PDA

View Full Version : Filtering Data



bink
July 24th, 2005, 01:33 PM
Let's say I have a Log file X and I'm using the "BEN factor-FJO/Autotune" map with my FJO Wideband. I have the "Empty Cell" = 50.Open Loop SD tune.
Now I want to filter :
1.)All noncell 22 data -i.e. All FTC Cell data less than 22 (Open Loop all decel is cell 21 and idle is cell 20).
2.) All TPS transients greater than 5% per 100 msec.

Both filters are setup. Can I do 2 passes of the logged data in sequence - one of each filter ?? Will this reliably filter out all decel, idle and transient data?
Cheers,
joel

TAQuickness
July 24th, 2005, 01:36 PM
If I understood your question correctly, no. You will need to set up one filter that will filter All noncell 22 data -i.e. All FTC Cell data less than 22 OR All TPS transients greater than 5% per 100 msec

It's similar in format to the filter screen shot in the autotune doc

bink
July 24th, 2005, 01:59 PM
Thanks.
I set up the filters. My question is can I run the logged data , for a given map (BEN map), sequentially through both filters. Basically two passes of the same logged data filtering out the FTC cells and the Throttle transients.

Cheers,
joel

Black02SS
July 24th, 2005, 02:50 PM
Not that I am aware of. If you want to filter out the data you can either do both, or one or the other. From my understanding and use, you can't apply one filter over the other. Only way is to combine both filters into one. What are you trying to do joel?

Tordne
July 24th, 2005, 03:57 PM
I have just read the original post in this tread, and it may be the most interesting post I have read since starting the Auto-Tune process.

I just want to clarify something though...

I just logged the FT Cell# on my drive home. I had never noticed the behaviours of these cell #'s (20, 21 & 22).

The cells to me see to represent the following (although this is observed from a single log session):
Cell 20 - Idle, but also when still decelerating
Cell 21 - Deceleration, but also when coasting at steady or non aggressive throttle position
Cell 22 - When the throttle position is increased, accelerating, or perhaps under increased load

Firstly, does that sound correct?

Secondly, why would you want to exclude cell 22 in this case? I was thinking I'd want to drop cell 20?

I'd appreciate your comments...

Cheers,

Black02SS
July 24th, 2005, 04:04 PM
You wouldn't want to Andrew, it says NONCELL 21. ;) He is wanting to remove all cells below 22, not including 22.

bink
July 24th, 2005, 08:52 PM
black02ss, Tordne and TAquickness - FTCs 20 and 21 are "nonload" cells and are used during idle and deceleration - when I'm in open loop.By "nonload" I mean when the engine is not under a load. We want to exclude "nonload" data, from the BEN/VE process, for accuracy.

We want to exclude throttle transients (Non-Steady State) for accuracy also. The AFR changes substantially during transients as the PCM adjusts for changing conditions.

I just want to filter out the garbage :D

Cheers,
joel

Black02SS
July 24th, 2005, 09:23 PM
I understand what you are wanting to do bink, I might have worded my response to Tordne wrong. I have been thinking about this as I had forgot about FTC's when I got rid of Edit. How much of a swing are you seeing on decel? If I don't have decel in mine, then there are some cells at low kPa that I can't hit. I am going to add FTC's to my log on the way home to see how much of an impact it has. I'll post my results shortly.

Blacky
July 24th, 2005, 10:34 PM
FWIW you can apply two filters sequentially.

Create a filter with the first criteria only, apply it and then select File->Save as and save the log file with a new name. That will save *only* the visible frames.
Then create a second filter and apply it to the saved data which contains only the visible frames from the first filter.

Regards
Paul

Black02SS
July 24th, 2005, 11:28 PM
I have posted 3 images. Each image shows the difference in one log file from the effects of decel and idle. From seeing this, I personally would use all three due to the fact I think idle(FTC 20) is needed for a good BEN at that MAP, decel(FTC 21) only appears to effect my car from MAP 30 down. If I didn't have that, I don't see how I could tune "some" of those cells. What do you guys think?

Dirk Diggler
July 25th, 2005, 12:33 AM
bink arent you glad you dont have to fool around with excel and access anymore ;)

GMPX
July 25th, 2005, 12:47 AM
black02ss, Tordne and TAquickness - FTCs 20 and 21 are "nonload" cells and are used during idle and deceleration - when I'm in open loop.By "nonload" I mean when the engine is not under a load. We want to exclude "nonload" data, from the BEN/VE process, for accuracy.

We want to exclude throttle transients (Non-Steady State) for accuracy also. The AFR changes substantially during transients as the PCM adjusts for changing conditions.

I just want to filter out the garbage :D

Cheers,
joel

Joel, why don't you just disable the DFCO when tuning?.
Another thought, look at what your DFCO settings are and set a filter to ignore any MAP values below 'x', where 'x' is the MAP DFCO disable value.

I have been using the filters for a few weeks now and I am so impressed, it's nice not having to alter or monitor your driving habits when logging.
In fact, I think to slightly alter a great line from the Blues Brothers movie "we have filters powerful enough to turn goats piss into gasoline", hey Paul, lets put that on the features list :lol:

Cheers,
Ross

GMPX
July 25th, 2005, 12:57 AM
bink arent you glad you dont have to fool around with excel and access anymore ;)

Excel is for accountants and book keepers, it has no purpose in the tuning world, but shhhh, don't tell Mr Gates.......soon on a shelf near you.
MS LS1Tune2005, can you imagine it!!

Cheers,
Ross

bink
July 25th, 2005, 01:05 AM
Ross- Great line...only you would have remembered it!!

I need to check my DFCO as I'm getting some random funky BEN values (0.19, 0.34 etc.).

Dirk - Yes and I will never forget how much time I spent trying to figure things out. You, of course, were a BIG help.

Blacky - Thanks. I tried it earlier. Works great.

Black02ss - I'm not convinced yet. I hit most cells witha bit more work back when I did the NoGo tune.

Cheers,
joel

Dirk Diggler
July 25th, 2005, 01:18 AM
Ross- Great line...only you would have remembered it!!

I need to check my DFCO as I'm getting some random funky BEN values (0.19, 0.34 etc.).




Are you telling the scanner to hide empty cells with a count of like 10 or something like that?

Black02SS
July 25th, 2005, 02:08 AM
black02ss, Tordne and TAquickness - FTCs 20 and 21 are "nonload" cells and are used during idle and deceleration - when I'm in open loop.By "nonload" I mean when the engine is not under a load. We want to exclude "nonload" data, from the BEN/VE process, for accuracy.

We want to exclude throttle transients (Non-Steady State) for accuracy also. The AFR changes substantially during transients as the PCM adjusts for changing conditions.

I just want to filter out the garbage :D

Cheers,
joel

Joel, why don't you just disable the DFCO when tuning?.
Another thought, look at what your DFCO settings are and set a filter to ignore any MAP values below 'x', where 'x' is the MAP DFCO disable value.

I have been using the filters for a few weeks now and I am so impressed, it's nice not having to alter or monitor your driving habits when logging.
In fact, I think to slightly alter a great line from the Blues Brothers movie "we have filters powerful enough to turn goats piss into gasoline", hey Paul, lets put that on the features list :lol:

Cheers,
Ross

Ross,
The pics I showed have DFCO disabled, so why the difference in BEN? If this didn't disable DFCO, then please someone tell me what I missed.

bink
July 25th, 2005, 02:26 AM
Ross- Great line...only you would have remembered it!!

I need to check my DFCO as I'm getting some random funky BEN values (0.19, 0.34 etc.).




Are you telling the scanner to hide empty cells with a count of like 10 or something like that?

Empty cell count is 50.

GMPX
July 25th, 2005, 09:27 PM
Ross,
The pics I showed have DFCO disabled, so why the difference in BEN? If this didn't disable DFCO, then please someone tell me what I missed.

What is the timing doing at this point?
I was logging today after messing around with the base spark table enable and found as the PCM transitioned from 26deg of timing in the Open Throttle Table to the Closed throttle table 17deg of timing the WBO2 went from 14.6ish to 16.5, just for about 1sec then settled back to 14.7AFR. Just a thought!!.

Cheers,
Ross

Black02SS
July 25th, 2005, 11:23 PM
Check this out. I looked at a log I just did now and this is what I came up with. When I decel a lot and just take my foot directly off of the gas, I get short segments where my timing goes from 39.5 to 34. I can see in the graph that everytime this happens, my afr goes lean from 14.5 up to 16.8 like you mention. After looking at a few tables, this is what I have noticed. Base Spark Table In Gear timing is 34* for that area.... Sometimes my TPA doesn't go directly to 0, it will hang at 0.4. At those values, I see 32-33*. So I will say that the timing is causing this issue. I don't know though how to fix the TPA issue of 0.4% as I have reset the TPA sensor before and it still ends up like this. I will try that again.

Thoughts??

bink
July 26th, 2005, 12:11 AM
Check this out. I looked at a log I just did now and this is what I came up with. When I decel a lot and just take my foot directly off of the gas, I get short segments where my timing goes from 39.5 to 34. I can see in the graph that everytime this happens, my afr goes lean from 14.5 up to 16.8 like you mention. After looking at a few tables, this is what I have noticed. Base Spark Table In Gear timing is 34* for that area.... Sometimes my TPA doesn't go directly to 0, it will hang at 0.4. At those values, I see 32-33*. So I will say that the timing is causing this issue. I don't know though how to fix the TPA issue of 0.4% as I have reset the TPA sensor before and it still ends up like this. I will try that again.

Thoughts??

Exactly. That is why you don't want idle or deceleration data ( FTC cells 20 and 21). The data should be "load only" with the associated timing tables.
FWIW. This is why I wanted to filter FTCs.

Cheers,
joel

Black02SS
July 26th, 2005, 12:17 AM
I altered my Base Spark Table in Gear to reflect 39* at those cells and I will be trying it as soon as the wife goes to work... (She doesn't know this yet... shhh)... I see totally where you are coming from now on this. I have two more pics for you guys to see the differences.

Bink - If the timing in the bast spark table in gear doesn't help the issue, how would/do you suggest to tune the low MAP values? I would think just to use the values for 30kPa and below and leave the 35kPa above without FTC 21. Make sense?

bink
July 26th, 2005, 02:50 AM
I try to hit as many lower MAP cells, as possible, using long sweeping hills and steep hills. Steep hill and minimal TPS (lower gear) seems to work well for low MAP/kPa. I would then use decel data for those I can't hit otherwise - and blend it. My .02.

Cheers,
joel

Black02SS
July 26th, 2005, 09:07 AM
I try to hit as many lower MAP cells, as possible, using long sweeping hills and steep hills. Steep hill and minimal TPS (lower gear) seems to work well for low MAP/kPa. I would then use decel data for those I can't hit otherwise - and blend it. My .02.

Cheers,
joel

Thanks joel, its flat here. :lol:

bink
July 26th, 2005, 12:46 PM
I try to hit as many lower MAP cells, as possible, using long sweeping hills and steep hills. Steep hill and minimal TPS (lower gear) seems to work well for low MAP/kPa. I would then use decel data for those I can't hit otherwise - and blend it. My .02.

Cheers,
joel

Thanks joel, its flat here. :lol:

Oh, sorry.
Head towards the hills......I'm in Cincinnati. We have a bunch of them.
Long and steep. Going down -low kPas. Going up - high kPas.

My car idles at 1125-1175 rpm and 70-75 kPa - I can't really hit the very low kPa and low rpm cells.

TAQuickness
July 30th, 2005, 12:12 AM
Thanks joel, its flat here. :lol:

same here, but the ship channel overpasses are great!



My car idles at 1125-1175 rpm and 70-75 kPa - I can't really hit the very low kPa and low rpm cells.

I'm almost afraid to ask - but, how much overlap do you have at 0.05?

Would it not be easier to apply a filter to exclude less than ideal timing?

bink
July 30th, 2005, 06:11 AM
Thanks joel, its flat here. :lol:

same here, but
[quote][quote] the ship channel overpasses are great!
[quote="bink"]
My car idles at 1125-1175 rpm and 70-75 kPa - I can't really hit the very low kPa and low rpm cells.

I'm almost afraid to ask - but, how much overlap do you have at 0.05?

Would it not be easier to apply a filter to exclude less than ideal timing?



the ship channel overpasses are great!
That sounds like an adrenaline rush. :mrgreen:

239, 251 on a 106 lsa. 0.570" lift. Installed straight up (not that it effects overlap). 33* overlap at .050".

They call it the "Grand Am cam".... I think of it as "The Old Man cam".


Funny thing is when I put this cam in the "gurus" told me it couldn't be tuned AND it would not make decent HP (they new better than the Great Grand Poobah aka GM). 2-3 years ago conventional internet wisdom, in the States, was - "LS1s don't like a lot of duration..like the SBC does".

Thanks to NoGo, GMPX gojo/jfPilla, and all the guys on these boards, it does pretty well.

Cheers,
joel

TAQuickness
July 30th, 2005, 09:56 AM
239, 251 on a 106 lsa. 0.570" lift. Installed straight up (not that it effects overlap). 33* overlap at .050".


:shock: I think I just wet my pants. - - Yeah, I did...

How's low speed driving <30 mph, school zones, parking lots, etc?

I bet that thing pulls like a mad arse up top.

bink
July 30th, 2005, 10:43 AM
239, 251 on a 106 lsa. 0.570" lift. Installed straight up (not that it effects overlap). 33* overlap at .050".


:shock: I think I just wet my pants. - - Yeah, I did...

How's low speed driving <30 mph, school zones, parking lots, etc?

I bet that thing pulls like a mad arse up top.

It's not a big cam nowdays. Many are running 240-250s at .050" with 0.600s lift - BIG stuff.

It does very well at lower rpm thanks to my Nick Williams 90mm TB (Sounds like a commercial :lol: ). Low rpm throttle response vastly improved with the TB. I have 4.10s and an M12 (2.97 1st?) which really helps keep it in the 3000, and above, RPM range. I can cruise in 4th, 5th and 6th down to 1500 RPM - it's amazing what can be done with compuerized EFI.

Above 4000 rpm the cam wakes up - you can feel it kick in. Good fun. :D

Dirk Diggler
August 12th, 2005, 02:57 PM
For your original problem you can filter "including all data" where FTC not equal to 21. THis will include 20 and 22 if i am understanding your questinon correctly after a few beers

bink
August 12th, 2005, 03:12 PM
For your original problem you can filter "including all data" where FTC not equal to 21. THis will include 20 and 22 if i am understanding your questinon correctly after a few beers
Thanks man. :)
But , I don't want 20 either. Just 22. I double filtered as Paul suggested. Been way too hot around here to do much in the way of scanning/tuning - 99*F this afternoon.

Dirk Diggler
August 12th, 2005, 10:57 PM
YOu can say include data frames where FTC equal to 22 and that will get rid of all the other FTC's this is the "and" condition the "or" condition will be exclude data frames where FTC not equal to 22. Double filtering the same PID is a little tricky as me and black02SS have observed, I think its because we cant use parentheses to group filtering criteria...

Blacky
August 13th, 2005, 10:42 AM
Double filtering the same PID is a little tricky as me and black02SS have observed, I think its because we cant use parentheses to group filtering criteria...

That is right, there is no way to parenthesise (is that a word?) the items in a filter. However, AND will always take precedence over OR.

For example:
A and B or C = (A and B) or C
A or B and C = A or (B and C)

Regards
Paul

bink
August 13th, 2005, 11:35 AM
YOu can say include data frames where FTC equal to 22 and that will get rid of all the other FTC's this is the "and" condition the "or" condition will be exclude data frames where FTC not equal to 22. Double filtering the same PID is a little tricky as me and black02SS have observed, I think its because we cant use parentheses to group filtering criteria...

I double filtered the log by 1.) Filter and save log 2.) apply new filter to saved log.

My terminology is wrong I suspect. :lol: