219 replies to this topic

[zanzarah]

very interesting. I'm keen to try and make a one way valve arrangement for the ICV, I reckon that is probably the first place to find improvements in the system.

This is a definite way to simplify the sequential system,as it will also allow you to remove the pre-spool dump valve.
There are several valves available already that are suitable to use.These are essentially non return check valve and come in various types/configurations.
I settled on a 50mm light spring (0.25 lbs) air check valve as I wanted confirmed closure when the secondary turbo was not spooling.

[acres]

I settled on a 50mm light spring (0.25 lbs) air check valve as I wanted confirmed closure when the secondary turbo was not spooling.

Does this valve you have used have a diaphram that lifts up and down or a gate that swings?

[tmh983]

This is a definite way to simplify the sequential system,as it will also allow you to remove the pre-spool dump valve.
There are several valves available already that are suitable to use.These are essentially non return check valve and come in various types/configurations.
I settled on a 50mm light spring (0.25 lbs) air check valve as I wanted confirmed closure when the secondary turbo was not spooling.

where did you source this check valve? I found some that are designed for spa bath systems that look like they could work. I've been thinking about designing my own, but that could be a lot easier said than done...

Have you installed this already? what sort of difference did it make?

[zanzarah]

Does this valve you have used have a diaphram that lifts up and down or a gate that swings?

Diaphagm with adjustable spring tension.
This valve comes in both design configurations, but the gate swing design was less reliable and slower in closing response.

[zanzarah]

where did you source this check valve? I found some that are designed for spa bath systems that look like they could work. I've been thinking about designing my own, but that could be a lot easier said than done...

Have you installed this already? what sort of difference did it make?

As a concept test I checked and bench tested quite a few valves.This included spa bath air valves and theses have proven to be the most suitable.Light weight,rated at 150lbs pressure with good bore sizes and easy to machine and modify to fit.
The valve is installed and performing very well.I originally tested it with the pre-spool/dump valve instu, but this has been removed as it becomes redundant.
Hose No 22 to PDS now comes directly off the 2nd turbo compressor boost port.

[Shadoxity]

Had a quick flick through the thread, great work and love all the info you are posting.
Now for the below

By pre-spool I mean when the solenoid in the black box begins to operate the positive pressure side of the actuator so as to feed the secondary a small amount of exhaust just to get it rotating before the valve is fully opened. I'm positive this occurs at around 3500rpm, up until this point the ecv is fully closed and boost is controlled via the wastegate. I would like to eliminate the use of the wastegate until the ecv is fully opened and instead activate the pre-spool at a much lower rpm

All of this you can change in the ECU (open source tuning) Depending on the conditions it will actually start feeding the secondary as low as 2000rpm!

Mine standard would start opening the ECV just after 2000rpm, and i changed this so it couldnt untill about 2800rpm just to experiment and that helped with the primary coming on to boost.

I am only in the middle of tuning my B4 at the moment, but i have the secondary turbo coming on at 3800rpm currently. Its pretty easy to modify the changeovers etc

Do you currently have a tune? Are you tuning it yourself or what?

Would be happy to lend a hand in open source tuning if you are interested.

The pre dual spool mode is set via a table which goes off engine load. You can make that start as late or early as you want to.

Hope this helps at all

[acres]

I wish it were that simple. My car is a gen2 so no obd2 and no open source tuning unfortunately. I will be purchasing a megasquirt 3 + ms3x in the near future though. This combo will have enough inputs and outputs to allow me to set it up and tune how ever I like.
Thanks for the offer though.

[Shadoxity]

I wish it were that simple. My car is a gen2 so no obd2 and no open source tuning unfortunately. I will be purchasing a megasquirt 3 + ms3x in the near future though. This combo will have enough inputs and outputs to allow me to set it up and tune how ever I like.
Thanks for the offer though.

Oh bummer!

Good luck with it

Im interested to read how this goes

[Shadoxity]

Hey Acres,

Just wondering if you made any progress on this? maybe a dyno graph?

[acres]

Hi. Haven't made much more progress, I have found myself very time poor the last 7/8 months however I have recently purchased a MS3X. Megasquirt isn't quite a plug and play unit though so the initial install and setup is proving to be a bit time consuming but it is getting there. I finally managed to get a clean crank and cam signal and have now moved on to adjusting the cranking and idle settings. This has gone well as it now fires up and hot restart's pretty quick and also has a pretty stable idle. Just need to wait for a few cold mornings to finalise the start/idle tune as the coldest morning so far has only been 20C.

Edit: Just had a thought. Since your open source tuning your B4, would you mind giving me a hand with something? Would it be possible for you to pull the "cranking injector pw/%" for the various coolant temps from your ecu? I would like to compare with what I already have and use them to set the cold start pw under 20C. That would be a big help and save me a lot of time as you only get 1 or 2 true cold starts a day and you basically have to guess the correct pw until it starts. Thanks

[Shadoxity]

Hi. Haven't made much more progress, I have found myself very time poor the last 7/8 months however I have recently purchased a MS3X. Megasquirt isn't quite a plug and play unit though so the initial install and setup is proving to be a bit time consuming but it is getting there. I finally managed to get a clean crank and cam signal and have now moved on to adjusting the cranking and idle settings. This has gone well as it now fires up and hot restart's pretty quick and also has a pretty stable idle. Just need to wait for a few cold mornings to finalise the start/idle tune as the coldest morning so far has only been 20C.

Edit: Just had a thought. Since your open source tuning your B4, would you mind giving me a hand with something? Would it be possible for you to pull the "cranking injector pw/%" for the various coolant temps from your ecu? I would like to compare with what I already have and use them to set the cold start pw under 20C. That would be a big help and save me a lot of time as you only get 1 or 2 true cold starts a day and you basically have to guess the correct pw until it starts. Thanks

Hey Man,

The definitions are not really complete for the B4 maps, I am thinking this will be the map you are after, but let me know. If not ill try to find another.
It was the only real map i could find that related to cranking vs rpm vs coolant



Uploaded with ImageShack.us


Let me know if you need anything else.

Im overseas at the moment, but cant wait to get stuck into my car when i get back (end of march) and get some more power out of her. Think ill be modifying some of the up pipes and whatnot.

[acres]

Thanks for the reply, appreciate it, but that's not really the map I was looking for. That looks more like an enrichment map for sub-zero temps based on crank speed. The map I'm looking for should just show injector pulse width in ms (or injector %) Vs coolant temp. The temps should range from about -40 to about 100.
Thanks
Aaron

[Shadoxity]

This will probably 1 of the ones you want.
These are about the only coolant related fueling maps in the b4 map.

[acres]

Hi.
Top one looks like an acceleration enrichment based on coolant temps. This is most likely added to the base acceleration enrichment.
Bottom map looks like an AFR compensation based on coolant temps. This most likely adjusts the AFR to prevent the code from trying to negate the effects of the warm up enrichment while in closed loop.
I have actually now downloaded the romraider editor but like you said the b4 definitions aren't complete. In my case when I tried to download them all I could open was the trouble code list. So I downloaded the definitions for a JDM GTB with Twinscroll assuming this would be a 2L too. These are the maps I pulled out of that:



The 2 top maps are the cranking pulsewidths based directly on coolant temp (although I have no idea why there is 2 different ones and how the ecu picks which one to use) and the bottom map is the warm up enrichment used after the engine starts. This one is a multiplier that is added to the base VE map.
These settings are actually pretty close to what I have so far in Tunerstudio so maybe I might just pick one of those maps to copy and see how it goes. It would be interesting to see if the B4 maps are similar to this one too.

Actually would you have the ignition timing maps too? I know the B4 runs pretty advanced timing from factory so it would be handy to see the timing limitations I could use for a base start. I have also just guessed these, except for at idle where I just kept adding advance until I achieved the lowest vaccum, then backed it off a degree or two.
Thanks
Aaron

[Shadoxity]

That picture is a tiny size, cant see any detail in it PM me your email address and ill send you an xml def file for the b4 if you like.
You will see that it is laid out very different to normal romraider maps, so it makes tuning a lot trickier!

The only other real 2d map i can see for coolant is closed loop lambda coolant temp comp.

Here is the timing tables. The ias open is during twin mode and IAS closed is during primary only.
Note that these are very bad and have way to much timing std... which is why most b4's ping.

If you take pretty much like 5-6 degrees off from any load over about 3 then you should be pretty close to something that wont ping.




Just a note about engine load, how it is calculated here is different to normal romraider maps. After a bunch of digging previously it pretty much works out to be double what the normal load is you would log on a romraider map.

So 4 load on these maps is roughly equivelant to 2 load.

[tmh983]

I think from my (limited) experience, the cranking pulsewidth only really needs to be enough to get the engine to fire cleanly. Its the post start enrich and cold running and base maps that you really need to get right, in order to get it to start perfectly every time. Here is the order I think you need to tune it:
1) Get the base fuel maps right first. There is no point trying to tune the cold start unless these maps are correct.
2) Roughly get the cranking pulsewidth into a region where it will reliably fire, i've found this is pretty forgiving, start off with roughly equivilant to the pulsewidth at 100kPa map pressure (absolute).
3) Rough up the cold running map, i'd start with about +30-40% fuel at 20deg and come down linearly to 0% extra by about 60deg.
4) Now get the post start enrich map right, this makes a big difference in getting it started cleanly
5) Now you can tidy up the cold running map a bit.

Overall it is sort of an iterative process, like u say, there is really only 1 chance every day to get the cold start right. Take a data log of the start every day (including afr if you can) and you can use this to help tune it.

With the idle ignition timing, pull it back at least 5-10 (maybe more) deg from the normal running value. I'm not sure if the megasquirt uses ign timing idle control, but you will get the best results using ign timing as the primary idle control, and the idle valve is secondary to that. For example my old sr20 motor ran about 35deg advance at light load/rpm, but we had the idle base at 15deg. This of course slightly increases the amount of air and therefore fuel you need to keep the motor running, but the idle will be heaps smoother and more stable, especially when the air con clicks on. This way if the load on the engine changes, the ecu only needs to advance the timing to compesate, which comes into effect on the very next spark event. I could get this motor to idle perfectly smoothly at 17:1 afr using ign control. With fixed idle ign, and using only the idle valve to control, the best I could get was about 13.5afr.

[tmh983]

those subaru maps are a great place to get a good idea about base values for timings, but they are measured in grams per cyl charge, and the i assume the megasquirt is using speed density tuning (ie. map sensor). So u will have to convert the g/charge into a pressure to be able to use the subaru maps. air is 1.225 kg/m^3 at sea level, 15deg. so 500cc of air at 100kpa is 0.6125g. Actually, although it says grams/cyl, i suspect it should say grams per total engine cycle. So 100kpa map is roughly equal to 2.45 grams per total engine charge, which is roughly midrange in the subaru map which seems about right.

Have you installed a inlet air temp sensor anywhere? Speed density tuning really needs temp input to work properly. Air at 35deg is 10% less dense than 15deg and therefore the ecu needs to apply the appropriate fuel correction based on air temp.

TLDR: put those maps into excel and convert g/charge to map pressure, then feed them into the megasquirt with a bit of timing pulled out of the high load bits. Instant base tune!

[Shadoxity]

those subaru maps are a great place to get a good idea about base values for timings, but they are measured in grams per cyl charge, and the i assume the megasquirt is using speed density tuning (ie. map sensor). So u will have to convert the g/charge into a pressure to be able to use the subaru maps. air is 1.225 kg/m^3 at sea level, 15deg. so 500cc of air at 100kpa is 0.6125g. Actually, although it says grams/cyl, i suspect it should say grams per total engine cycle. So 100kpa map is roughly equal to 2.45 grams per total engine charge, which is roughly midrange in the subaru map which seems about right.

Have you installed a inlet air temp sensor anywhere? Speed density tuning really needs temp input to work properly. Air at 35deg is 10% less dense than 15deg and therefore the ecu needs to apply the appropriate fuel correction based on air temp.

TLDR: put those maps into excel and convert g/charge to map pressure, then feed them into the megasquirt with a bit of timing pulled out of the high load bits. Instant base tune!

Close but the load values in the b4 map is essentially double the normal load calculation. So 2 load normally = 4 (or close enough to) on the b4 map.

But yes, good info and when i had the sm4 in my track car 10-15 degrees at idle would be about right to get her to idle.

[acres]

Oops. That pic didn't work too well at all did it. I just took a screenshot and uploaded it to photo-bucket as a jpeg - I'll try again.
Those bad timing maps are great as now I know where the bad limits are to set a base map. Thats great info Tim on converting MAF density to Speed density as you are correct Megasquirt uses a MAP and MAT sensors. It does have provisions to use a MAF sensor instead but as if you would. I hadn't thought about how to convert the load readings yet so thankyou as that has saved me a bit of work there. I do have the MAT sensor but I haven't bothered properly installing it yet. Right now it is sitting in the passenger footwell temporarily wired in. This is all good as if you look at the calculations megasquirt uses, air temp only has a minor effect on the final outcome. When I install it properly it will be positioned just behind the air filter just like the MAF.

In regards to the cranking pulsewidths, megasquirt ignores the base VE maps and all other enrichments until after the engine fires. The pulsewidths used are exactly what you tell it to use at each specific temp. You are given a 10 point table for cranking whereby you can enter any coolant temps and any pulsewidth for each temp. There is then a linear interpolation placed between each adjacent temp point, creating a non linear curve from the lowest set temp to the highest. I have found that the pulsewidths required have to be pretty spot on to get the engine to fire. Even just 0.5ms to rich or lean and the engine wont fire.
Once the engine fires and rpm increases past the user set cranking rpms, then the code will access the base VE maps, add the warmup enrichments and then add the afterstart enrichments on top of that. I have found these 2 enrichments to be pretty forgiving as they don't seem to need to be so accurate.

There is quite a few options for idle control including an closed loop PID idle control, 2 seperate idle specific VE maps and an idle advance feature but for now I am just using the most simple open loop control which just sets an idle valve duty dependent on temp. This still works quite well though as it has an idle up and delay feature for both the cooling fans and AC. I see what you are saying about using timing as idle control though and I will eventually implement this once I switch to PID control. For now though it is just accessing the base timing map which is about 26* in that region.

[Shadoxity]

On a side note...

If you are still using the std injectors (440) you could probably look at a wrx map for the injector pulse widths (cause they are 440's to) and its a pretty similar engine.