Xantia V6 Activa --- Performance Engine Upgrades

[CitroJim]

So, the bottom end of an ES9 is good for 400bhp then?

Interesting about ES9J4 heads being better..

[Koukku]

Don't know about the bottom end. He did have it at about 300bhp with NOS and stock internals tho.

[Stickyfinger]

Are there any more details on the build ?
I would love to see how he fitted the compressor and what injectors/engine management he is using.

[Deanxm]

Looks like a centrifugal supercharger located over to the left of the engine bay, much more compact and easy to get into a tight space than a Roots type.

D

[Koukku]

I haven't seen a project thread about that procharger build, but there are some pics of it at photobucket: http://s822.photobucket.com/user/paivto ... t=3&page=1

Here's he's current project, mid-engine 406 coupe with a twin turbo ES9J4. Google translator doesn't work very well with Finnish, but the pics start at about half way through the first page...
http://www.sportpug.com/foorumi/yleiset ... i-projekti

EDIT: In the mid-engine project thread he said that FWD/supercharger configuration had 600hp/600Nm, but the gearbox couldn't handle it with sticky tyres (and LSD). So I guess we now know the limits of the ML5T.

[Koukku]

Anyone here have any experiences with the VEMS PnP ECU for the ES9J4?

https://www.dp-engineering.nl/vems-pnp- ... 4-xfz.html

[Stickyfinger]

Very interesting....looks much more practical than the other options

[Mandrake]

Anyone here have any experiences with the VEMS PnP ECU for the ES9J4?

https://www.dp-engineering.nl/vems-pnp- ... 4-xfz.html

No experience of using that particular unit, but I've owned two Xantia V6's now and both have at times had intermittent lack of power, and after investigating those problems in minute detail I've come to a couple of conclusions of what can be improved on the ES9J4, albeit to do it properly you would need a fully customisable ECU like the one you have linked to.

The biggest problem the ES9J4 seems to have is an "over sensitive" knock sensor mapping. The engine is designed for 98 octane and any lower octane than this will cause timing advance to be removed in response to knock sensor events - up to a maximum of 12 degrees. However I have observed on both cars significant amounts of timing being pulled out at times at lower RPM (below 4000 rpm, and below 2500 in particular) even when running on 99 octane fuel, even when the engine is not being worked hard that I can't find a good explanation for.

This causes a loss of both overall power at lower RPM, as well as a "laggy" throttle response, due to the instantaneous timing retard (when you rapidly open the throttle) being excessive. When the symptoms are not there the throttle response on this engine is razor sharp and instantaneous with no lag.

The OEM ECU and Lexia simply don't give enough data or instrumentation to figure out exactly why this is happening, by how much and when, so my findings are partially anecdotal. You can see timing advance on the Lexia but you cannot directly see how much knock retard is being applied to reach that total advance figure, and how much of the advance is coming from the normal timing map.

The problem was much much worse on my first Xantia V6, where it was losing the maximum 12 degrees nearly all the time below 2500 - killing the performance below this rpm. (I confirmed this with a scope by comparing injector and ignition timing)

On my current Xantia V6 the performance is much much better and the effects are a lot more subtle, however I see the exact same timing retard related characteristics but to a much lesser degree and much less frequently. Most people would drive it and not find anything wrong with it but because I am attuned to the symptoms from the previous car I notice the more subtle but similar behaviour on this one.

The underlying cause of excessive knock retard I have not been able to confirm, partly due to lack of proper instrumentation and partly because it is intermittent - some days the performance will be absolutely top notch, no problems, other days I notice a bit of throttle lag at lower rpm and a "power bump" at 4000 rpm.

This power bump at 4000rpm is a commonly reported characteristic of this engine, in fact the torque curve does peak at 4000 rpm by design, however there is more to it than that. Around 3500-4000 rpm is also a threshold in the knock retard mapping table. In simpler ECU's like this there are only 2 or 3 rpm "bands" in the knock retard table with different knock sensitivity thresholds. Because the engine gets mechanically "noisier" at higher rpm the knock sensor sensitivity is reduced at higher rpm so that it takes more "knocking" before timing starts being pulled back. If this wasn't done the normal mechanical noise of the engine would falsely trigger the sensor at higher rpm.

Unfortunately this is not done progressively but is comprised of 2 or more bands. So for example it might have one band from idle to 2500 with a sensitivity of say 10, a band from 2500 to 4000 with a sensitivity of 8 and a band from 4000+ with a sensitivity of 5. Apart from the 4000 threshold I've plucked those other figures out of the air for illustration as I don't know what the bands or sensitivities actually are.

If the source of noise that is triggering knock retard is not actual knocking due to detonation but mechanical noise, and that noise is loud enough to exceed the threshold in the 2500-4000 band, but not loud enough to trigger in the 4000+ band then what you will end up with is lots of timing being pulled out from 2500-4000 when it shouldn't but normal timing above 4000. Hence as you accelerate you get a sudden jump in power as you pass 4000 rpm.

The more the performance at lower rpm is suppressed by excessive timing retard the bigger and more sudden this jump in power becomes - when symptomatic I've seen times in the wet where accelerating in 1st gear started off a bit sluggish and then went into full blown wheel spin the instant it hit 4000 rpm, the jump in power can be that large and sudden. I have also never observed a loss in power over 4000 rpm on either car. When the car is behaving normally there is still a power jump at 4000 rpm but it is much more subtle and just a small "bump", as the official torque curve shows.

While I'm pretty confident the issue is knock retard related, I'm unsure as to the exact cause or whether there are multiple causes. A few possibilities are:

1) Mechanical noise in the engine - "false knock". It's possible that the original factory tuning for the knock retard sensitivity is a bit too "cautious". This may be fine on a nice new quiet engine, but as the engine ages and becomes nosier as all engines do, the mechanical noise eventually reaches a point where it starts triggering the knock sensor. This could be something like noisy hydraulic lifters, worn gudgeon pins, worn bearings etc... anything that is hammering and passing that vibration into the cylinder block.

My old V6 had an engine that was considerably noisier mechanically than the one I have now, and I think it's no accident that it experienced this problem a lot more. From the drivers seat the V6 I have now is about half as noisy as the old one.

The problem with any false knock due to mechanical noise is that it very quickly removes the maximum permissible amount of advance - which is 12 degrees on this engine. The reason for that is that true knock caused by detonation will stop as soon as the ECU removes 1-2 degrees of timing at which point it stops removing timing. False knock will not go away as timing is removed so the ECU keeps removing more and more timing trying to solve knock that is not really knock, and in a very short time the maximum 12 degrees is taken out and the performance is killed. So any false knock is very detrimental to performance.

It can also be very specific to certain RPM's as the mechanical noise will tend to be worst at certain speeds, so it can cause holes in the performance at certain rpm and loads that feels like a hesitation as you accelerate.

2) Another possibility is misfiring. Lean misfires can cause the cylinder to "ping" in a way that can be detected as knock, so any time you are having lean misfires knock detection is a possibility. So as well as the power loss from a misfire you get timing pulled out as well. One possible cause of this is the ECU is not designed to correct the open loop mixture for different fuels like E5 and E10 - it was designed in the mid 90's when there was no ethanol added and when fuels still had things like sulfur in them. Ethanol blends have a different stoichiometric ratio than regular petrol which means for the same injection time it will burn leaner.

In closed loop mode - which includes idling (after the oxygen sensor has warmed up - which takes about 90 seconds from cold) and cruising at a light to medium constant throttle it will correct the mixture using the oxygen sensor and everything will be fine, even if you're running on E10.

However in open loop mode - which includes cold start and idling for the first 90 seconds, any hard acceleration, and momentarily any time you depress the throttle, even by a small amount the ECU defaults back to pre-programmed tables that do not compensate for the different fuel characteristics. In these occasions the engine will run leaner than intended if you use E5/E10 and this can result in a hesitation/stumble, or a misfire that triggers the knock sensor and causes timing to be pulled out. The end result is that sharp throttle response is lost both due to running too lean and with reduced timing advance. It feels like "throttle lag".

Believe it or not, after close observation I don't think this ECU incorporates any form of long term fuel trim - whereas long term fuel trim which can be used in open loop mode is pretty standard on more modern OBD-II ECU's. It seems to rely entirely on pre-programmed tables for open loop mode and short term fuel trim only for closed loop mode.

So I believe that with a fully custom ECU like the one you linked and a proper tuning session by a tuner who knows what they're doing, whilst probably not increasing peak power or peak torque much, I think the torque, throttle response and consistency of performance in the <4000 rpm range could be significantly improved, by a more sophisticated and better tuned knock retard table and algorithm. It could also incorporate a true long term fuel trim that would correctly compensate for E5/E10, and better still you could probably use a wide band oxygen sensor instead of the original narrow band sensor to get closed loop running over most of the engine's operating load range.

Would I be willing to pay that much for that custom ECU for a car that is worth less than half that amount ? No chance!

The things I have done to mine to try to work around these two issues are:

1) Increase the fuel pressure regulator from 3.0 bars to 3.5 bars. This makes the open loop mixture a few percent richer but still allows the ECU to maintain the correct mixture in closed loop mode - so you still pass the emissions test and still get good cruising economy, but the throttle response is a bit sharper and maximum power is slightly increased. (only a few percent, but I do notice it)

2) I connected a carefully selected resistor in parallel with the knock sensor at the ECU end to de-sensitise it slightly. This is somewhat equivalent to fitting a washer between the knock sensor and the block, a tweak that some people perform on other engines, but which is nearly impossible to do on this engine due to access to the V tunnel under the manifold. I don't recall the value but it was somewhere in the order of 1-3 Megohms. I do not recommend doing this! If the resistor value is too low and de-sensitises the knock sensor too much it could cause real knocking to go unnoticed and cause piston damage during detonation that could eventually destroy the engine.

Tuning knock sensor sensitivity should only really ever be done on a dyno by a professional who is listening to the output of the knock sensor audibly ("detcans") and testing the engine over the full range of load/rpm and temperature. I took a bit of a gamble by gradually reducing the resistor until I started noticing some improvement in performance then stopped at that point. So far it has been OK a year later.

A combination of boosting the fuel pressure (thus making the open loop mixture richer, since the ECU is unaware of the change) and de-sensitising the knock sensor has made quite a big difference - the loss of timing below 4000 rpm is mostly gone and performance is nearly always excellent. Incidences of less than optimal performance are few and far between, so I think a combination of knock retard timing and lean mixture was the cause of less than optimal performance.

A custom ECU professionally tuned would do a lot better job and probably achieve even better results and more safely, but I really don't think it's worth the cost unless you had a pristine car and plenty of money to throw at both the hardware and a custom tune...

[CitroJim]

Sian, an excellent post and essential reading for any V6 owner

[Koukku]

Great piece of information! Not sure if I agree on the lean mixture part though. I accidentaly filled my V6 with 25 liters of E85 once (the rest was 98 E5), as that's what I use in my Mini R53. Didn't notice any difference in how it ran. Cold starts were fine too. I just assumed that LTFT took care of it... but if there is no LTFT...

Any idea if knock retard is a learned value in the ES9J4 ECU, or is it always based on the current knock sensor value? Just thinking if it's a learned value would it be possible to reset it. My Mini's ECU learns it over time, but it uses more than just knock sensor input for calculating it.

For me that PnP ECU would actually be cheaper than the value of my car... we have a huge tax on cars here in Finland. It would also allow me to run bigger injectors and E85. And with E85 being so cheap I would break even in about 3-4 years.

[Mandrake]

Any idea if knock retard is a learned value in the ES9J4 ECU, or is it always based on the current knock sensor value? Just thinking if it's a learned value would it be possible to reset it. My Mini's ECU learns it over time, but it uses more than just knock sensor input for calculating it.

Knock retard is learned over time and saved in memory. Like any knock retard table, while the engine is within a given RPM/Load cell in the table timing advance is removed very quickly (fraction of a second) in response to knock events, but only added back in very slowly and cautiously in the absence of knocking. (Can take several minutes of running within that specific cell for the value to creep all the way from -12 back to 0)

Also if enough knock events are detected over time and a variety of load range the ECU assumes you are using a lower octane fuel and switches to a different base timing map designed for 95 octane fuel - once you end up in this lower octane map it is quite difficult to get out of it again - filling up with 98 would take a few days of driving for it to switch. The quickest way is to disconnect the battery for 10 minutes to reset it.

On this ECU the knock retard table is reset if you disconnect the battery for at least 10 minutes. In fact with this issue of false (?) knock retard its common for the performance to come back to normal immediately after an ECU reset but then deteriorate over time - which helps point the finger at learnt knock data as the cause of power loss, either the knock table, or the ECU has reverted to a lower octane map.

For me that PnP ECU would actually be cheaper than the value of my car... we have a huge tax on cars here in Finland. It would also allow me to run bigger injectors and E85. And with E85 being so cheap I would break even in about 3-4 years.

Here in the UK even a good Xantia V6 is only worth around £400-600 - less than the cost of that unit.

[Koukku]

Looks like the knock retard table learning acts similarly as in the Mini. Except in the Mini I can reset it with a laptop. It also gains some performance after a reset, but only for a short time, unless the root cause is fixed of course.

Here in the UK even a good Xantia V6 is only worth around £400-600 - less than the cost of that unit.

Here the asking prices for a roadworthy Xantia are around 500-1500€ depending on condition. V6 would be at the upper end. I paid 1800€ for this beat up V6 a year ago, but it's an Activa. A well kept V6 Activa sold for 3900€ just recently...

[Green fluid]

In France a very nice V6 can be sold more than €5,000... (while Turbo CTs still stay at €1,000 if not perfect...).

E85 is cheap in France (sometimes less than €0.55/litre) and many people use it. On the turbo CT you need to change injectors, on the V6... as long as you don't put more than 75% E85 in the tank, it works correctly - but it's better perform the init process to keep the engine smooth.

[Stickyfinger]

In France a very nice V6 can be sold more than €5,000... (while Turbo CTs still stay at €1,000 if not perfect...).
.

Maybe I should take mine one back to France....ohhh, it is going up, who would of thought a Xantia goes up in price

[CitroJim]

In France a very nice V6 can be sold more than €5,000... (while Turbo CTs still stay at €1,000 if not perfect...).

Is that a V6 Automatic or a V6 Activa? In the UK we can only dream of such prices... A good 2.0 TCT Activa will fetch perhaps £1,500 on a good day but even then it has to be very good... More ordinary Xantias go for very little indeed... They can be worth more as scrap metal...

E85 is cheap in France (sometimes less than €0.55/litre) and many people use it. On the turbo CT you need to change injectors, on the V6... as long as you don't put more than 75% E85 in the tank, it works correctly - but it's better perform the init process to keep the engine smooth.

Now that is interesting, especially with a TCT which is very keen to drink petrol I'm not sure if E85 is readily available in the UK yet and even if it is I expect it will cost as much as regular 95RON petrol and knowing the UK, possibly more as it will have some 'green' label attached to it as an excuse to raise the price!