rusefi.com

All over the internet there are stories about aftermarket wideband kits with their LSU 4.9 sensor failing very fast. Most people are lucky if they last a year. Some are failing after hours. From the information I have gathered on the net it seems that the strategy of driving the heating element is what kills them. Bosch states it clearly in their pdf that the sensor heating should never be turned on before the engine starts. On many occasions condensation is pooling inside an exhaust system and when you first start your car all this water is blown out the pipe. If these water droplets hit the sensor on their way out and the sensor is at it's operating temperature of 780 C then the thermal shock from the rapid contraction of the ceramic sensor element is enough to kill it. Bosch says to always start the heating of the sensor after the engine has started. At the beginning, when the temperature of the exhaust manifold is still below the water dew point, the heater voltage is limited to 2V. After that initial step it goes to 8.5V and from there it rises the voltage with a rate lower than 0.4V/s until it reaches 13V. 13V is the maximum voltage that is allowed and that can only be applied for max 30s. Max continuous voltage is lower than 12V.
What is the heating strategy in rusEFI? Does it start heating up on boot up or it waits for engine start?
Here is a nice read I found http://www.nzefi.com/bosch-lsu-wide-band-airfuel-ratio-lambda-sensors-fail-often-aftermarket-performance-applications/

rusEfi uses duty cycle, not voltage for sensor heating

rusEfi only uses 15% duty max while engine is not running 14v*0.15=2.1v (but it's duty cycle )

Bosch speaks of "effective voltage" in their datasheet that I guess is their way to say PWM driven with a duty cycle that equals to 2V DC. I think it is safer to not drive the heater at all before engine start like they say though. 15% PWM sounds great for warm up after engine start. To make things smoother someone could use a coil and a freewheeling diode to effectively make the drive circuit into a buck regulator and drive the sensor heater with DC current instead of square wave pulses.
In a video where CJ125 board is working with rusEFI I think I saw the duty cycle of the heater being only 3% since the exhaust gasses help to heat the senso. Is that correct? Was that during idle or load?
Edit. Just watched the video again. Seems like the duty cycle is reported as 0 to 1 value with 1=100%. This makes sense now.

How hard would it be to add a heater delay? I would think the heater would be limited to some small duty % if the RPM has not been greater than 500 for least 1 minute and also it should be limited if the engine temp is below something like 40C. Is something like that already a ticker? If not should that be a ticket?

hmmm ...
Vems start heater immediately. And we have no problems with LSU 4.9 sensors.

Too rich AFR will kill sensor also.
If you start to tune your engine, and run on 11.5 AFR / or less, for safety ...
.. then you cut lifetime of your sensor ..

I too know many people that run aftermarket widebands without issues. It's not like this a 100% failure rate but its always better to be safe than sorry. After all Bosch is the designer and manufacturer of the sensor, they should know better. Condensation damage has a lot to do with the car setup. I know for sure that on my car it would never fail like that because I never get condensation and my EGT is over 100 C just 10 seconds after engine start. It would be a totally different story on a turbo Impreza where the sensor is a long way down from the engine and the downpipe helps to make a pool of water just before the sensor. We don't actually gain something with starting the heater before the engine since you don't need short term fuel trim on a stone cold engine on idle anyway.

MHTSOS wrote: ↑

Mon Jun 03, 2019 5:25 am

We don't actually gain something with starting the heater before the engine

I hope you do realize that the engine can be started in an extremely cold weather conditions (like -30 degrees celsius)? And hot exhaust gases can potentially "poison" the sensor if it's not already pre-heated. Let me quote the LC-1 manual: "An un-powered oxygen sensor will be quickly damaged when exposed to hot exhaust gases."
I'm trying to say that we need a compromise: not to heat the sensor too early and too much, and not to leave it stone cold when the engine is running. So 15% preheating seems a good value to me (but of course we can move it to the Tunerstudio settings, if you ever want to tweak it)...

In Bosch technical information datasheet is says this

4.3.2 System measures:
- Never switch on sensor heating resp. control unit before engine start.
- Delay of sensor heater start or reduced heater power (see section 1.6)

The heating strategy is explained in section 1.6. I can't copy the image.

Here https://www.innovatemotorsports.com/support/manual/MTX-L_Manual.pdf it states this

Do not pre-warm the sensor before starting the engine, simply start the engine as normal. Allowing the sensor to pre-warm before starting the engine will increase the possibility of damaging the sensor from shock-cooling.

What you are saying is also correct amd it is stated here http://wbo2.com/lsu/

Pump Cell sensors (like the LSU) should not be exposed to an exhaust stream and left unheated. The internal diffusion chamber will readily clog with carbon deposits that would be "burnt off" in normal operation.
If your wideband unit is disconnected for any length of time (> 10 minutes), you should also remove the sensor and replace the nut in the bung.

If you run the engine for a long time without the sensor heated to its working temp it gets coated with a layer of soot/carbon just like the inside of your exhaust. Then when you eventualy power on the sensor heater this carbon deposits start burning off when it reaches over 500 C and this can damage the sensor.
By having the heater off for the first 10 to 30 seconds after engine start you wont get the sensor covered in carbon. If you have so much unburned carbon in your exhaust gasses then you have much bigger problems to worry about like your spark plugs shorting out all the time. I really think that having the sensor pre warmed is a bad idea.

..I have some time to test cj125 ..
but same results as Abricos ..


Can anybody help ..??

from cj125.cpp .. this is working SPI setting for cj125 controler ..??

I could remember it wrong but was there something about cj125 requiring 12v to communicate? Please post a picture of your setup, please confirm that +12v was there.

You would need a logic analyzer to troubleshoot this further.

I never had cj125 working or documented unfortunately

So now I have connected second cj125 board on ST Discovery, via jumper wires,
connected on "SPI2" ..
and it look's like working one ..

No 12V, no LSU4.2 connected ...

..strange ..

...and first instaled one .. have any problem...

Heater LED is on .. but, no FET, no 12V, no ST discovery, wired frankenstein board only ...

So, small hardware documentation for CJ125 and Frankenstein board.

CJ125 must be connected on SPI2 interface
PB15 - SPI2 MOSI
PB14 - SPI2 MISO
PB13 - SPI2 SCK
PD9 - CS_CJ125

UA and UR must be connected to any analog input, directly to cpu. (no via OP-AMP)
PA3 - UA
PC4 - UR
( .. I'm think so .. )

Heater Pin (DIAHG) set on PB11
(.. in my setup .. )

Minimal +5V pin and GND are required

Would @mck1117, @AndreiKA or @darxfame can tell me ..

How can I setup "any debug mode" .. .??
or AFR analog input for "closed-loop" .. ??

JRD McLAREN wrote: ↑

Mon Jun 10, 2019 8:25 pm

Would @mck1117, @AndreiKA or @darxfame can tell me ..

How can I setup "any debug mode" .. .??
or AFR analog input for "closed-loop" .. ??

What is "any debug mode"? Base Engine -> Base Engine -> Debug Mode -> CJ125? Also https://rusefi.com/wiki/index.php?title=Manual:Debug_fields

Please consider using Tuner Studio, console is not so great to change settings.

These posts are what I'm meaning ...
https://rusefi.com/forum/viewtopic.php?f=4&t=1052&start=90#p29388
https://rusefi.com/forum/viewtopic.php?f=4&t=1052&start=90#p29377
https://rusefi.com/forum/viewtopic.php?f=4&t=1052&start=90#p29375

What is "any debug mode"? Base Engine -> Base Engine -> Debug Mode -> CJ125?

I found it on TS this evening, after I post my question.
So, thanks for answer.

Please consider using Tuner Studio, console is not so great to change settings.

I'm using TS also, but when I can make printscreen of "any settings", it does not matter if it is console or TS ..
( .. or not .. ?? )

I have no idea, how it works ...

Please post a picture of your setup.

I do not recognize debug f15, where is this file coming from?

It looks like cj125 in rusEfi is not ready for end users to be using it

russian wrote: ↑

Tue Jun 11, 2019 7:56 pm

I do not recognize debug f15, where is this file coming from?

Nevermind, this was a defect with how logs were written to SD card. Fixed now https://github.com/rusefi/rusefi/commit/6448de52ba57f01d0c1038f299bf317a0f81a5cf - thank you for uncovering SD card logs issue.

But no sure how to help with CJ125 I know it has only idle output if engine is stopped, warm-up only happens if RPM is above zero.

Yes, is some bug in sd logger ...
In Excel / LibreCalc are many "debug f1" columns ..

and here is packed logs .. http://195.28.90.62/dohc/rusEFI/rus_log11_06_2019_JRD.7z

Just for record, I have LSU4.2 sensor ..

..as I see ...
Heater Led D1 and R21 are "very bad idea" to monitor heater function ..

IT IS UNNEEDED SHORTCUT FOR SERNOR'S HEATER

JRD McLAREN wrote: ↑

Wed Jun 12, 2019 8:33 pm

..as I see ...
Heater Led D1 and R21 are "very bad idea" to monitor heater function ..

IT IS UNNEEDED SHORTCUT FOR SERNOR'S HEATER

Why do you think so? Please articulate some arguments.

What is R21 value? What percentage of current travels via D1+R21 versus the current travelling via the heater? Is not it below 1%?

russian wrote: ↑

Wed Jun 12, 2019 8:52 pm

What is R21 value? What percentage of current travels via D1+R21 versus the current travelling via the heater? Is not it below 1%?

So ,, try to tell to CJ125 Chip ...

We don't need any parallel current to sensor's heater.
DIAHD and DIAHG pins are used for heater (temperature) measurement .. (via heater resistance)

Here are logs from first drive / try, without R21 and LED ..

..
and something is still wrong, with SD logger, some columns are "jumped" ..

Here are some videos from CJ125 controller, with +12 +5 GND connected only.
No heater, no SPI, no LSU sensor ...
Just voltage pins.

+12V rise form 4.5 to 12 .. and as you see ...
RED LED (heater led) rise they lightning ..

when RED LED light, then we have the current ...
http://195.28.90.62/dohc/rusEFI/DSCN4849.AVI
http://195.28.90.62/dohc/rusEFI/DSCN4850.AVI
http://195.28.90.62/dohc/rusEFI/DSCN4852.AVI

WE DON'T NEED ANY CURRENT ON DIAHD and DIAHG PIN, WHEN LSU SENSOR IS DISCONNECTED
WE DON'T NEED ANY PARALLEL / CONCURRENT CURRENT ON LSU HEATER PINS

LED can be connected via Q1 Gate pin and GND...

Please set LED D1 and resistor R21 on REV 0.4.1 as DNP ...

JRD McLAREN wrote: ↑

Thu Jun 13, 2019 7:53 pm

WE DON'T NEED ANY CURRENT ON DIAHD and DIAHG PIN, WHEN LSU SENSOR IS DISCONNECTED
WE DON'T NEED ANY PARALLEL / CONCURRENT CURRENT ON LSU HEATER PINS

YOU DO NOT HAVE TO BE LOUD, YOU NEED TO PROVIDE SOME CLEAR ARGUMENTS PLEASE

Why do you think that the LED is any sort of a problem? Please articulate your vision.

russian wrote: ↑

Fri Jun 14, 2019 1:25 am

Why do you think that the LED is any sort of a problem? Please articulate your vision.

Cause I'm desolder them, and it started to work ..??

Ask yourself, why the led is on, when no LSU is connected. ..??
(5V 12V GND only)
No heater pulse ...
Zero duty cycle.

I'm repeat myself.
We don't need any parallel connection to LSU heater.
We can't have any other parallel connection to LSU

That's it.

How can CJ125 "detect" heater shortcut or cut,
when we have "parallel" LED connection on their pins ..?

JRD McLAREN wrote: ↑

Fri Jun 14, 2019 7:17 am

Ask yourself, why the led is on, when no LSU is connected. ..??

I'm assuming your talking about D1 on CJ125_board schematic. I would think you have a problem with Q1 or the gate drive of Q1. Hmmmm, what is the leakage current of your Q1. What did you use for Q1? AKA a MOSFET will leak some current, which might be enough to make the LED bright, even though there is not enough current to cause any significant heat in the heater.

Hmmm, also why doesn't that have the second schottky diode which would provide a protection against a higher voltage. In this case that's not your problem, but it should be there to make it more rugged.