Could I get a copy of the eeprom and the logs, please? Would be interesting to see, if changes applied by MLV are in about the same range as the average EGO correction calculated by EcmSpy.

We had good success tuning the closed loop that way, but it requires several test runs to get there.

_________________
"It is often said that before you die your life passes before your eyes. It is in fact true. It's called living."
Terry Pratchett

Yeah

_________________
08 Specialized Langster

After riding the S1 around for the last few months, the Uly is clearly lacking in response, and the interesting thing is that I don't think the Uly is that bad at the moment compared to any other XBs, apart from my Firebolt.

So I started with some fuel system cleaner and a TPS reset. The TPS reset didn't change the stored voltage, which on the IB310 is stored at x0012 (low bits) and x0013 (high bits), from 0.073V. The TPS was right.

I set the idle up and blipped the throttle a few times and the engine felt very unresponsive off idle (compared to the HSR42 on the S1), so I looked at the ignition map:


and pushed the ignition timing up by 5 degrees:


This had given me good results on the Firebolt, especially in that area just above idle where every once in a while, you go to pull away and the engine feels like it misses a beat and almost stalls. Note that I changed to 5 deg on the front and rear maps.

I changed the timing live, with the engine running. The engine tells you that it prefers this increase in advance by increasing in speed, such that you need to wind the throttle stop out again to bring the revs back down to where it bounces either side of 1000rpm.

Next things on the list are the OL AFRs to be set, so this weekend I will be strapping the wideband O2 sensors on and logging data. Then after that, I will be playing with the acceleration enrichment - I want to get the Uly wheelying on the throttle like the S1 does...

_________________
08 Specialized Langster

Well, I put a RH scoop back on again to go to Spain worried about 40deg C and stuff, and again, the bike ran worse with it on, so it will be coming off...

_________________
08 Specialized Langster

Pash I run with a Remus and notice a difference with 97 octane Vs 95. Surely the 97 will have an effect on the calorific value and the running temp? have you tested with both?

Octane number is an indication of resistance to knock, it is not related to calorific value.

I can't tell the difference, however that is likely to change when more Ethanol is blended with U95 to make E10 plus assuming they don't add it to super unleaded.

My S1 makes 95hp on supermarket unleaded, it has an XB top end and cams with 10% less lift than an XB. On the same dyno, yours will struggle to make more than 85...

_________________
08 Specialized Langster

Back to logging data again. Wideband sensors went in. Not particularly successful as the connection kept dropping out and towards the end I think I broke my rear wideband sensor, so I ended up with about 8 runs as I stopped to check then restart the log. I stitched the data together into one long run.

I thought I might try and get the WOT set up, so did lots of WOT roll ons. The data ended up looking like this:


So, not looking great, pretty lean everywhere...

But with this we can create a new top line for the fuel maps to drive it to Lambda 0.9:


However, Houston, we have a problem. The maps are made up of 8 bit bytes, so I can't have a number greater than 255 yet I need 288. There are a couple of ways around this:
1. Scale the WOT correction - this is currently set to 110% and would need to be set to 124%
2. Scale the whole map and let the AFV correct it

For now, I will keep the WOT correction and map scaling as they are and just increase those cells to the maximum value of 255...

On another note, from cold, Uly used to run hesitantly for a few minutes before she nearly cut out then recovered (I guess when the lambda sensor started working). I had been meaning to try a fix for a bit, so I modded the table of data that factors the front cylinder fuel demand as a function of rear head temperature as it is a great tool to make the front cylinder richer than stoich at low temperatures.

I started out with:


Which means when the bike is cold it makes the front mixture leaner than the rear. No idea why they would want to do that...

I can understand why it might want to be richer at higher rear cylinder temperatures cos you are guessing that the front cylinder is running colder.

So, I thought maintain what they have in the normal operating temperature range and add a bit for when the engine is cold.

And ended up with:


And when I took the bike for a ride from cold, it ran seamlessly throughout warmup...

_________________
08 Specialized Langster

Surprising, how your wideband data comply to this

(image taken from my airbox tests).

Image data as a list:

Load RPM O2 Accel Corr.
26 2653 0.816 0.8
26 2632 0.826 0.5
31 2614 0.787 1.5
49 2629 0.337 9.8
115 2685 0.782 14
171 2731 0.787 13
223 2773 0.841 13
255 2850 0.836 8.1
255 2901 0.792 2.2
255 2948 0.65 1.1
255 3008 0.445 1.1
255 3081 0.117 0
255 3130 0.088 0
255 3175 0.054 0
255 3229 0.044 0
255 3285 0.054 0
255 3347 0.044 0
255 3396 0.039 0
255 3457 0.044 0
255 3499 0.039 0
255 3554 0.044 0
255 3611 0.039 0
255 3670 0.034 0
255 3705 0.054 0
255 3759 0.044 0
255 3809 0.068 0
255 3857 0.059 0
255 3924 0.684 0
255 3960 0.762 0
255 4000 0.797 0
255 4049 0.797 0
255 4091 0.811 0
255 4139 0.811 0
255 4171 0.826 0
255 4222 0.821 0

The drop in O2 voltage at WOT and 3000-4000 rpm seems typical for all XB12.
Might it be worth a try to fix this with the help of acceleration enrichment?

_________________
"It is often said that before you die your life passes before your eyes. It is in fact true. It's called living."
Terry Pratchett

Maybe, but a lot of that data was collected by rolling on in high gears where I would have thought the accel enrichment should not have been active.

I'll try the corrected map and see what happens. The accel enrichment is there to mimic a carb which has to work over a wide range of air flows with a large venturi. FI can get around this by metering more fuel than can be extracted by the venturi effect at low flows. Saying that, the accel enrichment could help by giving me the delta over 255 that the lambda sensor tells me I need.

I'll also try and use the Bluetooth transponder in preference to the cable I was using yesterday, in a bid to get rid of the transmission time out errors.

Looking forward to the next ride...

_________________
08 Specialized Langster

Bit more analysis to give more of a response to Gunter. This plot shows the lambda when accel enrichment is triggered:

It shows that I did 4 roll ons, but it is far from helpful with the lambda...

The factors applied:


And the AFV. Interestingly the ECM did not decide to add any more fuel in OLL:


And with this AFV being 89%, I can scale the whole map by 0.89 to allow more fuel in the areas where I am limited to 255. I really should have looked at this before

_________________
08 Specialized Langster

vs.


I'm a bit confused here, language wise ...

My intention was, to increase accel enrichment for more wall wetting, so it takes longer to evaporate the surplus of fuel. The ECM does not identify the lean area as an "acceleration condition", but from my understanding it is, because load is (much) higher than drag would require, even in highest gear. Therefore the bike will accelerate until load/drag is balanced or it reaches redline.

_________________
"It is often said that before you die your life passes before your eyes. It is in fact true. It's called living."
Terry Pratchett

A roll on in my book is twisting the throttle wide open in a high gear and low speed. The data is quite consistent once there is no accel enrichment...

Kevmean *may* have a different meaning...

_________________
08 Specialized Langster

_________________
2001 X1 2009 1125CR 2006 Uly X3 2010 Uly a fully Maz'd S1 and a 1991 RS1200 Westwind.

Both maps multiplied by AFV of 0.89 and top lines updated...


And this is where I am:



There is some work still to be done with the rear map as AFV does differ a little on a run:


But a little bit more data mining can polish this up...

_________________
08 Specialized Langster

This weather is rubbish (plus had visitors over the weekend) so I haven't had much chance to do anything, apart from to prepare for the next run...

So, since going out last, I have done the following:

1. Reset the TPS
2. Marked the throttle for 175 TPS so I can do '175 TPS' roll ons
3. Got my Bluetooth dongle working with my Dell D610 laptop (what a PITA that was...) - hopefully get rid of the data drop outs
4. Multiplied the maps by 0.89 (the AFV when I was doing WOT roll ons)
5. Updated the WOT data
6. Set the AFV to 100% - I really had to doing this, but there is a method to the madness
7. Fixed the AFV to 100%, max and min set to 100%


The last two mean that I can set the maps up without any interference with the WOT data.

I also had a look at my lambda sensors, the one in the rear cylinder was playing up with an error code of 8 (it is an Innovate LC-1) which suggested that the sensor is fooked. So, on the safe side, I bought a new one (unused secondhand off EBay for £11.50) with the same type of plug - always a give away. Saved me a small fortune.

It came and when I tried to fit it, I found it wouldn't plug in. The tab to the left of the connector, as you look at it in the photo, was in the way, so out came the file... The rest is history, just like the tab . Suffice to say it fits now...

Anyway, I did a recal on the old lambda sensor and it appears to work OK. The new one calibrated OK too. So, I will use the old one and if it gives me problems, I will fit the new one.

Looking forward to logging some more data on my test strip (A370 heading out of LA) when the weather improves...

_________________
08 Specialized Langster