knock sensing

[zguy36]

In the section on knock sensing, it is outline how to hook up a knock sensor using the gm ESM module. It seems like with the new software that reads knock voltages, that this module isn't needed anymore. Also in the section on knock sensing, it is mentioned that the small circuit needs to be built because the voltage from the module is too high to be directly inputted into the processor port. What is the limit on the voltage for the processor port? Thanks!

MSII, V3 Board

[grippo]

The voltage limit into MS II is 5V - DO NOT exceed it or you will blow out the ADC port at the least. But its trivial to make a voltage divider out of a pair of resistors. As far as the smart knock module you definitely want to use it if you have one - you can't just detect knock without doing some processing, and you are unlikely to get better processing than something made by OEMs for your specific car.

[zguy36]

Isn't this what the new software does v2.5? When I was looking at the setup for knock threshold, you could chose voltage values for the knock being sensed. The knock module that is shown in the knock sensing page doesn't give any feedback as to the magnitude of the knock being sensed, since it only pulls to zero when any knock is sensed. This doesn't seem very useful since it cannot be easily adjusted.

The stock knock processing was done in the ecu, which is scrapped. There isn't a separate module for my application.

[grippo]

The v2.5-2.6 code allows you to set thresholds as a function of rpm, but it doesn't do any processing of the knock signal. I believe there is a board (sold by Boris) that does some processing that will work for many systems. It was designed to work with MS so you could try inserting that between the sensor and MS.

Now if someone has data or a technical description of a system or module or sensor which provides a voltage that is proportional to the severity of knock, then the code could be changed to back off timing in proportion to the amount by which the knock signal exceeds the threshold. This would probably make the response of the system somewhat quicker and smoother.

We plan to do real time processing of the raw sensor data at some point, but this will have to wait until the router board hardware is done. Then we can integrate and threshold the signal only during that period when each cylinder is sparking, and then it will only be necessary to retard spark for the particular cylinders which are knocking.

[scirturbo]

Not a module but a chip, the tpic8101.
 
SFC
 
 
Now if someone has data or a technical description of a system or module or sensor which provides a voltage that is proportional to the severity of knock, then the code could be changed to back off timing in proportion to the amount by which the knock signal exceeds the threshold. This would probably make the response of the system somewhat quicker and smoother.

We plan to do real time processing of the raw sensor data at some point, but this will have to wait until the router board hardware is done. Then we can integrate and threshold the signal only during that period when each cylinder is sparking, and then it will only be necessary to retard spark for the particular cylinders which are knocking.



This post is at: http://www.msefi.com/viewtopic.php?p=110325#110325



Posted by email.

[Skiericski]

That TI chip looks awesome. Have any of you electrical folks looked at the datasheet? Everything looks pretty straightforward to me except for the serial programming interface. It would be nice to be able to let MS do the configuration of the chip so that the filter parameters could be set via Megatune, but I don't know how difficult it would be to interface the MS micro with the TI chip. If that would be a major undertaking then possibly a standalone programmer could be built.

The benefit of a signal proportional to knock severity and a programmable filter seems pretty worthwhile.

For those interested the datasheet is here:
http://www-s.ti.com/sc/ds/tpic8101.pdf

[masterx81]

I'm doing a project with this chip...
It's a little bit triky, but nothing impossible...
The led outputs are the SPI outputs (Casual thing? ), so the module can be atteched here, but it need at least one more pin (INT/HOLD) for work. The ms2 pin count is a little low...
The read can be done via the spi bus (Leaving free one a/d port).

[scirturbo]

Spi com can also be done with 'normal' i/o pins. I'm also testing the chip and it works just fine with my spi less 80c552 board.
 
SFC

----- Original Message -----
From: masterx81 (enrico@je.net)
To: megasquirt2@msefi.com (megasquirt2@msefi.com)
Sent: Saturday, May 20, 2006 9:07 AM
Subject: knock sensing



I'm doing a project with this chip...
It's a little bit triky, but nothing impossible...
The led outputs are the SPI outputs (Casual thing? ), so the module can be atteched here, but it need at least one more pin (INT/HOLD) for work. The ms2 pin count is a little low...
The read can be done via the spi bus (Leaving free one a/d port).



This post is at: http://www.msefi.com/viewtopic.php?p=121104#121104

Posted by email.

[83quattro]

This interface should be perfect for MS-II!

http://www.linkecu.com/products/Ancilla ... KnockBlock


O.J.

[83quattro]

This interface should be perfect for MS-II!

http://www.linkecu.com/products/Ancilla ... KnockBlock

It has 2 inputs, 0-5v out, and a headphone for manual detection!!


O.J.