Subpage of MembersPage/MarcellGal/PowerAndTraction
s2central or sjmautotechnik talks about differences between AAN and 3B. (TODO: find the link).
- fuel pressure
- AAN FP=4bar
- 3B FP=3.5bar
That means sqrt(4/3.5) = 1.07, so 3B injectors flow 7 percent less, therefore require req_fuel of 7% more. (also roughly 7% more cranking pulsewidth).
Cranking pulsewidth EXPERIENCE
- cwh=13 (1.9msec)
- cwl=2D (4.5msec)
- req_fuel=1E (3.0 msec)
- 50L E85 and 30L RON98 (we could call this E53, requires appr 15% more fuel. 727cc/min injectors at 3.5bar are only 680cc/min, too small for 406 Hp with pure E85)
- cwh=16 (2.2msec)
- cwl=34 (5.2msec)
- req_fuel=24 (3.6 msec) (3.5msec was a bit lean at low-load. Ideally VE retune needed because low-load requires different change than high-load).
Cranking pulsewidths (cwl/cwh, in msec) OLD SPECULATION
- Jorgen's AAN config had 15/12 msec
- I doubt that this is tuned properly. I suscpect he just started the engine at CLT~20C with appr 13.6 msec pulsewidth and left there
- Marcell started both AAN and 3B with 17/11 msec. Cold-start was reasonable. Warm-start not so good. Much better warmstart with 17/7 msec (7 msec in effect)
- Teppo and MembersPage/MiskaPeippo suggests it is drowned from too much fuel, "hell of a lot", "I think I've used ~2ms or so ... with factory injectors ... alternate mode"
- in one of their configs, 2.2/1.6 msec is used in batch mode (during cranking, thus alternate=14), req_fuel=4.5 msec and VE=77 (for RPM=3000, MAP=100kPa), thus 3.47 msec. These must be big injectors !
- someone used 5/1.5 msec in batch mode (during cranking, thus alternate=14), but that was with req_fuel=4.2msec and VE=83 (for RPM=3000, MAP=100kPa), thus 3.49 msec. These must be big injectors !
- so, for AAN with req_fuel=7.5msec and VE=109 (for RPM=3000, MAP=100kPa), thus 8.18 msec
- 16/6 msec might be a good next setting to try (in alternate mode). Or ... maybe lower PW in batch mode would be the way to go.
- so, for 3B with req_fuel=8msec and VE=109 (for RPM=3000, MAP=100kPa), thus 8.72 msec
- 17/7 msec worked fine when warm, but ... maybe lower PW in batch mode would be the way to go.
- for the siemens Deka (even racetronix keeps changing the specs) injectors, I apply the following changes (to start, warm up to retune VE)
- cold cranking: 5 msec => 3.2 msec (4/1.8 + 1 = 3.22)
- warm cranking: 2 msec => 1.5 msec (1/1.8 + 1 = 1.56)
- req_fuel: 5 msec => 3.1 msec (5/1.6 = 3.13, hopefully on the safe side)
- 668 * sqrt (3.5/3) = 721.52 cc/min, instead of 400 cc/min (factory injectors?) 721/400 = 1.8025
Brave men do like this: in audi 5cyl 20vt thmb rule: cc/per injector = max power in hp. 440cc = 440 HP
I think it's brave. But with tuned boostcontrol it can be tried. If going up slowly, one will know when running out of injectors flowrate.
I think there is an error on http://www.rceng.com/technical.htm they swapped the BSFC for turbo and supercharged. Turbo is better (lower). Below I calculate with BSFC(gasoline)=0.55l / (HP * l), consistent with some other sources.
- For factory 220 Hp 220 * 0.55 / (5*0.8) * 10.5 = 317cc/min (they used 350cc/min)
- Teppo Kalske used to have 460cc with wich ended up 100% dc at 5000 rpm with reasonable afr (580Nm/3800, 440hp/6800). Prolly maxed out near 400 HP:
- for 400 HP, even with 100% duty: 400 * 0.55 / (5*1) * 10.5 = 462 cc/min needed
- <Jorgen__> It's recommended but not needed.
- or in other words 360 * 0.55 / (5*1) * 10.5 = 415 cc/min (enough for 360 HP) is enough to start with but will not get max power
- 420 HP with 80% duty and pessimistic BSFC=0.63: 420* 0.63 / (5*0.8)* 10.5 = 694.58 cc/min (so the 599cc/min@3bar injectors are enough for this engine)
Safe injector sizing
If purchasing new injectors, it is best to have it sized high enough to be enough for sure, for the times to come.
The injector sizing should be rather 420 HP with K26/27 turbo (even if we'll more likely get only 370 HP).
- emil recommends [L107FM] to most people, except the insane ones.
- 5 x L107FM injectors INSTALLED
- earlier they had this info: "Static Flow Rate: 60 lb/hr @ 43.5PSI (300kPa) = 630cc/min"
- by now they changed to: "Static Flow Rate @ 43.5PSI (300kPa) w/Gas: 63.20 lb/hr = 668 cc/min = 475 g/min"
- At 350kPa fuel pressure, this is 668 * sqrt (3.5/3) = 721.52 cc/min
- Wide angle spray pattern (is that good?)
- May I get your (L107FM) injector settings for my database MembersPage/KevinBlack/InjectorSpecDatabase when you are satisfied with the result?
- [old Bosch-style Siemens injector] has Static Flow Rate (GAS): 57 lb/hr @ 43.5PSI (300kPa) = 599cc/hr
- Pencil beam spray pattern (is that good?)
- this Bosch-style injector has MUCH wider body than the slim L107FM. Also, 1mm longer! (that should not cause a problem, right?)
- Price is slightly higher: 5 * $57.49 USD + shipping
- both injectors sit in 14mm drill, and have AMP connector.
Kalske's hints in this case is to use 044 pump in PnP? install inside fuel tank
- 895201531A strainer
- 895201779H reservoir mount
- N 90330701 banjo union
- N 0138128 seal: 2 pieces needed
- N 0110691 domed cap nut
- I got a 130 EUR quote for above parts, I think that's a bit too high. But my mechanic did a good job, installed the fuelpump in the tank with some improvisation
First two are the most important ones, they are used fitting the pump in place of the original one. Banjo, seals and nut are used for fuel line connection.
It is weird that banjo is used inside the tank, for the fuel pressure line. Doesn't the 90 degree break obstruct the flow ?
- avtoshop (Russian) drawing (for the factory pump? the domed cap nut seems different):
Fuel pressure sensor
Currently not installed. Would be nice though. Not sure what is a good place to install it on 3B engine. Obviously, before FPR. But I just don't like the idea of messing around fuel tubes.
Fuel Pressure Regulator MAP reference
- The car was running well for weeks during the NA breakin
- the wastegate springs were out most of the time (yes, this had 2 springs inside, looks like tweaked by the previous owner)
- than we installed the tubings for the questioned (to be measured) charge cooler
- we limited boost to MAP=160 kPa (max 164kPa in the logs) during the first boost-run
- watching the lambda continuously and listening to engine
- MAT was very low. Obviously MAT showed at least 6C lower than real, but something else might be weird here. Truly, the engine is protected very well to prevent heat-soak to keep the intake air cool (TODO: pics of the engine bay with non-metal pressure tubes and lotsof Al-foil), but it's hard to believe we were so efficient
LogAnalysis of the boost-run revieled a problem though:
- near 100kPa, appr 10.0msec/100kPa pulsewidth was needed (with the 3B factory fuel injectors, and likely the factory FPR). This means req_fuel*VE/100 = 7.5msec * 133 / 100
- below 50kPa, we had to remove some VE
- above 140 kPa, we had to add some VE. Eg at 162kPa, appr pw=18.1msec was needed, that is 11.2 msec /100kPa
With the 30V transient diode flyback you just don't expect that much difference. The analysis revieled that the FPR reference was connected to athmosphere. Stupid mistake, should be connected to MAP.
Connecting to MAP makes the
- VE table flatter
- the low-kPa loadsites easier to tune (longer pulsewidth)
- allows more fuel in the high-kPa loadsites
- with the FPR reference forgotten, 160kPa @ 6072 RPM used 18.3msec pulsewidth, that is injector duty=93%. The factory injectors are small anyway (220Hp application, good for appr 270 Hp), this made them even smaller :-)