mspiegle

not as a result of this project, but i'm sure once these things are out of the way, work on the 85/86 can get moving

Louie928

I probably would have stayed with the LH/EZ-K. However, a good aftermarket system, well tuned, will probably outperform the LH/EZ-K combo. There are two aspects of the LH/EZ-K that caused me to move away. Well, at the time 3 aspects because tuning them wasn't an option. One was the batch fire injection. The ever present wiggle in the power curve above 5500 always bothered me and I had thought it was because of the intermittent heavy demand on the fuel system and pressure pulsing, due to batch fire. Marc Thomas' Motec driven engine didn't show that instability above 5500 and I thought it was due to the sequential injection. Now, I think I was wrong about that. Even with batch fire at 5500, the injectors are open for 80% or so of the time and I doubt if pressure pulsing makes a huge difference, but you still may need the pressure dampeners to keep pressure stable at the lower RPMs which teh sequential system doesn't seem to need. I've found that the more likely reason is that the LH processor can't keep up at high RPM so some extra variance in injector pulse width creeps in. The LH has an ingenious method to deal with too high duty cycle. It switches to fire the injectors every other revolution rather than every revolution thereby eliminating one open/close period from the total. John's LH tuner permits changing the swichover RPM to control duty cycle on custom size injectors. You can also do this at the low power end where a large injector may have a too short open period to be accurate. AFAIK, none of the aftermarket ECUs will do this. Maybe Motec, but I can't afford that.

The other thing was that I wanted to eliminate the distributors and use either coil on plug or wasted spark ignition. The EZ-K can't do that. I believe have figured out a way to separate the two EZ-K outputs into four. Each output would drive one stock 928 coil amp and those would drive 4 wasted spark coils. I haven't done it, but think it's possible.

The big thing in favor of keeping the LH and EZ-K is that you can start with some map that will probably run the engine. Change the map for your needs. John now has in his latest LH software, options to change the throttle tip-in enrichment in both amount and time and a bunch of other little goodies that give just as much flexibility as most aftermarket ECUs. The amount of initial setup that has to be done on an aftermarket ECU is daunting.

Another thing is that I wanted to eliminate the MAF and go to a manifold pressure sensor because of the limiting out factor in the standard MAF. I'm not so sure that is necessary anymore. Having to route your air through a MAF makes the air plumbing job a little harder, but a MAF also automatically compensates for air density changes which can only be done by mathematical formula approximation based on air temperature/pressure when using the MAP sensor. John's SuperMAF takes care of the stock MAF limiting out problem and Dave Roberts figured out how to do something similar for his SC setup. I think if a person was still bent on using a MAP rather than MAF, the LH maps could be edited using the ST to work with MAP voltage rather than MAF voltage. You'd have to come up with some method of altering the MAP voltage according to intake air temperature because the MAF driven LH doesn't have that input. Not much point in using the MAP I think.

Lot's of new developments in the 928 world and things change.

BC

I think most Aftermarket ECUs deal with MAP/IAT ratios pretty well. Add in baro correction for outside air, and you won't be dissapointed.

mspiegle

I knew the DIS was an important point for you, but I actually kind of expected you to say that you would have used the DTA anyways. That definately says something about the quality of tuning tools we have now.

Regarding the sequential injection, it seems that above a certain RPM (say 2500-3000), the valves are opening faster than an injector can inject all of the fuel for that cycle (causing the injectors to be open longer in order to maintain proper AFR). If you move a little farther up in the RPMs, I think it gets to a point where sequential injection doesn't offer any more precision than batch injection. The other point you mentioned however of the LH possibly not being able to handle higher engine speeds is very interesting. Didn't the LH-tuner provide data on pulse widths? Did you see fluctuations?

Bill Ball

Louie:

I saw the Hall sensor mentioned in a Porsche write-up of the EZK knock sense retardation. I'll relocate it later tonight. I think the EZK makes a calculation from the Hall sensor signal timing and RPM as to which cylinder is knocking, assuming knock happens at a certain time range before TDC.

Louie928

Well, I may have used something like the DTA, but now I can't think of why other than at teh extreme top end the LH sort of runs out of stream.

You are righ tabout no noticeable performance improvement with sequential injection. I wanted sequential because I think I can get rid of the fuel pressure dampeners. That's a copuple less things to fail. Sequential does improve idle and low speed plus reduce emissions. I wanted as good low end as I could get with the cams I have so I thought sequential may be a benefit. Yes, injector pulse widths are available. I didn't choose to log them though. They do vary a little, but that's the result of the variation in RPM and load cells that are being chosen due to (I'm told) the processor not keeeping up. The load cells change by two or sometimes three when RPM doesn't change that much and you know the air mass isn't changing that much either.

Louie928

Yes, I think they do and you can get teh mixture pretty close within a wide range of temperature/pressures with the proper compensation. My point was that the MAF does all this automatically. I'm not sure of the reliability factor MAF vs MAP. I suppose the MAP is more reliable

Bill Ball

For what it is worth re: Hall sensor and knock sensor operation.

Porsche Service Training P210005 Fule & Ignition System 928 S4 928GT
"To enable the control unit to detect the cylinder affected by the knock condition, a Hall sensor
mounted behind the camshaft gear of cylinder bank 1 - 4 attributes the corresponding cylinders:
Should the knock sensor or the Hall sensor become defective, ignition timing of all cylinders is retarded
by 6" for safety reasons."

John Speake

>>>>>>>>>>>>>>>>
Yes, quite correct Bill, and this has been shown to be the case by analysing the code, plus bench testing.

All these aspects are covered in the extended version of the ST User Manual, which if DR agrees, could be available on his website for download.

Yes, I am starting preparations for a LH 2.2 project. (for 85-86 USA 32v and also 84-86 Euro S2) That is to say I am constructing test jigs for the LH and the EZk. This is so I can start offering a repair service for these units. The test jigs are also an essential tool for analysing the code in the ECUs.

This potentially could lead to a SharkTuner for these earlier cars, but as this is a very substantial committment for both Niklas and myself, a final descision on this has yet to be made.

I have just worked out the "profit" margin on the ST project so far, and it just as well we enjoyed the challenge and satisfaction of producing something worthwhile for the 928 community.

This project would have bankrupted any commercial company :-)

John Speake

Louie said :-
" It's possible that sampling at 10 per sec missed some knocks. I didn't think of that at the time. I don't know whether there is an internal buffer to hold knock count until the next sample cycle takes place or if a knock happens between samples, it's lost. Possibly a much faster sample rate should be used. Something to check next time. John has done a huge amount of bench testing so he undoubtedly knows how this works."
>>>>>>>>>>>>>>>
THis is some comments from Niklas -

The answer is that knocks are never lost. Even if you sample at a very low
rate, say 1 sample per second, it will always count the number of knocks
that occur between samples so you will never lose any knocks. This was an
important design decision I took - I thought it was important to make sure
the knock counting is accurate.

He also wondered why the ignition curve looked jagged. This is because
there are knocks and you can see the retard on the cylinder(s) that
knocked. If you have no knock at all, the curve will be quite smooth."
>>>>>>>>>>>>

Louie is correct in responding to Bill's enquiry that the Hall sensor is used to determine which cylinder is knocking, and also that if the Hall signal fails, then everything is retarded 6 degrees (at high loads). This is the reason that the fault indicators are shown on the monitor of the ST. It is impossible to remap the igntion correctly if there are either Hall or knock sensor faults. If a knock sensor fails, then the same 6 degree reatrd also applies.

Vilhuer

No question about that considering fuel prices we are paying and how many miles Niklas has spend on the road developing and debuging the system. Not to mention wear on his S4.

John Speake

......and all those miles on my test jig, but at least that doesn't use any petrol :-)

John Veninger

Lorenfb

"For one, the S4, with a good MAF, is close to optimum already." - Louie928 -

Right on! As was said perviously, the results question the value of using "performance"
chips, which should be the case for ALL engines without major mods that use a
knock sensor closed-loop feedback ignition system, especially engines designed by
Porsche/Bosch for max performance.

Good effort!!!!!!!!!!!

Lorenfb

John deserves a lot credit for this device and the other other tuner (LH AFR optimizer).
Not only for their utility for tuning but also as an educational/learning device for many.

His clone of the 928 Hammer version (Spanner) has also been well received. I wish, though,
he would incorporate the Rennlist "posted" 964/993 diagnostic software, as those
guys are clamoring for a tester & would surely buy it versus having to make one
themselves. Hopefully, there's no copyright problem with Porsche.