Board logo

Alpha-n vs Speed-Density
flak monkey - 8/10/08 at 10:33 AM

Ok, so I am running TB's and long duration cam which points to the fact i will get a poor vacuum at idle so Speed-Density looks like it isnt going to work too well so I need to go for the Alpha-N approach.

Are there any disadvantages to this over the speed density? I am trying to understand why everyone doesnt just use Alpha-N in the first place....

Got my ECU talking to the laptop last night and updated the firmware sucessfully. Also found out that I needed to change some jumpers and a couple of bits on the board to trigger off the coil rather than a VR.

David

paulf - 8/10/08 at 03:02 PM

I am running speed density on my zetec and also did on the crossflow before that with no problems.It is a bit dependent on the size of the throttle bodys as if you go to large the resolution dissapears .I have38mm GSXR bodies and have all the vacuum takeoffs teed together then connected to a small resevoir via a .6mm mig nozzle to act as a restrictor to damp the pulses.This works well and I have 40kpa at idle so have a range of roughly 60kpa in which to tune. I find light throttle cruising is about 60 to70 kpa and overrun is 25 kpa. I get 30 mpg on a run and good performance but 20 or less if driven hard.
I tried Alpha -N but did not get on with it as well as S D.
Paul.

MikeRJ - 8/10/08 at 05:28 PM

Alpha N calulates engine load from speed and throttle opening. Since engine load changes a lot in the first few degrees of throttle opening, it can be problematic to get enough resolution and repeatability in this area. It also won't compensate for any changes in volumetric efficiency or air leaks, even something as simple as exhaust back pressure changing, or a dirty air filter could cause driveability issues.

Speed-density is a far superior solution for a road car if you can get a stable enough vacuum signal, but with big cams alpha-n is sometimes the only usable solution.

One possibility (which I believe MS supports) is a hybrid approach that uses alpha-n at idle so that the wildly fluctuating vacuum signal don't cause problems and then switches to speed-density to give better driveability.

flak monkey - 8/10/08 at 07:00 PM

I am running GSXR 600 TBs (38mm) on a pinto with FR33 cam fitted.

I would like to use speed density, as it sounds like its the best of the two. But what are the chances of it working?

MikeRJ - 8/10/08 at 10:14 PM

Suck it and see to be honest, but you have reasonably sized throttle bodies and it's not a ridiculously hairy cam so I'd say you have a fighting chance.

BenTyreman - 8/10/08 at 11:35 PM

The vacuum pulled at idle is not nearly as important as the range of vacuum available between light cruising and WOT. I have a MAP of 60 kPa at idle, but as soon as you start rolling the lowest MAP available is 90 kPa. This only gives a range of 10 kPa to cover every condition except idle and over-run.

After spending weeks attempting to make SD work, switching to Alpha-N made an immediate difference.

I got it mostly tuned on SD but always suffered from leaning out on throttle re-application after gear changes as well as difficulty blipping the throttle. 100 miles after switching to Alpha-N and the car is in the best tune it has ever been.

With regards to insufficient resolution of throttle on MS1, the ADC is 8 bits, giving a theoretical range of 0 to 255. Allowing for even a moderate realistic range of 200 steps, that's still 0.5% increments. More than enough resolution even for low throttle openings.

MS1 has a minimum of 8 load sites. At least 5 of these can be used to cover 0% to 25%, as the VE table tends to be quite linear for large openings.

MikeRJ - 11/10/08 at 09:08 PM

quote:
Originally posted by BenTyreman
With regards to insufficient resolution of throttle on MS1, the ADC is 8 bits, giving a theoretical range of 0 to 255. Allowing for even a moderate realistic range of 200 steps, that's still 0.5% increments. More than enough resolution even for low throttle openings.


The ADC resolution is not really the issue, as you say 8 bits is plenty. The problem is the relative lack of load sites, though the fact they they can be moved is of great benefit with the MS.

I've helped to map my brothers Elise using an old-ish Emerald system with fixed load sites, and getting good progression from idle has proved almost impossible since so much happens between a single pair of load sites. Biasing the throttle pot with an external resistor to give a shallower slope at low throttle openings is a possibility we have yet to explore.

BenTyreman - 11/10/08 at 10:17 PM

I'm running MS2Extra code with 16 load sites and most of them are redundant. I have half the load sites dedicated to below 20%. I suspect that you only need the one at 100% to complete the curve, the response is so linear.

It's a relatively simple task to plot a Bezier curve to the known points and examine the % error as the number of load sites is reduced to 12 or 8. My original idea was to work out the optimum placement of load and RPM bins automatically, but it seems to make so little point on MS2 that I never got round to it.