Ford Falcon Spark Advance

[jeremy]

I am trying to gain a better understanding of how spark advance is calculated in ford falcons and I do not seem to be able to get my calculations to match what I am receiving via the scanner. If anyone is able to shed some light on this topic or notices anything wrong then please do correct me. I assume that load is calculated based on map ( map/101 in kpa.) and all axis in the tables are the same as the sensor readings from the scanner. As I do not have expected lambda I base the lambda off the lambda lookup table values. I am certain that more corrections for lambda are present and this is not exactly what is demanded but I would expect it is in the ball park for most operating conditions as long as nothing is too far from the norm, no sharp jabs of the throttle ,no engine or cat protection doing its tricks etc. I am also aware that the obd2 port may not be the quickest to have exact matches as the speed of obd2 port can provide a little lag between each reading.

Is timing calculated using:

Borderline knock + base lambda correction + (IAT spark base*IAT spark multi) + (ECT spark base*ECT spark multi) + (IMRC spark base*IMRC spark multi) + VCT Correction + emissions retard - knock retard - torque reductions- other factors + other factors =total timing.

Is this correct??

Timing will be removed or added back if knock is present or has been present very recently, again if torque reduction is present, I am also guessing if cat protection or engine protection is also present, also probable other factors I am not aware of then timing will be removed. I am also assuming that MBT is used only in the torque modelling and is used to show how much torque is missing based on how far spark advance is from MBT figure and how much timing to remove or add to achieve the torque change required.

Is open loop a factor in timing calculations? Is there more factors and influences I am missing or possible tables.

I just realised I am not able to upload the spreadsheet as it is nearly 2mb in size. I will have to look at another way to make it available so see if anyone can see any errors.

[IH8TOADS]

The boarderline calc and the mbt calc are compared and the lower of the two values are used. Your calc is pretty close tho and should give a close comparison. In regards to knock, on the factory calibration quite a bit is added back in after a knock event, something like 8 degrees so I reduce this to 1 and have about 3 degrees max retard. If you're getting knock retard and you know it shouldn't, start increasing the gain to de-sensitize the thing!
Edit, cat protection involves the PCM dumping fuel to cool it off, spark is generally only referred to on a cold start when it is used to fire off the cat.

[IH8TOADS]

The boarderline calc and the mbt calc are compared and the lower of the two values are used. Your calc is pretty close tho and should give a close comparison. In regards to knock, on the factory calibration quite a bit is added back in after a knock event, something like 8 degrees so I reduce this to 1 and have about 3 degrees max retard. If you're getting knock retard and you know it shouldn't, start increasing the gain to de-sensitize the thing!

[jeremy]

I have attached a link with what I am trying to do, if you can see any issues with the calculations is it possible for you to let me know. My visual basic skills are poor and that is generous. I found some code and then modified the code, so it may have some issues with the parts I modified.

Spark Advance.xls

I am placing the tables used in the log in table 1 and then the modifying the tables in table 2 so I can see what effect the changes will have on the timing by changing one of the values in the table and then seeing which spark advance values would have been affected. As the tables are complex in the way they interact, I decided it would be easier this way. But I noticed when I compare my calculated values with the log they are not similar. I also notice that the spark advance is larger than in the MBT value I calculate which is different to what I have heard to be true.

[IH8TOADS]

I have attached a link with what I am trying to do, if you can see any issues with the calculations is it possible for you to let me know. My visual basic skills are poor and that is generous. I found some code and then modified the code, so it may have some issues with the parts I modified.

http://members.westnet.com.au/christ...arkadvv0.1.xls

I am placing the tables used in the log in table 1 and then the modifying the tables in table 2 so I can see what effect the changes will have on the timing by changing one of the values in the table and then seeing which spark advance values would have been affected. As the tables are complex in the way they interact, I decided it would be easier this way. But I noticed when I compare my calculated values with the log they are not similar. I also notice that the spark advance is larger than in the MBT value I calculate which is different to what I have heard to be true.

That's a good spreadsheet and a good start. BUT your tune is very average and it would be very slow. You need a lot more timing. Do some data connected scatter graphs of rpm (x axis) and timing (y axis). You should log MBT and make sure it's higher than logged spark advance. Also log MAP (as you have) and BAP (barometric absolute pressure) if you can. Then do a calculation in excel of boost in PSI = (MAP - BAP)*0.49109778

With the load i'm seeing i'd say you knock sensor is fairly active too. What is the max retard and adder after an inferred knock event?

Attached is a boost and spark log (from dyno) of a 310rwkw tune. Normally I would have more timing at 4000rpm and on-wards. 4 degrees more at 4500rpm wouldn't cause ping in the case of this car. This style of spark curve is what I aim for on-boost, normally its a little more flat though.

[jeremy]

The tune is not my tune, it is a old log and tune I found on this website. The graph should be easy to generate from the data in excel, i will just have to chose a range. I use baro of 101 as a default. I am not able to log retard, MBT and i do not think baro either. Do you know if the new betas have added any more values that are able to be logged. I will add the knock tables to the spreeedsheet tonight that will be easy to do and these can then show the possible knock retard if knock is detected. When knock is detected is the timing retard by the maximum amount not a percentage depending on the strength of knock and then increment back if knock is no longer found over time period.

[IH8TOADS]

Oh ok....that tune should be removed then! Not sure about the logging I use an SCT flash2 and it's livelink software. The knock sensor senses the noise when the engine is firing and compares it to when the engine is not firing.....the sensitivity / loudness of knock is then used to determine how much to retard. Yeah it increments it back in as you said. I set the increment to 1 and max retard to 3 across all RPM.....I do vary it a bit tho depending on what im tuning. The factory BA sensor is very active so if you increase boost and add some hardware mods then pretty much max knock will be pulled at high loads. You will need to raise the sensitivity to combat this.

[jeremy]

I would never suggest anyone use the tunes posted on the net with out having a look at them or asking the correct people, it is noted in the post with the tune that knock is present. Is the ignition always retarded to maximum in the knock table when knock is first detected or is there a scale depending on the level of knock.

[IH8TOADS]

level of knock...its scaled. I didn't have a real good look but im surprised that tune knocked, it had no timing at peak load.

[lukearmstrong]

i had det in a log once with a noisey idler pulley bearing tune had 4 degrees at peak load at about 4 grand it'd knock, 16 psi, althugh had a close look changed pulley and problem solved

[IH8TOADS]

Could you hear it knock? The BF has two sensors as opposedto one in the BA so it's much more reliable and easy to calibrate. Sounds like a false knock reading much like that seen in the ba or if you run noisy aftermarket internals. Good u got it sorted, sounds like a real nice car :-)

[jeremy]

Thanks Luke for clarifying the knock, i hope you do not mind me using your log. I am not trying to look at the quality of the tune just trying to validate the theory of the spark advance calculation. If the calculation is as simple as it seems it should be possible to replicate from the logs. It seems strange that it is not so easy to replicate in my calculations (possible it is my fault). I am assuming that a+b=c but it looks like there is also d and I am not able to see d. If the tune is bad but avoids detonation it should not matter in the calculation as no one has raised any other possible factors in the advance calculation. I am starting to think the knock sensor is playing a role in the timing changes.

Will hopeful have an update to the spreadsheet soon with a graph to show the advance figures and knock added to show the possible maximum retard at that point. I may try to look at boost over the long weekend. With limited logging it will be hard. May have to go back and start to learn about PID's to get a much better understanding of the theories and what I need to look at.

[IH8TOADS]

Focus on WOT timing when you're in open loop. You need to flag open loop so you know when the adder is working. IMRC is only active above 4000rpm iirc and the EGR adder is not relevant. There is Also an adder for emissions for low rpm mid load. There is also the tip in retard and decel spark that come into play when conditions are met And these don't form part of the spark calculation! Just keep it simple and still to wot runs initially. That log doesn't have really low loads like you see on the street iirc so it's most likely a dyno run.

[lukearmstrong]

Could you hear it knock? The BF has two sensors as opposedto one in the BA so it's much more reliable and easy to calibrate. Sounds like a false knock reading much like that seen in the ba or if you run noisy aftermarket internals. Good u got it sorted, sounds like a real nice car :-)

was until it was smahsed in the rear bu a R33 on a antilag cruise ruined my whole life really might trade her on FG, so much beeter base to work from target will be 400-420 RWKW ifi end up buying one itll be auto too.... cheers

[jeremy]

I have posted a update to the previous excel file under the previous link, it now shows open loop based on the table in the fuel calculation 1 for open loop and 0 for closed loop. I have added a graph to show an example of what can be graphed. I have also added a graph to show what changes in table 2 will have on the total calculated timing based on the formula discussed in previous posts. If you can think of anything that should be added or is missing then let me know or anything else that seems incorrect.

[IH8TOADS]

Don't over complicate it. All I do is set-up a spread sheet with all the main spark tables and then have a final spark table that has the load out puts on the Y axis and RPM on the X (i.e. same as the BKT and MBT tables). I then calculate the final spark (i assume open loop) for each load/RPM point, taking into account all the adders and modifiers. I also know what air temp and coolant temp my car runs at (from numerous data logs) so I use constant values for the air and coolant temp correction. I then graph it and make relevant changes to the boarderline knock table. Load, RPM and Spark advance is what should be graphed as these are easily logged and can be compared to your 'simulated' spark advance.

[jeremy]

I was intending to simplify the changes and not just make blind changes with no way of managing the changes or certainty of the influence the changes would have had on spark. Many tables influence one another in creating a final spark advance value, I wanted to be sure that the formula I am using is correct and that is why I was trying to simulate the calculation before putting too much faith in the formula. The intention was to use the tables with a 1 in the name as the tables that were used in the log. Then copy the tables with a name of 1 to tables with a 2 in the name and adjust the tables in 2 until I reached results that looked like what I wanted and then do another run with those values and repeat the steps until I have results that I am happy with in the logs . The 1st graph has many values so that it is possible to see what factors are influencing spark. Graph 2 shows the calculated spark from tables with a 1 and the calculated spark from tables with a 2 to allow comparisons and see the impact of the changes.

I am not sure that creating a table of timing values is easily done as I would need Load, TPS and RPM to use the calculations I have used. I am guessing you could create histogram of TPS in the scanner using load on one axis, RPM on the other and then just look up TPS based on load and RPM and use that value for the lambda calculation as lambda can have quite an influence of the timing values.

[IH8TOADS]

Yeah you have definately got the thinking part right, it's just that your using a very average log! If i'm looking correctly, I see 14 degrees max knock retard lol. You need to work with a decent tune and a decent log. The actual spark advance is terrible because you can't quantify the knock retard and if the gains and/or sensitivity of the KS has been left factory than its going to cause carnage on the spark curve at WOT which would explain the mess! I know I sound like an ar$e but I hope that helps your cause.

To be able to simulate a spark outcome accurately you definately need to know load, rpm, IAT, ECT and knock retard. You will need to know exactly what load is at particular RPMs.....so yeah histogram style. Logging your car will give you this info. I log outside of HP tuners as I can get ford specific PIDS / DMRs which helps. But I do all my logs at WOT so I know it's max throttle.

You won't need to know the precise TP for lambda correction as the correction only occurs in open loop (i know the switch to OL is TP based but ignore this for now). Focus on WOT because that's what is most important, in which case just look at your commanded lambda in your base fuel table for a particular RPM and then look up the lambda correction table and you can determine the exact correction. You have to do so interpolation between breakpoints and I make the assumption that the PCM uses a linear relationship.

[jeremy]

I am happy for any advice and guidance, I do not think you sound like an ar$e. If all you said was you are wrong and offering no help, I may think differently. You have been very helpful and patient. It seems as though logging is limited in HP tuners and it may be worth looking at other loggers.

When I was talking about histogram, I was talking about using a calculation to create one table for timing from all of the tables using the formulas I have been using with RPM and load as scales and advance as values. The values would be for one IAT and one ECT and represent the calculated advance at that point. The values would not be high on the accuracy scale but would be a guide and allow obvious problems to be spotted. I may go ahead and create an example so you can see what I am talking about as it would be easier to show than write.

I have heard of another spark advance added after 4000 rpm but I will have to investigate that.

[IH8TOADS]

Yeah thats what I do just make tables that looks identicle to the boarderline knock table but has values IAT & ECT corrections, variable cam correction, lambda correction etc and then you have a final spark table that simply adds all these up and gives you a simulated output for each load and rpm. That way you can compare what yopu put into the boarderline knock table (the one you change the most) and what you would see at the engine! It's also good when you're tuning an unfamiliar model and you want to compare the factory timing to an edited tune.