predictor algorithms

[zguy36]

What is the difference between the ignition predictor algorithms? I couldn't really see much info on the advantages of each. Trying to diagnos a stumble at high rpm, I switched from 'last interval' to 'first derivative, second derivative'. When I made the switch, the engine has detonation problems that it didn't have before. Ie detonation at 10psi compared to 20psi. I don't have a timing light readily available, but seem to think that the timing has probably been advanced due to this switch, since detonatoin is such a problem now.

[zguy36]

So, it seems as though this post is a lame duck. I'll answer my own question, at least for what I wanted to know. I stuck a timing light on the car and ran it up through the rpm. To make it easy, I set all entries in the table to 10 degrees advance. All the settings seemed to work pretty well, except for (1st derivative high, second derivative low). This one made the tach jump around and it ran like "crap". The others seem to retard the timing about two degrees at 6000 rpm, not really a big deal.

The big deal however, is that when switching between the different predictors, the base timing changed as much as 20 degrees. I had no idea that this happened. When I first got the squirt up and running, I had it set to the default and based the timing there. It ran like "crap" so I switched it to 'last interval'. This made my timing 13 degrees retarded from the original setting. I didn't realize this, and you can imagine how happy I am not that I found it. Oh so much more power! (I had been driving it like that for a month!) Anyway, watch out for this so no one else gets burnt.

[daniel505gti]

I have had similar issues. That is interesting information to know.

For me Last Interval was the only stable setting as everything else was to erratic.
I am going to move the trigger to the flywheel rather than the distributor just to make sure there is no play in the drive.

[Mike_Robert]

Zguy36, Please note in your post if you are using MS-II/MS-I/MSnS-E, the code version, MegaTune version, as well as V1.01, V2.2 or V3 main board, etc. What engine/vehicle is this? How are you doing ign triggering? I'm goint to go out on a limb here and guess it's a Z

It's not so much a lame duck as much as info is still being collected and looked at ref prediction algorithms. It's great that you have posted your observations and I hope you continue to do so. The first and second derivative algorithms load the code down more due to more math being done at every tach interrupt. The question remains, "how much is too much?". This varies broadly depending on platform variables. For example, A V8 does more ign calcs than a 4 cylinder, a 2 stroke more than a 4, etc. Depending on where the trigger is, the code can operate in either this-cycle or next-cycle mode. In a rapidly accelerating engine with a light flywheel, errors in timing can occur if the prediction routines used aren't optimal for your install.

[jsmcortina]

Mathematical theory says that during rapid accel at low revs you can get a very large timing error with "last interval" and that "2nd deriv" will be needed to get any accuracy.
The fact that you are observing the opposite is very strange but interesting and I'd like to know your engine, triggering method, code, MSQ and a datalog.

What happens to the timing when you blip the throttle?

Going back to 2003 ish, the author of "MegaSpark" observed large timing fluctuations when blipping the throttle until he added 1st deriv calcs. The maths for those calcs is the same principle in MS1/Extra and MS2. The 2nd deriv takes it a step further.

Anyone else with physical data to report on the different algorithms, please post.

James

[m0ntecarloss]

Did you happen to catch this thread?

http://www.msefi.com/viewtopic.php?p=10 ... tor#106411

I'd like to experiment with this whenever I get a chance to play with MS2 again

[jsmcortina]

There has also been discussion on the beta list amongst other code stuff.

However, this topic seems as good a place as any as two posters have reported actual results - far better than the scientific speculation the rest of us have offered!

James

[zguy36]

Thanks to whomever moved this post to a section where it would get some more traffic. This post sat for a few days in the MSII forum with no hits. Here's what I am running:

'83 nissan 2.8l turbo
optical distributor
MSII v3 board, v2.3firmware

The only tests that I did involved slow changes in rpm, but at a very high rpm. I am experiencing a slight drop in power around 5000rpm for a range of 200 rpm, then again at 6000rpm again for a range of 200 rpm. I thought this might be due to a timing fluctuation since the AFR stays steady. The derivative algorithms haven't given me very good drivabilty and make the tach jump erratically at idle.

[jsmcortina]

Please post your settings (file SaveAs in Megatune) and do a short datalog (ALT-L) to illustrate the jumpy rpm.

James

[Philip Lochner]

When I first installed my MS-2, I installed it ONLY for ignition.
 
I have a VR pickup (Fixed centr+vac dissy @ 6deg BTDC), going into MS-2 (no IGBT), going to TP100 going to std coil.
 
With "1st HighRPM, 2nd low" being the default, I simply could not get a steady timing position with my timing light @ idle.  It was jumping so much that it was difficult to know what the timing actually was, much like a dissy with points with a badly worn shaft. This led to idle just not being nice an smooth but just a little rough. The higher the revs, the more stable the timing became but still jumping around. 
 
Eventually I thought to give the other algorithms a try and found that Last interval offered stable timing at idle and gave the required timing at higher (but stable) revs as per the ignition map.  I have not yet bothered trying the other algorithms - the urge to get to where I'm now (full control of the engine) was just too great.
 
I'm pretty sure I posted on this finding at the time.  Since then I've seen others arriving at the same conclusion - that Last interval seems to be the only "useable" algorithm.
 
I have read that timing systems using trigger wheels offer better timing during acceleration having more timing intervals to calculate with.
 
Kind regards
Philip
MS-2/V3/2.35/MT 2.25/Innovate LC-1/LSU4.2
Land Rover 4.6 std, 4-2-1 exhaust headers, IAC, MS-2 controlled
'80 Jaguar XJS V12 5.3 (10:1 CR) is waiting for its turn 


Posted by email.