1997 F-Body ECM

[Tom H]

My current working assumption of the module pinout is:

PIN 1 +5V
PIN 2 ANALOG OUT (LEVEL)
PIN 3 MODULE ID
PIN 4 +10V ANALOG
PIN 5 KNOCK DETECTED
PIN 6 GROUND
PIN 7 INPUT
PIN 8 N/C

I had expected (based on some patents GM holds) that a "KNOCK WINDOW" signal would be sent into the part to define the detection window. This isn't the case (I think) and if windowing is used, it it all inside the TPU. Re-considering the system, this is probably for the best.
So that revises my guesses: Analog out is used to determine module and sensor are there only (?). Long path for this... analog out is measured by the A/D on Tside and passed to Eside through SPI. Again I think this is only for monitoring that the components are there and working. Knock detected is set when filtered levels of the knock frequency are at some threshold above the noise. Knock detected is counted by TPU when inside a knock window (?). +5V and Module ID are there only for manufacturing and test (?). Input comes from a pre-amp on the Tside (96/7 at least). Should two knock sensors be used as an option, there are some controls into the pre-amp. The two signals are combined there.

One other possibility for knock window in the system is that TPU pin 11 is routed through the ribbon (pin 25). The gain of the pre-amp might (?) be modulated OR the output zeroed .

Lots of things to check through. While I am trying to figure out how to simulate the knock sensor, there are tests to see how the TPU responds to knock out that can be run.

-Tom

[Tom H]

I wonder if anyone has made an audio recording of the knock sensor output? The sensor is piezo type contact microphone and may require some attenuation along with AC coupling to get levels right. Would just love it if there were an audio file I could just use.

-Tom

Edit: A little later I found https://theknockbox.com.au/the-knock...nd-recordings/
Does this sound right? not what I expected.

[NomakeWan]

My current working assumption of the module pinout is:

PIN 1 +5V
PIN 2 ANALOG OUT (LEVEL)
PIN 3 MODULE ID
PIN 4 +10V ANALOG
PIN 5 KNOCK DETECTED
PIN 6 GROUND
PIN 7 INPUT
PIN 8 N/C

I had expected (based on some patents GM holds) that a "KNOCK WINDOW" signal would be sent into the part to define the detection window. This isn't the case (I think) and if windowing is used, it it all inside the TPU. Re-considering the system, this is probably for the best.
So that revises my guesses: Analog out is used to determine module and sensor are there only (?). Long path for this... analog out is measured by the A/D on Tside and passed to Eside through SPI. Again I think this is only for monitoring that the components are there and working. Knock detected is set when filtered levels of the knock frequency are at some threshold above the noise. Knock detected is counted by TPU when inside a knock window (?). +5V and Module ID are there only for manufacturing and test (?). Input comes from a pre-amp on the Tside (96/7 at least). Should two knock sensors be used as an option, there are some controls into the pre-amp. The two signals are combined there.

One other possibility for knock window in the system is that TPU pin 11 is routed through the ribbon (pin 25). The gain of the pre-amp might (?) be modulated OR the output zeroed .

Lots of things to check through. While I am trying to figure out how to simulate the knock sensor, there are tests to see how the TPU responds to knock out that can be run.

-Tom

Wow! If this pinout is correct, it jives with the random photo I found on the internet of someone who rigged an LT4 module to work on the $DA3 MEMCAL. As there was only this one small photo from this angle I couldn't confirm which pins were which, but knowing what I know about the MEMCAL layout and using the pinout you listed above, it does appear to check out perfectly. Black ground, white knock sensor, green knock signal, red +12V (apparently through a 110 ohm resistor). I'll have to measure pin 1 and pin 4 for you this weekend to confirm PCM output voltages on the 94-95.

[Tom H]

Can anyone tell me the year that GM started using these knock filters? I don't mean the just the 8pin but the MEMCAL with the ESC. Looking through patents and setting a range of years will narrow things...

[NomakeWan]

Can anyone tell me the year that GM started using these knock filters? I don't mean the just the 8pin but the MEMCAL with the ESC. Looking through patents and setting a range of years will narrow things...

Moates states that their MEMCAL adapter works on 1986-1993 TPI and LT1 vehicles, so I would assume 1986. That said, the 5-pin ESC module used on the 1985 Corvette (and others) was apparently based on the same design, though again, I've never seen one cracked open to prove it, nor can I figure out how to open mine.

UPDATE: Apparently MEMCALs used to not include a knock filter. The 5-pin module is listed by GM as being a 1985-1989 thing, so 1990 would be the first year of the integrated knock filters. My apologies for the confusion.

[Tom H]

OK, somewhat unsure if I should post conjecture/notions and stuff that can't be proved yet but I would like to get more discussion about the knock control system going...

Using an 8 pin knock module I did some experiments this morning. I fed analog from my PC sound card into the analog input of the module. I powered the module with a quiet 10v. At first pass, it looks like the analog output represents the level of the input. My guess is that the Tside converts this to digital and the Eside processes the result. Again just guessing... I think this forms part of an Automatic Gain Control (AGC) loop. This makes sure that differences in knock sensor and effects of temperature and so on don't have any effect on the knock output (digital).

I am now going to dig through the Eside to find if this makes sense. I hope/expect to find that the A/D converter result is passed to the Eside over SPI. The AGC loop will include an output from the TPU which will be PWM and control the gain. I know that earlier modules (94/5) did not include some of the pre-amp chips that the later (96/7 ) boards have. I expect that there was some improvement that brought all this about....TBD

Does anyone have a different thought on how this works?

-Tom

[kur4o]

I cracked open a busted GM tpi IgnitionControlModule and there is the same chip inside as the knock modules have, It have the same 6x3 pins configuration and the same appearance. Could this be some kind of processor or some kind of signal converter. They are definitely related or the same part with similar functions.
I will try to get a pic of that ICM. You can figure something interesting of it.

Older pcms have external knock module that looks like that.
http://www.gearhead-efi.com/Fuel-Inj...module-for-TBI

The knock sensor goes in and there is one signal out to pcm.

Some more pics and info on knock modules here
http://www.gearhead-efi.com/Fuel-Inj...1-and-16181333

[Tom H]

Hi kur4o,

I don't think the knock module chip is a processor... I don't see a clock. I can check later to see if there is some sort of relaxation oscillator, but expect not. Just a guess, but I think it is custom mixed analog/digital chip. There is some sort of level sense for the knock frequency and a digital output to the TPU. By disconnecting this in the ribbon cable, I can cause the TPU to count single events.

For the later cars (96/7) the pre-amp is also of interest. There is a line (in?out) of the TPU connected to it. Could be to adjust gain--> serial word?pwm?digital output. This is what I am working on now.

The analog output from the module is (i now think) just for diagnostics. Not for sure because there could be links in the accumulated ceiling and floor levels to the rest of the code.

The Eside PRU sets operation for dual or single knock sensor. This goes into the pre-amp, I have not yet tested the operation.

Still wondering about the analog VCC for the knock module on 1994/5 cars. Mine has a simple supply made up of a power resistor and two zeners along with a bunch of caps. This powers the pre-amp, the knock module and bias to the knock sensors. Wonder if the early cars even need a supply. With no pre-amp, the power will be less so perhaps it is just ignition voltage??

This system is complex and will take some time to figure...

-Tom

[Tom H]

Schematic of knock system... updates/corrections

[Tom H]

One section of code that was trouble, now think I understand it (?)

In the late cars (96/7) there is an additional chip in the knock section. I have long struggled to understand the function. There was a connection from TPU pin 11 through the ribbon that drives an inverter made up from transistors&resistors. WHY? To start with, it seems like the chip is likely a pre-amp and/or perhaps an analog switch. Since it has a VCC of 10v, a regular inverter won't work. They used a transistor and three resistors to both level shift and invert. The updated schematic I posted the other day shows this.
I have been looking at the knock filter code for a bit and speculated (with myself) that a curious section might be involved with this line. Here is that code...

9438  12 4E 02 0B 	BRSET  @$4E,$02,$9447	; TOGGLE

943C  14 4E 02    	BSET   @$4E,$02		; 
943F  CC 00 00    	LDD    #$0000		; 
9442  FD 14 CE    	STD    $14CE		; TIO SHARED RAM
9445  20 09       	BRA    $9450		; 

9447  5F          	CLRB			; 
9448  86 40       	LDAA   #$40		; 
944A  FD 14 CE    	STD    $14CE		; TIO SHARED RAM
944D  15 4E 02    	BCLR   @$4E,$02		; 

9450  39          	RTS			;

I wrote more code to test and TPU pin 11 is controlled by $14CE BIT 14. It is this bit that is input to the inverter and sets the level of the pre-amp input (pin 9).

I believe that where two knock sensors are used this code is used to toggle between the sensors. The @$4E,$02 branch just toggles back and forth each time it is called. The $14CE location of the TPU seems to have no other function.

The pre-amp/switch just connects a different input to the knock filter when the line is changed.

-Tom

[kur4o]

Some pics of the icm module.

So you are sure that the preamp is just for 2 vs 1 sensor being used.

That is quite interesting. I think 94-95 b-bodies used 2 sensors stock. Finding a knock module from these will be good to confirm it.

I think that in 94-95 code all the data goes straight to tpu that feeds eside. tside as a buffer will lag the signal enough to be unreliable. Could be that both sides get the signal just like the low res.

[NomakeWan]

Just a guess, but I think Tom is referring to the knock section on the PCM, not within the knock filter? It sounds like for the 96/97 PCMs with dual knock sensors, the PCM flips between which sensor's output gets sent to the knock filter per tick. But the filter itself is the same, which is why LT4 modules work fine in earlier cars, and earlier modules will work in later cars. The filter didn't change, only the PCM function did.

[kur4o]

The pic didn`t get through, here it is.

I thought he was talking about the preamp within the knock module and its function. The 8pin chip that is missing on 94-95 knock modules.

I am sure he will clear it out.

[Tom H]

I have caused some confusion...

First, to fix what I wrote: The so called pre-amp/switch is on the Tside PCM. I changed between talking about the filter and board without saying so. I will try to take more care...

Second, to update: I think the Tside PCM pre-amp/switch is in fact a switch. Hope to get the scope on it later (today?). I believe the switch is a simple quad bilateral analog switch. To select the filter to pass up to the knock module, there are two select inputs. One select input is passed from the Eside TPU and is used at I mentioned the other day. The second input is connected to the Eside PRU, where it is defaulted by software.

The investigation goes on...

[Tom H]

One more thing regarding the knock module...

The input is VERY sensitive to knock sensor input below ground. It latches up very easy and if your supply isn't current limited, you will let the smoke out. I made an error by not biasing the input to 5V, thinking that AC coupling would be ok. The input signal to the knock module then moved above and below ground. Knock filter module was very unhappy about this. If you plan to set up on a bench and connect your PC sound card to it (as I did):

- Power the knock filter with 10V
- Bias the input at +5V through a resistor
- AC couple the PC output to the biased knock filter input

More results on this to follow.

Question... Has anyone recorded the knock sensor output? This should be very easy to do... If someone is willing I can suggest an easy way.