triggered much earlier.
id073897
Wheelie King

Posts: 236
Joined: Tue Jun 12, 2007 9:29 pm
Location: Frankfurt a. M. / Germany

 

[AZmidget91]

Would you mind sharing with us what controller you have? Actually I'd like to know everything you have in your setup as I am about to purchase one for my Uly. If possible I'd install two wideband sensors connected to one controller and from it to the ECM. If this setup gets too complicated I can just leave the stock NB in there to feed the ECM and use the WBs only for tuning.

I have 2 LC-1's which won't work with the ECM. I had these before I ever had a Buell, so I just use what I have. If I had the extra money or needed a whole new setup, I would get the DLG-1 which uses a LC-2. From what I have read, it comes with the 4.9 sensors, which would work with the ecm. Even if it only had 1 simulated narrowband analog output, it would be fine for a 03-09 Buell, since they only have 1 narrowband O2. Dean is most likely looking for the dual narrowband output for use with EBR motorcycles.

 

[d_adams]

This would be correct. I may try the DLG-1 if I can get someone at Innovate to answer the dang phone.

 

[lowkey]

You are exactly right in your thinking! That is what I'm saying, if you want to bump the fuel up a bit go to a mid point of .5v or higher, this will also effect the open loop fuel bumping it richer across the map as well.

I still don't understand the thinking of what the tuner was trying to accomplish you referenced who changed the mid point to .47v

Found a good discussion around o2 voltage in here:

http://badweatherbikers.com/buell/messages/32777/383786.html?1218319578

The info in this link supports my above statement, seems people have data logged and proven that it does in fact change the values for fueling. I can't say one way or the other for sure as I have never done this or knew anything about it before yesterday. You should be able to verify this by data logging a ride on stock 02 target values with stock map then change target 02 value to .51 and do the same data logged ride. Compare the cell values between the two logs and the .51 log should be asking for more fuel compared to the other. If you do try this report back with your findings.

Posted by Gunter, one of the guys behind ECMSpy, in another forum:

QUOTE
From the Screamin Eagle manual "Closed Loop Bias adjusts the AFR control point from nominal. Note: Lower values (O2 voltage bias) will cause a leaner AFR. Higher values will cause a richer AFR."
QUOTE

That's pretty much plain bull****. Because it's impossible.

NB-O2 sensors are not made to and do not - I repeat: DO NOT - signal a specific mixture except lean (= O2 partial pressure higher than stoich) or rich (= O2 partial pressure lower than stoich). They won't as they can't. Especially not if it comes to a rich mixture. A NB-O2 sensor will also never ever signal constantly 0.50 volts or some other midpoint voltage , unless it's broken. Period, over and out. We have thousands of miles of wideband logging showing clearly, that there's no relation between NB-O2 voltage and mixture except the qualitative determination of being above or below stoich. Run away from people telling you something else, as they prove only having no clue of what they are talking about.

NB-O2 signal varies with O2 sensor temperature and O2 sensor age. A lean voltage of 0.12 volts (typ.) increases to about 0.20 volts, a rich voltage of 0.75 volts (typ.) drops to 0.65 volt. So it seems only reasonable to use an algorithm which is capable to cope with this situation. Every ECU I know of, evaluates only the voltage jump to trigger a EGO correction: if voltage is above rich threshold (~ 0.58 volts) fuel will be reduced, if voltage is below lean threshold (~ 0.42 volts), fuel will be increased. If voltage is between the rich and lean threshold, the signal will be evaluated as erraneous and be ignored. The impact of moving the "error range" on mixture is neglectable, because
a) NB-O2 sensors do not signal a voltage near the thresholds (unless defective) and
b) the integral factor of the PID regulator is chosen that way, that a voltage jump/mixture change is triggered much earlier.
id073897
Wheelie King

Posts: 236
Joined: Tue Jun 12, 2007 9:29 pm
Location: Frankfurt a. M. / Germany

Not sure then... this user was also in the thread you linked above and no such statements were made there. Having had a hand in making ECMspy, you would hope he has insight as to why the values can be modified and what is accomplished by modifying them.

 

[mcaff]

The info in this link supports my above statement, seems people have data logged and proven that it does in fact change the values for fueling. I can't say one way or the other for sure as I have never done this or knew anything about it before yesterday. You should be able to verify this by data logging a ride on stock 02 target values with stock map then change target 02 value to .51 and do the same data logged ride. Compare the cell values between the two logs and the .51 log should be asking for more fuel compared to the other. If you do try this report back with your findings.

I can do that. How exactly should I tweak the O2 voltages on the ECM? Rich and lean upwards, making 0.51v be the new midpoint? Let´s say, increase rich voltage more than I increase lean voltage?

 

[mcaff]

One of the first discussion we had in the begining of this post was about the relevance of the EGO correction numbers in the open loop cells of the map. AZ also explained that EGO was still "there", it was just a matter of the ECM discarding it while in open loop.
However, I have just realized that this info is perhaps wrong. At least that is what I just figured while playing back my last friday´s datalog.
I´ll try to explain you what I saw but it´s quite difficult to me as a non native speaker of english. I´ll do my best.

Check below one screeshot of the datalog, a window that show EGO correction in black, TPS load in blue, RPM in red:


As you can see, TPS is at 212 and climbing...one 1 second later and it goes all the way up to 255 (max). Take a look what happens to EGO from that point on. It "freezes" at 96.900, which was the last correction applied before the WOT condition. And what happens on the History Tables, which we use to apply corrections to the map? Check below:


The 96.900 EGO correction populates the History Table and make us believe the ECM was truly applying this correction at that time, which is not true. It just repeats the 96.900 value because it was the last example of EGO available, but in fact EGO is clearly unavailable at that time.

This made me realize that tuning based on EGO is definetely worth only inside closed loop and absolutely worthless for the open loop. It does not give any idea on what is trully going on there.

And just to give some more subsidies to our discussion regardind the o2 sensor, look what the stock narrow band showed during that exact same time, while EGO freezed at 96.900:



o2 voltage, in red, kept quite steady at around 0.7-0.8v while TPS was full at 255.

 

[d_adams]

And the official answer from innovate is.... NO. No simulated narrow band output at all from the DLG-1 and also, none of their products are weather resistant. Guess I will be looking elsewhere.

 

[d_adams]

Bought another Wego-3 dual wideband today with a remote waterproof display. Pics to follow in a few weeks or so.

 

[lowkey]

Hi guys here we go again discussing ECM things.

I´ve been researching the tuning procedures on Buells for more than a year maybe to feel safe enough to start tuning mine. By the way I ride a 2008 Ulysses completely stock except for a K&N air filter and the exhaust valve locked open.

I managed to go for a great (dataloged) ride last weekend, about 1 hour riding with AFV locked and after that opened the log on ECMSpy and applied the average EGO corrections to both rear and front maps. The bike improved a lot and now when I run my EGO corrections stay pretty closer to 100 than before.

Well this is the first part of the tuning process I think, as it tuned only the closed loop area of the map, plus learn and idle areas. The ECMSpy EGO maps are populated with values for the closed loop area only, so even if I wanted I couldn´t see what happened when bike entered open loop area.

I always read that narrowband stock o2 sensor will not let us tune the open loop area. However, I have noticed that when I load my log file to MLV/VE Analyze it applies changes to map cels that are on open loop area also. Tunerpro is the same, when I dataloged with the acquisition plug-in from TunerPro it populated the History Table (LOAD_RPM_EGO) with values that looked like EGO corrections in open loop as well. I am aware that the ECM does not make EGO correction in open loop, so where does this data come from?

Recently I read a thread here in this forum from 2014 in which our friend ReadyXB explained to another friend that if AFV was limited to 100%, all EGO corrections shown on TunerPro log could be applied to the whole map including open loop.

I would like to understand this in full because sometimes it makes no sense to me. Should I consider these open-loop-area-EGOs and apply them to my map? Or it is better not to touch the stock open loop area and just have it tuned on the closed loop?

Hey not leaving you hanging here, I just have work to do so somewhat delayed. I got a bluetooth dongle from Cole so I can do some of this research myself as well now. I highlighted both programs from your first post of this thread, can you reply with what version of both programs you are running? Did you buy mega log viewer? If you found it can you provide a link? I have ECMspy (not mono version) and TunerproRT, I thought I had megalog but I can't find it.

Also I found some very good info from this link http://www.buellxb.com/forum/showthread.php?5165-ECMspy-megalog-tunning/page8 specifically the first post and page 8. I know you have read over the tuning guide but there is some really good info in there as well. http://www.ecmspy.com/tgv2/guide2.shtml#7.%20Methods The theory in the tuning guide states that if 100% fueling is met this is equal to 14.7:1 AFR's which is the target for the narrow band / closed loop. When open loop is triggered the value goes to 105% with an AFR value of 14.0:1 and WOT is a value of 110% which is 13.5:1 AFR. Also interesting is the narrow band voltage graph in the tuning guide is similar to the one I posted a few posts back except the value for stoich is an almost vertical line (also indicating there is no value adjustment to play with target AFR), you will see this in the linked tuning guide. Most of this info is in the appendix portion.

Also Have a look at this info explaining how the ECU uses closed loop learn to calculate open loop fueling as well as open loop looking at AFV value and NOT the EGO value to make adjustments.

5.1.3.2 Closed Loop Learn
The Closed Loop Learn region is a subset of the Closed Loop region and is where the global correction, the AFV, is calculated to account for all other uncertainties not addressed, these may include; ambient pressure and sensor deterioration. The Closed Loop Learn region occurs at about 40 – 70mph with a steady throttle and is shown in Figure 6.
The fuel is metered the same as in the Closed Loop region, except after 23 iterations of a difference between EGO and AFV, the AFV is reset to equal the EGO correction. Note that the AFV is only calculated in this region when the engine temperature is between [ Calibration Mode Maximum Engine Temperature ] and [ Calibration Mode Minimum Engine Temperature ], see section 18.
Top of Section

5.1.3.3 Open Loop
Open Loop operation occurs outside the Closed Loop and Closed Loop Idle regions, see Figure 6.
Open Loop fuel is metered as in Closed Loop, substituting the AFV instead of EGO correction and applying an open loop factor, [ Open Loop Default Correction ].
The Open Loop region covers the TPS and RPM which are encountered during a transition from Closed Loop to Open Loop WOT and during deceleration. For the former, the AFR needs to provide a smooth transition to WOT and for the latter, the AFR needs to be very lean to ensure the engine returns to idle quickly, to increase the engine braking effect and to reduce the chance of popping and banging in the exhaust.

5.1.3.5 Open Loop Learn Function
A neat little ECM function is the Open Loop Learn.
This uses the O2 sensor to detect rich or lean running conditions and operates in two ways:
If, during a decel, the ECM detects a rich mixture, the AFV is temporarily reduced.
If, at WOT, the ECM detects a lean mixture, the AFV is temporarily increased. This is shown clearly in Figure 7.
For decel operation, the following parameters are applicable:
[ Deceleration Learn Maximum RPM ]
Upper RPM for Open Loop Learn (decel)
[ Deceleration Learn Minimum RPM ]
Lower RPM for Open Loop Learn (decel)
[ Deceleration Learn Minimum Duration ]
Minimum number of cam revs for validity
[ Deceleration Learn Minimum Readings ]
Number of decels where mixture measured as rich before AFV is reduced
For WOT operation, if the mixture is lean, the AFV is increased. The following parameter is applicable:
[ Open Loop Enrichment Delay ]
Time at which mixture is lean before increasing AFV
This function can be enabled/disabled through the [ System Configuration ] Byte, see Figure 8.

Check corresponding figures 6 and 7 for great visual to go along with this data.

 

[mcaff]

Hello buddy.

Well I have Ecmspy for mono as my bike is a 2008 and the older version doesnt work. My TunerPro is the RT version and I downloaded the latest version from its website. Also you have to download the Buell I/O plugin at TunerPro website and definition files for your ECM at ecmspy.com.
Megalogviewer is available at the developers website and yes it's not a free software. You have to pay a license to gain full access. I just kept the demo version which is usable, but lately I havent been usin its VE analyzer for tuning. Thats because it comes with a default AFR table for Buells which is not suitable for the maps, it wants to make it richer than 14.7 inside closed loop.
So because of this I've been datalogging with ECMdroid and just reading the log in ECMspy, found it easier than datalogging with a laptop to be able to run on TunerPro.

 

[mcaff]

Bought another Wego-3 dual wideband today with a remote waterproof display. Pics to follow in a few weeks or so.

Adams would you mind sharing which exact WB you bought?

 

[d_adams]

Sure. Daytona Twin-tec WEGO 3. I went with the smaller version this time, no built in display.

http://www.daytona-twintec.com/WEGO3.html

"111004
WEGO IIID Wide-band Exhaust Gas Oxygen Sensor Interface Kit (Dual channel interface unit with 0-5V analog AFR outputs same as above, except harness length 42" for both channels. Includes two Bosch LSU 4.2 oxygen sensors, and two 18 x 1.5 mm weld nuts)
$488.00"

I also ordered a dual display gauge to mount on the handlebar.

http://www.egaugesplus.com/dual-wideband-air-fuel-ratio.html

I ran the dual display unit on one of my 1190's for a couple of years with no problems. I wanted the smaller one for the SX though, it has to fit under the seat somewhere since I don't have a fairing to hide it behind this time. A small gauge I don't mind, but a huge display, not happening this time around.