Simon's new Xantia V6 and Leaf blog

[Mandrake]

Ok ES9J4 owners, I have a little homework assignment for you.

The next few times you drive your car, try the following when starting it:

Turn the key fully on but don't start. Wait a couple of seconds then fully depress the throttle for at least a second then release it. Now start the car.

Take careful note of any change in throttle "responsiveness" during initial driving after the startup, especially light throttle pickup at lower speeds.

I've been trying this on mine and the result is quite surprising, suggesting that the ECU is not correctly self calibrating the TPS operating range without a little "help".

After a morning start, especially if I drive the car gently it often feels almost as if it is sluggish/unresponsive to the throttle or is being "held back" then after a while it will come right.

The TPS on this engine, like many is "self calibrating", where a zero/idle calibration is done every time you turn on the ignition, and a range/full throttle recalibration is done on the fly during driving when full throttle is detected, which means TPS calibration can potentially change in the middle of driving.

Depressing the throttle fully with the key on but engine stopped forces a full range calibration of the throttle for that "key on session" so that the ECU is forced to calibrate idle and full throttle immediately.

Any significant difference in throttle response during driving after doing this pre-start calibration suggests a calibration problem and perhaps a worn TPS. (The TPS has not been changed on this car)

Anyway I'd be interested to see if anyone else notices a difference doing this.

Just a follow up on this.

The issue of intermittent sluggish throttle response especially after a cold start has continued for the last 3 months, what I found was that if I very quickly floored the throttle and released it soon after starting the engine or while first driving away that it would usually "rectify" the problem for a while and make the throttle more responsive again. Without doing this it would feel decidedly flat and unresponsive to the throttle and with a significant throttle lag sometimes, so it became a habit to always do this very quick press to the floor of the throttle to "clear it up" after starting the car.

I've been suspicious all along that I was seeing classic symptoms of a "scratchy", worn out carbon track potentiometer, which is all the TPS is. On a scratchy potentiometer twiddling it quickly from end to end sometimes helps it temporarily and this is all that flooring and releasing the throttle quickly would have been doing. Sometimes contact cleaner can help for a while (although i think its a sealed unit that you couldn't get any into anyway) but if the track is physically worn out replacing it is the only option.

So on Sunday I finally fitted a new Intermotor TPS for the princely sum of £11. It turns out to be the best £11 I ever spent on this car.

The intermittent sluggish, laggy throttle response is completely and utterly gone. It has totally transformed the throttle response to how I think it should be. So much so that in these cold mornings with dry salty roads I'm constantly finding myself spinning the inside wheel when I try to accelerate away left or right from a T-junction as the power is instant and strong right from idle. I really have to moderate the throttle carefully on an angled take off so as not to spin the inside wheel. And this is with good almost new Michelins. This is how the V6 should respond.

Pulling below 2500 is now very strong and the throttle response is very sharp almost instantaneous, and best of all it seems to be consistently good all the time now whereas before it was fairly good most of the time (especially if I had done the daily throttle flick) but had spells where it felt a bit sluggish or where there was a large (up to half a second) delay in throttle response. I do about 20 miles a day so I've done 60 miles since Sunday now and it has been flawless the whole time.

When I changed the TPS I made sure not to disconnect the battery, nor did I turn on the key while the TPS and air intake temperature sensors were disconnected to avoid any fault codes being set - basically anything that might cause a false comparison due to an ECU reset or a fault being logged. I didn't want any uncontrolled variables ruining my comparison of old TPS vs new. I've also stopped flicking the throttle after starting the car as it shouldn't need that anymore and I wanted to confirm that the throttle was still responsive without having to do that.

The improvement after replacing the TPS was immediate and overall performance seems to have further improved over the following couple of days as well. So I'm very happy indeed.

It makes sense when you think about it - the TPS and the MAP sensor work together. The MAP sensor controls the overall fuelling based on the air pressure (and engine RPM and MAP tables) however the MAP sensor by its nature is slow to respond - the ECU has to filter out the pressure fluctuations caused by individual cylinders opening and closing their intake valves so there is a certain degree of low pass filtering/smoothing applied to the MAP reading before it is processed by the ECU.

So if the MAP sensor was the only input the ECU would be slow to react to the throttle being opened quickly which would cause a delay in power response and also could cause a lean hesitation due to the mixture not being enriched quickly enough. The lean hesitation/misfire can be detected as pinging by the knock sensor which may pull the timing back causing further loss of power. So the MAP sensor is "reactive" and responds to changes after they have occurred.

The TPS on the other hand is "predictive" in the sense that if you snap the throttle open quickly the TPS will register that change in throttle plate position instantly with only a small sampling delay in the ECU, and before the airflow has actually had a chance to increase since air being sucked into the engine has inertia. So the TPS movement gives the ECU early warning that it needs to enrich the mixture RIGHT NOW to avoid a lean hesitation, which the ECU does by giving the injectors an additional "out of cycle" injection pulse. This happens before the airflow gets a chance to build up thus the lean hesitation is avoided.

One other improvement I notice is that the gearbox kickdown seemed a bit intermittent sometimes - it would sometimes be very stubborn about kicking down, I might be doing 35mph in top gear at about 1500rpm, come across a hill, push the throttle down over half way and it would still not kick down.... Now the kickdown seems a lot more sensitive and responsive than before. So I can only assume a poor TPS reading was causing issues with gearbox kickdown as well, as the gearbox ECU does take the throttle position information from the engine ECU to use to work out when to kickdown and when to change up.

Anyway I'm very happy with the performance now - one by one I've fixed a few niggles here and there including coil pack etc and performance has got better but it is really top notch now. A carbon potentiometer has a limited lifetime and should really be considered worn out after 20 years in a hostile environment like an engine bay and for the sake of £11 to replace it it might well be a worthwhile thing to do to make sure your V6 is running in tip top shape.

The TPS didn't fail outright nor did it ever set any fault codes, however the difference in behaviour and performance before and after is undeniable.

The only thing I would warn about in doing this job is that it is a bit of a pig to get out unless you have the correct Allen key shaped Torx driver due to the close proximity of the LHM tank. I managed to struggle by without one using an Allen bit from my socket set turned with a spanner but it was very awkward, and I found that both bolts were tight nearly all the way out and back in again despite adding grease. So the correct Allen torx key is definitely recommended! Then it would be an easy 10 minute job.

[Skull]

£11 well spent ... sounds like a worthwhile investment for a scheduled service

[CitroJim]

Well done Simon I'd never have expected a worn TPS pot too have such a dramatic effect. An excellent piece of research!

Mind you, there was a similar problem with the Flapper AFM on the LU2-Jetronic system as fitted to the 205GTi... That was just a big pot and when it was worn and scratchy it could cause no end of strange problems.

And knowing all about old, warn crackly pots in vintage electronics you're dead right about the quick fix of rapidly rotating them from end to end helps them for a while... I have an amp where I have to do that regularly as I can't be bothered to replace the pots...

Would it be possible to spray the track of the TPS with a bit of Servisol or is it fully sealed?

Makes me want to check mine on the Activa now...

[Hell Razor5543]

Makes me want to check mine on the Activa now...

Be careful doing that Jim! You might find out what the 0 - 60 MPH time really is!

[CitroJim]

Makes me want to check mine on the Activa now...

Be careful doing that Jim! You might find out what the 0 - 60 MPH time really is!

That's true James Even now it goes rather well with the Manual Boost Controller... It has quite a bit of pep as a result and very pleasant it is too...

[white exec]

Simon,

If you look at the Hydractive 2 Consultant Technician B7 Training document, in the section D - Accelerator Pedal Travel Sensor, there is a detailed description of how the (suspension, in this case) ECU reads the 'pedal pot', both at IGN switch-on and starting, and as the car begins to be driven, modifying its reading of the sensor as it goes along.

This tallies with much of what you wrote above.

Although the XM 2.5 throttling (injection ECU) is controlled by a separate pot (at the end of a throttle cable), it wouldn't surprise me if the operation were similar.

[Mandrake]

Well done Simon I'd never have expected a worn TPS pot too have such a dramatic effect. An excellent piece of research!

Me either Jim. I was pretty sure it was a little bit iffy but didn't expect such an obvious change nor expect the intermittent symptoms to vanish completely. I did replace the TPS on my old V6 as well when it was having problems (albeit with a Bosch not an intermotor) and I seem to recall that it didn't really make any difference there, but with all the other issues that car had with the engine and gearbox I'm not surprised it wasn't a miracle cure...

Would it be possible to spray the track of the TPS with a bit of Servisol or is it fully sealed?

I think it's sealed although that's not to say it's liquid tight and that you might not be able to get something into it around where the shaft goes in. However I'd have to take it off to do that anyway.

For £11 new and with the original one 19 years old and under suspicion I felt it was better to just replace it that way if there was no change I could be confident that it wasn't faulty.

Simon,

If you look at the Hydractive 2 Consultant Technician B7 Training document, in the section D - Accelerator Pedal Travel Sensor, there is a detailed description of how the (suspension, in this case) ECU reads the 'pedal pot', both at IGN switch-on and starting, and as the car begins to be driven, modifying its reading of the sensor as it goes along.

This tallies with much of what you wrote above.

Although the XM 2.5 throttling (injection ECU) is controlled by a separate pot (at the end of a throttle cable), it wouldn't surprise me if the operation were similar.

Yes, the automatic calibration heuristics are quite interesting.

On the TPS closed throttle is typically about 0.6 to 0.7 volts and full throttle about 4.5 volts with a supply voltage to the potentiometer of 5v. When you turn the key on the current voltage reading is immediately calibrated as a fully closed throttle, but only if the voltage is below about 0.9 volts. If it's above about 0.9 volts (I don't actually remember the exact figure) then a closed throttle calibration is NOT done and the previously stored value is used.

This is to prevent it from falsely calibrating the closed throttle position if you turned the key on with the throttle pressed, but give it enough leeway to calibrate for small differences in potentiometer and throttle plate/limit screw adjustment. So anything less than about 0.9v at key on will be accepted as the closed throttle position and trigger a calibration.

You can actually observe this behaviour with a Lexia - turn on the key with the pedal not pressed, take note of the TPS voltage - probably about 0.66v, and the throttle opening percentage, which will be 0%. Press the throttle slightly until it increases to 0.8v and note the opening percentage which will be something like 5%.

Now turn the key off and turn it back on with the throttle held exactly where it is - you'll now see a TPS voltage of 0.8v as before but a throttle opening percentage of 0% Now increase the throttle slowly and you will see both voltage and percentage increase. Reduce it slowly and it will reach 0% at 0.8v again. However now release the throttle and it will drop back to 0.66v and still say 0%, now increase the throttle to 0.8v again and now it will say 5% (or whatever) again instead of 0% - it has done an "on the fly re-calibration" while driving the car due to a voltage outside the previously established minimum and maximum limits being encountered.

You can use this technique to discover what the highest voltage it will accept as a closed throttle is. Beyond a certain voltage at key on the reading will immediately have a non zero indicated throttle opening percentage which indicates it did not try to calibrate this voltage as closed throttle.

This is where one of the problems occurs with a scratchy TPS potentiometer - a poor contact between wiper and track can cause the wiper to go momentarily open circuit as it turns due to dead spots, a lack of tension on the contact, or play in the pivot that allows the wiper to lift off the track. This momentary loss of connection will cause the voltage to drop to 0 and trigger an on the fly closed throttle re-calibration to 0v (or as near to 0v as the A/D converter can digitise) which screws up the calibration. When it comes right again 0.66v is now considered to be a partially open throttle instead of a closed throttle and this throws everything out.

The on the fly calibration process is one way only - it will widen the accepted range in response to a voltage outside the previous limits but it will not shrink the limits again. The only way it will shrink them again is by doing a key off/on cycle with the throttle released so it can re-learn the correct throttle closed calibration. This is why sometimes pulling over and turning the engine off then starting it again will temporarily solve this problem as it forces a valid re-calibration.

So anything that causes the voltage to go outside the normal 0.6v to 4.5v range (or whatever the values are for your specific car) even for a fraction of a second will mess things up until the next time the key is turned on.

[white exec]

A really clear analysis, Simon. Explains the variable performance from start-up you described earlier.
Twenty years of what could be almost daily scraping is a lot to expect from a carbon track.

[Mandrake]

Last night after driving home I decided to put the car on the Lexia to check for fault codes and I also wanted to do an auto-adaptive reset on the gearbox.

While I was poking around I discovered the measured air intake temperature on the engine was 65 degrees C. Given that the outside temperature was 12 degrees at the time I was a bit surprised. Of course the air filter box does get quite hot in the engine bay, so its normal for the air intake temperature reading to get up to about 30-40 degrees when the engine bay is hot since the hot filter box and piping does warm the air, and I have seen readings around this figure on the old V6, but I have never seen anything approaching 65 degrees on the other car.

Out of curiosity I got my IR temperature gun and measured the outside of the black plastic air filter housing in the vicinity of the air intake temperature sensor and sure enough, the air filter box casing was close to 65 degrees.

Here's the thing though - there is no way that the 12 degree air flowing through the intake is reaching 65 degrees even though the box is that hot - so the sensor seems to be measuring the temperature of the box instead of the airflow.

I then remembered that the replacement temperature sensor I fitted some time ago had one key design difference to the original:



The original with the broken clips is on the left, the replacement on the right. Notice the difference ?

The original sensor tip protrudes further into the air stream but more importantly the actual sensor itself - the gold coloured stem in the middle is directly exposed to the air flow. So whilst it would tend to measure the heat radiated from the box if there was no air flow, if there is a significant amount of airflow the sensor element should be cooled much closer to the actual air flow temperature, and somewhat isolated from the temperature of the grey sleeve or the housing in general.

On the sensor on the right the sensing element is not exposed meaning that it is measuring the temperature of the inside of the grey plastic sleeve - which is probably very hot from heat directly conducted up the stem of the sensor from the housing. And any cooler airflow would have a much harder time cooling this grey sleeve than the direct sensor element.

In short the replacement sensor is badly designed. I did note this difference in design when I first replaced it and I wasn't too keen but the original sensor had broken lugs so wouldn't stay in position and kept falling down.

In a speed density system like this air intake temperature is one of the key parameters that controls injection time as it needs to know the air temperature to work out air density since hot air is less dense. (For the same pressure reading air at a higher temperature has less mols of air)

Therefore all other things being equal a higher air temperature reading will reduce injection time. If the air temperature truly is higher this is what you want to maintain the correct mixture. But if the air temperature reading is a lot higher than the actual air temperature by the time it reaches the cylinders because its measuring the temperature of the casing the reduced injection time will cause the open loop mixture to be leaner than it should be.

So I'm going to try to find another air temp sensor with the correct exposed sensor element design like the original. The question of course is whether all the after market versions of this sensor are the same design as the one on the right.... Anyone happen to know which style of sensor their V6 has fitted ?

[Mandrake]

This one appears to be of the correct open slot design, but I wonder if the part actually shipped would turn out to be this way ? I often find parts from ebay aren't quite as depicted...

http://www.ebay.co.uk/itm/Citroen-Xanti ... SwP0RXh9K8

And here is an interesting much more exposed design!

http://www.ebay.co.uk/itm/Citroen-Xanti ... Sw9NdXsU5G

To me this looks like a design that would measure the air flow temperature without being influenced by the housing temperature, even better than the original OEM part ? Only thing is it appears to be a push tight fit rather than having the snap on lugs.

The fly lead is pretty nice though as the original type is quite hard to unplug and plug in with the filter housing in place.

[white exec]

Just as an experiment, could you move the domed sensor out of its box-side location, and position it somewhere in the middle of the airflow it's supposed to be sensing - and see whether Lexia shows a better reading of the actual intake air temperature?
Just a thought.
Think you are right about the sensor design. Under-bonnet temperatures can go pretty high.

[Mandrake]

There's not really anywhere else I could put it - in theory the sensor tip should already be in the middle of the pipe that goes to the idle control valve - they put it there because that's the only path of air flow at idle when the throttle plate is shut. And when you open the throttle the ECU also commands the idle control valve to open wide to ensure that a high enough percentage of the air flow still passes by the sensor via the ICV to give an accurate reading.

I really like the design of that second ebay sensor with the fly lead - if I were to design an airflow sensor to work as well as possible within a hot housing I would put the sensor bulb out in the open in the middle of the air flow and keep it well away from the stem of the support, which is exactly what they have done.

Not only should that be more accurate, it should be much faster responding to changes in air temperature. When the housing is very hot the air intake temperature will actually change quite quickly with throttle - at a light throttle and low air flow the air will be warmed up quite a bit due to the housing, but at a wider throttle and greater air flow the warming effect of the housing will diminish, so when you accelerate hard the air will actually cool and become more dense.

But a sensor like the one in there now with the sensor element not exposed will be quite slow to respond to changes in temperature due to the thermal insulation provided by the plastic covering over the sensor element compared to the exposed sensor element of the other design.

I'll think I'll order that one with the exposed sensor and fly lead and give it a shot - it looks to me like it should work even better than the original OEM design, and for a tenner its a reasonable impulse buy...

[Stickyfinger]

worth nipping the end off that "new" one and see how that works....I wonder why they would cover a heat reading sensor like that ?

[white exec]

- to keep it clean, maybe? I presume it's just a NTC or PTC resistor. Be interesting to compare the resistances of all three (when you have the new one), and see if they're the same at the same temp. Wonder if a replacement needs calibrating, or whether the ECU somehow does that itself (not sure what it would check it against).

Wonder what would happen if the sensor were moved right "down front", so it was actually measuring the atmospheric air intake temperature.

Agree with your point about suddenly changing air flows, and a sensor with thermal lag. Doesn't make much sense, having the engine respond to conditions that no longer exist!

Just a thought: Perhaps the OE set up works splendidly under laboratory/emissions-testing conditions, when the engine will be run for prolonged periods at steady revs, when it would probably perform not too badly.

[CitroJim]

Most interesting Simon I'd always seen how hot the intake air seemed but never thought deeply about it...

I'm pleased you have as the same may apply to the TCT in the Activa too...

Excellent work... You are like Sherlock Holmes of the V6 Xantia world!