Some ECU related queries
Discussion
Two queries actually, that I'm sure the bright Piston Headed People will be able to advise and/or answer.
This is with regards an ECU I'm using that I'm trying to improve performance. It's on a race car, so not worried about economy or smoothness of idle or gentle overrun performance.... The ECU uses a TPS to judge load. I have a wideband O2 sensor in the exhaust that I log, but it doesn't talk to the ECU - i.e. open loop only.
1. Which is better - a linear load site column (e.g. 0%, 5%, 10% 15%... 95% 100%) or a non-linear load site column to approximate the airflow at different throttle openings (e.g. (0%, 0.8%, 1.7%, 3.5% 8% 20%.... 100%)?
I can use either, but I want to know if one is considered better than the other or not.
2. I'm logging Air:Fuel, and obviously I want to run with the optimum for performance when not idling. Ignoring the AFR number aimed for (as that seems to be a much discussed and never answered topic!!), what sort of AFR trace should I be aiming for? How stable can it be made?
At the moment, on a smooth acceleration it isn't too bad - a bit lumpy in places, but not that bad. But as soon as I lift off it goes rich, and as soon as I reapply throttle it goes lean. Can that be improved? I realise that the acceleration enrichment can help with the second aspect, but can the overrun AFR be stabilised at all, or should I not even bother.
You see, I sometimes get a misfire. When I change gear (a process that takes about 0.3 seconds, and sees me lift the throttle to about 50% to unload the gearbox) and the mixture goes rich-weak very quickly, sometimes the engine misfires afterwards until I release the throttle for a moment and get back on it again. At that point the AFR goes very lean, but as it's misfiring this is normal... The misfire has also struck when the wheels spin a bit - either in the wet or over bumps, despite being nowhere near a revlimiter and on a constant throttle. Plugs, coils, leads etc have been changed.
I'd love to be able to get rid of the misfire, but I'm not sure how. Should I reduce the overrun fuelling in the map (to avoid enrichment when I lift off) and increase the accel. enrichment? Should I leave the accel. enrichment because it isn't too bad out of corners (or is it? See attachment below) and try to add fuel via the main map at the load/speed site that best corresponds to the RPM/throttle openings as I get back on the throttle after the gear change?
As you can see, I'm having to learn as I go, but so far I've not cured the misfire... Turning off overrun fuel cutoff helped a lot - but that wasn't my idea, so I won't take credit for it.
Attached is (or should be) an image of my datalogging. RPM (blue), AFR (orange) and throttle position (pink). I realise that at some load/speed sites I need to enrichen, and at others I need to lean off a bit. But how much 'better' can the AFR trace be got, or will it always be a wobbly mess no matter what I do?
Cheers!
Or ignore all of question 2, and advise how I should go about improving the above a bit.
This is with regards an ECU I'm using that I'm trying to improve performance. It's on a race car, so not worried about economy or smoothness of idle or gentle overrun performance.... The ECU uses a TPS to judge load. I have a wideband O2 sensor in the exhaust that I log, but it doesn't talk to the ECU - i.e. open loop only.
1. Which is better - a linear load site column (e.g. 0%, 5%, 10% 15%... 95% 100%) or a non-linear load site column to approximate the airflow at different throttle openings (e.g. (0%, 0.8%, 1.7%, 3.5% 8% 20%.... 100%)?
I can use either, but I want to know if one is considered better than the other or not.
2. I'm logging Air:Fuel, and obviously I want to run with the optimum for performance when not idling. Ignoring the AFR number aimed for (as that seems to be a much discussed and never answered topic!!), what sort of AFR trace should I be aiming for? How stable can it be made?
At the moment, on a smooth acceleration it isn't too bad - a bit lumpy in places, but not that bad. But as soon as I lift off it goes rich, and as soon as I reapply throttle it goes lean. Can that be improved? I realise that the acceleration enrichment can help with the second aspect, but can the overrun AFR be stabilised at all, or should I not even bother.
You see, I sometimes get a misfire. When I change gear (a process that takes about 0.3 seconds, and sees me lift the throttle to about 50% to unload the gearbox) and the mixture goes rich-weak very quickly, sometimes the engine misfires afterwards until I release the throttle for a moment and get back on it again. At that point the AFR goes very lean, but as it's misfiring this is normal... The misfire has also struck when the wheels spin a bit - either in the wet or over bumps, despite being nowhere near a revlimiter and on a constant throttle. Plugs, coils, leads etc have been changed.
I'd love to be able to get rid of the misfire, but I'm not sure how. Should I reduce the overrun fuelling in the map (to avoid enrichment when I lift off) and increase the accel. enrichment? Should I leave the accel. enrichment because it isn't too bad out of corners (or is it? See attachment below) and try to add fuel via the main map at the load/speed site that best corresponds to the RPM/throttle openings as I get back on the throttle after the gear change?
As you can see, I'm having to learn as I go, but so far I've not cured the misfire... Turning off overrun fuel cutoff helped a lot - but that wasn't my idea, so I won't take credit for it.
Attached is (or should be) an image of my datalogging. RPM (blue), AFR (orange) and throttle position (pink). I realise that at some load/speed sites I need to enrichen, and at others I need to lean off a bit. But how much 'better' can the AFR trace be got, or will it always be a wobbly mess no matter what I do?
Cheers!
Or ignore all of question 2, and advise how I should go about improving the above a bit.
On an N-Alpha load sensing system (N- rpm, Alpha = throttle plate angle) you will be better off with a non linear load axis, as the throttles airflow is not linear with opening angle (the "rate of increase" of throttle area reduces as it opens) so to get enough resolution at low throttle angles, i suggest biasing the load axis to the low throttle openings. (also, remember at low rpm the effective WOT angle will be very low, typically on large multibutterfly race cars, at 2000rpm WOT can be as little as 30 degs open!)
Regarding the transient fuelling: if you are sure that your base mapping is correct, i.e at steady state the fuelling is correct for all throttle angles, then you need to revisit your transient gains and clips. Basically the systemm will add extra fuel on tipins (where the increaseing runner air pressure causes fuel to puddle on the walls, and hence the cylinder charge to go lean) and remove fuel on tip-outs (where the falling pressure evapourates the fuel puddle mass, which is ingested into the engine and causes a rich firing event)
Generally, you are better turning off fuel as soon as you are fully in dashpot mode (i.e. throttle fully released but engine speed above idle speed)which will prevent misfire and partial burns, although a lot of people deliberately add fuel at light throttles at high rpm to get flames out the exhaust !!!
(you might need to look at your effective throttle position under trailing throttle also, if you have insufficient air you can get partial misfires (depends upon your throttle /air bypass hardware etc)
Regarding the transient fuelling: if you are sure that your base mapping is correct, i.e at steady state the fuelling is correct for all throttle angles, then you need to revisit your transient gains and clips. Basically the systemm will add extra fuel on tipins (where the increaseing runner air pressure causes fuel to puddle on the walls, and hence the cylinder charge to go lean) and remove fuel on tip-outs (where the falling pressure evapourates the fuel puddle mass, which is ingested into the engine and causes a rich firing event)
Generally, you are better turning off fuel as soon as you are fully in dashpot mode (i.e. throttle fully released but engine speed above idle speed)which will prevent misfire and partial burns, although a lot of people deliberately add fuel at light throttles at high rpm to get flames out the exhaust !!!
(you might need to look at your effective throttle position under trailing throttle also, if you have insufficient air you can get partial misfires (depends upon your throttle /air bypass hardware etc)
Edited by anonymous-user on Thursday 9th September 21:15
tristancliffe said:
Or ignore all of question 2, and advise how I should go about improving the above a bit.
Take it to a dyno and get professional help ?As for throttle resolution. Expand it where you need more resolution, and decrease it where you dont.
And constantly targetting AFR's isnt always the best plan. The AFR you might think you want, isnt necessarily the optimum. Again the dyno will be useful there.
The car was mapped on a dyno by a professional (the same person the programmed the ECU and software). It has a non linear throttle table (which I wanted to confirm was the best way, as many other maps I've seen (from other ECUs) use linear tps sites).
The mapping was apparently done to lambda 0.88 as that's the value he has found best.
Many thanks so far though. My next race is this weekend, so I plan to spend some laps gathering data - full throttle runs from low rpm, coasting runs on closed and partial throttle, and steady state runs by holding the speed constant on the brakes.
I don't really mind what the AFR does, as long as the throttle response is good and the misfire on gearchanges and bumps disappears. In terms of outright performance, the engine is very strong. Without the misfire we'd be even quicker.
The mapping was apparently done to lambda 0.88 as that's the value he has found best.
Many thanks so far though. My next race is this weekend, so I plan to spend some laps gathering data - full throttle runs from low rpm, coasting runs on closed and partial throttle, and steady state runs by holding the speed constant on the brakes.
I don't really mind what the AFR does, as long as the throttle response is good and the misfire on gearchanges and bumps disappears. In terms of outright performance, the engine is very strong. Without the misfire we'd be even quicker.
If you are having issues like that, then really the only way to sort it, is to have someone work the laptop whilst you drive. They can make changes to see what cures the problem.
You could do all the logging in the world, and it still might not help. Trial and error really is the only way for throttle response tuning etc.
You could do all the logging in the world, and it still might not help. Trial and error really is the only way for throttle response tuning etc.
What i do to try to nail down transient fuel issues is as follows:
Pick 2 throttle positions, 1 partially closed (say 15%) one fairly open (say 65%), then hold the car in gear against the brakes at the first throttle positio (15%).. Then snap the throttle open to the second position in as rapid a snap as possible, but (and here's the skillfull bit) at the same time left foot brake to stop the engine rpm increasing. With a bit of practise you can get nice square inputs and hold rpm within a couple of hundred rpm. Then look at the AFR during these events and modfy the transient fuelling as possible.
You can also hold the car at a constant throttle position (say 30%) and use the brakes to change the rpm (you get more of a triange shape rather than square due to the cars inertia) to see what effect engine speed fluctations have.
You might want to adjust the max throttle stop to the 65% value to enable repeatable results (i have a dbw system so i just tell it to input perfect square throttle movements)
Pick 2 throttle positions, 1 partially closed (say 15%) one fairly open (say 65%), then hold the car in gear against the brakes at the first throttle positio (15%).. Then snap the throttle open to the second position in as rapid a snap as possible, but (and here's the skillfull bit) at the same time left foot brake to stop the engine rpm increasing. With a bit of practise you can get nice square inputs and hold rpm within a couple of hundred rpm. Then look at the AFR during these events and modfy the transient fuelling as possible.
You can also hold the car at a constant throttle position (say 30%) and use the brakes to change the rpm (you get more of a triange shape rather than square due to the cars inertia) to see what effect engine speed fluctations have.
You might want to adjust the max throttle stop to the 65% value to enable repeatable results (i have a dbw system so i just tell it to input perfect square throttle movements)
Many thanks for all your help.
I did a few of the things suggested - attempting to snap the throttle between two points at a constant speed (against my brakes), as well as various constant throttle/rpm runs to gather data. I then tweaked it a bit, and made it quite a lot richer at high rpm with smaller throttle openings to avoid it going so lean.
Made a lot of progress, and I think we might finally have cured the misfire. Perhaps too early to really tell, but I'm fairly confident. The AFR trace looks a lot better than it ever has, with the peaks and troughs much closer to that AFR we want at all times. Made a few more tweaks (one percent longer injector openings here, or shorter there etc), and it should be even better at the next race at Croft in a fortnight.
Thanks again for your inputs.
I did a few of the things suggested - attempting to snap the throttle between two points at a constant speed (against my brakes), as well as various constant throttle/rpm runs to gather data. I then tweaked it a bit, and made it quite a lot richer at high rpm with smaller throttle openings to avoid it going so lean.
Made a lot of progress, and I think we might finally have cured the misfire. Perhaps too early to really tell, but I'm fairly confident. The AFR trace looks a lot better than it ever has, with the peaks and troughs much closer to that AFR we want at all times. Made a few more tweaks (one percent longer injector openings here, or shorter there etc), and it should be even better at the next race at Croft in a fortnight.
Thanks again for your inputs.
tristancliffe said:
The car was mapped on a dyno by a professional (the same person the programmed the ECU and software). It has a non linear throttle table (which I wanted to confirm was the best way, as many other maps I've seen (from other ECUs) use linear tps sites).
The mapping was apparently done to lambda 0.88 as that's the value he has found best.
0.88 is a bit weak for a race engine.The mapping was apparently done to lambda 0.88 as that's the value he has found best.
http://www.pumaracing.co.uk/combust.htm
Any misfiring is an indication of a seriously badly mapped system which any competent dyno operator shouldn't have let happen. I suggest you go elsewhere in future.
RE lambda targets, I think it depends largly on the engine type, specifically combustion chamber shape, compression ratio, tumble, squish, and fuel atomisation. I found more pwer and better response by going from 0.85-0.88. This was on a 4 valve engine producing 100bhp/lt 80lb/ft litre. So not a low tune engine.
The key is give the engine what it wants, not what you think it wants. I also agree with Pumaracing, go somewhere else!
FWIW I rear somwhere (Don't recall?) that modern F1 engines are running Lambda in region of 0.95. But that is something else...
Steve
The key is give the engine what it wants, not what you think it wants. I also agree with Pumaracing, go somewhere else!
FWIW I rear somwhere (Don't recall?) that modern F1 engines are running Lambda in region of 0.95. But that is something else...
Steve
Edited by stevesingo on Monday 13th September 14:14
Edited by stevesingo on Monday 13th September 14:14
eliot said:
tristancliffe said:
The engine is a 2.0 4-Cylinder Toyota. I'm not prepared to say who makes the ECU.
Not that either are relevant to my problem really. Just pretend I told you the name of an ECU manufacturer.
So its an emerald then.Not that either are relevant to my problem really. Just pretend I told you the name of an ECU manufacturer.
Edited by stevieturbo on Monday 13th September 22:09
stevieturbo said:
eliot said:
tristancliffe said:
The engine is a 2.0 4-Cylinder Toyota. I'm not prepared to say who makes the ECU.
Not that either are relevant to my problem really. Just pretend I told you the name of an ECU manufacturer.
So its an emerald then.Not that either are relevant to my problem really. Just pretend I told you the name of an ECU manufacturer.
Edited by stevieturbo on Monday 13th September 22:09
stevieturbo said:
eliot said:
tristancliffe said:
The engine is a 2.0 4-Cylinder Toyota. I'm not prepared to say who makes the ECU.
Not that either are relevant to my problem really. Just pretend I told you the name of an ECU manufacturer.
So its an emerald then.Not that either are relevant to my problem really. Just pretend I told you the name of an ECU manufacturer.
Edited by stevieturbo on Monday 13th September 22:09
Gassing Station | Engines & Drivetrain | Top of Page | What's New | My Stuff