Mongo

I asked about that too (in PM) but didn't see a response to using 2 sensors. My theory is that the system is sensitive enough to only read and respond to one knock sensor signal. However, I'm gonig to be running this car on the street with 91 octane gas. The chances of detonation on a properly tuned setup are slim.

Ben Allison

Only for a pair of cylinders.

Louie928

The S4 EZ-K routinely retards some cylinders due to knock sensing using the stock timing map and 92 octane fuel. The stock intake is bad enough that cylinder filling isn't the same and the differences occur at different RPMs. That is, some cyls will retard spark at 3800, some at 4500, and some at 5500. Some at 4000 and then again at 5500, etc. You can retard timing enough so that no cylinders knock, but you lose quite a bit of power at those settings. The engine works best to let the EZ-K retard the spark on individual cyls as required.
I use a J&S Electronics ignition controller to provide individual cylinder spark retard on my GT with a DTA ECU that has no knock sensor input. http://www.jandssafeguard.com/Vampir...S_Vampire.html
It works fine. On units that have only one knock sensor input, you should still use the two S4 sensors, but connected in parallel to the single knock sensor input. I did some knock sensor testing and found that if only one sensor was used on the S4 block, you get uneven sensitivity from one end of the engine to the other. Both sensors are needed, but one sensor would be better than none if you are working with the '85/'86 block. Don't mount the sensors on the cylnder heads. The head gasket blocks the acoustic path from the block where the detonation resonance occurs, plus the cams and valve train make a lot of background noise.

Mongo

When you say run the sensors in parallel Louie, do you mean splice the 2 together on S4/GT cars and make them into 1 harness?

Wow I also didn't realize the EZK did that much work on continuously retarding due to the design of the intake manifold. I am hoping the kit isn't a PITA to tune on the S4.

Louie928

Yes, connect each wire in each sensor cable together so you have one sensor cable. Connecting the two sensors together reduces the overall output, but there should be a sensitivity setting that can be set to compensate. The reason for using two sensors is to get relatively even output from the combined sensors for all cylinders.

I'm only guessing that the knock retard on quite a few cylinders is due to the uneven cylinder filling. If you set the mixture rich, there are fewer retards, but that costs some power too. I'll try to dig up some Sharktuner log files to show what I mean on the spark retard.

Carl Fausett

Ben has it exactly right.

Because each coil fires two cylinders at once (wasted spark system) the ignition adjustment that the knock event will cause will effect two cylinders at once. Being that it is a wasted spark system, retarding the other cylinder is of no consequence.

I selected to develop on the TecGT system because I feel it is more likely to be the selected replacement for the stock system. It already has higher resolution and greater capabilites than stock, and plenty of knock and boost controls. Yes, the TecGT system will take out timing and can add fuel at each knock event within the parameters you set.

If you want to step up to the Tec3 system, and run 8 separate EGT's and true sequential injection and all that, you can. But for most of us, there is no gain and no need for that complexity.

Here is a dyno chart from my last dyno with the TecGT system under about 12 pounds of boost and running up to 6500 rpm. Note the dead flat air/fuel ratio even with the climbing boost pressures as the engine accelerates. The engine noise (knock sensor) data log was so low we had to lower the threshold further and still could not find any knock events.

With an A/F line like that, you can see why. The red dashed line across the A/F table represents 12.1:1 A/F, which was our target.

Mongo

Carl would that mean I wouldn't have to splice the 2 knock sensors together for the TecGT ECU to read????

dprantl

You definitely want input from both knock sensors. There are two of them purposely so you can hear knock adequately from all 8 cylinders, regardless of what engine management system you are using.

Dan
'91 928GT S/C 475hp/460lb.ft

Mongo

so splicing together wouldn't disrupt the accuracy of the signal then? If not, than I would be even more happier and determined to pick this up!

Ben Allison

The TecGT and Tec3r only have provisions for one knock sensor.

Carl Fausett

You can splice the two knock sensors together. Keep in mind most V-8s have a single knock sensor.
I think the requirement for 2 knock sensors may have come from Bosch to satisfy the LMB, and not from Porsche. Even the 3.0L 4 cyl 1994 968 that I am working on has 2 knock sensors. It seems to be what the LH-Jetronic wanted, not what the car needs.

Also keep in mind the higher resolution we now have, so the system can better separate noises from the sensor. I am confident one knock sensor is fine - but we have confirmed you can splice the two together if you feel the need.

hacker-pschorr

Generation III LS Chevy motors have two (this includes the trucks), so does every Ferrari V8 since the 355 came out in 1995. I would be very surprised if the BMW & Audi V8's didn't also have two.

Porsche / Bosch were ahead of their time back in 1986

DR

Carl,

That is great info that will help other people, thanks for sharing!

Carl Fausett

Before this spirals out into a full-scale "Oh My God - there is only 1 knock sensor!" moment, we are making wiring adapters and pinouts for the Porsche 928 anyway, so if you want, we can make a 2-into-1 knock sensor plug-in adapter for you and provide it with the kit.

I think if you check at the LMB (LH computer) although the old system had two sensors, they were combined into a single IO at the computer. It listened to a single signal, not two.

But whatever. If you require two to be comfortable, I can make that happen.

Louie928

The testing I did indicated that if we had a single knock sensor located in the center of the valley, it would probably be ok. With our sensor pads located one near the front and the other near the rear, two are needed. The rear pad doesn't pick up the front cyls very well and the front pad doesn't pick up the rear cylinders very well. Both sensors get all the cylinders quite equally. The J&S Electronics unit keeps the background noise from each cylinder saved separately so it's not so important that each cylinder have equal sensitivity. I don't know how other systems work on analyzing and storing each cylinder background and knock signal values.