Knock Sensor Location - EFI Conversion

[mikedsilva]

I've seen photos of people using a single knock sensor on the case on the stud that supports the throttle bellcrank plate. My question is, is it acceptable to mount the sensor to the TOP of the plate, seeing as I have 2 threaded holes that could easily be utilised?
I made a short vid to explain my question...

[mikedsilva]

Out of curiosity, I mounted the sensor in various locations and tapped the motor "gently".. measuring the AC Volts registered.. Not sure if I did it properly, but here is the results...

[stownsen914]

It's an interesting question. One would think that as long as the mounting is solid with no gaskets, that there will be some attenuation but not much. I like the idea of bench testing to see the difference in responsiveness, that's a solid idea.

I wonder if the inconsistency in voltage readings you saw in the first test might be due to responsiveness of the voltmeter. Perhaps an oscilloscope would be better for measuring small, transient voltages, as when measuring output from a crank position or cam sensor.

[rswannabe]

My gut says the sensor should get a stronger “signal” when mounted directly against the case. I would theorize that beaks in the sound path from bolted connections would reduce the sound “signal” to the sensor. BUT the factory did not mount knock sensors directly to the case. Instead they mounted them to metal bridges that were bolted to the heads. So seems a bolted connection can’t be too compromised.

[targa72e]

Here is an interesting alternative mounting Knock bridge used on 964/993 on earlier engine.

https://forums.pelicanparts.com/porsche-911-technical-forum/1152445-new-csf-porsche-911-930-turbo-engine-oil-cooler.html#post12155429

john

[mikedsilva]

Originally Posted by targa72e

Here is an interesting alternative mounting Knock bridge used on 964/993 on earlier engine.

https://forums.pelicanparts.com/porsche-911-technical-forum/1152445-new-csf-porsche-911-930-turbo-engine-oil-cooler.html#post12155429

john

yes, have seen that. It is indeed an alternative that no one has mentioned before. I guess it must work in that location or a shop like that wouldnt use it.

[dannobee]

Originally Posted by rswannabe

... BUT the factory did not mount knock sensors directly to the case. Instead they mounted them to metal bridges that were bolted to the heads. So seems a bolted connection can’t be too compromised.

No they didn't. They bolted the bridges to the BOTTOM of the cylinders, nearest to the case.

Just because people drill and tap the heads for the bridges doesn't make it right, and in fact Bosch specifically says NOT to mount the knock sensors in this manner because of valvetrain noise possibly giving erroneous signals.

[mikedsilva]

This is what I have decided to do...

[cgarr]

I’ve machined a lot of heads for the bridge and seem to work well from what I’ve heard.




Sent from my iPhone using Tapatalk

['78 SC]

Mike,

I have the Bosch knock sensors mounted on the 993 bridges same as shown in Craig's photos. I did not move them around to replicate your scenario (sensor on throttle console), so I tried tapping at the console to see how the sensor responded. Not exactly what you're looking for, but somewhat relevant.


[sensor on 993 bridge; tapped throttle console]



[sensor on 993 bridge; tapped inside exhaust port]

I was simulating knock by dropping a small ratchet extension onto the throttle console and tapping inside the exhaust port, so no way is the energy applied consistent. Consequently I wouldn't read too much into the signal strength. I think it is significant that the signal is similar (frequency and damping).

Conclusion? It appears the signal propagates across the block. I even tried the same experiment using the sensor on the opposite bank and got similar results. I suspect the ability to detect knock is going to come down to the ECUs capability to detect and analyze the knock's acoustic signature in a sea of noisy mechanicals (like the valve train, as noted above). Part of the tuning is setting the characteristic knock frequency based on the cylinder diameter (D):

Knock frequency (kHz) = 1800/(Pi * D)


I'm going to use the ECUMaster EMU Black ECU. Version 3 of the firmware (now in beta testing) is claimed to be able to attribute knock to a specific cylinder (by knowing where each piston is in its cycle when knock is detected) and respond on a per-cylinder basis. We'll see.

Steve

[AndrewCologne]

Just tapping on the console does not simulate knocking. As dannobee said, there are also many other noises which do affect the overall reading of a knocking sensor.

You can do the following cheap set up:

1. Simply connect the two wires to the knock sensor and lead them to the cars interior, here connect these wires to a cheap LM 368 amplifier where you can connect common in ear phones. Those tiny amps can simply be fed by a 9v battery.


https://www.ebay.de/itm/255283194260

And then you can enjoy how your engine really sounds, its nice and beside knock detection "by ears" its also a very good tool to hear if anything else in the engine could sound odd. Then you can connect that sensor to your ECU of your EFI system.
I installed my sensor at the middle of the top of the crankcase behind the alternator, see pic above. And btw for me personally its the most secure check for detonation at lower and higher rpms, as I can hear it --> by myself.

[Mike Billings]

Instead of the separate amplifier board, I routed my knock sensor output to the aux in on the car stereo! Did that on a couple cars.

[rswannabe]

Originally Posted by Mike Billings

Instead of the separate amplifier board, I routed my knock sensor output to the aux in on the car stereo! Did that on a couple cars.

I like that; just swap to the aux line in any time you want to check on the motor. I might try that.

[AndrewCologne]

Nice ;-) but ... its totally different perception if listening to the inner engine sound with almost encapsulated in ear phones where no other environment noise/sound will affect the focus on the inner engine noise itself. Not mention the outer engine sound from the back to the front of the car when listening nock signals though the car stereo – at lets say 6000 rpm :-)

[PeteKz]

Steve, Andrew and Mike: Thanks for that info. Cool stuff. Now I have another "nice to do" item on my project list!

[Cloggie]

I am doing the bridge on my 3.4. I had a J&S Safeguard on my 3.2 CIS SS which had the knock sensor on the throttle plate.

I had a dynamic readout built into the clock (just a 10 LED display from green to red) and the knock sensor certainly was active, even if I did not actually hear knock with my ears.

I think both will work well enough....but I think the bridge will be a little better.

Time will tell...

D.

[Speedy Squirrel]

First, I want to share the datasheet for the Bosch sensor to emphasize the importance of the tightening torque, the imperative to not use any washers between the sensor and the engine, and the need for a clean, flat mounting surface.

https://www.bosch-motorsport.com/content/downloads/Raceparts/en-GB/51782795118019851.html#/Tabs=51794187/

The sensor is a quartz crystal that produces a charge output when mechanical pressure is applied to it. The charge gets converted to a voltage with a precision capacitor. Ideally, the mechanical pressure is applied by a mass of metal that is moved by the vibration of the engine structure associated with engine knock. If you mount it on a non-flat mounting surface, such as a washer or a dirty mounting pad and/or you tighten it too much it will produce an offset voltage that is always there, whether the engine is knocking or not. So there is that.

[Speedy Squirrel]

With regard to mounting location, the block is by far the best choice. You can try other locations and determine their suitability by trial and error, but the block is virtually guaranteed to work. The problem is that every part of the engine has its own natural frequency. Blocks and heads are obviously very stiff. Brackets and sheet metal not so much. When excited by engine operation, the knock sensor with pick up that vibration and could misinterpret it as knock. So there is that.

[mikedsilva]

Originally Posted by AndrewCologne

Just tapping on the console does not simulate knocking. As dannobee said, there are also many other noises which do affect the overall reading of a knocking sensor.

You can do the following cheap set up:

1. Simply connect the two wires to the knock sensor and lead them to the cars interior, here connect these wires to a cheap LM 368 amplifier where you can connect common in ear phones. Those tiny amps can simply be fed by a 9v battery.


https://www.ebay.de/itm/255283194260

And then you can enjoy how your engine really sounds, its nice and beside knock detection "by ears" its also a very good tool to hear if anything else in the engine could sound odd. Then you can connect that sensor to your ECU of your EFI system.
I installed my sensor at the middle of the top of the crankcase behind the alternator, see pic above. And btw for me personally its the most secure check for detonation at lower and higher rpms, as I can hear it --> by myself.

Brilliant!

[mikedsilva]

Originally Posted by Speedy Squirrel

First, I want to share the datasheet for the Bosch sensor to emphasize the importance of the tightening torque, the imperative to not use any washers between the sensor and the engine, and the need for a clean, flat mounting surface.

https://www.bosch-motorsport.com/content/downloads/Raceparts/en-GB/51782795118019851.html#/Tabs=51794187/

The sensor is a quartz crystal that produces a charge output when mechanical pressure is applied to it. The charge gets converted to a voltage with a precision capacitor. Ideally, the mechanical pressure is applied by a mass of metal that is moved by the vibration of the engine structure associated with engine knock. If you mount it on a non-flat mounting surface, such as a washer or a dirty mounting pad and/or you tighten it too much it will produce an offset voltage that is always there, whether the engine is knocking or not. So there is that.

Interesting.
The reason I used an aluminium standoff/spacer is because the body of the sensor is larger than the relief that is around that particular stud. So this will result in an erroneous reading?