Hi,
i have a V8 engine with 4 knock sensors from OEM side…they are mounted between cyl1/2; 3/4; 5/6 ; 7/8.
I at the moment i use the ones between 1/2 and 5/6. Obviously the signals of especially 4 and 8 then are quite quite. With the settings i got all on the same level, but i am worried that i miss events on 4 and 8. Knock is the biggest enemy of my engine.
So i was researching about possibilities to use all 4 sensors.
I read that some people put 2 knock sensors in parallel and throw them on one input. Is this really a good idea and is this working? Can the EMU handle this?
My first thinkings say that this results in a messy signal, which probably isn’t worth anything anymore. Does anyone have expiriences with that?
That is an interesting idea.
We never tested something like that.
It has the potential to work.
I am unsure about parallel connection because it means that one knock sensor is also a load for the other. The signal will likely be partially attenuated.
I think series connections might be better. However, this would have to be tested because these are just my initial thoughts.
There is also a proper way to add signals together, which is much more complex. It would require an operational amplifier circuit that outputs the sum of input signals. This allows for the sensors to be completely independent while adding the signals together.
Thx for ur opinion.
So probably then i will play around a bit with 2 sensor in series after my current setup is finished. Sadly i don’t have the possibilities to analyze this deeply. So if u and ur team is bored, would be cool if u can have a look at this in the future.
What do u think about putting the two knock sensor in parrallel and adding a Diode behind each? Then i think always the higher signal of both should reach the ECU. Do i think to simple or u think this could introduce some additional effects?
I will build a new harness and want to be prepared for different options to try.
The Knock sensor provides an AC signal.
Adding diodes will make it DC.
I don’t think that will work well.
It can introduce all sorts of problems.
After some breakes i came back to this topic and did a lot of testes.
Due to V8 setup i can always compare the signal with a clean one from the other bank. That is my reference.
Two Sensors in series: Doesn’t work well…it results in just having a useful signal for the second sensor…the first is always much quiter than the second.
2024_1110_1551_rev90_knock_ana_34_12_Signal.emuprolog (2.2 MB)
2024_1110_1605_rev90_knock_ana_12_34_signal.emuprolog (1.7 MB)
Parrallel shows better results. Both siganls are equal….but sadly there are even with maximum gain pretty quiet.
2024_1110_1612_rev90_knock_1234_parrallel.emuprolog (1.5 MB)
For me it seems single events somehow just get lost. Thats why i also think this way is not good.
Next step i started to play with amplifier circuits…to simply just add the signals.
I used a simple summing amplifier like that:
Also not working. I needed 1MOhm as resistors to have a signal (without gain)…and the output signal respectively the reading from emu were implausible. I later read, that this circuit is most likely just working properly for DC.
I my researches showed that most likely a AC coupled inverting amplifier is the way to go, where both sensors first seperatly routed through an amplifier, to have a stable signal and then add them with another amplifier. Sadly here my simple electric knowledge reaches its border cause there are some more capacitors and resistors to be specified. There i start to get lost…does someone have expiriences with this?
I throw so much effort and thoughts in this topic cause my engine will get an compressor update soon and i am quite sure i will kill it without a suffient knock control. This engine is really sensitive in terms of knocking.
You’ve got me thinking more about it, and now I am sure it’s not possible to do what you want.
Since two sensors are mounted at different physical locations, a phase offset exists between the signals from those sensors. One sensor sees the signal later than the other.
Both sensors see exactly the same frequency, but with a delay.
If the delay ends up being 180 degrees and the amplitudes are equal, adding those signals together results in destructive interference, and the outcome is a signal equal to zero. This is a scenario where the knock is actually present and the signal is visible on both sensors independently, but adding them together gives zero.
Unfortunately, there is no way to get around that.
P.S.
The adder circuit works for AC signals, but you must supply the op-amp with a symmetric supply (positive and negative voltages).
Thx for ur answere. This is also something i was worried about a long time…those unknown interference.
But i expirienced on my engine, that if i want to listen to cylinders 3 and 4 with the knock sensor placed between 1 and 2 the signal gets massively quiter (from 3 and 4). So the amplitudes must significatly decrease with distance. So i was thinking about to combine signals of the diagonal knock sensors…so 1/2 and 7/8, 3/4 and 5/6. I have the hope the amplitudes of the relevant frequencies are more or less no more present on the other side of the block due to long distance and so the risc of destructive interference is very low.
I just came up with this thoughts when i got bad results with my tries. So i didnt tried yet, but i definitly want to .
But thanks for the hint with the symmetric supply…that sadly adds a lot of complexity.
Do u maybe have an other idea to somehow get a good anti knock startegy for V8 or even bigger engines with just 2 sensors?