AF-XIED

Moderator: Roger 04 RT

Post Reply
Flashviper
Posts: 33
Joined: 21 Sep 2019, 20:17

AF-XIED

Post by Flashviper »

Hi . Can we have some information on how these dongles work . And what target they can reach .
knutk
Posts: 23
Joined: 23 Sep 2019, 08:47

Re: AF-XIED

Post by knutk »

Since no one has answered yet, I'll give it a shot.

Newer ecu's depend heavily on the O2 sensor for their continuous fine tuning of the fuel delivery.

Looking at the curve for the Narrow Band (NB) O2 sensor, we can clearly see that it offers an output where the voltage output rises with a lowering AFR, thus high voltage means rich, and low voltage means lean.
Image

The manipulator, such as the AF-XIED simply intercepts the signal from the O2 sensor and brings a signal to the ECU where it generally sends a lower voltage to the ECU than what comes from the O2 sensor.
This means that the ECU thinks the mixture is leaner than in reality and riches up the fueling accordingly to compensate.

The long version, for the nerds:
You may think that simply putting in an voltage divider vil do the trick. No such luck...

The ECU is programmed to check the level and quality of the signal from the O2 sensor. It runs a startup check, and if voltage goes below apprx. 0,2V or above 0,95 V it will throw a fault code and go into Open loop (i.e disregard signals from the O2 sensor)

So, then, how does it handle all this:

Here is a typical readout from a stock, unmanipulated ECU:
Image

The curve shows voltage from the O2 sensor that varies between 0,2V and 0,80V, corresponding to AFR between AFR 16 to 14. The timeline is 0,5 seconds between each vertical line, thus the engine runs lean-rich-lean at 1 - 2 seconds intervals. This continuous varying lean-rich - lean causes the engine to feel a bit nervous, especially at low to medium throttle settings. (Moderator: See Roger 04 RT post below.)

Manipulated O2 signal:
Image

The curve above shows maipulation: Blue curve is voltage from the O2 sensor, while the red curve shows what the voltage sendt from the manipulator to ECU. The first dip is during warmup, but there after the manipulation starts.

Take notice to the fact that the blue and red curve follows each other at the top voltages, but ones the voltage starts to drop, at a defined voltage the red curve drops below the blue curve. At this point the manipulator sends a lean signal to ECU. The response from the ECU is to richen the signal, and we can see that the blue curve stops dropping, and starts to rise again. And when reaching a defined voltage, higher than the drop-point, the red curve once again joins the blue curve. And the sequence keeps repeating it self.

By adjusting the drop-point voltage and the rise point voltage it is possible to force the ECU to produce a more even AFR curve.
Image

In the curve above, the voltage from the O2 sensor is 0,9V, corresponding to AFR 13,5 +/-. (Now, the red curve represent O2 signal and blue line signal to ECU)
The manipulator tries to produce a given voltage from the O2 sensor. However, looking at the O2 sensor curve at picture #1, we may see that the NB sensor is not very accurate in this range, and depending on sensor temperature, the AFR for 0,9V may be anything between AFR 13 and 14.

It is possible to push the AFR voltage up to 0,95V without throwing a fault code, corresponding to AFR 13,5 to below 12.

Due to the inaccuracy of the O2 sensor at the higher voltages, manipulating the NB sensor will not be highly accurate.

Testing the benefit of manipulation, it seems like the best way is to make the engine to run rich enough to stop the lean/rich varying. This happens when approaching a 0,8V - 0,85V reading from the O2 sensor corresponding to a steady AFR below 14.

Signal analysis is performed by a micro computer that converts analog signal from O2 sensor, and after adjusting signal, it sends the digital result to an Digital to analouge converter and amplifier and transmits to the ECU.

To conclude, O2 manipulation will not win a race, but it is a simple solution to improve the running of the engine, and may easily be installed and removed DIY.
Last edited by knutk on 06 Oct 2019, 09:48, edited 2 times in total.
knutk
Posts: 23
Joined: 23 Sep 2019, 08:47

Re: AF-XIED

Post by knutk »

admin wrote: 06 Oct 2019, 05:19 Thank you for that excellent explanation, and a heads up for the image size that has now been repaired, you may upload a bigger image now, to be more readable, as we all know here, we like things to be seen clearly and in total transparency. :) :) :) :)
Sorry that I failed to be able to edit my post, so I reposted it and added larger pictures. As of now, I may not edit a post once someone else has made post afterwards.

Feel free to delete my first post, or minimize the text to a point where all the threads still make sense.
I know that the PHP forum does have the ability to allow for editing, but I also know that there are lots of options to choose from, so I regard this as a teething problem :D
User avatar
Bigtune
Site Admin
Posts: 73
Joined: 21 Sep 2019, 14:55

Re: AF-XIED

Post by Bigtune »

All fixed Knut slowly but surely we will get there. :)
Roger 04 RT
Posts: 20
Joined: 22 Sep 2019, 13:54

Re: AF-XIED

Post by Roger 04 RT »

Knutk,

That’s a really good explanation of how it works with great graphics. Simply, the AF-XIED uses a microprocessor to measure the O2 signal, and along the lines of your explanation, changes the point at which it sends a switching signal to the Motronic ECU.

However, the Motronic still varies the fueling, just as much at any AF-XIED setting. Because of the slope of the O2 sensor voltage (shallow at lower AFRs, higher voltages) you just don’t see it on your scope. This can be demonstrated by using an lc-2 wideband O2 to measure AFR while riding.

It is not the varying of fueling that causes surging. I’ll come back with the reason why it does reduce surging later today.
Roger 04 RT
Posts: 20
Joined: 22 Sep 2019, 13:54

Re: AF-XIED

Post by Roger 04 RT »

The principal cause of surging/rough running is AFR and power imbalance and in a two cylinder engine that effect can be pronounced. A good article on this can be found at gami.com, an aviation engine tuner. The article is “Back to the Future”. (I’m also a pilot and aircraft engine tuner.) You can operate quite lean if everything is balanced, without surging.

This occurs at higher AFRs because there is excess air in the cylinder charge, which can be converted to a power difference by any fuel difference between the cylinders. In BMW Oilheads with one tps and one o2 sensor, you can work hard to balance airflow but fuel imbalances cannot be adjusted. When you richen the mixture with an AF-XIED, there is less available O2 and therefore less power difference from fueling differences even though at richer mixtures the fuel is still being varied by the Motronic.

The importance of the above is attested to by BMW’s addition of two O2 sensors on the R1200 and two TPSs and two O2s on the R1250.
User avatar
Bigtune
Site Admin
Posts: 73
Joined: 21 Sep 2019, 14:55

Re: AF-XIED

Post by Bigtune »

Brilliant and to the point as usual gents, always a pleasure to read and learn. :D :D
Roger 04 RT
Posts: 20
Joined: 22 Sep 2019, 13:54

Re: AF-XIED

Post by Roger 04 RT »

There are a few more points:

—In addition to a smoother engine, lambda shifting for richer mixtures also adds a small amount of power throughout the lower 70-80% of the fueling map. A part of the map often ignored by conventional tuners.

—You can achieve a shift in the lambda switching point with either an AF-XIED or Innovate Motorsport LC-2. The LC-2 has a wider range of possible shifting and can go leaner if desired (for economy but who would). It requires programming and calibration.

—When lambda is switched richer, exhaust gas temperature drops due to the cooling effect of the extra fuel.

—Also, richer mixtures mean a cooler catalytic converter due to the lower EGT and because oxygen stored in the cat is depleted and the cat no longer oxidizes CO and unburnt fuel as well.
Roger 04 RT
Posts: 20
Joined: 22 Sep 2019, 13:54

Re: AF-XIED

Post by Roger 04 RT »

Hi Knutk,

“ The ECU is programmed to check the level and quality of the signal from the O2 sensor. It runs a startup check, and if voltage goes below apprx. 0,2V or above 0,95 V it will throw a fault code and go into Open loop (i.e disregard signals from the O2 sensor)”

Can you tell me exactly when in the sequence this test is made? E.G. Is it made immediately at key on, or some hundreds of milliseconds later, or after the start button is pressed?

Thanks,
Roger
knutk
Posts: 23
Joined: 23 Sep 2019, 08:47

Re: AF-XIED

Post by knutk »

Roger 04 RT wrote: 06 Oct 2019, 13:20 .....
However, the Motronic still varies the fueling, just as much at any AF-XIED setting. Because of the slope of the O2 sensor voltage (shallow at lower AFRs, higher voltages) you just don’t see it on your scope. This can be demonstrated by using an lc-2 wideband O2 to measure AFR while riding.

It is not the varying of fueling that causes surging. I’ll come back with the reason why it does reduce surging later today.
Interesting indeed. :D

But how does the ECU Control the varying fueling if it can not see it? If the O2 sensor is too slow to give feedback, do you reckon that the ECU simply runs a lean-rich pattern based on a built in timer that cycles rich-lean blindfolded? And then adjust cycles based on feedback from O2?
knutk
Posts: 23
Joined: 23 Sep 2019, 08:47

Re: AF-XIED

Post by knutk »

Roger 04 RT wrote: 06 Oct 2019, 17:06 Hi Knutk,

“ The ECU is programmed to check the level and quality of the signal from the O2 sensor. It runs a startup check, and if voltage goes below apprx. 0,2V or above 0,95 V it will throw a fault code and go into Open loop (i.e disregard signals from the O2 sensor)”

Can you tell me exactly when in the sequence this test is made? E.G. Is it made immediately at key on, or some hundreds of milliseconds later, or after the start button is pressed?

Thanks,
Roger
When I made my manipulator to work with Euro 4 I ran tests and observed the faultcode. Using ISTA+, the faultcodes describes more details, and ISTA+ also describes the O2 check criteria, and that's where I got the voltage numbers from.
Further, it seemed like the ECU ran a test whenever the engine was started from cold, but I never managed to establish for how long the engine needed to be cold prior to run a test. For various reasons, the engine was allowed to cool off for a couple of hours, and this was enough for the ECU to throw another fault code.

I noticed that there was never sendt any active test signal from the ECU towards the O2 sensor in order to test it, thus , the ECU runs a passive test only, which means it observes the signals from O2 sensor only. As the sensor start out as a device of high resistance, there is no point in starting any test prior to engine start, at which time the O2 sensor starts to warm up.

The curve from the startup of a none-manipulated engine:
Image
The curve is spanning 20 seconds and is compacted in width.
It starts out with ignition off, curve rise with ignition on. Engine start shows a short drop, and then the curve remains steady at the bias voltage of 0,485V until O2 sensor warms up, and once the O2 start to warm up the voltage rise to a AFR Rich condition that seems to be > 0,6V, and from there on engine continues in closed loop.

Here is the AF-XIED startup:
Image

Blue curve is signal from AF-XIED to ECU, red curve is the voltage from O2 sensor, unloaded, i.e not connected to any bias voltage.

Startup as before, but take a look at 16seconds point, where the blue curve makes a drop. The fault code is thrown shortly after this point.
The ECU does not get a direct feedback from O2 sensor confirming it is warmed up, but as a working O2 sensor warms up in less than a minute, I would expect a time window of a couple of minutes at the most for accepting the O2 sensor signal.

Anyway, if the startup sequence is copied, no faultcode is thrown, and it looks like this.
Image

Red curve is signal to ECU while blue signal is voltage from the O2 sensor.
The rise of the curve is slightly different from the stock curve as the O2 sensor voltage not loaded with bias voltage.
I tested by loading the O2 sensor with bias resistors in order to make the curve rise more shallow, as in stock configuration. However, the ECU did not care about the difference, so I dropped it.

So what is the trick to all this:

I use a PIC with built in OpAmps, AD Converters (ADC) and DA converters (DAC). It has the capacity to handle both cylinders.
The manipulator consists of this chip only, a 5V regulator, capacitors for the powersupply and Diode and polyfuse for input protection. O2 sensor is hooked up directly to the PIC, as is the output.
There are also LEDS for Power and functions, and a pushbutton for changing AFR.

During start the output pin is used as an input pin, and it reads the bias voltage from the ECU. This voltage is stored in memory, and the signal from O2 sensor is monitored.
After reading Bias voltage from ECU, the input pin changes mode inside the PIC, and is now reprogrammed to be output, driven by an internal OpAmp buffering the internal DAC. The output is now ready to push the voltage to the ECU, but is allowed to float (acting as it is not connected) until the voltage from O2 sensor starts to rise. As soon as the O2 voltage is a matching the Bias voltage from the ECU, the OpAmp is being activated, and the PIC is now delivering Biasvoltage to the ECU. As the voltage from O2 is rising, so is voltage from PIC, and the ouput replicates the O2 signal until the voltage from O2 sensor has exceeded a minimum voltage above 0,6V and has initiated a drop. After that, the PIC in the manipulator starts to work it's magic. In the graph above this happens at 24 seconds into the recording.
Treetoptuning
Posts: 18
Joined: 22 Sep 2019, 14:07

Re: AF-XIED

Post by Treetoptuning »

This is brilliant info . When we get some time maybe we can cross reference this info in the Bosch Funktionsrahmen. We could translate some pages from that module and send it to both of you .
Roger 04 RT
Posts: 20
Joined: 22 Sep 2019, 13:54

Re: AF-XIED

Post by Roger 04 RT »

knutk wrote: 07 Oct 2019, 00:13
Roger 04 RT wrote: 06 Oct 2019, 13:20 .....
However, the Motronic still varies the fueling, just as much at any AF-XIED setting. Because of the slope of the O2 sensor voltage (shallow at lower AFRs, higher voltages) you just don’t see it on your scope. This can be demonstrated by using an lc-2 wideband O2 to measure AFR while riding.

It is not the varying of fueling that causes surging. I’ll come back with the reason why it does reduce surging later today.
Interesting indeed. :D

But how does the ECU Control the varying fueling if it can not see it? If the O2 sensor is too slow to give feedback, do you reckon that the ECU simply runs a lean-rich pattern based on a built in timer that cycles rich-lean blindfolded? And then adjust cycles based on feedback from O2?
You clearly understand this stuff so I’m surprised by your question. The Motronic or BMSX/K doesn’t see your O2 sensor once you add your AF-XIED-like box, it sees the output of the AF-XIED which still varies high and low, sensitively to the O2 signal. The Motronic does the same thing whether it is looking at an NBO2 or AF-XIED output.

As you know, in closed loop operation the ECU has a convergence algorithm that ramps fueling up or down (depending on the current state of the signal it sees at its O2 inputs) until it sees a 450 mV threshold crossing in one direction or the other, then waits a time, then steps the fueling halfway in the opposite direction (not all ECUs do the half step), then continues to ramp more in the new direction until another threshold crossing is seen, then repeats the cycle in the opposite direction. As a result, once convergence is achieved, fueling varies by about +/- 2% around the O2/AF-XIED setpoint. Eventually short term and long term trims are developed so that: a) the convergence is faster, and b) the trims can be applied to open loop areas of the map (if there are any).

Thanks for all the other information on timing of the O2 integrity test.
knutk
Posts: 23
Joined: 23 Sep 2019, 08:47

Re: AF-XIED

Post by knutk »

Roger 04 RT wrote: 08 Oct 2019, 13:17
knutk wrote: 07 Oct 2019, 00:13
Roger 04 RT wrote: 06 Oct 2019, 13:20 .....
However, the Motronic still varies the fueling, just as much at any AF-XIED setting. Because of the slope of the O2 sensor voltage (shallow at lower AFRs, higher voltages) you just don’t see it on your scope. This can be demonstrated by using an lc-2 wideband O2 to measure AFR while riding.

It is not the varying of fueling that causes surging. I’ll come back with the reason why it does reduce surging later today.
Interesting indeed. :D

But how does the ECU Control the varying fueling if it can not see it? If the O2 sensor is too slow to give feedback, do you reckon that the ECU simply runs a lean-rich pattern based on a built in timer that cycles rich-lean blindfolded? And then adjust cycles based on feedback from O2?
You clearly understand this stuff so I’m surprised by your question. The Motronic or BMSX/K doesn’t see your O2 sensor once you add your AF-XIED-like box, it sees the output of the AF-XIED which still varies high and low, sensitively to the O2 signal. The Motronic does the same thing whether it is looking at an NBO2 or AF-XIED output.

As you know, in closed loop operation the ECU has a convergence algorithm that ramps fueling up or down (depending on the current state of the signal it sees at its O2 inputs) until it sees a 450 mV threshold crossing in one direction or the other, then waits a time, then steps the fueling halfway in the opposite direction (not all ECUs do the half step), then continues to ramp more in the new direction until another threshold crossing is seen, then repeats the cycle in the opposite direction. As a result, once convergence is achieved, fueling varies by about +/- 2% around the O2/AF-XIED setpoint. Eventually short term and long term trims are developed so that: a) the convergence is faster, and b) the trims can be applied to open loop areas of the map (if there are any).

Thanks for all the other information on timing of the O2 integrity test.

My question was related to the statement :
"However, the Motronic still varies the fueling, just as much at any AF-XIED setting. Because of the slope of the O2 sensor voltage (shallow at lower AFRs, higher voltages) you just don’t see it on your scope. This can be demonstrated by using an lc-2 wideband O2 to measure AFR while riding."

The "Motronic still varies fueling, just as much at any AF_XIED setting" had me thinking that the fueling cycles rich-lean even if the NB sensor shows a steady rich (as would be the case for AF-XIED setting #8).
The manipulator transmits several lean pulses to ECU in order to make ECU turn up the fueling. Are you suggesting that ECU will cycle according to the lean pulses?

What I am expecting from the ECU would be as follows:
The fueling from an engine not manipulated shows a certain slowness in response.

Graph from an untouched engine, at idle:
Image

The ECU allows for several 100 mS rich before going lean, and this graph makes me think that the ECU has a built in slowness in the response.
Hence, the statement of varying fueling that varies quicker than possible to be picked up by the O2 sensor would mean that there must be some other means of controlling the fueling in addition to the feedback from O2 sensor, which according to the statement was interpreted by yours truly as visible if using a WB sensor, with it's quicker response time.

At 1200 RPM, close to idle, the engine fires 10 times pr second, meaning that conditions in the exhaust will cycle slightly rich -lean while waiting for the next opening of the exhaust valve. Have you checked the pulsing from the reading from WB sensor vs rpm? could it possible be that the sensor reacts to the pause between each time the valves are opening rather than actually rapid changes made by the ECU?
Roger 04 RT
Posts: 20
Joined: 22 Sep 2019, 13:54

Re: AF-XIED

Post by Roger 04 RT »

You understand the process so I think there's just a bit of miscommunication. Close Loop control by either the BMSK or Motronic relies only of the signal coming from the AF-XIED (or O2 sensor on a stock bike). Even though the WBO2 is very fast, one sees no difference in the cycle rate.

This is literally the sequence:

"As you know, in closed loop operation the ECU has a convergence algorithm that ramps fueling up or down (depending on the current state of the signal it sees at its O2 inputs) until it sees a 450 mV threshold crossing in one direction or the other, then waits a time, then steps the fueling halfway in the opposite direction (not all ECUs do the half step), then continues to ramp more in the new direction until another threshold crossing is seen, then repeats the cycle in the opposite direction. As a result, once convergence is achieved, fueling varies by about +/- 2% around the O2/AF-XIED setpoint."

As a result, the engine does fire several times before changing from rich to lean or lean to rich. The ECU algorithm has a built in delay, and exhaust transit time is a factor. On the R1150GS, its single O2 sensor is located at the front of the catalytic converter between the entry of the left and right cylinder header pipe. That bike has a 5 SECOND PERIOD for Closed Loop cycling at idle. As you advance the throttle, the closed loop period speeds up and eventually hits about a second. Because of the sensor location, that bike is harder to lambda-shift.

Related to your other point, although the NBO2 shows steady a steady signal on AF-XIED 8, the fueling is actually varying just as much as at AF-XIED setting 1 or no AF-XIED at all. The NBO2 output looks steady because it only changes a few 10s of millivolts for a +/-2% fueling change.

I will try and clip a WBO2 output plot that shows lean rich cycling. You can clearly see the fueling ramp up and down even at lambda=0.92 or lower. The cycling speed is unaffected by in magnitude or time, at different lambda settings.

Let me know if I'm unclear or not getting to the heart of your question.
User avatar
Bigtune
Site Admin
Posts: 73
Joined: 21 Sep 2019, 14:55

Re: AF-XIED

Post by Bigtune »

Simply brilliant being able to have the best of both worlds, a real tune by Treetop and also AF-XIED these are 2 amazing ways of getting the bike better fuelled, we all know the wethead runs a little lean due to emission standards so these 2 solutions really do work and sort the problems, no imaginary invisible stuff, will be interesting to see later what the Af-xied guys will do with the 1250, as that bike is near perfect fuelling, so will be an interesting topic. ;)
Post Reply

Return to “AF-XIED”