Good luck with the dragons in Alpha-N land Tuomo. Keep us posted.!

 

Richard - the software is an alpha release but the system Tuomo is running is SMAF - not ST-Alpha.

Alpha-N isn't suitable for turbo engines, that's why we are developing a MAP based system..

 

Hi John,

I was wondering about this..!!

Ok, I get it now..

Say, can't you just do a conversion scaled interpolation voltage curve from the SMAF to give a log converted value for Kpa voltage in a delco sensor?

(My guess is the values are log not linear, so that is where the issue is. However, I think you can reverse engineer the density ratio formula and chart the scale to air mole weight with a correlation to the TPS for load. I wish I could code, this is an interesting issue to solve, even if the LH is batch fire..)

Let me know how the MAP development goes, as it would be really interesting to test once ready.. With the MAF, I will surely be limited on the inlet restriction for the compressor as we discussed last I visited you.

Hope things are well in Cambridge. Cheers from across the channel!

Best,

 

Not wishing to hijack Tuomo's thread, but we have the basics of the MAP system all worled out, it has been evaluated with a twinscrew system but has an annoying little trick which we are having problems nailing dowm. The proto system is being installed on another engine configuration to try and throw some light on this. I will definitely publisise when the system is available !

 

Alpha-N could be used for mildly boosted turbo motors, as long as the barometric pressure compensation has enough range. The problems only appear at higher boost levels when the density correction to load is no longer linear.

Alpha-N corrects for pressure variations. The new speed-density (MAP) systems incorporate the throttle position sensor. As far as the modern Alpha-N and speed-density systems go, they aren't that different anymore. The only difference I can think of is that it's called speed density if the pressure sensor is downstream of the throttle (between the throttle and the port) and it's called Alpha-N if the pressure sensor is upstream of the throttle (between the filter and the throttle). Not a great difference IMO.

This brings me to a product idea for 928. Why not build a small hardware box that implements both Alpha-N and speed-density for LH and EZK? JDS's Alpha-N already has a box. By box I mean a general MAF load emulator. The inputs to the box are throttle position sensor, MAP sensor, and temperature sensor. The output from the box is emulated MAF load signal. Inside the box is a very small logic piece that amounts to one computation and one user configurable table. The one computation is the computation of density from pressure and temperature, a simple function to implement. The one user controlled table maps the (computed) density and throttle position into the emulated load signal, interpolating between the cells as usual. LH wouldn't be any wiser to the fact that it's not listening to a MAF sensor load signal but instead to an emulated signal. The only thing inside the LH that would need to be changed is undoing the engine speed division in the mapping from MAF voltage to load signal, as the emulator output is already in the per-cycle units instead of per time units.

The single table could be used to implement either Alpha-N or speed-density system. For a pure speed density, the throttle position would be ignored and the table would have the same values regardless of the throttle position. For a pure Alpha-N, the density dimension would only implement a linear density correction.

For a hybrid system, which would be the best, the table could be simply empirically calibrated for the initial calibration by driving the car in normal use with both the emulator box hooked up and the MAF hooked up, recording the emulator box inputs and the MAF signal load from LH. The initial calibration for the emulator box would then simply be the table that minimizes the mean squared prediction error between the actual MAF load signal and the emulated MAF load signal. One could even calibrate a different initial table for installations that have the MAP sensor upstream of the throttle and those that have the MAP sensor downstream of the throttle.

Any thoughts?

 

With a turbo, throttle position by itself isn't enough to give the required "load" signal due to the fact that manifold pressure can be a variable for a particular rpm/throttle pot voltage ? Are you assuming a "lagless" system ?

 

Throttle position itself isn't enough to give the required "load" signal even in a normally aspirated motor. One has to have a "barometric pressure compensation" which in practice means a pressure sensor and a temperature sensor which are used to perform a density computation that adjusts the throttle position signal. This way, on a hot day in the mountains the load signal is lower for the same throttle position that it is on a cold day at the sea level.

You of course know this, your Alpha-N manual reads: "The system operates as an Alpha-N fueling system. This uses a precision throttle position potentiometer giving a measurement of engine load, in combination with air temperature and atmospheric pressure sensors for the necessary air mass corrections."

Now, let me ask you this question: What is the difference between pressurized air due to elevation (or lack thereof) and pressurized air due to a turbocharger? The answer is that there is no difference, other than the magnitude of the pressure variations. Consequently, one should be able to use the Alpha-N density correction logic to deal with low levels of boost pressure. Just hook up the pressure and temperature sensors to pressurized part of the intake tract upstream of the throttle.

Note that I am not assuming a lagless system anywhere here. The density correction in Alpha-N is agnostic to the causes of the pressure upstream of the throttle.

 

I didn't say tuning was easy. But we can learn - we are human. I said the user interface is user friendly. Meaning it's learn able. I have seen them and interacted with them.

 

Hi John,

I agree with Tuomo, in that the hybrid A-N/Speed Density system would be best as opposed to a pure A-N or pure SD setup.

Let me know what is ready and when...

Also don't mean to hijack the thread...!!!

 

SPAM^

 

The first indications are that the stiffer wastegate spring allows the car to build somewhat higher boost. Testing over 18 psi at 6000 rpm on the highway is obviously limited, so the next dyno session will tell the full story.

We'll also be switching from the expensive Dynapack that is far away to a nearby Dynojet. The closest Dynapack that can hold both the rpm and the torque that we may generate is far away. This is a cost control measure, as the load holding Dynapack is really only necessary for doing large fuel and ignition mapping project. The bulk of the fuel and ignition maps are basically in a good shape now, and all we're doing is working with the boost control and mapping fuel and ignition for those few virgin LH/EZK cells where no man has gone before. As a by product of using the Dynojet for the next couple of tests, we'll get a second opinion on where we are in terms of the level of power, although so far tuning and not the accurate measurement of power level has been the focus.

 

More testing and more bottle necks. Still can't keep the boost above 18 psi at the top end. Tracking down the source of the problem from the mass air flow, boost pressure, and turbocharger speed data. The main question whether it's a physical limitation/problem or a logical control problem. The car did put down 750 dynojet ft-lbs of torque in the mid range rpm during the debugging, but that's with too many knocks and therefore it doesn't "count".

 

The SMAf won't be the bottleneck, it is capable of much more power than you are making at present. ..
5.5v out at about 1000HP

 

I didn't mean that the super duper MAF would be a bottle neck.

What I mean is that the MAF sensor, manifold MAP sensor, and turbocharger speed sensor can be used to overidentify the location on the compressor map. With the location on the map overidentified, it's possible to see inconsistencies that will then point towards specific problems. For example, the turbocharges speed sensor data puts us on a specific curve on the compressor map. If that curve implies a combination of higher boost and higher flow than what we observe from the MAP and MAF sensors, then we may have a boost leak between the MAF sensor and the compressor outlet.

 

OK ! Eagerly watching for further reports....