Interpolate Extrapolate and Smooth.

[ayjayef]

Spent the day on the dyno and got some good load numbers in the Fuel Table.

I'm not sure I "trust" the 1500 column as it looks too fat and the dyno cavitated a lot at everything under 2000rpm

The 6 cells in 0%-2.5% TPS and 1000-2000rpm is the idle window, intentionally leaner than the rest of the table.

I'll be back at the dyno to do some pulls into the higher rev ranges but it would be good to have something sensible in there so I'm only tweaking AFR's instead of running a super lean plasma cutter.

 

So questions are:

1: Under 6000 rpm - do I highlight a few known good cells and extrapolate left to fill-in the missing areas I couldn't load tune?

2: Higher than 6000 rpm -  extrapoate right or should I make some guesses in the 14000-15000 columns and interpolate between them and the known good?

 

 

Interested to hear people's thoughts.

 

Andy.

2017 Kawasaki H2-014.pclr

[ayjayef]

anyone got an opinion?

[Tim D]

I was waiting for someone else to respond, but in my experience, tuning my fuel table I couldn't have predicted any of the cell values, I believe you need to 'access' each and every cell to do it properly.

[Adamw]

To start with you will notice there is very little difference in the fuel table numbers from about 30% TP up.  So I would delete a whole lot of those rows to make your initial tune much less work.  Then only add a new row if after initial tune you see a large change between two adjacent rows that doesnt appear to be linear.  On an ITB engine I would normally start with TP breakpoints something like 0,2,4,7,10,15,20,30,50,70,100% and I rarely have to add any more than that.

 

On 11/17/2018 at 9:17 PM, ayjayef said:

Under 6000 rpm - do I highlight a few known good cells and extrapolate left to fill-in the missing areas I couldn't load tune?

Due to the reason above it looks like you could just "fill down" the rest of the 2500 and 3000 columns.  Maybe add a little extra on so you are starting on the safe rich side and then pull it back using logs of mixture map.

 

On 11/17/2018 at 9:17 PM, ayjayef said:

2: Higher than 6000 rpm -  extrapoate right or should I make some guesses in the 14000-15000 columns and interpolate between them and the known good?

Your numbers will usually increase proportionally to torque so it should ramp up towards the peak torque RPM, then flatten off from there towards redline.  I wouldnt have a clue where an H2 will peak so you will probably have a better idea than me. Again I would make the numbers bigger than you expect they need to be so you are starting on the safe side.

 

Also, these things are supercharged arent they?  If so, make sure you have the equation load source set to MAP, the open loop Lambda table turned on (assuming using traditional fuel equation) and make sure the lambda target table has MAP or MGP on the axis.  These settings are needed to ensure that boost is properly compensated for (if for instance boost pressure is changed at a later date).

[ayjayef]

2 hours ago, Adamw said:

something like 0,2,4,7,10,15,20,30,50,70,100%

Makes sense, I'll do that.  It's already non-linear in the tiny openings and as you said, not much changes after 30%

 

2 hours ago, Adamw said:

usually increase proportionally to torque so it should ramp up towards the peak torque RPM, then flatten off from there towards redline

I have some old ramp runs done on this engine with the factory ECU so I'll use those to build the direction.  Ta, great tip that increase is proportional to torque.  I should be able to guess the peak torque area values and then fill-in the blanks.  I'll spend a bit of time getting that to make sense.

 

2 hours ago, Adamw said:

supercharged arent they?

Yes, and this one has a bigger supercharger in it.  

 

2 hours ago, Adamw said:

equation load source set to MAP

Currently set to: Load=BAP/MAP Xover.  I'll change it.

 

2 hours ago, Adamw said:

open loop Lambda table turned on

Current Fuel Equation Mode = traditional.

Current Open Loop Lambda Table = off

The Help Browser for Open Loop Lambda Table shows "Can't Reach this page error" so I had left it off.  Can't find it in the search function of the help file either.  I'll turn it on.

 

2 hours ago, Adamw said:

lambda target table has MAP or MGP on the axis

Currently set to: MGP

 

Thanks for the guidance Adam.

[Greg W]

"Also, these things are supercharged arent they?  If so, make sure you have the equation load source set to MAP, the open loop Lambda table turned on (assuming using traditional fuel equation) and make sure the lambda target table has MAP or MGP on the axis.  These settings are needed to ensure that boost is properly compensated for (if for instance boost pressure is changed at a later date)."

 

Adam , The load source set to MAP rather than BAP/MAP x over , how much difference will it make with various boost levels ? this setup we are both using  is based on my turbo bike  , which i use both in Australia at 200m altitude  and at Bonneville at 1200m , and so far works great up to where the turbo starts loosing effiancy and i have generated a small map table to remove fuel to compensate 

I have 6 stages of boost and it seems to work well  ?

[Adamw]

 

1 hour ago, Greg W said:

Adam , The load source set to MAP rather than BAP/MAP x over , how much difference will it make with various boost levels ? this setup we are both using  is based on my turbo bike

I possibly spoke without thinking through all scenario's with that comment.  Load=MAP or Load=BAP/MAP are both fine to use.  In fact, BAP/MAP Xover may be preferable if there are really large camshafts or something at play that gives a very unstable MAP although I have never needed to use that mode.  The important thing is Load = BAP is the main one we want to avoid. 

[Greg W]

Ok thanks

[ayjayef]

On 11/23/2018 at 1:51 PM, Adamw said:

I possibly spoke without thinking through all scenario's with that comment.  Load=MAP or Load=BAP/MAP are both fine to use.  In fact, BAP/MAP Xover may be preferable if there are really large camshafts or something at play that gives a very unstable MAP although I have never needed to use that mode.  The important thing is Load = BAP is the main one we want to avoid.

While I have large camshafts I think my MAP looks pretty stable.  

At which point would it look unstable (mostly at idle?) and how much change are we talking about?

Attached is a log of idle, 4000rpm load, and idle again.  MAP seems to be following TPS and would have guessed I shouldn't need BAP/MAP crossover?

log20181116-4k-b-load-fuel-tune.llg

[Adamw]

31 minutes ago, ayjayef said:

MAP seems to be following TPS and would have guessed I shouldn't need BAP/MAP crossover?

I agree, that looks completely usable to me.

[ayjayef]

Well that answers that then!  

I should have the table roughed-out and back on the dyno in a week odd to start some ramp runs into the "unknown".

 

Thanks Adam, MAP it is.

 

[ayjayef]

Made the suggested changes to the Fuel table axis, load source set to MAP and open loop Lambda table turned on (still can't open the help file for it though)

 

I had some ramp-run dyno graphs from last year with the factory ECU and...

6,000rpm - 50ft/lbs torque

11,000rpm - 135ft/lbs torque

So I multiplied the 11k column by 2.7 and interpolated between the known good 6,000 and the fat 11,000

Sure made a funny looking fuel map but I guess once back on the dyno it will take shape from there.

(tough getting dyno time in a town with 1 dyno at Christmas when everyone gets their bike out of the shed)

 

[Adamw]

2 hours ago, ayjayef said:

Sure made a funny looking fuel map but I guess once back on the dyno it will take shape from there.

Yep, looks a bit extreme but it will be a safe starting point.