worf928

Yes it does. 20 n-m without a washer.

I have removed knock sensors torqued up to 200 n-m.

Back to the question: would a seriously over-torqued knock sensor still function correctly? I think not - but would like someone who knows to chime in.

If the sensors will not work correctly if over-torqued then I think it very wise to check knock sensor bolt torque before assuming that they function without other positive evidence - such as your soon-to-be-born tool - before trusting them to detect knock in new engine configurations - such as a supercharger.

Jim bailey - 928 International

I do recall reading that over torque is BAD for Piezo units !!

Z

I don't have any equipment to change the mapping (yet), but I can easily change the ignition timing to the point that the EZK timing retard would start to kick in. Just a twist of the **** one way to advance it, and a twist the other way to retard it.


Supposedly the torque can make a difference. Too loose and the ability to pick up the knock/sound/vibrations would be reduced. I suppose another possibility might be that over-torqueing might damage the knock sensor in some way.

Here's a signal from a knock sensor. This is NOT specifically a 928 knock sensor, or for that matter may not even be a Bosch one, so may not be a very good representation of what a 928 knock sensor signal would look like. I'd guess that the shape of the actual signal from a 928 knock sensor might be close to the same, but from what John said the voltage spike is much lower on the 928 one.

blau928

Over Torque = Bad for Piezo sensor...

Klaus has it right..

Hot spots in the chamber can cause Det.. So can incorrect heat range spark plugs when the engine is boosted..

Temp dependent things like H2O temp, and oil Temp play a factor, but so can a lot of other things.. If you hear pinging, you are late, it's been happening for many cycles...(hmmm, 3000rpm = 50 rev/sec = ? pings per combustion event = LIFT FOOTwhen you hear pinging...!)

Get the timing, spark, and air charge temp & density right, and the pinging will be non existent.. One thing that the SC folks have not mentioned too much, is the effect of density or the density ratio of their setups.... Pressure spikes from the detonation causes the ring lands to crumble, not the heat itself...... Heat is the cause of the detonation starting..... Pressure at specific points of the crankshaft rotation is bad for the piston......

As far as the code in the EZK, I would be willing to ask my engineering guys in india to see if they would be willing to entertain a debugging/decoding session for a nominal fee... This is not difficult, and if John is willing to share his understanding of how it works, I'm sure my guys can crack the stuff..

John, if you want me to introduce you to my friends, i can arrange it.. Director of Software Engineering at an IIT campus, with numerous projects in all sorts of stuff... Real time constraint programming over distributed networks, and some quantum computing advanced research as well..

Let me know...

Cheers,

Z

It's actually a combination of the two. Normally there's an unburned boundary layer on the surface of the combustion chamber and piston top. That boundary layer acts as an insulating layer keeping heat out of the piston and heads. The sudden violent action of the detonation blasts away the boundary layer, allowing the heat to get to the normally more protected surfaces, and being transfered to them more. The resulting increased heat in the pistons makes the aluminum softer. That softening combined with the increased pressure is what causes the piston to break. Besides being stronger than cast pistons, forged pistons are also denser, making them more thermally conductive. They are able to transfer that increased heat away from the top of the piston faster than cast pistons. Getting that heat out of the piston to help keep the aluminum from getting weaker is why oil squirters that spray the bottoms of the pistons are used on some engines. The purpose of sodium filled valves is the same. To get the heat out of the valve head faster. The increased heat transfer also goes into the head when detonation occurs for the same reason. Usually the piston is the weaker link and therefore the first to give way.

Lorenfb

No Guessing, Please!

As I said in one of my above posts, you just need to read (the manuals or the circuit
boards) to determine how the 928 LH & EZK work and what technology is used. With
all the discussion on the subject, I'm sure that the following has been disclosed:

1. The LH & EZK use the basic same uP technology as 964 DME;
a. Intel 8051 (8 bit N-Channel)
b. 256K EPROM (32Kx8)
c. a standard off-the-shelf A/D converter (ADC0808)
d. CMOS static RAM (volatile) for the adaptation variables

2. The LH & EZK each have 8051s whereas the 964 has 2 8051s in the DME unit.

3. Both the 928S4 and the 964 have the same knock sensor circuitry;
an analog multiplexer feeding a custom A/D converter with a serial output
feeding the 8051. The 964 deals with knocking in EXACTLY the same manner
as does the EZK, i.e. progressive 3 degree retards on EACH cylinder. Since the 911
3.2 DME also uses an 8051 to process both fuel & ignition data, the 8051 in the
EZK is more than capable of processing just ignition, i.e. additionally digitally process
the knock sensor data.

Calculations for EZK:

Givens -
a. 8051 @ 5MHZ (.2us per clock cycle) & 10 clocks worst case per Intel instruction
or 2us per instruction
b. EZK worst case output - spark every 2.5ms @ 6000 RPMs

Then -
The 8051 can process about 1000 instructions between sparks to;
a. calculate RPM
b. use load data from LH unit
c. monitor engine temp
d. determine knock status
e. determine crank position
f. lookup timing from ignition maps


4. The standard A/D converter handles;
a. the temp sensor
b. the idle/WOT switches
c. misc. inputs

5. A separate processor handles the diagnostics and "K" & "L" interface on both the 928 & 964
and not the 8051s, e.g. 964 uses a 68705 uP.

6. The 928 lacks sequentual injection that the 964 has. The 964 runs as does the 928
(all injectors fire @ once - bank fuel injection) when the CMP (cam sensor) fails or
is disconnected. Also, as mentioned the per cylinder retard no longer functions without this sensor.

Side Note: The EPROM code can easily be disassembled using the old DEBUG command in DOS.
The output is in HEX and can be converted to the 8051 instructions, thereby determining the
LH & EZK firmware.

John Speake

Z - useful box of tricks ! So you could advance until you achieved knocks, but if several cylinders start to knock at once, then the results may be difficult to determine ?

Loren, #1-4 still don't give any confirmation of individual cylinder ignition retard on the 928.

#5 is incorrect for the 928 LH

John Speake

Z - I have a Pico logger now, so I could make a better measurement of the 928 knock sensor output voltages. I am sure that over torqueing the knock sensors will reduce the output voltage.

The Tech info for '87 has an interesting piece of info -

The front knock sensor senses cylinders 1,2,5,6 and the rear 3,4,7,8.

We know that the igntion is grouped 1,7,6,4 and 5,8,3,2. The fuel injectors are also grouped (via the igntion protection relay when fitted) the same as the ignitiion.

I agree with Z that the tech info does appear to imply that individual cylinder retard is used, but it would be interesting to prove it.

Blau - the EZK BIN files are readily available if you have someone who feels like some homework over Christmas :-)

docmirror

What, no guessing? Are you the guessing police?

I though I did pretty well, considering.
Anyway, your experience with this particular ignition controls is deeper than most. I've only just bought mine 3 months ago, and (knock wood) my ignition works flawlessly right now. However, even the non-guess method is a bit flawed. Your calculations for the sparks per sec needs to be massaged. I made a similar mistake, but caught it later.

6600RPM = 110RPS = 1 rev @ .0091S; .002272727...

The EZK also handles things like MAP, idle/full, temp, etc. Instruction branching can also have a significant effect on instruction count. I've been thinking of cracking the PROMs on these, but there was a rather spactacular negative post on the evils of IP theft. I guess it's taboo or something.....

Lorenfb

"Your calculations for the sparks per sec needs to be massaged."

What?

6600RPM = 110RPS = 1 rev @ .0091S; .002272727...

That's the same as 2.5ms @ 6000

6000/6600 x 2.5 = 2.27ms

"Loren, #1-4 still don't give any confirmation of individual cylinder ignition retard on the 928."

The Porsche 928 manual and the EZK circuitry both confirm the same.
Open the EZK unit and check it out.

"#5 is incorrect for the 928 LH"

The 8051 doesn't directly interface with the "K" & "L" lines in either the LH or EZK units.

John Speake

"The 8051 doesn't directly interface with the "K" & "L" lines in either the LH or EZK units."
>>>>>>>>>>
True
>>>>>>>>>>>


"A separate processor handles the diagnostics and "K" & "L" interface on both the 928 & 964"
and not the 8051s, e.g. 964 uses a 68705 uP.
>>>>>>>>>>
Not true

Z

If several cylinders did start to all knock and receive timing retard at the same time, it might make it a little harder. As long as all of those cylinders weren't receiving the ignition signal from the same ignition control module, it would confirm individual cylinder timing control. If any of those cylinders were receiving different amounts of retard than any of the others it would also confirm it.

I'd think it would be pretty unlikely for just the cylinders from one ignition control module to all start knocking at exactly the same time, and to all require the same amount of timing retard, but it is a possibility. If that were to occur, I'd have to make sure that one of those cylinders either knocked sooner or later than the others. Maybe just pull off one spark plug wire so that there's no way that cylinder could be knocking, but I'd still be able to measure the ignition timing for that plug wire to see if it was also being retarded with the others where knocking was occurring.

Lorenfb

So what's the issue?

1. the 928 manual indicates such
2. the technology used & signals available make it possible
3. the actual circuitry indicates such
4. the same technology is used in the 964 DME which has the same writeup in the 964 manual
5. It's the logical implementation to do, i.e. not to retard banks of cylinders but only the appropriate one

This should not be an issue 15+ years later. I'm surpised no one raised the issue before now.

Z

That's what I'd think too. But then on the other hand, look at how many professional mechanics still think that the 928 timing belt is automatically tensioned.

Jim bailey - 928 International

Perhaps what is says is how well the ignition system works on the 928 so little need to fix or examine what is not broken.