odd wideband behaviour

[rac]

I'm getting some odd behaviour that just comes and goes randomly. I have 2x LSU 4.9s set right next two each other, just in case one failed I could switch which bank I was reading from.

1. Sometimes one, or rarely both sensors read lambda 1 (default sensor fail value is not set at this value), and then randomly start working.
2. Sometimes one responds much much slower relative to the other one.
3. Sometimes they both do something odd for no apparent reason (just deviate while idling and no fuelling changes).
4. Their behaviour with ignition on differs on a different day as well, sometimes one or both go ultra lean prior to cranking, while other times they stay around lamda ~1.

What are the symptoms of dodgy sensor?
Much has been discussed about lsu 4.9 heating strategies. It appears the heating strategy is on when ignition is on, so if I spent too much time working with the ecu while the engine is cold pre-start, is it possible the sensors get heated up prematurely then are damaged from condensation?

[rac]

cant upload logs;

"Sorry, the board attachment quota has been reached."

files were only 270KB? what is the limit?

[RICE RACING]

It appears the heating strategy is on when ignition is on, so if I spent too much time working with the ecu while the engine is cold pre-start, is it possible the sensors get heated up prematurely then are damaged from condensation?

On a S8 this is how long the heater is on. This is on a NTK sensor, the test right after that is another NTK but driven by an external controller. Around 30 seconds. The red is another NTK stand alone meter powered by PDUx3.

[TimH]

LSU 4.9 and NTK heating strategies are very different, and the NTK is robust and very difficult to kill. The LSU 4.9 is the lambda sensor equivalent of an Impreza 5 speed gearbox...made of chocolate

The LSU 4.9's heater should not be on if the engine isn't running and once the engine is running there are very strict guidelines on how to ramp up the heater volts during the "condensation phase". There are dire warnings throughout the sensor datasheet concerning failure mechanisms.

I'm guessing this is on an S6+ or S7, but I have no experience of using the LSU's with these as I've always suggested NTKs to customers for whom I've done looms. But I do know I killed 3 or 4 LSU 4.9's while developing software for one of our products! When they failed they did give very odd readings but they never recovered - but the software was "green" at the time so I absolutely wrote them off lol.

The fact that the behaviour you see is random suggests, to me, either suspect wiring or sensor(s) that are failing/intermittent.

[rac]

Thanks Tim.

The heating being on with ignition on has always bothered me, there should be an enable with run mode option.
Its an S6+ pnp, the provided cal drives the heaters at ~47% duty at "0" seconds declining to ~40% duty once they are warmed up (which is about 30 seconds also like Peter's examples). The heater calibration is plotted against run time so it stays on the run time = 0 axis until started.

My sensors are also well down from the turbo hotside so there would be quite some delay for the exhaust temps to come up.

I have often had the ignition on for well over 30 seconds uploading new data logging or calibration files etc., or fighting the usb/ethernet connection issues (which I have recently bypassed by using the wifi).

Not sure I understand why bosch specify 1.5v during "condensation water phase" and then immediate step change to 8.5v at t = 0.
Why have any voltage applied at all until cranking? Is it a function of minimum possible voltage from the controller?

I might experiment with 0 duty cycle at time = 0, and have a gentle increase in duty over the course of 1min or more.
Is there any downside to having an extended warm up duration? sensor fouling couldn't be an issue in a few minutes of running, and I don't really need to be closed loop in the first minute either.

my understanding is the ecu itself is solder bridge set to run the lsu 4.9s so switching to ntk's is out of my skill set.

Rod.

[TimH]

47% duty sounds very high - data sheet calls for an effective heater voltage of <2V during the condensation phase. If Vbatt is 12.5V at that time then the duty should be around 2% if I've calculated it right.

47% is around 8.5V for the same battery voltage which is where it should start after the condensation phase and then ramp up in a controlled manner to a maximum of 13V (effective) for no more than 30s, and then drop back under PID control to maintain the correct temperature.

The $1m question is how long the condensation phase is and this may be where the LSU 4.9 "suffers" when used as an aftermarket wideband is it probably needs to be characterised on a per-engine, per-exhaust system, basis.

My take is that the heater should not be fired up AT ALL until the engine is running (Link CAN lambda, for example, receives engine RPM and won't start the heater until the engine is running) and then keep it at that 2% duty for maybe 8-10seconds?

Random websearch for the sensor spec:

http://www.industry4zero.it/index.php/i ... -info/file

[rac]

Yep, that is the $1m question. Given most oem applications are short headers close to the engine and heat quickly, even if post turbo. Compared to many aftermarket applications with long runners, low points for condensation collection and long down pipes post turbo hotside before getting somewhere you can physically put a sensor practically.

Might see what I can do to measure egt in the same position from cold start to get an idea if I need to consider this delay seconds or minutes.

[rac]

I've ordered some thermocouple bits and pieces to log both the egt and the pipe surface temperature at the wideband install location. will be interesting to see what the environment is like at cold start.

[TimH]

I've ordered some thermocouple bits and pieces to log both the egt and the pipe surface temperature at the wideband install location. will be interesting to see what the environment is like at cold start.

Will be very interested in the results of this - please share if you're willing

Makes me wonder about enhancing our mini expansion unit (MEU) to make use of the EGT circuitry as part of the LSU control strategy - perhaps using a strap-on sensor to give an indicative reading

[RICE RACING]

I despise Bosch LSU's they are garbage
Was forced to make them work back when Innovate Motorsports made the first LM1 meter but honestly a bigger piece of shit sensor you will not find!
Best place for them and the company is in the rubbish bin!

NTK is 1 million times better