Page 1 of 2

Far out idea that fred is likely to shoot down immediately

Posted: Tue May 24, 2011 2:23 pm
by dandruczyk
EDIT: Moved this from another thread in another section, posting this half of the post, unchanged, as an idea here.

NOTE: This will require a wideband O2 sensor.
Do you anticipate a self-tune operational mode (optional mode to be enabled by the user for limited periods or time), such that as long as they can get it to idle, and have a lambda table created, that gentle increases/holds of rpm,load and let the ecu hone in to the lambda value and automatically fill the VEtable cells for the user. For example: set lambda table to X (1.0, 14.7, 12.5 whatever) everywhere to start, adjust vetable idle region to something that allows the engine to start, manually tune that areas until lambda values are stable, then enable lambda control and auto-tune. In this mode lambda authority limits would essentially be ignored (since the vetable is not fully configured), and an iterative weighted average (or some other algorithm) is used to adjust fueling until the o2 reading is where we want it, thus gradually increase rpm/load and allow the vetable to auto-adjust within the firmware, it would be nice if the ecu could report what it's actually doing to the tuning software. (easily doable via the packet subscription model). Theoretically this would allow a significant portion of the table to be done automatically which may suit a subset of lesser skilled users, who don't understand what a fuel table should really look like.

Re: Far out idea that fred is likely to shoot down immediate

Posted: Wed May 25, 2011 2:10 am
by jharvey
So where did this come from? Link to the original thread please. Also the thread name is of poor form. You may not like it, but if I were djandruczyk I would consider not posting again in this forum, as you like to modify, and delete content.

A more appropriate name would be something like "djandruczyk self tunning theory" or "Fred feels like being a dink" :) Both would better more descriptive of the content of the thread.

Re: Far out idea that fred is likely to shoot down immediate

Posted: Wed May 25, 2011 10:50 am
by Fred
The thread title, Jared, is made entirely of Dave's words. Not only that, but I requested permission to move his post from the other place and remove the unnecessary content that we had discussed and resolved in IRC which detracted from the original thread. He suggested I delete the whole thing. I preferred to keep this, as although I don't agree, and will respond later, it's valid discussion for this forum section. Just not where it started its life. Again, his words, with his permission...

Re: Far out idea that fred is likely to shoot down immediate

Posted: Wed May 25, 2011 11:05 am
by johntramp
I don't think anything like this should be in firmware. It should be part of the tuning software.

Re: Far out idea that fred is likely to shoot down immediate

Posted: Wed May 25, 2011 11:34 am
by Fred
johntramp wrote:I don't think anything like this should be in firmware. It should be part of the tuning software.
I completely and totally agree, possibly with some small exceptions. Maybe I'll provide a more elaborate reply at a later date, but not now. Thanks johntramp.

Fred.

Re: Far out idea that fred is likely to shoot down immediate

Posted: Wed May 25, 2011 3:41 pm
by dandruczyk
Funny how people overreact (myself included).

Jared: I made that post on another subforum as fred was asking for comments (sorry, I don't remember the exact URL), in my typical rambling manner I went off topic, and posted a bunch of wrong information initially (due to lack of understanding regarding fuel tables, lambda tables and the VE concept which fred/baldur straightened me out on). Fred asked if he could move my post to a more suitable location (due to my rambling off topic), and I said to just delete it, since I couldn't. He said he'd delete the top part which was just plain wrong (due to my initial lack of understanding) and move the rest to a different spot as he saw some value in it. (I figured it wouldn't matter to keep it or not as my idea there is not well fleshed out). The thread title came exactly from my post, as I expected fred to shoot it down immediately, as I'm of the opinion that he'd rather not have that sort of logic/feature within the firmware.

So NO he didn't just go randomly editing some post, he ASKED ME FIRST, and I gave him permission to do so. He's the moderator and contrary to popular opinion, forum's are not free-for-all's and violation of guidelines/rules about content and abberant behavior can and will be edited out by moderators.

Johntramp/fredcooke:
I still think this is a value add within the firmware, think of it as a self tuning mode, its something that can run MUCH faster than the tuning software can, as that has the overhead of the datalog, buffering/processing/averaging of data, then the decision making portion, whereas within the device it can do that simply enough and adjust and hone much much faster.

Re: Far out idea that fred is likely to shoot down immediate

Posted: Wed May 25, 2011 3:52 pm
by Fred
djandruczyk wrote:Johntramp/fredcooke:
I still think this is a value add within the firmware, think of it as a self tuning mode, its something that can run MUCH faster than the tuning software can, as that has the overhead of the datalog, buffering/processing/averaging of data, then the decision making portion, whereas within the device it can do that simply enough and adjust and hone much much faster.
Fair point.

Re: Far out idea that fred is likely to shoot down immediate

Posted: Wed May 25, 2011 3:54 pm
by baldur
jharvey wrote:So where did this come from? Link to the original thread please. Also the thread name is of poor form. You may not like it, but if I were djandruczyk I would consider not posting again in this forum, as you like to modify, and delete content.

A more appropriate name would be something like "djandruczyk self tunning theory" or "Fred feels like being a dink" :) Both would better more descriptive of the content of the thread.
This was completely uncalled for.


As for auto tune, it is well at home in the firmware but there should be a separate table for auto learning values, this way you can set a limit to how much authority the auto tune has and you can see how much it has changed and revert the changes or accept them as modifications to the main fuel table.
Obviously you don't want it to be writing the auto tune table to flash all the time so that will have to be kept in RAM and either written to flash using logic to select a good time to write (and keep it on a separate page also to not corrupt other data if the write is incomplete) or kept in EEPROM or NVRAM.

Baldur

Re: Far out idea that fred is likely to shoot down immediate

Posted: Wed May 25, 2011 4:00 pm
by SleepyKeys
I think the best way to get a quick and usable rough tune is with an air meeter.

After getting a log of your engine running with an air meeter you can generate a ve table and fine tune it in SD mode.

As for firmware based auto-tune it think it should be very limited and renamed to something like a auto-trim.

-sean

Re: Far out idea that fred is likely to shoot down immediate

Posted: Wed May 25, 2011 4:05 pm
by Fred
Baldur, +1 for separate table and limited scope and eeprom/ram/etc. This is great, I don't even have to type, and all the good stuff in my head ends up on the forum. Hooray.
Sean0 wrote:As for firmware based auto-tune it think it should be very limited and renamed to something like a auto-trim.
Good call, re the naming! Naming is EVERYTHING! :-)

Fred.