[Gmecm] BLM cell confusion
Gary Evans
gary
Mon Nov 14 21:41:19 UTC 2005
Small nit:
The integrator does not need to reach the high clamp point to
increment the BLM. If learning is enabled, any deviation of the INT
more than 2 will begin to move the BLM after the specified wait time
has bee reached. The further the INT deviates from 128, the sooner
the BLM will update. See previous post on this subject.
Also, the INT does not reset on a BLM value change... at least not on
$88 or $a1
Other than that, you are correct.
On Nov 13, 2005, at 3:40 PM, WopOnTour wrote:
> Well first you need to operate it so it's IN cell 2 (In gear with
> some RPM above idle??)
> In any cell, as integrator climbs small amounts of injector pulse
> width/delivery (bpw x 1.001?) are occurring. When integrator hits
> it's high clamp point (eg 138) it will increment BLM +1 and reset
> integrator to 128. This would result in an increase from the base
> delivered fuel or base pulse width x 1.01 This cycle will continue
> if the O2 feedback continues to show leaner than stoich. So in your
> example if BLM in any cell achieved 128+5 (133) when it finally
> achieved a stoich ratio this would equate (depending on
> calibration) to say bpw x 1.05 This value will then remain "stored"
> in this cell so that next time the vehicle is at the same load
> (same cell) it will commence immediately with a 1.05 factor. Of
> course the opposite occurs if O2 feedback is indicating RICHER than
> stoich.
> Well that's the way I understand it anyway
> Hope it helps
> WopOnTour
>
> ----- Original Message ----- From: "Ryan Hess" <rgmecm at yahoo.com>
> To: <gmecm at diy-efi.org>
> Sent: Sunday, November 13, 2005 10:52 AM
> Subject: Re: [Gmecm] BLM cell confusion
>
>
>> Any other ideas about getting cell 2 to learn? I'm
>> really stumped here.
>>
>> Or - maybe you can explain the relationship between
>> INT and BLM? I think I read somewhere that one INT
>> step is 1/2 a BLM step? So could I say an INT of 138
>> would correspond to the current BLM + 5?
>>
>> Ryan
>>
>>
>> --- Ryan Hess <rgmecm at yahoo.com> wrote:
>>
>>> That helps immensely, Gary, thanks!
>>>
>>> Since all the other conditions are being met, I'll
>>> just have to assume it's my BPW that isn't above
>>> 1.08.
>>> It's hard to tell when the ALDL data only puts out
>>> integers...
>>>
>>> But in starting to tune the main VE table, I guess I
>>> would want cell 2 'learn enabled' while I hold the
>>> engine at some specified RPM, and "anchor" the VE
>>> table at that point with a BLM of 128 by changing
>>> the
>>> BPC vs EGR constant, then go open loop and start
>>> tuning the VE via wideband.
>>>
>>> I guess the problem is in enabling block learn. The
>>> BPW you mentioned of 1.08 is in the low hysteresis
>>> constant... I don't understand hysteresis to begin
>>> with, but I assume if I changed it to 0.7, I would
>>> want to change the high hysteresis value to around
>>> 1...
>>>
>>> Ryan
>>>
>>>
>>> --- Gary Evans <gary at garyandliz.com> wrote:
>>>
>>> > The $A1 code uses 3 BLM cells:
>>> >
>>> > Cell 0 = idle
>>> > Cell 1 = part-throttle
>>> > Cell 2 = "overrun" aka no load
>>> >
>>> > As you have discovered, cell 2 is entered on the
>>> > conditions of high
>>> > (ish) RPM and low map, such as coasting with the
>>> > throttle closed. I'm
>>> > not sure what the point of the cell is other than
>>> > for emissions
>>> > purposes.
>>> >
>>> > From the ARUR bin, the vales required to enter
>>> Cell
>>> > 2 are:
>>> >
>>> > Map less than 32 kPa
>>> > BPW less than 1.95 mSec
>>> >
>>> > There are also a number of conditions that must be
>>> > met for block
>>> > learn to be enabled. MAP, BPW and RPM must all be
>>> in
>>> > a certain range.
>>> > For the ARUR bin, the conditions are:
>>> >
>>> > Closed loop conditions met (Coolant temp up,
>>> engine
>>> > run time met, no
>>> > fault codes, o2 sensor ready)
>>> > Charcoal canister not purging.
>>> > RPM between 400 and 6375
>>> > MAP above 22 kPa
>>> > BPW above 1.08
>>> >
>>> > If you are just revving the car in neutral, it is
>>> > likely that one of
>>> > the conditions is not being met. The time it takes
>>> > for the BLM to
>>> > update is also dependent on how far the INT is
>>> from
>>> > 128. The table
>>> > looks like this:
>>> >
>>> > ###########################################
>>> > # F58 TABLE #
>>> > # BLOCK LEARN DELAY TIME INTERVAL #
>>> > # VS. DELTA BETWEEN INT AND 128 #
>>> > # TABLE VALUE = SEC*20 #
>>> > ###########################################
>>> > 8997 52 82 F58A 4.1 SEC
>>>
>>> > 0 INT-DELTA
>>> > 8998 4C 76 3.8 SEC
>>>
>>> > 2
>>> > 8999 3C 60 3 SEC
>>>
>>> > 4
>>> > 899A 32 50 2.5 SEC
>>>
>>> > 6
>>> > 899B 2C 44 2.2 SEC
>>>
>>> > 8
>>> > 899C 28 40 2 SEC
>>> > 10
>>> > 899D 24 36 1.8 SEC
>>> > 12
>>> > 899E 20 32 1.6 SEC
>>> > 14
>>> > 899F 1C 28 1.4 SEC
>>> > 16
>>> > 89A0 18 24 1.2 SEC
>>> > 18
>>> > 89A1 14 20 1 SEC
>>> > 20
>>> > 89A2 14 20 1 SEC
>>> > 22
>>> > 89A3 14 20 1 SEC
>>> > 24
>>> > 89A4 14 20 1 SEC
>>> > 26
>>> > 89A5 14 20 1 SEC
>>> > 28
>>> > 89A6 14 20 1 SEC
>>> > 30
>>> > 89A7 14 20 1 SEC
>>> > 32
>>> >
>>> > So, a delta of 4 (INT at 124, for example) will
>>> not
>>> > start to move the
>>> > BLM for at least 3 seconds.
>>> >
>>> > Each calibration will be a little different, but
>>> > that is the gist of
>>> > it. With so few cells, the BLMs will naturally
>>> vary
>>> > with driving
>>> > conditions. If it varies a lot then there is room
>>> > for improvement in
>>> > the tuning of your VE tables. In a perfect world,
>>> > the VE tables would
>>> > perfectly represent the exact airflow for every
>>> > condition. In the
>>> > real world it's never gonna happen, so don't spend
>>> > too much time
>>> > chasing "perfect" 128 values. Even filling your
>>> car
>>> > with a different
>>> > grade of gas can throw the values by a few, so a
>>> BLM
>>> > of 124 is
>>> > perfectly acceptable, nay, even excellent.
>>> >
>>> > If you do not have it already, I highly recommend
>>> > getting ahold of
>>> > the $a1 disassembly file and making an effort to
>>> > understand the
>>> > things the computer is looking for in its
>>> operation.
>>> > Even if you do
>>> > not understand all the computer language stuff,
>>> just
>>> > looking at the
>>> > tables and comments can be very enlightening.
>>> >
>>> >
>>> > -Gary
>>> >
>>> >
>>> > On Nov 10, 2005, at 8:07 PM, Ryan Hess wrote:
>>> >
>>> > > Yes, exactly. Entering cell 2 (by reving in
>>> > neutral)
>>> > > swings the INT to 137-ish, but the cell 2 BLM
>>> > doesn't
>>> > > change off 122. I don't know if learn enable
>>> has
>>> > some
>>> > > kind of qualifier that I'm not meeting in
>>> neutral
>>> > > (MPH? MAP? BPW?) or what, but cell 2 appears to
>>> be
>>> > the
>>> > > only one that doesn't move.
>>> > >
>>> > > An interesting side note is that my BLM data
>>> > varies
>>> > > widely depending on the driving that I do. But
>>> -
>>> > I
>>> > > guess that's to be expected if the $a1 truly
>>> does
>>> > use
>>> > > only 2 BLM cells (idle and off idle). I don't
>>> > know
>>> > > that it does. I can say that BLM cell 2 does
>>> > appear
>>> > > to be calculated from the cell 0 (idle) and cell
>>> 1
>>> > > (seems to be a changing TPS). I never see cell
>>> 2
>>> > > actually change, it only changes when you're in
>>> > other
>>> > > cells... "behind closed doors" so to speak.
>>> > >
>>> > > Now, from what I have seen in my data logs, it
>>> > appears
>>> > > as though the INT has to be off from 128 by 4
>>>
>> === message truncated ===
>>
>>
>>
>>
>> __________________________________
>> Yahoo! FareChase: Search multiple travel sites in one click.
>> http://farechase.yahoo.com
>> _______________________________________________
>> Gmecm mailing list
>> Gmecm at diy-efi.org
>> http://lists.diy-efi.org/mailman/listinfo/gmecm
> _______________________________________________
> Gmecm mailing list
> Gmecm at diy-efi.org
> http://lists.diy-efi.org/mailman/listinfo/gmecm
More information about the Gmecm
mailing list