Davidv

One thing that I've been curious about lately is advancing cam timing at part throttle for better fuel economy. 
So I did some tests, where I set the timing all to 0 degrees, then 10 degrees, etc, and then went for a drive at a set rpm on the same stretch of road with cruise control and closed loop lambda turned on.

Interestingly, 0 degrees advance clearly gave the best results. At first I felt satisfied with this, but then the nagging problem in the back of my mind... Toyota documentation says that advancing the cam "about half way" yields best economy.

So I took a closer look at the logs. When the cam advances, at same throttle angle, the MAP sensor reading goes up! In one case, up 60% higher. 
So the ECU is of course trying to dump fuel in, and pulls the ignition timing back which is why the economy was notably worse even though Closed Loop Lambda was trying its best to salvage the situation.
Why does the map sensor value jump up? Because of internal EGR, when you introduce cam overlap, the low pressure in the intake manifold and high pressure in exhaust pulls exhaust gas back into the inlet manifold which raises its pressure.
So this is obviously why the factory ECU uses a MAF sensor rather than MAP - A map sensor is including some "dirty" air in its readings (which has no oxygen left in it) where as a MAF only reads fresh air coming in.
Since the factory ECU only has a narrowband sensor, using the internal EGR method allows it to run a greater airmass to reduce pumping losses while still operating at 14.7:1 as the recycled air has little or no oxygen left in it.
Pretty clever.

So I thought I'd wire a MAF sensor back into the car. 
But this presents the next problem, how do you get the 0-5v signal of the MAF into a grams/sec that the ECU needs for a calibration curve.
So for starters I was just logging raw MAF voltage output, so depending on airflow it spits out somewhere between 0-5v to ECU
My ECU currently has load source as MAP sensor, and one of the values it logs as part of the modelled fuel calculation is "Grams of fuel per cylinder Estimated"
So we need to turn this into a grams per second, so some maths to create a custom field in Megalog viewer and now I've got Grams per second which I can compare to voltage:

Which I can then use as an axis on a scatter plot, which shows me a very rough outline of a MAF curve starting to form...

So I rough out a voltage vs grams per second to put in the calibration in Link

And then go for another drive and do same thing again in megalog viewer...

Starting to look better! 

(It would be cool if PClink allowed Maths functions like this... Just saying)

Then from here have updated the MAF curve again to suit the trend seen there. 
I think another 1 or 2 iterations of this and I'll have that low airflow area cleaned up.
If not, I will just switch to map or alpha N based tune around the areas where it sucks.
But from here, once the MAF sensor data is accurate I'll build secondary ignition and fuel tables which have MAF as the load axis. 
Then I can start experimenting with cam timing at part throttle some more, without my load axis going bananas (map sensor value changing a lot)
In order to find the sweet spot for economy though I really need to play with a few variables at once. 
As when you're introducing EGR gas, you might start getting misfires at 16:1 where as this is most economical if you are running no overlap. 
And when you introduce EGR gas you need more ignition timing as it slows the burn.

So I think I'm gonna make a little box that communicates over the CAN network that has a few potentiometers which log as virutal 0-5v which I can use as trim tables for ignition table, cam angle, goal AFR, and maybe a little display that shows fuel economy.
So I can very quickly go through a lot of combinations while someone else drives with cruise control turned on.
It's a lot of effort to make only a small iterative improvement to the car's economy, as it's already getting 7-8L per 100km if driven nicely. But it's always bothered me that I've had no way to quantify how to best set the part throttle cam timing so I'm thinking there's something to learn here yet.

I'll post the results once I've got my MAF curve dialled in nicely and I've put together a CAN box.
Also, at the same time as wiring in a MAF I thought I'd wire in an exhaust pressure sensor. 

I wasnt sure what sort of pressures I would see on an NA car, so I bought a 30psi sensor. 
I drilled a hole in a spare wideband bung, which then goes to a line of copper tube to cool the gas, then to a rubber line, then to the pressure sensor. 

Results are interesting! At high RPM (Or high mass airflow more specifically) the measured pressure actually drops. Down to 91kpa which was quite literally the last thing I was expecting hahaha.

Thoughts as to why? 

I think the airspeed past the hole in the wideband bung is creating a venturi effect and pulling the air out of the hose, rather than telling me what the pressure is.
So I am thinking that having an angled bit of pipe internally that either faces towards or against the flow will prevent this from happening. Or maybe upsizing the diameter of the tube that I am using as it's very small. (maybe 2mm ID)

The resolution isnt very good with a 30psi sensor so I'll perhaps switch it out for one from an NA car instead. I've got a spare Toyota one here somewhere. It doesnt really show any results that I was expecting though, so I might just ditch it from here.
Was worth it just for curiosity's sake though! 
More useful on a turbo car though of course.
 

 

@CamB you can set the byte order to MS first or LS first to swap the order when setting up incoming frames 
(In G4+ at least)

Hi Marc
Good to hear back from you. How much does your AFR oscilate? Where is the wideband located? Depending on position, you need to adjust Lambda rate and Gain. 
You should do some testing. Tune your VE table with CLL off  as good to your Target AFR as possible. Adjust your target AFR table to the rich side with one task by the amounts listed in the gain table 0,33 0,66 etc. And log, with a high logging rate, how long the sensor needs to get to a stable reading. 
Lets say it needs 0.3 seconds. This means your frequency in the lambda rate table may be max. 1/0,3  = 3.333
This test should be done at different loads and rpms to work out best CLL settings. Best place to do that is a load bearing dyno. 

From my messing around with it, all the raw data in the Generic stream is the second parameter + 256 * the first parameter (in the pair). Interestingly when you set a custom stream the order is reversed... 
Is there a reason not to just use the standard convention? (I guess efficient coding?)

Regarding Lambda control error correction table. I spend alot of time to tune these. The base map is adjusted the wrong way around. Because the error correction tables is a % corretion of actual error, you want big corrections like 15%  on small errors (0.03 lambda error) and small correction (like 5%) at the biggest error on the table. This is because a fuel film built up first in the ports when big correction are applied. This means it needs several burn cycles to get the whole change applied and measured. This means lambda control applies big changes two or three times for big corrections which leads to Lambda oscillation.  On small changes fuel film built up is much less.  Lambda change is done and measured much faster and within same burn cycle. This means the Lambda correction can be set much higher, because the change in AFR is measured instant.
I have attached a tuned example.  

Hi David,
I dont have all answers and our documentation around MAF is a bit thin and barely anyone I know has used it, so I just done a few quick tests on the simulator to get some basic answers.
With a MAF since you are measuring air mass directly so you dont need air temperature (to approximate the mass via density).  If you select " MAF IAT" then the fuel equation is not modified by IAT. - It really just becomes a parameter that is logged and has no effect.
 
I dont know this one confidently but I suspect "Off" is the most appropriate.  I cant see how BAP would affect the measured air mass so I dont think that is useful.
 
It appears to me that if you use Modelled mode then "estimated" is used, if you use traditional mode and have a MAF assigned then "measured" is used.  That means to me that traditional mode is a more true MAF strategy but you then miss out on some of the fuel pressure/density comps.  In modelled mode if I do a test at a fixed mass air flow and change something like "engine capacity" I see "estimated air/cyl" change and pulsewidth change, so it seems the "air density model" is still trying to do its work in the background.
 
If you wanted to use modelled mode then put the charge temp table to zero and make sure your IAT is set to "MAF IAT" so charge temp is taken out of the equation.  Provided the other air density related settings like "engine capacity" are left alone then it should provide a good MAF strategy.   

I asked over at another forum where you asked this, but this is probably a better place... does the change in backpressure have any relativity to the change in cam timing?

Correct, DI9 & 10 just need to be set to off to allow CAN through.

Hi
Another one from another brand of ECU that I find useful (Omex / DTA)...
An external box that plugs into the Laptop that allows you to alter fuel / ignition in the cells you are currently in, then "enter" them into the map. So you can quickly turn a knob to get the fuel and timing how you want it very very quickly.

Great thing about this is that you can assign the knobs to various parameters, like fuel, ignition, boost, injection timing etc etc.
Who else thinks this would be a great tool for tuning the Link?
Cheers, Grant

Hey people, 
Just thought I'd post up a quick note about something I did recently that worked out well.

I was wanting to optimise ignition timing for cruise, so using some switches on my dash to trigger a combination of datalogging, 4D ignition, 5D ignition and the 2nd ignition table set to overlay mode.

With the idea that I could add or remove timing from the main table in varying amounts without having to stop the car, and datalog the whole lot easily.

Like so: 



Since you can turn on more than one ignition trim table at once, using those three you can get a combination of timing settings which I then marked on the switches. 

So +1 degree, + 3 degrees, +5 degrees, etc. 
I completed a run on a particular stretch of motorway that has lots of ups and downs, with cruise control turned on at a speed that's at 3250rpm in 6th gear.
Then flicked the first switch, did it again.
Flicked second switch, did it again, and so on. 

When home looking through the data, bringing up a time plot with instant fuel consumption and throttle angle it was very easy to see which timing gave best economy. 


However a secondary method of checking fuel consumption overall is to create a "statistics" page and bring up wheel speed and instant fuel consumption, and look at the mean values:


Then from here I've made a quick excel sheet that converts it to Litres per 100km:



Then from here, collated the results from each run.


So based on this it's pretty clear that an additional 9 deg advance made the engine pretty happy on those particular cells, so updated my ignition table and readjusted some of the surrounding cells to more sensible values too.

It was a fairly time consuming exercise but it's amazing to see how much fuel I have been throwing down the toilet just based on under advanced ignition. 
It was also interesting to see that at 100kpa my car only has 14 deg ignition at that rpm, but then by 70kpa it's wanting 33. (The goal AFR changes though, to be fair... 15.2:1 goal AFR for cruising)

Since changing the timing the car is a lot quieter too! 
I am guessing because when you dont have enough timing, the flame front is still expanding when the exhaust valves open. So instead of having energy push the piston down, it's coming out the exhaust as noise and heat. 

Here is where you adjust that:

The logging will still be limited to the 100Hz or 40Hz limit regardless of the data source.
Although the ECU can still receive, process and use the data much faster if it is used for internal calculations - it will only be time stamped and written to the logging memory chip 100 times per sec.   
The maximum rate for transmitting CAN messages out of the ecu (for say an external logger) is limited to 200Hz.
 
 

The only thing I can see that's varying similar-ish to your RPM spikes is changes to your dwell time



Maybe try change the value in the table so it's a bit more stable at idle.

Hey people, 
Just thought I'd post up a quick note about something I did recently that worked out well.

I was wanting to optimise ignition timing for cruise, so using some switches on my dash to trigger a combination of datalogging, 4D ignition, 5D ignition and the 2nd ignition table set to overlay mode.

With the idea that I could add or remove timing from the main table in varying amounts without having to stop the car, and datalog the whole lot easily.

Like so: 



Since you can turn on more than one ignition trim table at once, using those three you can get a combination of timing settings which I then marked on the switches. 

So +1 degree, + 3 degrees, +5 degrees, etc. 
I completed a run on a particular stretch of motorway that has lots of ups and downs, with cruise control turned on at a speed that's at 3250rpm in 6th gear.
Then flicked the first switch, did it again.
Flicked second switch, did it again, and so on. 

When home looking through the data, bringing up a time plot with instant fuel consumption and throttle angle it was very easy to see which timing gave best economy. 


However a secondary method of checking fuel consumption overall is to create a "statistics" page and bring up wheel speed and instant fuel consumption, and look at the mean values:


Then from here I've made a quick excel sheet that converts it to Litres per 100km:



Then from here, collated the results from each run.


So based on this it's pretty clear that an additional 9 deg advance made the engine pretty happy on those particular cells, so updated my ignition table and readjusted some of the surrounding cells to more sensible values too.

It was a fairly time consuming exercise but it's amazing to see how much fuel I have been throwing down the toilet just based on under advanced ignition. 
It was also interesting to see that at 100kpa my car only has 14 deg ignition at that rpm, but then by 70kpa it's wanting 33. (The goal AFR changes though, to be fair... 15.2:1 goal AFR for cruising)

Since changing the timing the car is a lot quieter too! 
I am guessing because when you dont have enough timing, the flame front is still expanding when the exhaust valves open. So instead of having energy push the piston down, it's coming out the exhaust as noise and heat. 

Slightly off topic but perhaps relevant to the intentions of this thread.

I did some tests with a thermal camera, with the fuel pump running and the engine turned off. 

In this case, what heats up the fuel is/was when the fuel flows through the hot fuel rail bolted to the head, and then circulating down to the fuel tank and back through hot fuel rail again and so on.

Which is even worse when you have a surge tank to the fuel rail, which is probably why people think its the pumps making all of the heat.

If you slow down the speed of your fuel pump, and the fuel is flowing slower through the fuel rail then you're still transferring the same BTU of heat. (Half as much fuel flowing for twice the time through the hot rail)
If I leave the fuel pump running with a cold engine, it took something like half an hour to increase by 5 degrees. 
But with the motor running the fuel temperature increases almost immediately. 
Insulating the fuel rail from the heat of the head and I never had fuel temp issues again.



Fuel rail is dark blue item near the left hand side after insulating it with thermal paint, and putting plastic washers under the bolts that hold the fuel rail to the head.

EDIT: 

Found a pic of a close up of the fuel rail, you can see the hot bolts (now insulated) against the rest of the rail (darker colour)


 

Mike have see the Injector dynamics PRI videos. Worth a watch if you haven't  
 

Hey people, 
Just thought I'd post up a quick note about something I did recently that worked out well.

I was wanting to optimise ignition timing for cruise, so using some switches on my dash to trigger a combination of datalogging, 4D ignition, 5D ignition and the 2nd ignition table set to overlay mode.

With the idea that I could add or remove timing from the main table in varying amounts without having to stop the car, and datalog the whole lot easily.

Like so: 



Since you can turn on more than one ignition trim table at once, using those three you can get a combination of timing settings which I then marked on the switches. 

So +1 degree, + 3 degrees, +5 degrees, etc. 
I completed a run on a particular stretch of motorway that has lots of ups and downs, with cruise control turned on at a speed that's at 3250rpm in 6th gear.
Then flicked the first switch, did it again.
Flicked second switch, did it again, and so on. 

When home looking through the data, bringing up a time plot with instant fuel consumption and throttle angle it was very easy to see which timing gave best economy. 


However a secondary method of checking fuel consumption overall is to create a "statistics" page and bring up wheel speed and instant fuel consumption, and look at the mean values:


Then from here I've made a quick excel sheet that converts it to Litres per 100km:



Then from here, collated the results from each run.


So based on this it's pretty clear that an additional 9 deg advance made the engine pretty happy on those particular cells, so updated my ignition table and readjusted some of the surrounding cells to more sensible values too.

It was a fairly time consuming exercise but it's amazing to see how much fuel I have been throwing down the toilet just based on under advanced ignition. 
It was also interesting to see that at 100kpa my car only has 14 deg ignition at that rpm, but then by 70kpa it's wanting 33. (The goal AFR changes though, to be fair... 15.2:1 goal AFR for cruising)

Since changing the timing the car is a lot quieter too! 
I am guessing because when you dont have enough timing, the flame front is still expanding when the exhaust valves open. So instead of having energy push the piston down, it's coming out the exhaust as noise and heat. 

Hey,
I think this might have been suggested before. But once you start to understand the wall wetting concept, it starts to explain a lot of other concepts which are described seperately in the current software. 
Example - accel enrichment compensates for changes in wall wetting, but in a fairly crude manner. 

Cold start cranking enrichment is essentially compensating for fuel film build up over the initial cranking period on a cold engine, and the long time period of "tau" on a cold motor.

Fuel cut on decel doesnt really go into this, but if there was awareness of the fuel film depleting during this period then "accel enrichment" would naturally increase again after decel fuel cut. 

It feels like if this was modelled many of the other enrichment features etc could be done away with or minimised. 

Here's a great article on it.

http://www.megamanual.com/ms2/xtau.htm

 

Instead of duty cycle have a look at the log of Injector Actual Pulsewidth
When the injector cuts you'll see a delay in the reading changing which I believe is for two reasons: 
1. The amount of gas coming past the wideband sensor under these conditions is very low so its slower to respond

2. There is still the fuel film build up on the port walls which is coming into the cylinder, although the amount of fuel is low the amount of air is very low too when the throttle is shut.

But I see a delay in the readings here, and look how long it takes to register again once the injectors turn back on.


 

Hi agree that there is some space for improvements on Accel enrich. 
If you enrich enought that you don't see a lean spike at the begining of rappid TP change, AFR's goes usually to low. Even if you set decay rate to 100% and enrich only on one event. 
I can only explain that with fuel film which is built up during extra accel enrichment injection which is then sucked aways on the follow engine cycles, because flow goes up with RPM. 
Some kind of asynchron enrichment may help to built up a fuel film. Or a full fuel film modell. 

Hey,
I think this might have been suggested before. But once you start to understand the wall wetting concept, it starts to explain a lot of other concepts which are described seperately in the current software. 
Example - accel enrichment compensates for changes in wall wetting, but in a fairly crude manner. 

Cold start cranking enrichment is essentially compensating for fuel film build up over the initial cranking period on a cold engine, and the long time period of "tau" on a cold motor.

Fuel cut on decel doesnt really go into this, but if there was awareness of the fuel film depleting during this period then "accel enrichment" would naturally increase again after decel fuel cut. 

It feels like if this was modelled many of the other enrichment features etc could be done away with or minimised. 

Here's a great article on it.

http://www.megamanual.com/ms2/xtau.htm

 

Cool man, sounds like you've been getting out there and enjoying the car. Isn't it funny how breaking the car miles from home opens the opportunity to meet like minded people and make new friends. Thats what I love about cars, the joy of machine and the fact it brings people together.
Quite like the idea of the K type on the brake disc, I would never have thought of that, you could always drill a small hole into the friction material of the brake pad and force the tip of the thermo in there?
I'd sat stick with the car, i've always quite liked the idea of taking a car and evolving it continously which is what you've been doing and thats always appealed to the engineer in me. You've made something unique and its truly yours, i like that.
Keep going with it buddy and next time get some on track footage 
  

Oh hey! 
Havent been up to much with the car lately... I moved to the South Island to take a job in the motorsports industry, so funnily enough I actually ended up moving right near that track shown on the post above. 
It was awesome down there, but some life complications meant I needed to come back.
Some interesting stuff has happened between now and the last time I posted though I guess...

Firstly I had the windscreen smash which was fun! 



Then I competed through the time atttack series which was good fun, cut a little bit short by me moving away. But I was determined to make my "Street class" car as streetable as possible... Drove it to-from events and even used the car for camping near the events.



At one of the events I had one of the outer injectors in my staged injection setup fail - due to a rusty fuel tank and blocked up fuel filter which obviously let some debris through. 
So I had to roadside retune the car back to primary injectors only, then set a new fastest lap once the engine was running right. 
Since then I switched to a Seimens 850cc "shorty" injector, this gives me more clearance to the bonnet which was needed.
I made up some MSpaint style stickers for the numbers that I needed for the race series, also at the event furthest from home (about 6-7 hours drive) I ended up having the front universal joint in my driveshaft explode which sucked! 



Ended up getting towed all of the way home by some friends who unloaded their car off the trailer so I could get home. Which was much appreciated.

On the ECU side of things, I've got a few changes to plan coming up. 
I have just been working through populating the table to run my tacho as a combined tacho and fuel economy meter, using virtual aux conditions to switch between them automatically depending on what the car is doing. 
So currently I am thinking that when RPM is between 2000 and 4000, rate of acceleration below (something) and speed is above say 60kph the tacho switches over to the fuel economy meter. 
I dont see a way to achieve this apart from wiring two auxiliary outputs to the same tacho input, and switching between them using virtual aux. But I've got some left over so that's fine. 
I also want to setup something using CAN, I have been meaning to double check the accuracy of my analog wideband input so I might kill two birds with one stone and buy the link Can wideband module some time soon.
I've got a lot of DIs left over so will be looking at setting up wheel speed sensors as well, it will be interesting to see how much my LSD is slipping or locking under different circumstances.
It will be interesting to have a play with the traction control system when this is setup, not that a car with this power level particularly needs it. Might come in handy on wet trackdays though! 

I saw a cool idea where someone was using a K type thermocouple dragging against a brake disc to measure brake temperatures. I've been meaning to get some temperature paint or similar to find out how hot the front gets, as I wear out brake pads quite quickly. 
So for sake of interest I might do something similar while I'm setting up wheel speed sensors. Or maybe even a K type just sitting against the brake pad, I guess that's really what you're trying to measure and prevent from overheating.

It might also be fun to setup 4x EGT or 4x wideband, one for each runner but would look at doing this with a CAN based solution.

Apart from that not much else going on, just driving the car and enjoying it.
I like having the car NA and fairly simple but realistically for the time attack stuff I've been doing this means my car is mid pack at best. 
It's heeaapppss of fun though, with some great people and I still wince at the extra engine bay complication of forced induction. As the whole point of this car is to keep things simple. 

I have been having some thoughts lately about starting a new build using maybe a Suzuki cappucino or a Honda Beat. 
Just because they are awesome, and I like the gimmick of tiny cars. Both options will most likely be slower than my existing car haha.
Maybe a Toyota MR-S or similar would be a more sensible starting point.
(Or just go on holidays to Fiji etc instead and enjoy the car I've got!) 

Keep coming back here to check for updates, what you been up to man?