Jaguar XJR lambda sensor replacement (Upstream bank B)

This page describes diagnosing and changing the upstream bank B lambda sensor in a 1998 Jaguar XJR, and in a 2001 Jaguar XJR. This sensor fails for a blown heater coil far more often than the bank A sensor, probably because it is hidden away in the engine far away from any cooling air flow, and possibly due to Murphy’s law because it’s much harder to get too!

The page also includes loads of other information about Jaguar lambda sensors that I’ve picked up along the way and which you may find useful.

Tools

It turned out that this is a job a starter mechanic can do, as long as you have the right tools for the job. My theory on tools is that having the right one for the job makes things so much easier and more enjoyable, they are always worth investing in if they’re not that expensive. For this job I used:

• Large flat-blade screwdriver
• Small mole grips
• Liquid extractor pump (see later)
• Multi-meter
• Sensor removal tool
• Socket set (8mm, 10mm, and 12mm sockets used)
• Irwin bolt grip set
• Right angle screwdriver attachment
• Breaker bar
• Small wire cutters

None of these tools were particularly expensive, all below £20 apart from the socket set which I already had.

Fault codes

I read the fault codes using an ELM327 wireless DTC reader, and found P1647, which is the upstream O2 sensor B. I fixed the pipe with superglue and silicone sealant, and cleared the DTC with the ELM327, but the P1647 code returned.

Diagnosing faults

A useful way to tell if it is the senor that is faulty is to switch the A and B banks over. The DTC for the A bank sensor is P1646. In front of the bulkhead of the car, you should be able to see the two connectors, one grey (bank B) and one black (bank A) which are right next to each other.

Swap these over. You will need a long screwdriver to take the connector off its mounting, and then disconnect them by hand.

Now clear all the DTCs and drive round for a bit, and at least twice turn off the engine and restart it. If it is the bank B sensor that is faulty, and it usually is, then you should now get a P1646 code instead. If you still get a P1647, then it is probably the engine management ECU or the harness between the ECU and the bank B sensor that is faulty.

Finding the lambda sensor

Since the upstream bank B sensor is buried deep in the engine, it is easier to check the sensor heater coil from the engine management ECU, which is under the cover in the engine bay on the left hand side of the car near the windscreen:

This is held on with anti-tamper screws, but by pushing down hard enough on them they are not too hard to remove. I replaced these with normal screws afterwards for convenience.

Testing for the faulty sensor

The electrical manuals for the 1998 and 2001 XJRs are here:

1998 XJR Electrical System

2001 XJR Electrical System

The most common cause of fault in this sensor is a burned out heater. The wiring diagram is shown below. This shows that the upstream O2 sensor B heater coil is connected to the engine management ECU is between pin EM85-02 and a ground which goes to pin EM85-08 (click on the picture to open it separately).

The guide also shows these connectors:

I disconnected connector EM85 and measured the resistance between these two connections and sure enough it was open circuit.

Buying a new sensor

Now the sensor heater resistance and the sensor’s connector is different between my 1998 XJR and the 2001 XJR, shown in this table:

For interest, you can have a look at the Denso catalogue for lambda sensors which explains quite a few things about the sensors. You can find it here.

Here’s the photo of the DOX-0428 from the Denso catalogue:

Many suppliers have supposedly “inter-changeable” parts. The table below shows various suppliers and their codes that I have found for the 2001 XJR. However, you are taking a risk if you do not refit exactly the same supplier and type as the faulty sensor you took off the car. There are variances in heater resistance, signal, and quality between these. Furthermore, most suppliers will not give a refund if the sensor has been fitted to a car, but of course you don’t know if the sensor will work until you have fitted it to the car!

From my experience, and reading other people’s comments on the forums (www.jaguarforums.com and www.jaguarforum.co.uk) you need to be a bit careful where you get your sensors from. I purchased one from ebay originally for £24. This turned out to have an open circuit heater coil too. The guy kindly sent me another one but this one also was open circuit. In the end he measured all the ones he had which were all open circuit so kindly gave me a refund. Stories from the forums also recount that most second hand sensors seem to have the heater coil open circuit too.

The Gendan part was not stamped with “Denso” on it and I think may be a copy part (although the latest Denso one I have also is not stamped on the part itself). On their website they only say that it is a “Replacement for Denso part 192400-3101”. Although it worked fine for a few miles, I then got a “P0174” DTC with the message “RESTRICTED PERFORMANCE” on the instrument cluster, which I’m guessing is because the sensor is not giving exactly the right signals. I’ve bought a real Denso sensor now and will update this page based on the results.

Therefore I would strongly recommend buying an original Denso part, and the first two rows of the table above are, I believe, Denso parts. Phone and check before buying.

Sensor types

The Denso DOX-0428 is an “A/F” type sensor, which is a little more complex than basic lambda sensors. Denso’s catalogue has the following information about this:

What is the difference between Zirconia Lambda Sensors and A/F Sensors?

In general A/F Sensors are more sensitive and efficient than conventional Zirconia Lambda Sensors. This is because of the way each type of Sensor measures the air/fuel ratio, and the different output signals they produce to indicate the result:

Air/fuel ratio

• A Zirconia Sensor indicates whether the air/fuel ratio is either above or below Lambda 1.00. The engine ECU alters the fuel quantity step by step until the Sensor indicates that the mixture is wrong again. At this point the ECU starts correcting again, step by step in the other direction. This method results in a relatively slow and constant ongoing correction around Lambda 1.00, never able to exactly maintain Lambda 1.00.
• An A/F Sensor indicates the exact air/fuel ratio value. This means that the engine ECU knows how far off the air/fuel ratio is from Lambda 1.00, and therefore also knows how much it needs to correct the fuel injection. This allows the engine ECU to correct the injected fuel quantity to achieve, and maintain Lambda 1.00 almost immediately.

Output Signal

• A Zirconia Sensor outputs a small voltage between 0V and approx 0,8V, switching from low to high at around Lambda 1.00 (Fig. 6).
• An A/F Sensor outputs a small current between -10mA and +10mA, where the output is a stable value, proportional to the air/fuel ratio (Fig. 7).

Overall, especially in changing conditions (sudden accelerations or decelerations) systems with a Zirconia Sensor will have an under- or overshoot in fuel, resulting in a less efficient Catalytic Converter.With an A/F Sensor, the engine ECU will notice fractional changes in air/fuel ratio even in changing conditions. The ECU is therefore able to make precise adjustments, considerably reducing any under- or overshoot in fuel. This results in optimal gas conversion inside the Catalytic Converter, resulting in cleaner air, lower fuel consumption and better drive-ability.

Replacing the sensor

The pictures in this section are from my 1998 XJR. However, the 2001 model is very similar, differing only in the mounting of the sensor connector, and the three screw heads that hold the exhaust shield on.

First, I took off the rubber cover in the centre of the engine bay, and the one on the left hand side of the car, exposing the coolant expansion tank:

First drain the coolant water out of the expansion tank, maybe by using a siphon tube, an old squirty bleach bottle top (the ones with the trigger), a turkey baster, or a liquid extractor pump (best):

These are only £14 from www.thisisitstores.co.uk and are very useful. I recommend it.

Once as much water is removed as possible, next disconnect the plastic water pipes shown by arrows in the photo. These should be squashed either side with a pair of pliers (or strong fingers) and pulled off. Be careful with them because the plastic ring at the end of the connector can easily break. Next undo the two 10mm nuts that hold the expansion tank in place. These are easy to find.

Next you need to remove the object shown in the next photo. I’m not sure what this is, maybe a valve of some sort, but it has four large rubber hoses going into it, two each side, and an electrical connector. First remove the hoses. To do this you will need to use a mole grip to squash the hose clamps together to hold them loose, then pry off the hose with a screwdriver. A squirt of WD40 first will help the hose come off. The upper arrow shows one of the hose clips, the lower arrow where the hose fits onto the object.

It’s worthwhile at this point putting rubber bands on the two upper pipes each side to make sure they don’t get confused with the lower pipes when you come to re-assemble.

Then undo the three 10mm nuts that hold this thing in place. The nuts are screwed onto threads on soft rubber mounts which allow the object to move around a little. The photo above shows the nuts already removed. There are two at the top and one at the bottom. They are quite easy to access and remove once the hoses are bent out of the way.

There is another pipe which is best to remove where it joins the side of the car (which will be nearest where you are working at the side of the car). This is quite easy to remove and will make the job easier to get it out of the way. It’s likely that quite a bit of coolant will spill out when you remove this one so be prepared!

The exhaust heat-shield is now exposed:

This is just a shaped steel plate fixed onto the exhaust downpipe with the three screws that you can see in the photo. On the 2001 XJR these are Torx screws, but are hexagonal head (see photo) on the 1998 XJR. A right angle screwdriver attachment like the one pictured is useful in the restricted space available.

These screws go into the exhaust, and the heat can cause them to seize. For the 1998 car with standard hexagonal head, a set of Irwin bolt grippers does the job on the rounded screws:

The heat-shield then just lifts off. The plate that held the lambda sensor on is easy to remove by undoing the two 12mm nuts, although it may not be necessary to, but you may need all the space you can get to undo the sensor:

The sensor can now finally be seen screwed into the exhaust downpipe:

In the photo above it is almost undone. Squirt some more WD40 on it to loosen it up, then use a sensor removal tool to undo it. A sensor removal tool is essential for this job:

I bought a set (two) for about £16. I used a breaker bar on the sensor tool. You can’t turn much angle with the breaker bar, but you only need to get it going.

On the 1998 XJR, the sensor’s cable has a cable tie about half way along its length which is easy enough to cut off with wire cutters, albeit quite a long way down into the engine bay. On the 2001 XJR the cable goes straight to a mounting plate in front of the bulkhead.

The other end of the cable terminates in a grey connector. This can be slightly tricky to disconnect, but not too bad.

Once that grey connector is disconnected the whole sensor can be removed. Note that the 1998 and 2001 connectors are reversed from each other. The 1998 sensor has a male connector and the 2001 sensor has a female, so you can’t fit the wrong type of sensor.

Re-assembly

The new sensor should have a little copper grease wiped around its thread (VERY IMPORTANT: make sure none gets on the sensing end). Denso sensors supposedly come with a little tube of copper grease. It should be tightened up to 40-50Nm according to the XJ308 Workshop Manual, or 20Nm according to the Denso fitting sheet – I suggest 20Nm! Use a torque wrench if possible (if you can get it in there). I couldn’t so just tightened it hand-tight using the leverage of the sensor removal tube. Make sure the cable runs along a path well away from the exhaust otherwise it could melt through the insulation.

When topping up the coolant, you will need to add a little more since you will have lost a bit when disconnecting the pipes. Fill it up slowly because it takes a while to trickle down into the empty pipes and will overflow if you go too fast.

Conclusion

Use the DTC reading tool to clear the P1647 code, or do a full reset (disconnect battery and connect the leads which went to the battery together for a few seconds – DON’T SHORT OUT THE BATTERY!!!), and it’s done.

All in all I’d say it took me about 4 hours to do the whole job (the first time), it’s really not that hard even for a very raw mechanic like me.

That leaves buying a new Denso one. The table below summarises my experiences:.