When does the ECM switch from closed loop to open loop mode?

[Totenkopf]

Originally Posted by ToBeFrank

BTW, interpolating cells is mandatory if you want a correct algorithm. An example: suppose your VE table has a TPS axis with 10 and 15 and a RPM axis with 2000 and 2250. Now suppose you log a data item at 12.5 TPS and 2125 RPM. Which cell does this data item contribute it's data to? The next data item is logged at 12.0 TPS and 2150 RPM. Which cell does it go into?

My understanding is that the Delphi firmware in the ECU does the cell interpolation real-time. You don't need to pre-smooth it.

[BVBOB]

Originally Posted by ToBeFrank

See below for what I define as spikey.



This is NOT what I'm doing. I am not smoothing the VEs after they are generated. I have completely replaced the Twin Scan VE generator software with my own algorithm. That algorithm includes a smoothing algorithm. There are no assumptions being made. The data logged from the bike tells it what to do.



So do you think this surface plot looks acceptable? This plot is what you get when street tuning with the Twin Scan software. BTW, Steve Cole agrees with me that the answer here should be "no" (supporting link posted earlier in this thread).





This is NOT what I'm doing.



When did I say "having a smooth color transition is the only thing that matters"? My point is if the algorithm to generate the new VEs is correct, you automatically end up with a smooth graph. And for the nit pickers, yes, you'll have some spikes, but in general the map will be pretty smooth.

BTW, interpolating cells is mandatory if you want a correct algorithm. An example: suppose your VE table has a TPS axis with 10 and 15 and a RPM axis with 2000 and 2250. Now suppose you log a data item at 12.5 TPS and 2125 RPM. Which cell does this data item contribute it's data to? The next data item is logged at 12.0 TPS and 2150 RPM. Which cell does it go into?

Ok, another question, when you say you are using the data from the bike, are you talking about the data from the factory 02's? Just want to be clear on this, if so, I would assume you are taking the VE vs VE new in the data mode of the SERT/SEST/tts type tuners and using that info to generate a new ve table, maybe converting that data to "comma delimited" and viewing in an excel spreadsheet.

[ToBeFrank]

Originally Posted by Totenkopf

My understanding is that the Delphi firmware in the ECU does the cell interpolation real-time.

Yep, but it can only interpolate the data it has in the map, the VEs. If that data is all over the place, as the graph I posted illustrates, the interpolation will be all over the place, and incorrect. Note that with a closed loop ECM, this is not a problem as the VE New values provided by the ECM are already interpolated. The map you generate is made from those VE New values. However, we're talking specifically about the Twin Scan here. We're not using the ECM's interpolated VE New values (on an open loop bike, there are none!). We are generating the VE New values ourselves from the AFR data we are logging. Essentially, what I'm doing then in my algorithm is emulating the behavior of a Delphi closed loop ECM.

[ToBeFrank]

Originally Posted by BVBOB

Ok, another question, when you say you are using the data from the bike, are you talking about the data from the factory 02's?

We're talking about the Twin Scan and generating maps from the logged wideband AFR data. If the bike is not providing VE New values (i.e. an open loop bike), the VE New values are being generated from the AFR data.

Just want to be clear on this, if so, I would assume you are taking the VE vs VE new in the data mode of the SERT/SEST/tts type tuners and using that info to generate a new ve table

That's one mode of the software I developed, but that is a different subject than what we are talking about here.

maybe converting that data to "comma delimited" and viewing in an excel spreadsheet.

As I told you in email, I am NOT using Excel at all. I developed the algorithm I am using, and it is fully contained within my software.

[AZFXSTB]

I told you guys TBF is a geek to the 10th power.....

Back to my corner with Clay


I have the sudden urge to make the Carlos Mencia retard noise

[ToBeFrank]

Originally Posted by AZFXSTB

I told you guys TBF is a geek to the 10th power.....

Just don't tell my girlfriend...

[BVBOB]

That's one mode of the software I developed, but that is a different subject than what we are talking about here.

And I thought we were talking about when the ECM switched form closed loop to open loop, severe thread drift I guess, but yet another question, have you logged afr data after using your smoothing software and compared it directly to the "before" afr, if so, what were your findings, and secondly, have you put this to the test yet on any dyno with it's own afr (or 4gas) sampling equipment for comparison, and if so, did it help performance?

[AZFXSTB]

Originally Posted by ToBeFrank

Just don't tell my girlfriend...

Hell Frank, who would I tell? I have never met the same girl twice with you...



Btw, Buffalo Soldier ride this Sat interested?
Info on ride

[ToBeFrank]

Originally Posted by BVBOB

have you logged afr data after using your smoothing software and compared it directly to the "before" afr, if so, what were your findings

Don't focus on the smoothing. That is just one part of the algorithm (although it is the hardest part to get right). I did not develop smoothing software. I developed VE generating software. Think of it as V-Tune for open loop bikes using the Twin Scan (I am not affiliated with V-Tune or TTS in any way). The other mode I was alluding to will allow SERT and Super Tuner users with closed loop bikes to street tune their bikes the same way V-Tune users are doing (with a couple of extra copy/pastes since I don't have access to the SERT/Super Tuner map data).

I got sidetracked. To answer your question... the only thing that determines the accuracy of the algorithm is this: do the generated VE values result in an AFR out of the tailpipe that matches the AFR target the algorithm was shooting for? The answer is yes. Graphing the logged AFR data after applying the generated VE tables verifies it.

secondly, have you put this to the test yet on any dyno with it's own afr (or 4gas) sampling equipment for comparison, and if so, did it help performance?

The real question here is does the Twin Scan with Wego produce accurate AFRs? This is completely unrelated to my algorithm as all the algorithm can do is assume the information from the Twin Scan and wideband sensors is reliable and generate the VEs accordingly. As I stated above, the algorithm is accurate in that sense.

As far as a performance difference, yes, there is a noticeable difference between the generated map and the base map. My algorithm generated a map that was within 4 HP and 4 torque of what my bike produces after being dyno tuned. Note that the dyno tune included tuning of the timing. You can't "tune" timing on a street tune. All you can do is get the timing just short of knock and then hope that is close to max performance.

[BVBOB]

The real question here is does the Twin Scan with Wego produce accurate AFRs? This is completely unrelated to my algorithm as all the algorithm can do is assume the information from the Twin Scan and wideband sensors is reliable and generate the VEs accordingly. As I stated above, the algorithm is accurate in that sense.

And that was going to be my next question, it still boils down to starting with reliable logged data from the twinscan, your method would have to trust the data from the twinscan anyway, otherwise you are not starting from a known point, without a reliable starting point, everything beyond that will be off also. As for the performance aspect, I would guess I did over a dozen bikes with the twinscan before I installed my dyno, I have put many of them on the dyno since and have pretty much seen the same thing as you have, a few lbs gain, mostly from optimizing timing though. I think it boils down to the twinscan being a great tool and if you have something to supplement an already great tool to get even better results and make map changes more user friendly I am all for it.

[nw_guy4_fun]

I am an admitted "tooner" not a SME by any means

reliable logged data from the twinscan

Of course regardless of any algorithms this data has to be trusted. Can it be trusted when sampled on a bagger with stock headpipes and free flowing mufflers that have equal flow potential out both?

[ToBeFrank]

Originally Posted by BVBOB

And that was going to be my next question, it still boils down to starting with reliable logged data from the twinscan, your method would have to trust the data from the twinscan anyway, otherwise you are not starting from a known point, without a reliable starting point, everything beyond that will be off also.

Note that this does not apply to just my software. It applies to the Twin Scan software, the V-Tune software, the stock Delphi closed loop system, etc. They all assume they're getting good data.

As for the performance aspect, I would guess I did over a dozen bikes with the twinscan before I installed my dyno

Can you post some surface plots from these maps? Specifically ones that were tuned with the Twin Scan on the street and were not modified after you checked them on the dyno. In other words, I'd like to see the surface plots you have from a Twin Scan street tune only.

[BVBOB]

Originally Posted by ToBeFrank

Note that this does not apply to just my software. It applies to the Twin Scan software, the V-Tune software, the stock Delphi closed loop system, etc. They all assume they're getting good data.



Can you post some surface plots from these maps? Specifically ones that were tuned with the Twin Scan on the street and were not modified after you checked them on the dyno. In other words, I'd like to see the surface plots you have from a Twin Scan street tune only.

Should not be a problem, being the geek that you are what's the best way to accomplish this, no tuning software on my internet computer and no internet on my tuning computer.

[claydbal]

Originally Posted by BVBOB

Should not be a problem, being the geek that you are what's the best way to accomplish this, no tuning software on my internet computer and no internet on my tuning computer.

load twinscan software on your internet puter, load files on thumbdrive, insert thumbdrive in internet puter and open with newly installled software.

[BVBOB]

Originally Posted by claydbal

load twinscan software on your internet puter, load files on thumbdrive, insert thumbdrive in internet puter and open with newly installled software.

I believe he wants the SERT VE graph.