Knock Sensor w. 3.3 CIS
 

Does anyone use knock sensors on a 965 3.3 turbo engine?

I am using the Clewett Engineering XDi2 dual plug crankfire kit, which has knock sensor retard capability. However, there seems to be a consensus that the 3.3 CIS engine is too noisy to effectively use knock sensors.

I would greatly appreciate any input.

Cheers.

You might want to read through this thread:

http://forums.pelicanparts.com/911-930-turbo-super-charging-forum/405586-knock-sensor.html

Mines not a 965 but it is an aircooled, boosted motor. I use a Bosche adapted from a 993 bridge across 4-5-6 cylinders, feeding my Adaptronic ECU.

How long have you had it and what is your experience with it?

Do you feel that it falsely retards due to noise that is not related to knock?

I have actually posted this thread after reading the one you are referring to.

Unfortunately, the other thread seems to be mainly based on hearsay rather than on factual experience. I would like to get feedback from people who either have or had it installed on their engine.

Cheers.

I posted a quote from Steve Weiner of Rennsport in that thread. Speak to the source if you consider it hearsay:

Steve Weiner
Rennsport Systems
Portland, Oregon
503.244.0990
E-mail: porsche@rennsportsystems.com

didn't he state that he uses it for tuning purposes only, but advises against keeping it on the engine?

I already have a system that can detect knock, but I don't have the sensors installed as there seems to be some controversy about the effectiveness of using them on the 3.3 engine.

This post makes it sound like closed loop knock control:

http://forums.pelicanparts.com/911-engine-rebuilding-forum/229355-thoughts-combo-please.html#post1998268

I've had it on about 6 months, including ~15 hours on the dyno altogether. I've found power falls off before I get knocking so how we have the timing and fuel mapped it doesnt interfere very often. I originally installed it because I didnt know thats how this motor was going to behave.

Thanks for the link. I will contact Steve and get his opinion whether this would also work on a 3.3

Cheers.

Knowing what you know now, would you install it again?

Definately. I'm upping the boost to 15psi now.

Excellent. Thanks for your input.

Les Garten has a Tec III, but I don't know if he has them enabled.

I contacted Steve re knock sensor and his take is that it is not feasible on a CIS motor. Oh well...

I know for a fact that Raymo also runs one in his CIS motor. He tapped the breather cover and installed it there. He used it for tuning, I've seen his logs and we talked about it this morning.

I would like to hear more about his setup.

I was told that the kock sensor would require either MAP or MAF along with throttle position, and RPM to be truly effective.

Did he say he tried it on a CIS engine? I'm guessing he said it wouldn't work because that's a mechanical system with no sensors to read.

The SafeGuard has a processor and onboard MAP sensor, so it reads load and RPM.

Decel or light loads can cause false knock, so software ignores the knock sensor if the onboard MAP sensor sees more than five inches of vacuum.

Also, the program has a detection window to listen for knock at the appropriate time.

I'm pretty sure Steve wasn't aware of these points.

There's a lot more I would like to say about the detection algorithm, but it's proprietary. Unfortunately, saying nothing gives the impression there is nothing to brag about.

Revealing details also has drawbacks. One, it gives the competition a starting point. Second, it gives the armchair detractors an opportunity to say "Why did you do it that way, here's how I would have done it..."

It's already happened to me on another forum. The original forum where I posted the information totally ignored it, so I gave away secrets for nothing, then got hit over the head with them later by someone that posts links to knock papers, hoping someone will design what he wants.

AIEEEE! (Marvel Comics scream).

NASIOC - View Single Post - The Knock Box

John,

are you saying that it does work with CIS?

I honestly do not understand too much about this topic and how the Safeguard works. I am using CIS with a XDi2 ignition system that has knock sensing capability. Can I use the Safeguard in combination with the XDi2 and how would it work w/o throttle position, RPM, and MAP/MAF inputs?

SafeGuard reads RPM and has an onboard MAP sensor to read load.

J&S Eight Channel Interceptor

I would have to check with Electromotive to see if it's compatible with their DFU.

If not, our Vampire system will work, but it requires MSD 8870 coil spacers. These install under the coils and allow access to the coil negative signal, required by the Vampire.

J&S Vampire

Ray replaced the lambda box with a GM ECU he cracked. It retards the signal to his MSD box. The motor bottom-end is entirely stock and the stock sensors, throttle body are all retained.

You do not need MAP or throttle, just ignition control. After the ECU performs the filtering/amplification and any other adjustments you desire it compares the "knock value" against a pre-determined value or algorithm and modifies the timing appropriately for a number of cycles or time period.

A basic system can simply compare

the knock sensor output, if it exceeds X then it pulls 10 degrees from ignition timing for 1 second, then goes back into "listening" mode.

Professionals who focus on this aspect of engine development can and have developed some very effective and sophisticated systems. I have never done business with him but John sounds like he's put some quality time and thought into it.

If your Ignition system has Knock control, (can retard timing based upon parameters set by user) then it comes down to fitting a sensor and some sort of signal conditioner.

This is as important as the ECU control. Frequency type, noise level and the ability to distingush between mechanical noise and knock is going to be your issues.

Call Performance Developments 949 646 7461. Ask Neil about the new Link G4 Knock block. They are also very knowledgeable about fitting sensors to the Porsche engine. They have done this many times over and its works really well.

If your ECU will accept an input from the Knock Block, then your set.

Yes, I am still trying to figure out, if this can be done on a 3.3.

BTW, I just read the manual for the XDi2 and it seems that I do have a MAP sensor (I am certain that I do since it is listed on my invoice!!!) and that the device gets a tach signal. So, I have MAP and RPM, but I am still missing throttle position. Hopefully, these two inputs along with knock sensors are enough to do the trick...

I will give Neil a call and get his input.

Cheers.

Pinsko:

The knock sensor connects directly to the XDi-2, so it has internal signal conditioning.

Don't know how well the detector works.

Please tell me more about the 'how well it works' issue.

Is it possible to connect your system to the XDi2 and would it do a better job at detecting knock and eliminating false knock detection?

Well, I didn't want to point this out (scroll to the knock sensor section):

Engine Sensors and Tools - Electromotive Engine Controls

Number one, I didn't want to bash another product, and two, the statement may lead people to think knock sensors in general are a waste of time.

The Interceptor could be installed between the XDi-2 box and the Direct Fire Unit (DFU). The Interceptor retards by extending the dwell time, so I would have to check with Electromotive to see if the DFU can handle more dwell.

The Vampire could also be used, but it requires MSD 8870 coil interface modules (spacers) under the coils. These have wires that allow access to the coil+ and coil- signals. The Vampire control wires would then connect to the coil negative signals.

The XDi-2 would continue to do most of the work, and the Vampire kicks in only during the retard event. The Vampire can even be disconnected while the engine is running, and it will continue to run.

John,

can you point out in layman's term how and what your system does better?

Cheers.

The two features I mentioned above (knock window and load qualifier) are easily implemented with a micro processor, but did Electromotive include them?

There are a few other features of my algorithm that are more important than those two, but I won't go into it for reasons stated in an earlier post.

I ran across a description of the Bosch system last year and was surprised at the similarity. I conceived and implemented the main feature of my detector in 1984 while working as a tech at Hughes Aircraft.

I've been at it a long time, but so have others. I recently ran across this Ford patent from 1983 with similar features:
Adaptive cylinder by cylinder knock ... - Google Patent Search

From the dates I have mentioned, one might think this is old worthless technology, easily surpassed by anything released in the last ten years. But try to find an aftermarket controller that does individual cylinder knock control or ignores false knock as well as this one. You will have to go stand alone ECU such as MoTeC and Autronic, and they didn't have it until last year.

Electromotive says the coil drivers are inside the XDI-2 box, so Vampire is the correct unit for your application.

It would connect to the ABC wires, so no MSD 8870's are required.

As far as reliability issues, I have one guy with a 944 and TEC-III that has been using one for two years. His installation required the 8870's because in the TEC-III, the ignition transistors and computer are in the same module.

For dual plug with two DFU's, you would need a six channel Vampire.

I have the 944 turbo with the Tec 3 and I use the J&S safe guard. I ran this combo for two years and am convinced it is a great system for any car that is turbocharged. I have gotten bad gas on the street and prior to the J&S heard audible knock. After installing the J&S If I was getting into any detonation above the set limit, I would get a knock indication on the led I put up in a gauge pod on the A pillar.

My car is a three litre and was a combo street/ track car. With the J&S system I was able to use the factory knock sensor which is tuned to the engine by Porsche and I also had a GM knock sensor hooked to the TEC 3 which could be datalogged. Even though the TEC 3 has provisions for knock control, the GM sensor was too sensitive to the mechanical noises inherent in the 944T engine with two balance shafts and the overhead cam.

After installing the J&S system and setting the system up, and having a good tune, I never had an audible indication of knock, but on occasion the J&S light would flicker a bit indication it was cutting timing. I really couldn't feel it and I never saw any indication on my EGTs (four cylinder) that would indicate it was cutting timing enough to raise EGT.

Two years and twenty thousand miles with excellent oil analysis results, no oil consumption and great looking plug readings tell me this has been a good addition to a high performance engine.

The only problem I ever had with the system is that the connectors in the MSD 8870 got a bit loose and I was getting a random loss of ignition usually corresponding with hard braking into a turn or on hitting a dip. Found the problem by taking it apart checking the connectors, tightened them up a bit and after that it was back to normal.

Hope this helps as a completely objective assessment of my results.

Correction: The Interceptor cannot be used with the XDI-2. The Vampire must be used. It's also an easier install.

Drawing for Vampire to XDI-2 with one DFU:


http://forums.pelicanparts.com/uploa...1266281558.jpg

Pinsko:

Have you played with knock sensors on your XDI-2 yet?

John

not yet. I finally picked up my car today and will drive it a few hundred miles before it goes back for final dyno tuning.