rwyatt365

I posted this exact same thread over on MSExtra, hoping that Matt Cramer would come an take pity on me, but I've gotten no bites. So I'm throwing myself to the MT wolves to see how big of a ****-storm I can build up. So, without further ado...

I've almost run out of things to tweak on my "trusty old" PNP2 so I thought I'd dive into seeing if I could get anything working with knock sensing. In the "MegaSquirt PNP Gen2 Manual", near the end, is a section that implies that setting up knock sensing is easy-peasy. So I cracked open the case, spotted J15 and moved the jumpers from the "run" position to the "test" position. Then I configured two new gauges to look at ADC6 & ADC7 as described in the manual and...nothing. Both gauges show zero at idle unloaded...DISAPPOINTED. I was hoping to at least see a few counts, but nothing. I was tempted to start fiddling with the "start" and "slope" pots, but it was getting late so I packed it in for the night.

My question is (and no snarky comments from the MS3 and Pro guys); Am I on a fool's errand? I've searched far and wide and have found any number of threads on different forums that say, "You can't do knock control native on the MS2". There are others that indicate that the procedure in the manual works, but that attempting control is sketchy at best. Then there are those that suggest external boxes, or CAN bus adapters - neither of which is something I want to do (more often than not, the cost of those external boxes would justify just popping for a MSPro).

So, again, I'm not looking to control spark with the PNP2. I'm just trying to determine if that elusive knock signal can be datalogged.

rwyatt365

No one here has replied, but I did get a response from Matt on msextra, basically saying, "It should work, check the 'gain' pot and see what happens". Well, I turned that pot (*** well as the "start" and "slope" pots) every which-way but loose and got no change - and the ADC6 & 7 gauges I set up were still pegged at 0.

So...Next steps; Check the physical wiring. I'm using the stock knock sensor and that and the wiring are intact AFAIK, but I'll check continuity up to the pinout in the MS. I don't have an oscilloscope, so I don't know how I can look at the actual signal (other than put it on a DVM). But I'll give that a look-see also.

Looking forward; If continuity is good, but I can't detect a viable signal from the sensor would it be worthwhile to look into a sensor OTHER that the stock Mazda part? Would something like that cause me more problems than I might be solving, or is there any benefit to doing that in the long run?

rwyatt365

For the sake of completeness - I hate those threads that start with, "I have an interesting problem...", and then after two or three posts - nothing. Well, here's what I've found out so far (with a little help from Matt Cramer over on msextra.com);

First, the wiring on the existing knock sensor was completely borked. The shield wire on the sensor side of the connector was broken, and the signal wire on the harness side of the connector was broken. I decided to replace the sensor with a new "OEM" sensor (i.e. an eBay part that claimed to be OEM) and wire in a new connector.

After that was done I determined that when the jumpers at J15 were in the "test" position I could adjust the level of the ADC6 line (the knock threshold) using the "start" and "slope" pots to be just barely above the max output from ADC7 (the knock signal). In my case, the max from the sensor was around 230-ish counts and I "tuned" ADC6 to be around 280-ish. The output from ADC6 bounces around like hell at idle, so the only way to get something reliable was to datalog a normal trip and guess-timate a value.

Now, it looked like I had a decent knock signal and a decent knock threshold so the only thing to do was to turn put the J15 jumpers back into "run" mode, turn on knock sensing and see what happened. I set the values in the Knock Sensor Settings VERY conservative (i.e. don't pull a bunch of timing when knock is detected - all I want to see is if anything at all is detected). My first effort yielded nothing at all so back to some coaching from Matt again.

Then I had an "aha!" moment after some more reading on msextra. I was assuming all along that the signal from the knock sensor on the stock ECU connector (2F, for those that are curious) was being fed into the knock detection circuitry. WHAT IF that was not the case...what if the MS2 is depending on the "knock" signal (pin 10) from the 26-pin auxiliary connector? So I took a leap of faith and ran a wire from 2F, to pin-10 on the aux input connector. After doing that, and making some tweaks in the Knock Sensor Settings and "voila" I was getting knock detection and knock retard events!

Can I guarantee that any of this is "real"? Nope. Will I stake my engine on the readings that I'm getting? Nope. Will I continue to explore this? You bet your ***!!

howangsNA

👌