scheides

iTrader: (10)

^Yup I believe you! I have not touched the maf smoothing table but the numbers present seem likely to affect fueling directly and not fractionally at all. Kinda crazy that this table wasn't used to scale the maf all along while the maf scaling table was.....pure coincidence I guess.

Ceddy

I've heard the MAF Compensation Table, has values directly from the MAF's Data Sheet.

The MAF Smoothing Table, is there to correct values that don't match the Data Sheet.



Little Off Topic,
I got SD working on the H8 DSMs, thanks very much for the help, MrFred.
http://www.youtube.com/watch?v=JB1CeaAV5gE

mrfred

iTrader: (50)

Yep, that matches with what I wrote. Glad you got it working with the H8. BTW, I read your email from several days ago, but have been too busy with work to think about anything in any depth except work. :-)

03whitegsr

iTrader: (8)

I tried out a modified method for tuning SD based on the idea of linearizing the IPW to MAFSOURCEMAIN

With the factory MAF tables, you get very non-linear correction based on MAF Frequency, which happens to be a calculation of the SD tables now. In a typical speed density system, the IPW is directly related in a linear fashion to the VE tables. In the standard method used on the EVO, this means a non-linear response below 500 Hz.

II tossed the MAF tables in the trash. Set the MAF Characteristic Scaling table to 211 across the board and the MAF Compensation scaling to 128 across the board. Now, MAP * RPM * MAP VE * RPM VE*MAT*(MAF Table [now constant]+a bunch of other corrections that drop to zero in steady state conditions) = IPW Just like a normal speed density system. Seems to be working well.

RPM VE starts at 87% and climbs to 105% then drops back down to about 95% by 7500. MAP VE is 1:1 above 90kPa and only marginally richer then that down low to help smooth out part throttle conditions.

scheides

iTrader: (10)

Ya I'm wondering if this cruising jitter thing can be ironed out with the MAF Scaling/smoothing table, let us know how it goes for ya

Got alky dialed in today, car rips again! virtual dyno room puts me right back where I was, around 420whp on evoM comparator mode. I had to mess with the load scaling to get timing where it was, but seems pretty good now!

So, I have 2 pending issues:
-knock on spool-up, or tip-in. I'll try messing with the tip-in table, but I haven't logged a lean condition yet and timing seems ok, so not sure what the deal is.
-2000-2800rpm jitter. Still searching, hoping to log it and find a work-around.

Just wanna say thanks to everyone that helped me get this far, I am SO happy with the new speed density setup, it is absolutely amazing.

l2r99gst

iTrader: (2)

I tried this months ago. It didn't work for the jumpy IPW. But, at least we are thinking along the same lines...makes me feel better about my ideas way back when.

Seriously though, go through my '96530706+DMA+SD+Livemap working' thread. I think page 4 starts where I noticed the issue. I posted several logs and pinpointed the area where it occurred. I also listed everything I had tried in the subsequent pages. It may save you guys some time.

I honestly don't think it's something that we can fix. It has to be something either with this ROM or the way the SD has been implemented. John goes into great detail about how he implemented it. He mentions that he replaced the MAF variable after a smoothing routine, so no smoothing is applied to the calculated SD variable. Perhaps that is part of it, perhaps not. I think mrfred was looking into it a bit, too.

But, whatever anyone does find, please post up. I think I was a little ahead of the curve and when I was asking these questions, not many people were running SD, so I couldn't get much feedback. Maybe I will switch back to SD again and start some more testing. I wanted to ever since mrfred found all of the IPW variables to log. I want to pinpoint the exact variable(s) that is causing the jitter.

03whitegsr

iTrader: (8)

I haven't noticed what you picked up on, but I haven't been specifically looking for it. I would say that if it's very apparent just driving it, then my car probably doesn't do it, as I've tried to hit every load and RPM combination you could dream up. Lugging the car in 5th at 1500RPM to 20kPa free coasting down a hill with just enough throttle to keep it from fuel cutting. All in open loop, of course.

I don't know if it could be related, but I did notice something that kind of surprised me today. I have been logging the accel and decel MUT channels, I believe 54 and 55. Doing an RPM sweep where you basically hold the MAP at a constant level (I did 80kPa), but not in boost and allow the car to rev out. The two channels are quiet until 2500-3000 RPM, then nothing, then around 4500-5500 again it picks up and then 0 all the way out from there. No noticeable change in TPS or MAP. IPW jumps around a small amount with the accel and decel. It's not as dramatic as what you saw, but it would go from about 4.2 msec to 3.8 msec.

My last issue to get ironed out related to the SD patch is a jitter issue though. In open loop, the car drives so damn smooth and the AFRs are within 3% of the fuel map pretty much under every condition on the map. Put it in closed loop though and it seems like the AFR swings due to O2 feedback induce a lot of choppiness. I dropped the STFT limits down to +/- 12% and it actually made a pretty big difference. Anybody know if there is a table to control how quickly it cycles? Right now it's like 2 cycles/second. If I could slow it down to 1 cylce every two seconds, I think it would solve the rest of my issues.

Without a doubt though, driveability in general is better then it's ever been. Absolutely no lift off jerk. Climbing hills use to be hell. Part throttle would cause the BOV to flutter like crazy. I thought it was compressure surge and the BOV combining to cause problems. Drives perfectly smooth now in those same conditions. It use to creep down on AFR too (even while in closed loop), into the 10:1 range at low loads and RPM while climbing steep hills. 14.7:1 right where I tuned it for. If it wasn't for the fact that I tin canned the car and had this ceramic twin disc, the car would be driving like a Cadillac. So smooth.

The only performance related complaint is that the "crispness" isn't there quite yet.

SeanV

I want to start using Speed Density on my ROM and will be making use of the of the latest 96530706 V7 Rom for Speed Density as posted by Phenem.

"Option 5 96531706 (v7 + Mods + Speed Density)"

I recall seeing a "Quick Start"/"Dummy Guide" thread for setting up Speed Denisty, with regards to the steps involved, but for the live of me I can't find that thread any more.

Searching isn't helping at all.... it is as if that thread doesn't exist any longer.
Does anyone have a link to that thread or a link to another one that will help me in getting started.

Thx

l2r99gst

iTrader: (2)

I think my logs show it the best because (1) I was using DMA logging to capture the data at very fast rates and (2) I was still on stock injectors at the time.

For example, with injectors twice the size of stock, you would expect to see half of the jitter range that I saw. The 'feel' of the jerkiness/hesitation may still be there, just less evident in the logs.

Now that I have 1250s, running E85, etc, etc, I was planning on testing everything out again to confirm. Especially with all of the new IPW variables and more people running SD to compare with. I think some of the tables are better defined now as well...but I can't remember. I was testing this about 5 months back.

03whitegsr

iTrader: (8)

I wonder if it has anything to do with the MAF Reset call. The attachment is a page out of the factory service manual that talks about the reset function.

• When the throttle position sensor output voltage
is low, the ECM causes the power transistor to be
"ON" to send an airflow sensor reset signal to the
airflow sensor. In response to the reset signal, the
airflow sensor resets the filter circuit and
improves the ability of the airflow sensor to mea-
sure the amount of air in a small air intake region.

I would also assume this has to do with the two MAF Filter settings in ECUFlash. Might be worth a shot looking into them anyway. I wonder if you can log the MAF Reset condition and compare it to your jitter.

My thoughts are that maybe the conditions you see the jitter at is the transition point between it using the reset signal or not. The ECU is maybe turning the reset off and on and expects the MAFSOURCEMAIN signal to have a similar change. The SD patch doesn't have the change though, so it misinterprets the data and and chops it up to some extent thinking that it is changing.

l2r99gst

iTrader: (2)

Yeah, that does sound like a good candidate for what is happening. In talking with mrfred over some PMs in the past, he said that he went over the code and there seems to be something going on for low airflow conditions. Perhaps this is it.

I'm sure we can eventually figure it out, but I sort of lost interest a while back without too many people involved. Also, not being able to read the ROM code myself, I was sort of stuck.

It seems like more and more is being discovered every day, even things like you just mentioned, so maybe it's time for me to test again. Hey, I'm off today ... maybe a good time to switch to SD again.

Happy Thanksgiving, btw!

scheides

iTrader: (10)

Well it seems we have a growing number of people using this patch now and ready to figure out the details for the little nuances it has.

Would it be possible to simply disable that low load maf reset signal? Otherwise I guess build something into the speed density patch to acknowledge it?

mrfred

iTrader: (50)

The accel/decel values you mentioned represent positive and negative changes in master load. Very small changes can be detected, so I wouldn't be surprised if small values showed up once in a while.

The cyclic response of the STFT depends at least in part on how fast the O2 sensor responds and the distance of the sensor from the exhaust valve. I suppose its possible that there some type of PI(D) feedback controller with an integration time that could be adjusted improve the response time of the STFT.

03whitegsr

iTrader: (8)

One thing about that though is that it's not just a small amount. My scaling is probably way off, but I only see a max value of about 4.5 from either one of those channels, and it's hitting 4.5 when there is no noticeable change in load or TPS. They do work as I would expect other wise though. Step on the gas, the accel channels jumps up then bleeds off very quickly. Let off the gas and the decel channels gets active.

I actually want to slow the feedback down. Instead of bouncing wildly back and forth between the max and min allowable values, I'd like it to just hover around within a few percent of the target AFR. It may have just been that the car was running leaner every where, but on my old GS-t, I'd switch to open loop and I'd pick up almost 5mpg. Just wondering if keeping the swings less aggressive would actually improve mpg while smoothing out cruise conditions.

knochgoon24

iTrader: (2)

So all you did was change the STFT limits, and that seemed to fix the problems with part throttle and the BOV randomly fluttering open?

I'm not running SD (yet) but I am using the stock 2g DSM MAF sensor in my car with the Evo 8 Ecu. It's not SD, but it is a different air flow metering device than what the ecu was programmed for. I'm having symptoms similar to what you are describing. My car is fine when running open loop, but has some problems in closed loop.

Can you please describe exactly what you changed? For a while, I've been thinking my BOV was defective. It would be nice if this fixed it.