Big injectors + MS hi-res question?
#1
Senior Member
Thread Starter
iTrader: (14)
Join Date: Nov 2004
Location: Ambler, PA
Posts: 1,275
Total Cats: 1
Big injectors + MS hi-res question?
Granted hi-res gives you better resolution for larger injectors, but my question is this:
If using, say 1000cc injectors, the req fuel will be very small and I know that this is just a muliplier, but I've also read that haveing too small a req fuel can cause issues with fine tuning. So is it better to run a larger req fuel number and have lower VE numbers in the fuel table (scale the table for a higher req fuel) or just go with the lower req fuel and usual VE numbers?
If using, say 1000cc injectors, the req fuel will be very small and I know that this is just a muliplier, but I've also read that haveing too small a req fuel can cause issues with fine tuning. So is it better to run a larger req fuel number and have lower VE numbers in the fuel table (scale the table for a higher req fuel) or just go with the lower req fuel and usual VE numbers?
#5
Boost Pope
iTrader: (8)
Join Date: Sep 2005
Location: Chicago. (The less-murder part.)
Posts: 33,022
Total Cats: 6,590
Right. Ideally, you want to take advantage of as much of the space as possible that's been provided for you in the VE cells- you've got from 0 - 255 to play with. I'd guess that is resolution is the concern, ideally you'd want to be well into the low 200's in the upper cells, to really maximize the range of tuning available to you.
It's ironic that this came up this morning, I just got done re-scaling my own VE table not 10 minutes ago to gain some more resolution...
It's ironic that this came up this morning, I just got done re-scaling my own VE table not 10 minutes ago to gain some more resolution...
#7
Boost Pope
iTrader: (8)
Join Date: Sep 2005
Location: Chicago. (The less-murder part.)
Posts: 33,022
Total Cats: 6,590
No. I decreased my Req. Fuel value by dividing the old value (6.1) by 1.4, giving me a new Req Fuel of 4.4. I then scaled the entire VE table by 1.4, resulting in larger numbers in each cell. IOW, I'm giving myself some more resolution to work with, causing each +/- 1 in the VE cells to be a finer increment.
#9
Boost Pope
iTrader: (8)
Join Date: Sep 2005
Location: Chicago. (The less-murder part.)
Posts: 33,022
Total Cats: 6,590
The results have been that since building the new table earlier this morning, I've pretty much sat here in my pajamas drinking beer and futzing around on the forum, instead of going out to the car and loading the new MSQ.
MLV. I played with autotune early on, but it has a serious flaw- Autotune only tunes the one cell that you happen to be closest to at any given time, and it does not scale its correction relative to how far from the center of that cell you happen to be in. MLV is much smarter- if a certain record happens to be between four cells, slightly closer to the top ones, then MLV will portion out its corrections appropriately, weighting correction more heavily towards the upper cells.
This was the same thing that used to **** me off about the "Sampling Map" that Greddy added to the 2.1x builds of the EMU software- it recorded only one value per cell, without regard for whether you were on the edge of the cell or right in the middle. When I built the 1.6 AFM map for the EMU, I basically did the same thing (by hand) that MLV does automatically.
Are you using MLV or autotune for tuning?
This was the same thing that used to **** me off about the "Sampling Map" that Greddy added to the 2.1x builds of the EMU software- it recorded only one value per cell, without regard for whether you were on the edge of the cell or right in the middle. When I built the 1.6 AFM map for the EMU, I basically did the same thing (by hand) that MLV does automatically.
#10
Boost Pope
iTrader: (8)
Join Date: Sep 2005
Location: Chicago. (The less-murder part.)
Posts: 33,022
Total Cats: 6,590
I've come up with a new theory- I'd like opinions.
I never got around to testing the rescaled map I did this morning, but I decided to change it again. This time, rather than pulling a number out of the air, I specifically chose 3.5 as my ReqFuel constant. Why? According to the MegaManual, the resolution of the injector PW computation with MSnS-E HR is 0.035ms. Assuming that my understanding of the PW calculation is correct, that ignoring enrichments and lag, PWactual = (VE/100)*ReqFuel, then there is now a perfect correlation in the steps between VE numbers and actual output- each VE increment of 1 should increment actual PW by exactly one step of its native resolution. This should, in theory, yield the most precise control possible.
Of course, my VE table (which I multiplied by 1.74) now has some fairly large numbers in it. Bear in mind that a lot of these cells are still untuned, but here's what it looks like presently.
Thoughts?
I'll take this map for a drive tomorrow and see how it goes.
I never got around to testing the rescaled map I did this morning, but I decided to change it again. This time, rather than pulling a number out of the air, I specifically chose 3.5 as my ReqFuel constant. Why? According to the MegaManual, the resolution of the injector PW computation with MSnS-E HR is 0.035ms. Assuming that my understanding of the PW calculation is correct, that ignoring enrichments and lag, PWactual = (VE/100)*ReqFuel, then there is now a perfect correlation in the steps between VE numbers and actual output- each VE increment of 1 should increment actual PW by exactly one step of its native resolution. This should, in theory, yield the most precise control possible.
Of course, my VE table (which I multiplied by 1.74) now has some fairly large numbers in it. Bear in mind that a lot of these cells are still untuned, but here's what it looks like presently.
Thoughts?
I'll take this map for a drive tomorrow and see how it goes.
#11
Elite Member
iTrader: (15)
Join Date: Dec 2007
Location: San Antonio, Texas
Posts: 4,847
Total Cats: 27
Man... You are on a thought provoking streak right now. First the dyno quest thread, then the WI thermodynamics thread, and now this... Great stuff, all of it
It is late so I'll check your pw numbers sometime later. But, details aside, this is an excellent idea IMO.
It is late so I'll check your pw numbers sometime later. But, details aside, this is an excellent idea IMO.
#12
No. I decreased my Req. Fuel value by dividing the old value (6.1) by 1.4, giving me a new Req Fuel of 4.4. I then scaled the entire VE table by 1.4, resulting in larger numbers in each cell. IOW, I'm giving myself some more resolution to work with, causing each +/- 1 in the VE cells to be a finer increment.
#14
Elite Member
iTrader: (21)
Join Date: Jun 2006
Location: Point Pleasant, NJ
Posts: 2,957
Total Cats: 2
Granted hi-res gives you better resolution for larger injectors, but my question is this:
If using, say 1000cc injectors, the req fuel will be very small and I know that this is just a muliplier, but I've also read that haveing too small a req fuel can cause issues with fine tuning. So is it better to run a larger req fuel number and have lower VE numbers in the fuel table (scale the table for a higher req fuel) or just go with the lower req fuel and usual VE numbers?
If using, say 1000cc injectors, the req fuel will be very small and I know that this is just a muliplier, but I've also read that haveing too small a req fuel can cause issues with fine tuning. So is it better to run a larger req fuel number and have lower VE numbers in the fuel table (scale the table for a higher req fuel) or just go with the lower req fuel and usual VE numbers?
jesus christ tom. did i miss a build thread? 1000cc... what do you have up your sleeve?
#17
2 Props,3 Dildos,& 1 Cat
iTrader: (8)
Join Date: Jun 2005
Location: Fake Virginia
Posts: 19,338
Total Cats: 573
curious question/comment: does the MS extrapolate values that fall below the table values or does it make them zero? I ask because you periodically DO get below 800 rpm (cranking) and below 23 kpa (compression braking) and I'm wondering if it makes the car go haywire when you drop below those #s.
corollary: if it doesn't extrapolate, the lowest column / row should be set to a number well below you ever see. ie zero kpa / zero rpm. the hydra has this hard coded so there's always a 0 on both.
corollary: if it doesn't extrapolate, the lowest column / row should be set to a number well below you ever see. ie zero kpa / zero rpm. the hydra has this hard coded so there's always a 0 on both.
#19
Boost Pope
iTrader: (8)
Join Date: Sep 2005
Location: Chicago. (The less-murder part.)
Posts: 33,022
Total Cats: 6,590
Success. The new table works nicely with the 3.5 ReqFuel constant. It needs some fine-tuning (nothing that MLV has been unable to deal with) but it's definitely operational. Took it all the way to 11PSI with nary a hiccup, and I'm hoping it's not my imagination, but my idle AFRs seem a tad more stable. I wonder if my recent troubles in finding a dyno have been the work of a higher power, denying me access until this hurdle was overcome...
What a happy coincidence that the fueling requirements of the Miata engine, coupled with the 440cc injectors that I just happened to have, allowed this rather unusual ReqFuel configuration to work.
y8s, the MS does indeed seem to extrapolate off the edges of the table. My idle, which is still not 100% perfect, occasionally dips down into the 500-600 range, and even then the fueling continues properly.
What a happy coincidence that the fueling requirements of the Miata engine, coupled with the 440cc injectors that I just happened to have, allowed this rather unusual ReqFuel configuration to work.
y8s, the MS does indeed seem to extrapolate off the edges of the table. My idle, which is still not 100% perfect, occasionally dips down into the 500-600 range, and even then the fueling continues properly.