GTX3582r Short Runner TwinScroll Setup
#106
Good news, mrfred verified I had the Omni 4 Bar setup correctly and I determined the warning buzzer on the alarm was going off at 28psi vs 30psi. I can only assume I didn't set up the warning correctly so that is definitely one thing less to worry about. Took some time to tune again today and upped the boost slightly since the motor has almost 300 miles on it. I figure the pistons have gone up and down enough that whatever break in that is going to happen mostly took place. I will wait to turn the boost up to 33-34psi at least for another 100-150miles.
I started today by turning up the boost slightly from ~27psi to ~29psi. Additionally, I increased all the ignition timing values from 280-320 load and 4000rpm and up. I also set my slow knock flash to 3 counts and fast knock to 5 counts. I then took it out and slowly worked up to revving to ~7500rpm making sure AFR was good and no knock count. I then made a 3rd gear pull from 2000rpm and when the car came up on boost the compressor surged pretty bad for a second and then the motor starting out flowing the turbo. You can see in the log below the compressor surge once I start hitting full boost. I was consistently seeing this and the car is hitting 25psi in 3rd gear at ~4000rpm. My old 2L would do the same thing, but with the GTX3576r, but never surged the GTX3582r.
I also have a video of the car surging and you can hear it when the RPM hits ~4000rpm, it sounds like 2 quick flutters and it causes the car to stumble. On the initial passes it was much worse and became less noticeable as the car heat soaked and the spool dropped off a little.
GTX3582r Surge
I was able to play with the gain on the Profec and it decreased the intensity of the initial hit and the car no longer surged, but I am not sure with the gain in this setting it will allow me to run up to 34psi. I might have to change the gain back as it affects how much boost I can run. The log below is with the surge gone. You will also notice that I have upped the timing a degree with the ultimate goal to run 0-1deg @ peak torque ramped to 8deg redline with 33-34psi.
Finally, the VD results for how the car sits at this time running ~28-29psi. I am hoping for ~550@wheels on 93 pump at 33-34psi so I think it will likely meet my expectations. The car pulls strong to redline so just upping the boost 2-3psi today made it feel much stronger than yesterday so can't wait for full boost.
I started today by turning up the boost slightly from ~27psi to ~29psi. Additionally, I increased all the ignition timing values from 280-320 load and 4000rpm and up. I also set my slow knock flash to 3 counts and fast knock to 5 counts. I then took it out and slowly worked up to revving to ~7500rpm making sure AFR was good and no knock count. I then made a 3rd gear pull from 2000rpm and when the car came up on boost the compressor surged pretty bad for a second and then the motor starting out flowing the turbo. You can see in the log below the compressor surge once I start hitting full boost. I was consistently seeing this and the car is hitting 25psi in 3rd gear at ~4000rpm. My old 2L would do the same thing, but with the GTX3576r, but never surged the GTX3582r.
I also have a video of the car surging and you can hear it when the RPM hits ~4000rpm, it sounds like 2 quick flutters and it causes the car to stumble. On the initial passes it was much worse and became less noticeable as the car heat soaked and the spool dropped off a little.
GTX3582r Surge
I was able to play with the gain on the Profec and it decreased the intensity of the initial hit and the car no longer surged, but I am not sure with the gain in this setting it will allow me to run up to 34psi. I might have to change the gain back as it affects how much boost I can run. The log below is with the surge gone. You will also notice that I have upped the timing a degree with the ultimate goal to run 0-1deg @ peak torque ramped to 8deg redline with 33-34psi.
Finally, the VD results for how the car sits at this time running ~28-29psi. I am hoping for ~550@wheels on 93 pump at 33-34psi so I think it will likely meet my expectations. The car pulls strong to redline so just upping the boost 2-3psi today made it feel much stronger than yesterday so can't wait for full boost.
#108
EvoM Guru
iTrader: (1)
Given all the power changes how do you run your cruise area of timing map ? High thirties ? Anything significantly higher ?
edit I didnt see the map.above. doh!
have you looked into the egr cruise timing maps at all they allow some additional timing under cruise conditions
edit I didnt see the map.above. doh!
have you looked into the egr cruise timing maps at all they allow some additional timing under cruise conditions
#109
Given all the power changes how do you run your cruise area of timing map ? High thirties ? Anything significantly higher ?
edit I didnt see the map.above. doh!
have you looked into the egr cruise timing maps at all they allow some additional timing under cruise conditions
edit I didnt see the map.above. doh!
have you looked into the egr cruise timing maps at all they allow some additional timing under cruise conditions
I see the map you referenced and it is zeroed out, but don't know why I would need to run additional advance in that map when I can just run the advance I want in the High Octane Fuel map. I run Shell 93 99% of the time so I don't hardly see a difference, at least on my EVO based on the Haltech work. I do see the octane difference on my GTR (also runs Shell 93) when trying to make 700@wheels so definitely the knock threshold can vary tank to tank pushing the car that hard.
I'll post all the maps when I'm done with the tune if you're interested.
#110
While I am having the scaling issues sorted I am looking at ways to more effectively analyze the data for the SD 2D tuning approach. I use LogViewer software by Epifan for all of my GTR tuning because it is very easy to select portions of your log and analyze various parameters. What I have below is a short cruise log where I typically vary rpm (by gear) and also play around with different loads to check the tune. What I have done below is plot RPM (x-axis) vs kPa (y-axis) and then STFT (z-axis) using the custom graph feature. What the software does is grab all the data for a given RPM and kPa breakpoint and display the data (min, max, avg, count, sum, etc) depending on what you want to see. I chose for it to take the average STFT based on how ever may data points it had collected. If you click on an individual cell it will display all of the associated data i just listed.
If I take a longer more extensive log I could sufficiently populate key areas of the graph of interest such as kPa between 30 - 100 and RPM between 1000 and 4000. Below is how I have my current breakpoints set.
So I know most recommend you adjust the RPM breakpoints first, but I can easily look at all of the data across the RPM range for any of my MAP breakpoints and see if there is an overall adjustment that can be made. So, for example, if all of the STFT values for 60kpa from 2000rpm to 4500rpm range from -6% to -15% then I know I can at least drop the 60kpa load value by 6%. This approach can then be applied to all of the breakpoints in the range of 20-80kpa. I would then update the tune and take another log to then apply the same approach to the RPM VE values. Pictured below is an example file from a short 3 min cruise I took yesterday so I imagine you can collect sufficient data with just a 10-15min log focusing on the 2000-4500rpm range and 20-80kpa range.
If I take a longer more extensive log I could sufficiently populate key areas of the graph of interest such as kPa between 30 - 100 and RPM between 1000 and 4000. Below is how I have my current breakpoints set.
So I know most recommend you adjust the RPM breakpoints first, but I can easily look at all of the data across the RPM range for any of my MAP breakpoints and see if there is an overall adjustment that can be made. So, for example, if all of the STFT values for 60kpa from 2000rpm to 4500rpm range from -6% to -15% then I know I can at least drop the 60kpa load value by 6%. This approach can then be applied to all of the breakpoints in the range of 20-80kpa. I would then update the tune and take another log to then apply the same approach to the RPM VE values. Pictured below is an example file from a short 3 min cruise I took yesterday so I imagine you can collect sufficient data with just a 10-15min log focusing on the 2000-4500rpm range and 20-80kpa range.
Last edited by 240Z TwinTurbo; Feb 11, 2019 at 05:11 PM.
#111
So I did another log, but for about 20 minutes. Don't let all the data scare you and understand this is averaged data that is not necessarily based on steady state conditions so we are looking for trends. Also, we need to understand the ECU is going to correct +/- 25% in closed loop so we don't have to be perfect, but hope to be in the +/- 10% range if possible. Based on this, we see the yellow, light orange, and light green are acceptable corrections because they are in the +/- ~10% range. Darker green and darker orange to red are areas that need help as they exceed 10% STFT.
As I looked at the data (it is cutoff in the pic so you can't see above 3063RPM) I see that the load of 80kpa is on average running lean so I richen up the load % by 6%. Don't see a clear trend for 40kPa - 60kPa so I leave those alone. However, for 2500rpm and 3000rpm I see a general trend these RPM VE numbers are running lean so I up 2500rpm by 5% and 3000rpm by 8%. From a calculation standpoint they could have taken a higher correction, but I usually go part way to see how it responds in the next log.
The reality is that tuning the 2D tables can only get you so far. Correct me if I am wrong, but does the 2D strategy assume the High Octane Fuel map is the pseudo VE table?
As I looked at the data (it is cutoff in the pic so you can't see above 3063RPM) I see that the load of 80kpa is on average running lean so I richen up the load % by 6%. Don't see a clear trend for 40kPa - 60kPa so I leave those alone. However, for 2500rpm and 3000rpm I see a general trend these RPM VE numbers are running lean so I up 2500rpm by 5% and 3000rpm by 8%. From a calculation standpoint they could have taken a higher correction, but I usually go part way to see how it responds in the next log.
The reality is that tuning the 2D tables can only get you so far. Correct me if I am wrong, but does the 2D strategy assume the High Octane Fuel map is the pseudo VE table?
#113
I got the base map from Aaron and he recommended the 2D SD. From my understanding you can actually tune a 2D ROM in 3D by modifying the High Octane Fuel map. I'm going to tune this out as best I can and perhaps in the future I will swap to the larger maps and 3D approach. Definitely the lack of breakpoints is a limiting factor for 2D for several reasons. One is MIVEC, which has a significant affect on VE for certain portions of the map that you just can't account for other than letting the STFT provide excessive (+/- 10%) corrections.
Last edited by 240Z TwinTurbo; Feb 12, 2019 at 12:15 PM.
#114
EvoM Guru
iTrader: (1)
my 3d map looked something like this. I was using a smaller ve map on 8858 which I put a lot of information from many sources together on. Probably worth a check..
i tried the logging of ve by spreadsheet driving around using maf. I didnt have a good understanding of how to use all the maf adjustments so I think I stopped doing it that way. Ultimately that is what I would think would be the best way to implement it
Last edited by 211Ratsbud; Feb 12, 2019 at 06:53 PM.
#115
my 3d map looked something like this. I was using a smaller ve map on 8858 which I put a lot of information from many sources together on. Probably worth a check..
i tried the logging of ve by spreadsheet driving around using maf. I didnt have a good understanding of how to use all the maf adjustments so I think I stopped doing it that way. Ultimately that is what I would think would be the best way to implement it
I've got a lot of time invested in the 2D stuff and I am getting to where the car is actually driving very smooth. Additionally, mrfred is helping me to address the logging issues so I feel pretty good about how this is coming together.
I'm going to read thoroughly through that thread so I at least have an idea of what you are doing and how it is working.
#116
Special thanks to mrfed and JohnBradley for the ECUFlash help and for answering all my questions.
When the tank went below empty at ~480miles on the new motor I decided it was "officially" broken in and time to beat on it. I celebrated by putting almost 7 gallons of 110L in the tank dropping my ethanol gauge from E9 to E1. First order of business was to formally tune out the knock filters. Below is how I approached.
Next order of business was to raise the boost to 33psi and get the AFR and Ignition timing roughly where I want it to be. I had a very hard time with this because I slightly increased the ignition timing to ~1deg below what I was running on my 10:1 motor on 93pump, but this presented challenges due to bad surging in 3rd gear. In order to get the boost up to 33psi I had to increase the gain subsantially, which makes the boost hit harder and thus the car surges. About every 1 in 5 passes the car wouldn't surge so I could get a decent log and then import into VD.
Here is a good log and the basis for the VD results below, but this was with the ignition timing 1 deg lower (4000 - 8000rpm) vs what I am currently running. I also have slightly leaned out the AFR, but have yet to get it completely to 11.5:1.
So for the next however many miles the 110L lasts I will continue to watch for any "Knock Sums" and I will get the AFR in the 11.3:1 range just to be a bit conservative for now. Ignition timing will be ~0deg ramped to 8deg@8K RPM. I also need to modify the MIVEC because it may be ramping out slightly too aggressively and once it does ramp out it doesn't ramp out aggressively enough so I want to see how this affects the VD results. What also makes me think it can do better is because peak torque is lagging peak boost hit by ~500rpm. I did a lot of work on the GTR understanding how cam timing affects spool and power so I think there is some gains to be found. I will post before/after once I get the MIVEC adjusted to my specific setup.
Following this I will fill up completely with Shell 93 pump and turn the boost back down to ~28psi and start to work it back up to 33psi and I hope it will take all of my current settings without having to touch the ignition timing.
BTW, if you thought the surging in 3rd was bad, 4th gear....forget about it!
33psi 4th Gear Surge
When the tank went below empty at ~480miles on the new motor I decided it was "officially" broken in and time to beat on it. I celebrated by putting almost 7 gallons of 110L in the tank dropping my ethanol gauge from E9 to E1. First order of business was to formally tune out the knock filters. Below is how I approached.
- Base tune thus far (0 - ~480 miles) was based on Aaron's knock filter values. Understand there is no way to predict what knock values are needed and each engine will be different.
- I then loaded the stock values I used on my stock motor with S2 cams now that I am running 110L. I have to assume the car would not otherwise knock with 110L based on the current ignition timing values.
- Proceeded to raise the filters based on any knock (1+ knock count), but never saw more than a 3 count
- Cannot exceed any knock filter value based on Aaron's table.
Next order of business was to raise the boost to 33psi and get the AFR and Ignition timing roughly where I want it to be. I had a very hard time with this because I slightly increased the ignition timing to ~1deg below what I was running on my 10:1 motor on 93pump, but this presented challenges due to bad surging in 3rd gear. In order to get the boost up to 33psi I had to increase the gain subsantially, which makes the boost hit harder and thus the car surges. About every 1 in 5 passes the car wouldn't surge so I could get a decent log and then import into VD.
Here is a good log and the basis for the VD results below, but this was with the ignition timing 1 deg lower (4000 - 8000rpm) vs what I am currently running. I also have slightly leaned out the AFR, but have yet to get it completely to 11.5:1.
So for the next however many miles the 110L lasts I will continue to watch for any "Knock Sums" and I will get the AFR in the 11.3:1 range just to be a bit conservative for now. Ignition timing will be ~0deg ramped to 8deg@8K RPM. I also need to modify the MIVEC because it may be ramping out slightly too aggressively and once it does ramp out it doesn't ramp out aggressively enough so I want to see how this affects the VD results. What also makes me think it can do better is because peak torque is lagging peak boost hit by ~500rpm. I did a lot of work on the GTR understanding how cam timing affects spool and power so I think there is some gains to be found. I will post before/after once I get the MIVEC adjusted to my specific setup.
Following this I will fill up completely with Shell 93 pump and turn the boost back down to ~28psi and start to work it back up to 33psi and I hope it will take all of my current settings without having to touch the ignition timing.
BTW, if you thought the surging in 3rd was bad, 4th gear....forget about it!
33psi 4th Gear Surge
Last edited by 240Z TwinTurbo; Feb 15, 2019 at 01:58 PM.
#117
EvoM Guru
iTrader: (1)
A little cold side boost control would solve that.. we had a 3582 on a snowmobile and we used cold side boost control to keep the turbo speed up and eliminate or rather keep in check surge.
you could use the stock ECU boost control to open or close a bleed line in that rpm range.
you could use the stock ECU boost control to open or close a bleed line in that rpm range.
Last edited by 211Ratsbud; Feb 15, 2019 at 02:35 PM.
#118
A little cold side boost control would solve that.. we had a 3582 on a snowmobile and we used cold side boost control to keep the turbo speed up and eliminate or rather keep in check surge.
you could use the stock ECU boost control to open or close a bleed line in that rpm range.
you could use the stock ECU boost control to open or close a bleed line in that rpm range.
#120
Ideally I would like to run the GTW3884R, but not sure it would fit because the center section is larger and pushes the compressor away from the turbine. That raises issues with physical room and I then have to likely redo the downpipe. For the GTX3584RS, the turbine is slightly larger, flows about 6% more vs normal 35r, has a max efficiency 8% higher vs a normal 35r, and I can cut my 1.06a/r T4 TS for it so a direct bolt on. Unfortunately, they don't post turbine wheel flow as a function of shaft speed so we don't get the complete picture. The larger compressor wheel should slow down the spool by ~500rpm in 3rd and jack up the topend potential.