I am having mapping issues with the duratec - its driveable as it is, but you have to drive it a very specific way to get it to run right which can be annoying.

Basically under moderate acceleration the AFR is 10 or less, which is obviously way too rich. However if I accelerate hard (foot to the floor) then the AFR is about right at 12.5. Under very light acceleration where I stay in vac then its also ok, mixture stays around 14.7 all the way up.

I have attached the MSQ and here is a link to the log: http://gtslocost.locostsites.co.uk/drive48.msl

Theres an example below....

I know there was a problem with the accel enrich threshold being set too low on this log, which I have now fixed.

If I lean out the on boost areas to give nice smooth gentle acceleration then its dead lean (20+ AFR) when you actually floor it on boost - I can counteract this for 0.5sec or so with AE, but as soon as the AE stops its back to lean again... I see boost on gentle acceleration when the load is not as great as under hard accel so tuning for one or the other seems OK, but not both? I have a BOV fitted, but obviously once the boost is there its held closed anyway.

I wasnt sure on setting the KPa values for the map because of the following:

The idle MAP value is around 45-55KPa, I can then light cruise at 60-65kpa at say 30mph in 5th gear. Then cruising at 40mph (2000rpm) is around 80KPa, cruising at 50mph (2500rpm) is around 85-90KPa, 60mph (3000rpm) is 100-105Kpa, 70mph (3500rpm) is around 105-108KPa - to be honest I was expecting to cruise on lower KPa than this... So I do need the resolution in the lower KPa areas of the map too I think to get a smooth cruise (which even now I have varying AFR's and slight misfires presumably due to large injectors?). The MAP signal is picked up from the standard injector positions in the throttle bodies (so valve side of the butterflies). Maybe I should look at running dual table and have double the resolution on the KPa?

If anyone has any suggestions on setting better values for KPA in both spark and VE tables I would most appreciate it

If I tune the boost areas for light acceleration to be ok then when you floor it at the same KPa its too lean (after AE) and visa versa.

Heres an example from the log:

Light acceleration:

RPM MAP TP AFR
2956 111.3 16 9.9
3053 108.9 17 9.5
3026 109.4 17 9.4
3069 109.6 17 9.4
3118 109.9 17 9.4
3116 110 17 9.4
3167 114.5 17 9.6
3187 114.7 17 9.4
3207 115 18 9.4
3325 111.9 25 9.3
3401 117.6 30 9.6

You can see boost slowly rise at around 20% throttle. But the AFR is stuck at 9.5

Now from the same drive and same msq under 50% throttle:

RPM MAP TP AFR
2852 109.8 19 14.5
2876 110.9 36 13.7
2886 112.8 41 13.4
2894 113.3 42 12.3
2921 113.7 43 12.3
2924 113.9 43 12.6
2964 114.4 47 12.7
3022 114.2 48 13
2948 114 48 13.4
2997 115.3 48 13.2
3006 115.4 48 13.3
3013 114.8 48 13
3033 114.7 49 13.2
3050 115 49 13.4
3067 114.9 49 13
3081 115.2 49 12.9
3096 115.4 50 13
3111 116.4 50 12.9
3120 116.6 50 13.2
3143 115.6 50 13.1
3161 115.6 50 13.2
3174 117.1 50 13.2
3187 117.2 50 13.3
3206 117.3 50 13.3
3224 116.1 50 13.2
3239 117.2 50 13.5
3250 117 50 13.8
3269 116.2 50 13.5
3295 116.2 50 13.2
3303 116.3 50 13.2
3334 116.3 50 13.3
3328 116.9 50 13.4
3343 117.2 50 13.5
3382 118 49 13.5

You can see the boost at each RPM is almost the same as previous slightly higher, but only marginally by around 1psi, but the AFR is now almost correct....


I also have a problem that if I accelerate hard to say 60-70mph and then back off the throttle the AFR drops to 10 or so, then if I lift off and put the throttle down again it returns to around 14. Looking at logs from when this happens it's because the boost stays there - I am currently working on the assumption that this could be a sticky BOV though. I dont think this is helped by the fact that once cruising at 70ish the MAP hovers around 100kpa or is just over 100kpa (7120 in attached log)

So where am I going wrong?

Thanks,
David

What ECU are you using?

Knew I would forget something!

It is Megasquirt 2 V3 using speed density (MAP based) code. Running the beta code (3.0.3r) for better MAP sampling.

David

David,

You know more about megasquirt than I do (as we prove whenever I post) but let me ask a question.

Are you running alpha-N, speed density, or Hybrid Alpha-N.

I assume the last of these. In which case I don't understand it but am interested in hearing how you get on.

Matt

[Edit] our replies crossed. So why not Hybrid Alpha-N?

[Edited on 12/4/10 by matt_gsxr]

I am running Speed density. Hybrid alpha_n isnt available in MS2, you have to go down the tuning of 2 seperate tables to do this.

Which is what I had thought about doing anyway, but then I was told by the developers that it wouldnt work. Infact you can see the discussion here:

http://www.msextra.com/forums/viewtopic.php?f=101&t=34791&p=236558

You then also get into the fact of how on earth you map 2 tables at the same time!?

[Edited on 12/4/10 by flak monkey]

sadly i know naff all that can help but i have messaged a man that can with some luck he may pop over to help.

good luck with the mapping dying to see this finished and realising its potential

quote:

---

Originally posted by dlatch
sadly i know naff all that can help but i have messaged a man that can with some luck he may pop over to help.

good luck with the mapping dying to see this finished and realising its potential

---

this is why i pay for my mapping and use omex lol

although i would like to be able to do it myself

quote:

---

Originally posted by flak monkey
Hybrid alpha_n isnt available in MS2, you have to go down the tuning of 2 seperate tables to do this.

---

I do intend to pay someone to finalise the mapping for me.

But like everything I want to get it as good as I can before that, just to see how close I can get it

I have had a very quick look and noticed a couple of things.
1. You need to damp out your map signal (use a small fuel filter) as it is very noisy
2. There is a lot of lag in the map signal as well (is the pipe to the ECU very long or connected to only one cylinder.
3. Maybe lean at part throttle around 2500rpm (could be wrong as I only had a very brief look)

Good luck

It is lean at part throttle crusing at 2500rpm yes, and thats now fixed.

MAP signal is being processed using the new sampling code in the alpha code. I will have another play and see if I can smooth it out any more.

Its sampled of all 4 cylinders, then to a 3 way splitter - one to the BOV, one to FPR and one to ECU. Theres no 'damper' in the system though. Do these really work or is it a myth?

David,

I now understand that this is way complicated.

What would simplify things would be if the MAP sensor were moved into the airbox.

Then (perhaps) you could use Alpha-N, but have the airbox pressure as a barometric correction factor (i.e. a simple scaling).



I hadn't appreciated that MS2 can't handle Hybrid Alpha-N.


As you will have worked out, I don't know anything about dual tables. But the approach of iterating (making many small changes in the correct direction for each table) should get there in the end. Easier said than done.

Matt

Yes the filter does work
I did some investigation when I had an ignition oly ECU.
See thread:
http://www.autosportlabs.org/viewtopic.php?t=1171

quote:

---

Originally posted by flak monkey Do these really work or is it a myth?

---

Hmm interesting. So on a 2 litre engine what sort of sized chamber would I be needing? And where should I look?

I dont think it will solve the underlying issue, but if it helps with noise, then I am all for it anyway. Heck knows where I'll put it either!

Thanks,
David

The narrow hose and fuel filter trick does get rid of ripple on the MAP line.

It will also delay a bit, but I think we know the priority.

I have a small (20ml perhaps) filter on my 1litre engine (this is the capacitance in the electrical equivalent circuit). This is at the end with the sensor. The hose to the sensor needs to be fairly narrow (this is the Resistance in the electrical equivalent circuit).


You could run closed loop in the problem areas. I agree, too much like giving up.


Matt

quote:

---

Originally posted by matt_gsxr
The narrow hose and fuel filter trick does get rid of ripple on the MAP line.

It will also delay a bit, but I think we know the priority.

I have a small (20ml perhaps) filter on my 1litre engine (this is the capacitance in the electrical equivalent circuit). This is at the end with the sensor. The hose to the sensor needs to be fairly narrow (this is the Resistance in the electrical equivalent circuit).


You could run closed loop in the problem areas. I agree, too much like giving up.


Matt

---

have you thought about an omex ?

I have no idea what you guys are talking about. But then again thats why I spend time reading posts like this and why im on this site.
I really want you to get a good map. I need one of these engines in the next 5 years!

At light throttle could the tps line be noisy triggering the trainsient fuelling?
(aceleration enrichment).
Darren

quote:

---

Originally posted by turbodisplay
At light throttle could the tps line be noisy triggering the trainsient fuelling?
(aceleration enrichment).
Darren

---

This is mine. Thinking about this, this shouldn't care about your engine capacity, more to do with cam timing as its function is to smooth out the pressure pulses.


my_MAP_damper

The interesting patterns (in the map) aren't strictly noise (i.e. it isn't random), but aliasing of the sampling rate with the pulsations.




I have looked at your traces. Am I correct in saying that at times you have the same TPS and RPM but the MAP increases.

Presumably this phenomenon is due to the supercharger pumping a constant volume per unit time, but the air not being able to get past the throttle plates, so you are effectively "pumping up" the inlet manifold (at 7258.3s).

Or is it just poor resolution of the TPS. You could do the parallel resistor trick to improve TPS resolution at the closed end of things. non-linear tps resistor

I know this isn't the big problem, but...


Matt

I'll have a look tonight and see if I have something suitable laying about.

I am wondering if this is a problem more to do with the sampling of the MAP and the averaging rather than an inherent problem with the set up like you say. Sometimes the pattern in the map signal changes to be almost square wave, I am not sure if there is an example in that log

I am going to have a play with the sampling later, maybe even switch back to an old code without the new sampling to see if it makes any difference.

Not sure if I know what you mean with the changes in MAP at the same throttle opening? At that time the throttle is virtually closed and its virtually at the idle map of around 50kpa.

Thanks,
David

Well a little update is in order of course. Been out in the car for a quick drive or 2 to try and see what the best way is to solve this problem of too much or too little fuel.

First thing I did was reload the current release code for MS (2.something) and o for a run to see if there was a significant difference in the map sampling which was causing a problem. There wasnt, if anything the MAP signal was worse on the standard code, which didnt suprise me as the whole idea of the code in the alpha is to make the sampling better! Essentially it allows you to specify a specific MAP sample window and range which requires some experimentation, but in the end you get a lower and more stable MAP signal when runnin ITBs.

I then had a little play with the map sampling in the latest alpha code to try and smooth out the noise that way. I lowered the MAP reading a little, by another 5kpa with some adjustments, but still a little noisy. Satisfied I had done everything I could here I have set about going down the dual table tuning route on the suggestion that it may well work and solve my problem.

This dual table allows you to run 2 different load parameters on 2 different VE tables, or allows you to extend the VE table to be 16x32 or 32x16 if you wanted to. The downside being you cant auto tune or use ve analyse any more as it cannot compute across 2 different VE tables. So its more than hybrid alpha_n in MS1. You can make the second table additive or multiplicitive depending on what you want to do.

I have attached the MSQ for those that are interested. You need tunerstudio to open it, wont work properly with megatune. You can see VE1 is a normal MAP load VE table with the full fuelling map. My plan is to get this sorted so that it is right for normal driving and moderate throttle. Then VE2 is all set at 100 at the moment so its having no effect on VE1. The load is TPS only and the RPM bins are the same as VE1. The code multiplies the look up from each table to get the final fuelling number. The second table is in percentage terms when in multiplicitive so 100 means multiplied by 1.

I am hoping that VE2 is going to be fairly simple at the end of the day when its all finished just adding in the extra fuel in the holes where its needed. (Mostly around 40% throttle and up at mid RPM's)

Time will tell!

The sampling of the MAP is a bit complicated as it is confounded by the sampling rate of the datalogs (which is much lower than that of the calculations).

The actual output will be based on all the MAP samples it sees, which (on average) will be a smoothed MAP curve (but with noise). I don't think that this could give you the bias in the AFR that you have seen under different acceleration conditions.

Sorting the MAP to give a nice smooth function can only help.


Have you considered monitoring the pressure in the plenum. There is an option for a second pressure sensor in the MS. I think that could be a useful hidden variable.

Matt

[Edited on 13/4/10 by matt_gsxr]

OK, I will look at finding a small filter of some description and see if it helps with the smoothing as well. Like you say it can only be a good thing anyway.

Compared to the current release code, the map in the alpha code is probably 50% smoother with the new algorithm. I dont have any comparison datalogs handy though.

I could probably measure the air pressure in the plenum too yes, I'll look into it.

Thanks,
David

Just spoken to Omex, and they run blended TPS and MAP when they see problems like this and they have on supercharged duratec and zetecs running individual throttle bodies. So it is the way to go by the looks of it....

I still may change to omex this winter depending on how the squirt goes this summer....

you'll now probably kick yourself to know we have recently finished a group buy on omex600 ecus, which were 499.99 inc vat and included 1 hour free mapping.

perhaps once our trader status and site advertising gets sorted out we could arrange another one on here if theres enough interest?

- colin.

Dammit, never mind though.

Am going to persevere with the MS for a little while yet, and then think about changing this winter if I need to.

I have just ordered a smaller supercharger pulley which should arrive tomorrow which should take me up to around 12psi of boost. Then I just need to sort out my tensioner to stop that slipping, change the oil and we are away for a full on mapping session.

David