Is my knock sensor good?

[fphil]

I just started the Knock setting process on the biturbo.  I was still relying on the detection of the Marelli ecu which is running together the Vems.  The first question was "Is the sensor good" since the car is about 30 years old.

I guess the accelerating sensor is the Marelli Sen14B  commonly find with the Lanca and Alpha of the eighties. It is srewd on the middle of heads (V block) under the admission and hard to seeand reach.
The SEN14B is certainly of piezo type with no electronics (impedance is > 1 M).
Engine running the signal peaks are about 100 - 150 mV and I hardly see the pattern sequence of the ignition.
When I (gently) ping with a bar near the sensor I scope the underneath response


(vertival div = 50mV)

When I set the ampli gain =1 the knock value 0,1,2 ..  are in the range 50-70.
When I compare with other data these levels seems to be quite low, knowing besides that the Vin_Min of the knock chip is 50mV.
What do you think?. Does someone know the standard sensitivity  of the SEN14B?

[gunni]

I suggest you run a log recording all the noise from all the cylinders and see what values you are getting as you have one sensor so some will be dead quiet compared to other ones.

If you get reasonable values I recommend adjusting the scaling so they all read the same, repeat until they do, then introduce knock at medium rpm to see if you get a spike in noise recorded on any of the cylinders. If that is the case then you can proceed to fill in the noise table with your threshold limits, then repeat the test and see if the ecu isnt accurately retarding the ignition.

Do not expect the primary knock frequency to the be noisy one, I just ran a test on the dyno yesterday and the second order frequency was FAR noisier when knocking vs the primary (i.e 6.5khz is the primary, 13khz is the secondary) Thus we disregarded the primary and used the secondary as the primary input, there was also good amount of noise at 2.5orders.

You can record on your scope the crank signal to get a feeling of where the noise was occuring (if you can record that fast).

[fphil]

Thank you for your answer. Yes this is a good procedure to follow, but first I would like to check the sensor.
As I said for an ampli gain =1 (done by the Texas chip) the knock values "Knock 0", "Knock 1","Knock 2" ..  are in the range 50-70 so I am wondering if this order of magnitude is any good. No use to go further and search for detonation if the sensor is out of specs and the S/N dead.
Right the 1st harmonic can be more significant.

[gunni]

Try the sensor across the rpm and load range to see what its outputting
the table has a max limit, if it doesnt reach that but to about 60-70% then you should be good if idle is very quiet.

[fphil]

My concern is that the VemsTune Output values of 50-60 that I get while running with gain=1  might be too low (not too high). The noise table shows value about 800.
What are the value average that you usually get?

[gunni]

The noise table needs updating based on what you get, are those 50-60 at high rpm and load?

[VEMS]

Knock setup procedure is covered extensively in help: http://vems.hu/vt/help/v3/v3_knock_control.html

[fphil]

As I said, this is not the matter of the setup procedure. Yes indeed, the tables and the parameters can be tuned and there is a doc for that :D
My concern is with the sensor itself. No need to tune anything if the sensor is not working good.

The sensor accelerometer is working as the picture above shows, and the "Knock average" does increase somewhat with the load (3300rpm, 100kPa), but  it seems to me that the output voltage of this sensor is quite low with a poor S/N. (I checked the line cable with a square signal good)

What I need to know is the level of the VemsTune data "Knock Average" (with which Knock gain parameter) one usually gets from a piezo accelerometer of the Marelli SEN14B type.

[gunni]

That is a variable for each and every engine, the mounting torque alone changes the noise values. distance from cylinders, block density etc all have effect.

[fphil]

Yes, this is true. Could you tell what minimum of the Knock average you have experienced for some car?