Here is the first beta source.
If you have a socketed pcm and wants to be a guinea pig go ahead.
My only concerns are for some typo errors in the patch section. Otherwise I am sure you will like the new controller interface. It is also resizable and can hide the less used stuff.
With delta ve control you can nail the overall off condition really fast.
I justflash a stock bin with the beta and evrything looks fine. There is no brick situation and the engine starts. I didn`t have the time to test functionality.
So you can safely flash with the beta. There will be no suprises.
I also noticed how to add info buttons to explain how to use the extra controls. Next on the list.
Now with stock bin I found that the engine starts harder, at the 3rd 4th attempt, but the cold idle was incredibly smooth and tamed. I haven`t heard the engine running that healthy from a long time.
So what was the difference. 4 less degrees advance, stock underspeed and maf tables and much more fuel being dumped. I guess I need to be back on the tunnig board.
Some log for comparison.
I'll probably pass on testing until I'm done getting my tune at a respectable baseline - it's coming along really well so I'd rather not get distracted with other stuff right at the moment. I've been switching between your fork and an old version of the original that I've customized.
Are you stock cubes but bigger injectors?Originally Posted by kur4o
I've been messing with that 12680 multiplier with the stock base pulsewidth table. For the sake of discussion let's call it a two byte table where both bytes must match (similiar to the target AFR tables). I followed your advice at first and used 0x44 (in both bytes). It seemed to be a little lean - hot restarts popped right off if I bumped the key and then hit it again after 5 seconds. So I went to 0x55. It hits every time but seems to be pretty rich there. I tried 0x4F and it still seems a bit rich so I'm working my way down from there. Whatever the case it's hitting on the third or fourth low res pulse every time now regardless of ECT.
You can share the new goodies and I can try to include them in the new release. The key shortcuts are very usefull, I still haven`t had the chance to expand them to more controls.an old version of the original that I've customized
The 2bytes table whenever is changed, the prime pulse tables must be adjusted to compensate for the new offset factor.
With stock settings at $6666 the stock table will look like this
When the 2 byte table is changed to $4444215.63 215.63
209.38 209.38
120.31 120.31
59.38 59.38
39.06 39.06
17.19 7.81
12.50 6.25
7.81 4.69
7.81 4.69
7.81 3.13
6.25 3.13
6.25 4.69
6.25 4.69
6.25 6.25
6.25 6.25
The table will look like this with new conversion factor (1.007028* X) + 0.000000
So whenever the 2 byte table is changed you need to readjust the prime pulse tables.138.97 138.97
134.94 134.94
77.54 77.54
38.27 38.27
25.18 25.18
11.08 5.04
8.06 4.03
5.04 3.02
5.04 3.02
5.04 2.01
4.03 2.01
4.03 3.02
4.03 3.02
4.03 4.03
4.03 4.03
With $4444 scalar the resolution goes to 1ms intervals vs 1.5 ms at stock $6666 settings.
I have done this on a cold start with the same result. Adding fuel didn`t help at all, so my assumptions it is actually rich.hot restarts popped right off if I bumped the key and then hit it again after 5 seconds
Hot restarts are tricky to tune and test due to some variables that can change on the fly. The main issue will be how much fuel left on the walls unburned with the engine switching off.
The prefect start will be at sub 0.3seconds only on the prime pulse fueling. Than the crank ve tables hit with 2 different afr tables based on low res pulses.
I found some 96 lt4 bin and explore the differences. The idle advance is set at 19* at cold coolant and gradually decrease to 14* while warming. On the lt1 f-body is reversed. Really interesting.
I have changed the injectors onlyThe ls` ones flows 24lb at 3bar, so no change with fuel flow. With the old some crap accell 24lb/hr I didn`t have any hard start issues even with completely stock bins.
So it could be the injectors needs some tweeking or the ls1 injectors patch I run needs some attention at starting routine.
I also tried the vette CL settings and the logs I made are on par with yours. Slight blm split and less crosscounts. The idle runs perfect in closed loop.
I have some theory how to tune the CL settings for more crosscounts, but need to test some of them to confirm how they work.
The off idle trims made huge difference at cruising around 1500rpm. I wonder how far we can go using them safely before the air stream equalize in the cylinders.
Stock settings are at 0-14% tps, <3000 rpm and no catprotect and PE.
On the lt4 bin the tps is at 24%.
I want to increase the tps as much as possible and add new conditions such as airflow or map thresholds.
There really isn't anything new or terribly worthy of sharing. I did recently fix a problem where the IAC override would set it to 4 (steps) and not change if switched to step mode. But that might have been a bug I created. Aside from the key accelerator stuff it was mainly making fields like knock count and stuff big enough this old blind guy could see them on the 11" screen.
I think I'm pretty clear on that. I suppose what I'm confused on is if I were to set 12680 to 0x4c4c why wouldn't the prime pulse table look like this with 1.16ms intervals:
???Code:160.034 160.034 155.396 155.396 89.294 89.294 44.067 44.067 35.950 35.950 17.395 17.395 12.756 12.756 5.798 5.798 5.798 3.479 5.798 2.319 4.639 2.319 4.639 3.479 4.639 3.479 4.639 4.639 4.639 4.639
You won't hurt my feelings to tell me why I'm being a bonehead as long as I can learn something from it.
You'll get no argument from me that restarts in general are tricky with different injectors, displacement, etc. I tested the above multiplier tonight (0x4c4c) and had a very obvious flood at 73c restart (fired then died, required pedal to clear on 2nd attempt). This is why I was curious if we have any idea if injector offsets are added to the prime pulses. Then we get to the subject of having good injector data. My ecm is seeing 10v average during cranking. that's about 1ms of offset for the stock injectors. If my offsets are wrong, how far off are they?
You should put my experimental ignition system on it so you have another unknown! (joking)
Mind sharing a screenshot of the BLM analysis?
I'm pretty happy with my VE table after tonight's log so I think I'm going to turn the MAF back on and fine tune a bit. Probably mostly on the damned prime pulse / restart issue. Beyond that I'd be happy to test any closed loop tweaks you would care to share.
If you can do that I can surely conquer startup fueling.
Last edited by spfautsch; 10-05-2019 at 05:09 AM.
It is exctly like that. I have nothing to add. What I mean is that you can fine tune the tables from there.why wouldn't the prime pulse table look like this with 1.16ms intervals:
So for now I want to end the disscusion on the pw scalar since it is all clear.
The 2 bytes table should be called. Multiplier to startup prime PW table.
The formula for the multiplier will be d={x*0.0152587891}/256
The multiplier decimal result must be entered in the crank prime tables conversion field with the following formula {x*d}.
Now you can start fine tunning with the new multiplier.
The InJ vol offset are for sure used during cranking. I improved that part of my patch with major improvement on the first start attempt.
I am starting to suspect that my ls1 inj flows more than specified at 100kp range. That explains the hard startup[rich condition] with stock settings.
I am getting rich right side at cell 6 only, 3-4 points split. I check logs with older settings and the split is also there at #6 cell. At other cells the split is negligible.
I also changed the off idle trims but it seems they don`t affect the split.
I experience zero split at idle so far. I tested it for a short period so it could be not very reliable.
I thought for a way how to find the break point for off idle trims condition.
Setting the trims on one of the side richer and look for the conditions where the blm splits equalize.
Than set the tps map thresholds.
I already made the map threshold patch but it interferes and interacts with the v5 patch so it is a little tricky to install. If you want to give it a try I will set you a bin. If it turns out bad I can change it to an air flow condition.
I suspect the $26fb table to be responsible for the crosscounts. It is also the cure for idle CL along with the swing voltage.
AT $26fb slightly increasing the values at cells 2-5[don`t touch the first cell- it is for idle]
As you can see I thought more is better but not for the first cell. For the first cell less is better, almost twice less at stock y-body vs f-body.y-stock ;f-stock ; my old settings
0.40 0.80 1.20
0.18 0.18 0.35
0.13 0.13 0.25
0.11 0.11 0.23
0.10 0.10 0.20
At cell 2-5 increasing the value can lead to higher crosscounts. Field discovered by mistake. Not sure how good is this for engine running healthy but you can give it a try.
$270c is also different but not in the idle region. It also can attribute to the crosscounts but needs testing.
The main difference is at cells 2-5
vette; f-body
0.3 0.5
and the swing voltage
380 for vette
vs
460 for f-body
Your problems with hot restart might be due to crank spark advance. I ripped the settings from different calibrations.
Interesting I am running 2 degrees at 80*C and 7* at 20*C stock. The closest to cammed car is lt4 setting at 5* and 11*, all other calibrations have 8-10 * and 13*.
I suspect this might be culprit with my issues of first cold start.
So it could be hot restarts needs less advance and first cold start needs more advance.
From post #108
I finally took the time to download and build this. Nice work on the controller window! I was able to utilize it thoroughly today. I did make some modifications to get the overall build to my liking - if you're interested in incorporating them I can send you the sources I changed. First off I made the fields and fonts for the dashboard data I like to be able to see bigger. Next I added the log timestamp to the knock events. This may or may not be useful if you have a car with knock - it doesn't tell you which log the knock happened in so it can be a bit of an adventure if you're analyzing multiple logs. Lastly I added a setFocus() call to each of the individual trim button's toggled event i.e.
This lets you jump directly to using the keyboard to adjust the trim slider with the left / right arrow keys after clicking one of the cylinder trim buttons. Keyboard = efficiency of motion!Code:void controller::on_cylcutcorr_1_toggled(bool checked) { if (checked == false) { parameter_indicator_display(ui->lcd_cyl1,false); return; } else { ui->slider_cylcorr->setFocus(); // <<< ADDED drop_cyl_corr(1); // int lcd_cyl1 = ui->lcd_cyl1(); int trim1_int = ui->lcd_cyl1->value() ; ui->slider_cylcorr->setValue(trim1_int); control->m4_cyl_corr (trim1_int); update_m4_raw_display(); parameter_indicator_display(ui->lcd_cyl1,true); } }
The only thing that would be better is the ability to set the prime pulsewidth multiplier in a similar way. It's the only thing I feel I'm going to spend time fine tuning on since every time I touch individual cylinder trims startup goes to crap.
Spent about an hour today working on cylinder trims. Idle BLMs are still split by 4% but idle feels and sounds good open or closed loop. I was having a bit of part-throttle surging that improved immensely with a couple trial and error sessions with the new off-idle trim controls. I'm holding off on calling it a victory though because a weather system has moved in dropping air temps and bringing a bunch of moist air with it so BLMs are taking fuel out almost everywhere but idle. I hate winter! :-\
I'm thinking the off-idle trims may take even more time & effort than closed tps trims because even though my surge seems much improved, I'm feeling a good bit of NVH when opening the throttle slightly without load.
Whatever the outcome I feel like what's left are minor tweaks and fine tuning. And I'm pretty sure there are no other bolt-ons I can try that I don't already own except forced induction. Hmm... :-)
Anyway, thanks to everyone who chimed in to help. kur4o in particular - I still have a diy-ltcc board with your name on it if you'd like. I'm sure it will drive the capacitive discharge coils I have and by my butt-dyno estimation they seem to be just as beefy as the truck coils everybody wants (and they're cheaper).
Last edited by spfautsch; 10-25-2019 at 04:51 AM.
I hate to keep this thread alive, but I made another discovery today that is pertinent to the idle BLM splits that were chronically showing the right bank lean and the left bank rich.
So some background - it's been rainy and cold all week and all I've been able to do is work with startup fueling and idle trims. I've been noticing the wideband maxing out lean periodically when both left and right integrators synched up for a lean swing. I've also been noticing an intermittent miss at part throttle that also seemed to briefly peg the wideband lean (presumably from raw fuel getting in the exhaust). I began to wonder if my individual trims are too extreme and are causing these problems. Today I had a bit of time to experiment with trims in open loop while trying to keep my wideband matching commanded AFR. Oddly, the individual trims seemed to make very little difference when I added or removed from one bank. But while I was doing that I noticed that when I forced AFR above 14.5:1, after a few minutes the right side O2 would start trending down around 150mv and then eventually flatline at 40-50 (this is with BLMs reset to 128).
Then, as I was looking over kur4o's nice new controller window it occurred to me I've never tried moving EOIT. Sure enough, when I got to 129* things smoothed out somewhat. I then switched it to closed loop and to my amazement there was only a 2 point split. Toggling the EOIT override while in closed loop I witnessed the BLM splits come back and then go in unison with the override. So I suppose my lean condition on the right wasn't an exhaust leak or so much an actual lean condition as it is raw fuel getting into the exhaust during overlap. So I guess I'm going to try using 129* EOIT and start all over with individual trims and see what happens. I expect my off-idle fueling will get way out of whack due to loss of wall film.
kur4o: after a bit of experimentation, I think your calcs are inverted on the display of the EOIT slider value. I think it should be:
Edit: disregard that - it's pretty superficial. I'd overlooked the ATDC label and I hadn't thought about EOIT terminology in ages.Code:void controller::on_slider_endofinj_valueChanged(int value) { //new end of inj // upd. text display always int degrees = (double)value * 5.625; ui->display_endofinj->setPlainText(QString::number(degrees,'d',0)+"*"); // upd. mode4 if(ui->check_endofinj->isChecked() == true) { control->m4_comm_endofinj(true,ui->slider_endofinj->value()); update_m4_raw_display(); } }
So what I was assuming to be 129* BTDC was actually 596* BTDC. So my testing wasn't reducing EOIT but in fact increasing it by 56 degrees.
Whatever the case, this is promising news and falls in line with what the LS guys were saying about increasing EOIT from "60 to 61-64".
This has absolutely no relevance to intake valve closing on the cylinder with the active injector. I'll have to do some degree wheel work, but I suspect it's more closely related with having an adjacent (fuel scavenging) cylinder on it's intake stroke while fuel is being sprayed on the "current" cylinder.
Whatever the case, increasing my 0x12df2 table values in the 44c and greater cells from 0x60 (540* btdc) to 0x6a (596* btdc) has reduced my idle BLM splits from 10-16 to 1-4.
I'll post more analysis when I figure out the whys and hows.
Edit: After quite a bit of hypothesizing I'm still not sure why this works. The only thing I can come up with is that perhaps it moves injection to a point where fewer left bank cylinders are on their intake stroke while even bank injectors are firing. Fuel being robbed by another on the same bank won't cause a BLM split.
Last edited by spfautsch; 10-31-2019 at 07:19 PM.
You still amaze me at discovering more stuff through experiments. I never thought about possible blm split by EOIT target. What you have done is move the target earlier to the actual intake valve opening. So there is more time for the fuel to turn into vapor. Stock settings of EOIT takes into account normal operating engine temp at 100-110C*, now when you run it at 80C* it takes considerably longer for the fuel to evaporate.
One reason this can happen is one of the head runs at substantial lower temp while the other is hotter. Not even temp distribution accross engine. The colder side can`t evaporate all the fuel and there is some left unused by the cylinder, during combustion the left over fuel evaporates and is robbed by adjacent across the board cylinders leading to one side running leaner and the other running richer.
Of course this is only speculation.
For the sake of confirmation you can heat the engine upto 100C* and confirm or deny the theory.
I also been considering improving the code of EOIT target by adding more tables, like MIN EOIT vs rpm or closed throttle EOIT table.
Think about it and let me know if there will be any benefit of doing it.
The setting in eehack is *ATDC, which is correct. If you need it at *BTDC just substract the result with 720*.
The shorcuts can get even better if you have some time to do it. Adding some more shortcuts like D+1...8 to set drive cyl and I+1...8 to set at idle cyl, D+0 and I+0 unset, and some more for enable and disable, maybe ctrl+I and CTRL+d might do it, I am sure you will find even some better combination.
Last time I played with the cyl trims controls it sucked really with that mouse and slider I even considered changing the design of it. But with some keyboard shortcuts it will play really nice.
I had a ps pump failure and abandon tuning so far, The last time I managed to run it I discovered that accidentally I have disabled the trims and they were set at 1 all the time, invalidating my last attempts. With 1 for trim the starting was a little weird sounding but still started almost good with no hesitation. When I enabled the trims it started again with healthy sound. Maybe your no start condition with off the chart idle trims can be the result of crank VE table not set right.
I also discovered that the injectors definitely flow more that my initial settings. I managed to increase the fuel flow scalar with much positive results. Also played with the warmup spark scalar table. At 20C* it idles best at 14 degrees and slowly recovering to 23-24 degrees at 50-60C*. Even 1-2 degree change at 30C* coolant change the sound alot. Still no clue how to tune it the proper way other than feel and smell.
Last edited by kur4o; 10-31-2019 at 10:44 PM.
Another thing that may be a factor is that the exhaust system is asymmetric between the header collectors and the x-pipe. One side might be generating more exhaust scavenging than the other at idle speeds. In working out the different valve events looking for ideas I found 596 BTDC moves the injection event (at 800 rpm it's only 3 degrees for ~2ms) so it's almost out of all overlap scenarios. But this may be completely irrelevant. I'll definitely try getting it hotter to see if your idea of uneven cooling could be a factor.
The only reason I can think of for having the need for more EOIT controls is if this change causes a BLM split at part throttle, and it will be a few days before I'll have a chance to drive it. I'm hoping to rework the closed individual trims tonight, but I'll probably get roped into handing out candy.
I'll look at the possibility of adding those key accelerators - I haven't tried modifiers like ctrl, alt or shift so no idea how involved that, or using a normal key like 'i' and 'd' as a modifier key will be.
Similar results to me then, extending the time the fuel sits in the port or forcing it further down the port so it has less chance of being sucked back out?
Have you tried a sweep test.
This was mine last year.
I think the LH inj pw staying the same is an issue I've had with flashing & logging with different versions.Never quite pinned it down.
EOIT can make quite a difference to the splits. I'm currently running at #63.
2018-12-13
Did another EOIT spread degree test with odd figures .........
Test B at 86+øC
Blm's running on pcm setting of &66 120 122
log went lean/weird around 74øC to 75.5øC 68C eoit = &50 80C=66
Left bank richer - rhs longer inj time ?????????????????????????????????
start 1200 Idle Fast
Time log ø table blm's blm's Inj PW SPLIT SPLIT
685 7170 146 66 120 122 1.79 1.79 pcm setting &66 + 2
832 8764 186 5F 115 125 1.79 1.79 +10
1110 11681 208 5B 113 130 1.77 1.94 +17
1225 12875 231 57 106 130 1.77 1.95 +24
1400 14711 253 53 104 130 1.77.1.92 +26
1540 16180 276 4F 99 134 103 139 1.75 1.95 +35 +37
1715 18010 298 4B 103 137 99 145 1.77 1.97 +34 +46
1890 19843 321 47 98 137 99 145 1.77 1.95 +39 +46
2030 21313 343 43 98 137 99 145 1.77 1.97 +39 +46
2135 22413 366 3F 97 137 99 145 1.77 1.97 +40 +46
2275 23881 388 3B 99 143 100 150 1.77 1.94 +44 +50
2590 27180 411 37 98 134 100 144 1.77 1.92 +36 +44
2730 28647 433 33 96 134 100 144 1.77 1.91 +38 +44
2870 30109 456 2F 96 136 100 144 1.77 1.92 +40 +44
3010 31574 478 2B 104 136 97 144 1.77 1.92 +32 +48
3150 33038 501 27 101 136 101 138 1.77 1.88 +35 +37
3255 34143 523 23 104 135 106 132 1.77 1.83 +31 +26
3465 36343 557 1D 104 132 104 132 1.77 1.82 +28 +28
3605 37807 579 19 104 132 110 132 1.77 1.85 +28 +22
3710 38908 602 15 99 132 99 132 1.77 1.86 +33 +33
3850 40373 624 11 96 139 104 132 1.77 1.83 +43 +28
4025 42209 647 0D 99 136 102 132 1.77 1.89 +37 +30
4200 44042 669 09 103 131 107 132 1.77 1.85 +28 +25
4340 45510 692 05 103 131 104 132 1.77 1.88 +28 +28
4480 46979 714 01 103 131 104 132 1.77 1.82 +28 +28
Restart at end of table
4585 48077 96 6F 107 118 111 120 1.75 1.66 *INJ PW swaps* +11 + 9
4725 49541 118 6B 116 121 116 120 1.83 1.56 + 5 + 4
4865 51005 141 67 118 121 119 120 1.77 1.60 + 3 + 1
4970 52105 163 63 118 117 119 120 1.77 1.63 - 1 + 1 <-<-
5145 53935 186 5F 117 117 119 120 1.77 1.66 0 + 1
197 5D 108 125 111 122 +17 +11
5285 55938 208 5B 107 130 102 139 1.77 1.82 +23 +37
5565 58321 231 57 107 129 105 139 1.77 1.79 +22 +34
5635 59054 253 53 104 129 101 139 1.77 1.85 +25 +38
Mitch
'95 Z28 M6 -Just the odd mod.
'80 350 A3 C3 Corvette - recent addition.
I haven't tried a complete sweep test yet, finding that my splits changed immensely was mostly by accident. It's been ridiculously cold this week and I've had limited time to work. When I have it's in the garage with the door open so the gas fumes are tolerable, so I haven't been able to be as methodical as I'd like.
Last night I worked on closed throttle individual trims. One of my classmates showed up to trick or beer me as I was working so I had to stop, but I was basically back at stock trims with just a few minor changes and BLMs were 127 left and 129 right. I'd really like to get this nailed down so I can also nail down startup fueling, but as it works everything seems to be inter-related. Out of curiosity, I did briefly try moving EOIT to well after my exhaust valve closing @ 323 BTDC but the BLMs went crazy. Once I'm more satisfied with individual trims I'll give a full sweep test a shot.
My theory on why this is working is based on the idea that fuel scavenging inside the intake only edit: (or, primarily) happens while an injector is spraying, and once the fuel is down on the back of the intake valve and port bowl the scavenging effect is dramatically reduced because the air from the idle feeds can't reach it there. I don't think it's related to having more time for the fuel to boil into vapor, but I do intend to test because my theory could be complete nonsense. I attached the worksheet I built to help visualize the relationship of all the different events.
It's extremely puzzling that your LBPW never drops below 1.77ms. I'm curious - what values do you have in 0x126d5 and 0x126d7 (min injector pulsewidth)?
Last edited by spfautsch; 11-01-2019 at 09:24 PM.
Hey Mitch would you mind recapping your methodology when you did the sweep tests? Did you reset BLMs when you moved EOIT? How long did you let it idle before recording the BLM split, or did you average?
I did a sweep from 540 to 624 BTDC this morning and in the process noticed my BLMs seemed to inversely follow coolant temp. I have a 71c thermostat and fan on setting was at 82c so coolant temp was varying by about 4c. At 83c trims were about 2-3 and at 79c they were about 4-7.
So I decided to try bringing temps up by digging out my 82c thermostat and dropping it back in. I had to bump fan on temps up some so they weren't running on high all the time. After a short drive the coolant temp at idle was steady at 91.3 with fans contant on high. I did another brief (i.e. not well documented) sweep test and found nothing compelling. Splits averaged out to about 3 but maxed at 7. I'm planning on doing another thorough sweep tomorrow.
This is all with almost stock individual cylinder trims. Overall it's running much smoother than when I was running radical trims trying to compensate for a ~12 point BLM split.
In the midst of this I'm pretty sure my wideband sensor is toast. In closed loop it's reading pegged at 16.39:1 (5v) quite a bit. Last week when air temps dropped, on the last leg of the commute the wideband was flatlined at 0v (11.39:1) the entire 62 mile drive. I thought the fuse powering the controller might have blown, but it wasn't. After installing the 82c thermostat I took about a 10 mile drive and had a chance to open it up for a few seconds where commanded AFR was 12.5:1, and the wideband was flatlined through the whole PE event, then went back to reading really lean. Curious if anyone else has noticed this type of behavior in cooler air temps, or if I should start looking for a new Bosch sensor.
Last edited by spfautsch; 11-03-2019 at 04:52 AM.
Posting Permissions
Reply With Quote
Bookmarks