BBY Fehler 17584 Lambda-Sonde nach Katalysator | Posts 16+

Hello,

Okay... First attempt at updating information with a local VW dealer.

Master's warning beforehand: "I must warn you, this often goes wrong (because it often breaks in the middle - for whatever reason), and then you'll need a new engine control unit for 1000 euros." When asked how often this happens, he gave the answer "3 out of 20" after the third inquiry.

Since it's also possible to reverse the process (and a new flash can be done if necessary with a chip tuner), and you can get these things cheaply on eBay, I decided to take the risk anyway.

But it didn't work, because they hadn't even gotten that far - it didn't even start and gave us an error message, "which we had never seen before". They suspect that the problem lies with the server at VW, and they can try again tomorrow.

I initially rejected their request and asked them to clarify what the error actually means (especially whether it's related to the car or the server), and then I will call them back.

Question 1 to the experts:

Do 3 of 20 tax devices really die during an update due to a failure?

Question 2 to the knowledgeable:

Can the workshop really not reset it in the event of a failure? Normally (see phone), the bootloader and operating software are clearly separated, and the second is flashed using the first - I wouldn't know why one would do it differently with an MSG.

Depending on the answer to question 1 and what I hear from you on the phone tomorrow, I am inclined to have a different dealer try again...

Best regards,

Jan

Hi Jan,

According to my research on the web, the processor in the control unit should be an ST10.
This has an internal serial bootloader, which allows even the actual bootloader of the control unit to be written. So sollte man also nicht zu 100 % wieder hin bekommen - natürlich nicht, wenn das Know-How auf das Klicken eines Buttons beschränkt ist...

Best regards, Rainer

Hi,

...that's how I thought it... but I don't think the relevant pins are being exposed. But regardless, flashing is actually not some kind of magic, but something that is done in many industries.

So the question remains: Did VW do a poor job here, or is it due to that particular workshop?

Best regards,

Jan

Quote:

So bleibt die Frage: Hat VW hier schlecht gearbeitet oder liegt es an der Werkstatt?

My take on this: I can't say that, because I can't assess the workshop's Flash issues. If there are problems with the server that manages the content, then it simply won't work. How likely is that?

Hi,

Moment... we need to distinguish between two things.

Today, it didn't work because, according to their statement (I will find out more tomorrow), the server was not reached. This could probably happen, as it's perfectly normal to do over the internet.

The other statement is that there were 3 out of 20 updates that were likely interrupted mid-way - the reasons were not specified. This could mean a problem with the MSG, it could have been the intern tripping over the cable, or it could have been the workshop cat... whatever. So, as they have presented it, it wouldn't be their fault, but rather that 3 out of 20 fail, which then requires a replacement of the MSG.

Regarding the update itself, I hope that only the check and download happen online initially, and the rest is then loaded onto the diagnostic device and only then flashed.

Best regards,

Jan

I have also already answered in the other post.

That shows how much expertise they have. This device can be flashed without requiring a server connection.

Hello,

Well... true masters of their craft.

If you look closely, it's quite unfair that a software update that fixes an emissions-related issue is, firstly, a paid service, and secondly, that the workshop apparently assumes the customer will pay if the update fails.

Best regards,

Jan

Hi,

Okay... Version 5084 is installed.

Car drives better than before, but:

In the error memory, after 2 x 12 km test drives, the error 17584 (monitor sensor limit reached) reappeared. The engine control lamp did not illuminate, but this also took some time with the old software, despite the error being present. Perhaps the new SW is actually more tolerant at that specific point...

However, there is a difference: I checked the monitor probe using RLF (i.e., the default setting Block 36, applying and releasing the brake). This had initially worked once with the old software, then stopped working, and now it worked again without problems twice in a row (I haven't tried it more often). The voltage of the monitor probe fluctuated by 0.1V and rose to slightly over 0.8V during the test, while it fluctuated by more than the required 2% each time. Therefore, the test was successfully completed, and the voltage fluctuates between 0.8V and 0.1V, which, in my opinion, is exactly correct.

The temperature sensor's heating element is also active according to the display (which should be sufficient for testing), but I'm surprised that despite the active heating element, the resistance value displayed in block 41 doesn't change significantly. The resistance of the upstream sensor is around 70...80 Ohms, while the downstream sensor is at 8-9 Ohms, which presumably corresponds to the cold value. Does anyone have any experience with these values? What resistance should an intact car in block 41 display for the heating element of the after-catalyst sensor?

Readiness is currently not fully complete (the KAT test has not been completed, presumably due to error 17584), but I have temporarily stopped attempting to set the readiness manually due to noise disturbance from the neighborhood.

The goal is now to first observe whether the readiness condition is achieved on its own or can be achieved, and whether the engine control light remains off.

According to RLF, the airtightness in the intake area still needs to be checked - I can also do this thoroughly again, although I'm wondering if this shouldn't also trigger other errors. At the very beginning, I had a problem after cleaning the throttle valve, which led to a whole range of errors, including... Emergency mode.

Regarding the heating of the Nachkatsonde, the above question is interesting. I also want to see if I can measure the voltage at the corresponding pins using my equipment (but I fear that without a ramp, I won't be able to get far, as the connectors for the Nachkatsonde are somewhere in the middle under the car, under the underbody panel). Therefore, I want to do more complex things only when the engine control light comes back on or the readiness is not set.

Best regards,

Jan

Hi,

Update: With the cold exhaust system, the field 3 in the MWB 41 shows a resistance of 4 ohms. After the engine starts, the value quickly rises to the above-mentioned 8...9 ohms. Something seems to be happening there. The VCDS tool tip here isn't very helpful... it says the target value should be 0...0.9kOhm.

Best regards,

Jan

Hello,

Update: Unfortunately, the software update did not help. It took (seemingly) longer for the light to turn on again, but it did so two days ago (driving distance approximately 150 km since the update). The error 17584 is appearing again in memory (it was also appearing shortly after the update).

I tried setting the readiness code again today. All the individual points worked, except for the "Kat" function. Of course, there's also the monitoring probe, and the problem likely lies within its environment.

However, I'm not entirely sure if it might have been due to my operation. VCDS and RLF, in particular, are concealing an important point. One must maintain the engine speed between 2000 and 3000 rpm when the CAT temperature is > 300 degrees, until the test changes from "Test AUS" to "Test AN". After that, it can take several minutes for the test result to appear. However, I haven't even gotten that far... it remained at "Test AUS". The conditions were met... the CAT started at 100 degrees, but after several minutes, I was already above 400. I also re-selected the function and waited for about 90 seconds (engine speed in the specified range) at > 300 degrees, but the test remained unsuccessful. Are there any experiences regarding how long it can take for the test to even begin? By the way, the previously deleted error 17584 has reappeared in this test, but the light did not go out.

The test of the monitor probe (Cooling water > 80 degrees, default setting channel 36, then press brake) continues to be successfully passed. It does take a moment for the Lambda values to shift from around 0.1 (with some fluctuation) to around 0.8 (also with fluctuation), but it does happen.

Does anyone have any ideas?

My further plan is (if there are no better ideas):

- Use brake cleaner to find leaks in the intake area
- Under the car, inspect the lambda sensor connector and try to verschandeln the contacts. Additionally, measure the resistance of the probe heater and the heater voltage during operation - the displayed resistance seems to be strange (see above).

Can a faulty lambda sensor pass the test with basic setting channel 36?

Best regards,

Jan

Hi,

Another update (without any changes or attempts):

On Sunday, the engine light indicator only appeared again 15 km after being turned off. However, and this is interesting, it did so at exactly the same spot where it had happened before (previously with the old SW). The following is how one drives (everything is as described): Approximately 500 m at 50 km/h (1500 rpm), then approximately 100 m at Tempo 30 (similar speed), and then one approaches a traffic light. One needs to stop briefly and then turn right (very confusing, so one should accelerate slowly, especially since there is Tempo 30 again behind). In both cases, I was certain that the light was still off when the car was rolling towards the traffic light, and that it started flashing after the turn.

Frankly, I find this very unlikely to be a loose connection. I need to investigate the possibility of an intake leak as soon as possible. But it's not quite straightforward with the BBY because you have to remove the engine cover (= air intake housing) to get to the air intake bridge - you then need to be careful that no brake cleaner mist goes into the throttle valve.

Does anyone have any further tips or ideas regarding the readiness issue from the previous post?

Best regards,

Jan

Hi,

The brake cleaner used to find leaks in the intake area also didn't work... but maybe my brake cleaner (Nigrin from Bauhaus) is also not suitable for that purpose (although it does at least say that it is highly flammable).

Therefore, I will probably have to try measuring at the connections of the probe itself next...

Does anyone else have any ideas about the issue with the sonde heater resistance?

I had written:

Quote:

Update: With the cold exhaust system, the field 3 in the MWB 41 shows a resistance of 4 ohms. After the engine starts, the value quickly rises to the above-mentioned 8...9 ohms. Something seems to be happening there. The VCDS tool tip here isn't very helpful... it says the target value should be 0...0.9kOhm.

For Probe 1, the resistance is significantly higher... over 100 Ohms, when the motor has been running for a while (the initial value needs to be checked again). However, since the probes are not identical, this doesn't necessarily mean anything. With Sonde 2, I have never seen values above 9 Ohms.

Does anyone have reference values for similar engines?

Best regards,

Jan

Hello everyone,

I've made a step forward: removed the floor panels, disconnected the lambda sondes, and taken some measurements:

Resistance of the heating element measured directly at the connector: 2.6 Ohm (should therefore be okay)
Voltage at the connections of the probe heater while the engine is running and the probe is disconnected: 12V (should therefore also be okay)
Voltage on the sonde's leads on the patient side without a connected sonde: 70 mV (Here, without the sonde, the resting voltage should be 0.4-0.5 V, so this is not okay!)

Therefore, the Lambda probe should be okay, or at least not the problem. That could either be in the lines (a break or short circuit against the heating line) or in the engine control unit itself.

Since I don't have a measuring adapter and would also have to build sufficiently long test leads, I'm stopping here for now. I haven't performed any resistance measurements against the mass, because "up" refers to the engine control unit.

Are there any experiences or data points regarding where this cable harness is likely to experience damage, or are there other locations where it's easy and simple to measure? Or other things to test? Or trivialities like power supply?

Currently, it appears that - if anything - there is more likely a short circuit towards ground, as the engine control unit indicates a voltage of approximately 70 mV (slightly fluctuating) for the sensor 2 in the corresponding measurement block. When there is a cable break, something else should be observed (namely the quiescent voltage of 0.4-0.5 V).

Therefore, a test could look like this: Disconnect the plug from the probe, disconnect both plugs from the engine control unit. Then the wires should be open on both sides, and there should actually be no measurable resistance between any wire pairs. Before I start, I want to gather some ideas so that I don't have to disassemble everything multiple times.

Best regards and thank you very much,

Jan

Hello Jan,

I don't really have an idea for that, but what's wrong with testing whether the lambda sensor needs to be replaced or swapping the control unit for a used part?

I know it would be better to find the error, but you can't really calculate the working time while searching...

Best regards, Rainer

Hi,

Lambda-sensor replacement: Assuming the RLF is accurate (which I assume it is), a new lambda sensor cannot make any changes. After all, you can measure the base voltage without a sensor. Visually, the monitoring sensor looks significantly less deteriorated than the surrounding exhaust system. So, it could very well be that this isn't the first time.

Used MSG can be found for around 50 euros on eBay. I was also considering this, but it brings the problem of updating the software back. This was already a major issue with the current device, and it is definitely not feasible in a car. Therefore, I think checking the lines is the lesser evil (since it's ultimately just measuring). However, if the lines are okay, then a different MSG would be the next attempt. This is exactly what is stated in the RLF: If the basic voltage is correct, replace the lambda sensor. If it is not correct, then check the lines. If everything is okay, then exchange the MSG.

The only thing that is unclear is that I may have mistakenly measured the resistance with the engine off instead of with the engine off and the ignition off. However, IMO this should not make a difference, because the RLF claims that you can detect a break in the circuit when the engine is warm by observing that the base voltage is present (break == probe removed). I can't imagine that the base voltage is present when the engine is off, but not present when the engine is cold, and then reappears when the engine is warm. Various manufacturers of probes state that with a cold probe, only the reference voltage is measured – which I see as an indication that it is always generated.

What kind of cable problem (e.g., frayed or damaged)? The error was initially intermittent, but now it's almost permanent (currently, it's more like it only occurs sporadically, meaning you can perform a diagnostic test).

That the engine control lamp sometimes takes days or even weeks to illuminate, I now understand. Or, I have a theory: The MSG (Motor Control System) can only reliably detect a problem when the engine is warm. Therefore, the error message probably only appears when the coolant reaches 80 degrees. In short-distance operations, this is not always the case (even our high-altitude tour ended with 79 degrees).

Best regards,

Jan

Hi,

Since it was warm today and it's also a holiday in North Rhine-Westphalia (NRW), I took another measurement, this time very precisely according to RLF.

Sonde heating has a cold resistance of 2.5 Ohm -> okay
12 V are present at heating connections when the motor is running -> okay

Test using default settings, Channel 36 -> not okay

The voltage in the MWB 36 is approximately 70 mV, and it fluctuates very slightly.

Okay, here's the translation:

Test Basic Tension

Measure at the connections of the sensor when the probe is disconnected from the cable (coming from the control unit). Should: 0.4-0.5 V.

Is: 60...70 mV.

The value is not 100% constant. When the engine is running, it is more in the range of 70 mV, and when the engine is stationary with the ignition on, it is more in the range of 60 mV. I have measured it when the engine is cold and warm, and when it is running and not running. The results are always the same. Therefore, the initial tension is INCORRECT.

According to the RLF, the line must then be checked. I removed the MSG and then measured the resistance between all four contacts (with the probe removed) as well as against ground. It was always "infinite". After the passage... the two wires have a resistance of less than 1.5 Ohm, as required.

The RLF ends at the point (Ground voltage not okay, wires okay) with "MSG replace".

Does anyone else have any helpful ideas? Could this be due to a faulty mass airflow sensor? The car shows no other noticeable issues, except for this one. I also had a sporadic error message regarding the current flow of the other lambda sensor (before the cat, this is the one after the cat) once. That didn't happen again.

The RLF also identifies air leaks and the probe itself as potential causes for error 17584, however, in both cases, the base voltage must be correct. The suction system does not affect the base pressure, nor does the probe, as it is being withdrawn during the process.

Best regards,

Jan