To all DIY_EFI

bcroe at juno.com bcroe at juno.com
Thu Nov 8 23:39:40 GMT 2001


That display is quite linear, as defined in the 
accompanying writeup.  Different gain on rich
and lean sides to compensate for the sensor.

The issue might be "What is linear?".  If your engine
pulls in some air, you add some fuel that is only 
a few percent by weight, and negligible by volume.  
You add fuel with pulse width, and the amount of 
air is essentially unchanged.  So maybe a display 
that moves linearly as you increase fuel (injector
pulsewidth) would be a practical "linear".  The 
number discussed here is Fuel divided by Air, and
the sensor is very linear in this relationship, except
the gain chancges at stoich.

Trouble is, A/F is the reciprical F/A, and the reciprical 
of a linear progression is not linear.  So I see making
the F/A readout linear as the main goal, and the A/F 
values will fall where they fall.  The display does that, 
so the A/F scale is not linear, but the F/A is.  You may
still write those familiar A/F values next to the correct
LED though, as listed in the table in the writeup.  It
won't be 3/8 A/F per LED except at one point.  It WILL
be about .00164 F/A per LED from end to end.

Of course a digital readout will need some kind of 
linear readout; some kind of lookup table is the usual
way of handling this.  Fitting an equation works if you
add enough computing power.  The KC-5300 idea is 
an attempt to scale existing and available stuff to this 
problem.

Bruce Roe
 
On Thu, 08 Nov 2001 13:35:09 EST A79coupe at aol.com writes:
> In a message dated Wed, 7 Nov 2001  2:59:51 PM Eastern Standard Time, 
> A70Duster at aol.com writes:
> 
> > My idea is to use the LM3914 10 LED Dot/Bar Display driver.  If 
> you look at the volts to A/F ratio, the curve is "fairly" linear 
> from 10:1 to 14.7:1.  Then the curve knees up and again is fairly 
> linear from 14.7:1 to 20:1.
> > 
> > By drawing some lines in to guestimate how a linear device would 
> respond to the curve, I see no more than 2% error.  The worst of the 
> error bring around 14.7.  I envision two of the LM3914 working in 
> concert.  Use one of the LM3914's in the rich region (green LEDs, 
> 1.3 to 2.5 Vout) and the other LM3914 working in the lean region 
> (red LEDs, 2.5 to 2.9+ Vout).
> > 
> > Some will say to use a voltmeter, but I'm an "analog gauge" fan 
> cause they are quicker to read.
> > 
> > The circuit could be added to the WBO2 board (with the three "open 
> chip" holes). 
> 
> Take a look in incoming at the WB disp.zip.  this is Bruce Roe's 40 
> led display for the WB.  Actually only 36 led's are actually used, 
> but it measures from 10.0 AFR to 23.5 AFR in those 36 steps which 
> means that each step is 3/8 of an AFR.  It is set up to be a zero 
> center around Stoich, with led's lighting in bar graph fashion as 
> you go rich/lean of stoich.  
> 
> Not fully analog, but IMHO much easier to read than a bouncing 
> needle.
----------------------------------------------------------------------------
To unsubscribe from diy_efi, send "unsubscribe diy_efi" (without the quotes)
in the body of a message (not the subject) to majordomo at lists.diy-efi.org




More information about the Diy_efi mailing list