JP86SS

iTrader: (1)

So the actual detection scheme is to read the output of the KS and amplify an output based on a "gain" setting, from there you need a "reset" for every "so many" DRPs.
IIRC the ECM code pulls spark by the magnitude of the input voltage from the KS control.
Not being a EE but couldn't this just be done with Schmitt trigger circuit to read the magnitude of the KS (tune w/ trimming pots), an op amp with some gain resistors, a timer/counter circuit to do a reset of the output?
Making it small enough to fit would be some work but it doesn't seem like a micro is really needed. Just thinking again.
Jp

roarin_mouse

The TPI scheme (presently) appears to be simple detection (rectification) with a small cap and a high value resistor both connected in parallel and from the diode to ground. The diode and cap serve to convert the AC to a DC level and the resistor serves to allow this level to bleed down. Without the bleed, the DC level would remain high well after the high level AC (knock) has passed. I don't think that there is a filter because the KS used in the TPI cars is somewhat narrowband (that's why the 305 and 350 KSs are different). A filter isn't therefore necessary and (probably) isn't used. I'm speculating a bit but I am an EE.

jwscab

you guys are on the right path, but it is much more complicated than that. The older knock sensor is just a detector, this is true. But it need a specific amount of gain, and a specific bandwidth. The knock sensors are small microphones, they have a very specific bandwidth, that means, they are very sensitive around one frequency. the problem is that they still generate a whole bunch of noise while the engine is running. the knock board contains some gain stages, to allow the signal to be processed. thats easy enough. then the signal is passed though a band pass filter, which has very specific requirements based on the engine. this is much more difficult to implement, as you only want a true 'knock' to be seen, becuase reducing timing afterall, is a performance degradation. the filter implementation is almost definitely a digital signal process, becuase a discrete filter would be very large due to capacitor values, ie, it wouldn't fit in the memcal. the signal is then detected, probably a peak detector, but it's width is measured, to make sure a glitch doesn't trigger a knock event. this is also not too difficult, but again, need some parts to make happen.

the newer cars use the micro to control when the knock circuit is active, basically, a certain number of degrees before a spark event, as it looks for a knock event prior to ignition. this way, some of the processing on the knock signal is easier. it just makes it tougher for us, becuase now we have to make it act like the older detectors. I'll probably play around with this when I get some time and get the other stuff nailed down.

Cyclone17

I've been following this thread and am also interested in buying some of these boards since I'm stuck with a pile of V6 memcals.

Blown87IROC

Great thread! I came across it while researching for a possible 165 to 730 swap.

According to streetrodstuff.com, “Resonance, which is characteristic of combustion detonation, occurs at about 6400 Hertz. So the pinging you hear is actually the structure of the engine reacting to the pressure spikes. This noise of detonation is commonly called spark knock.”

As noted earlier, 5.0 and 5.7 knock sensor modules are different. This is true whether mounted on the memcal or firewall. I’m guessing the difference is due to the size and weight of the pistons (and rotating mass), cylinder wall thickness, and anything else that changes the engine’s harmonics; or instead maybe, or possibly in combination with, combustion chamber shape, diameter, or volume—any of which affect flame propagation time, and hence frequency. If so, doesn’t it stand to reason that if you bore your engine or install forged or different shaped pistons, perhaps changing to aluminum heads—anything that would change the resonance frequency—a different knock sensor calibration would be required? My thought is, unless a circuit can be designed that can be tuned to a specific vehicle, unless bone stock, it’s not much use worrying what knock sensor module is used, 5.0, 5.7, 3.1, whatever.

I’ve attached the most relevant portions of 165 and 730 schematics in a jpeg. On the 165, the knock signal from the module (KNOCK#) comes in on J2/B7 and goes straight to the circuit roarin_mouse mentioned above, and does not use CAL32 (what would be the output from the memcal). On the 730, the signal from the knock sensor (ESC#) comes in on J3/F9 and goes to a different circuit. A branch of this circuit (CAL29) goes to the input of the knock circuit on the memcal, then the output (CAL32) goes to the circuit roarin_mouse mentioned. This circuit has a branch (ESCIO#, which doesn’t appear to be used) that is identical to the 165’s KNOCK# branch.

Using or changing to an external module is certainly an alternative, and the swap should be pretty easy. The cheapest price I found so far for external modules is $53 at AutoZone for the 5.0 and $66 at Rock Auto for the 5.7. You would still need a connector, which are about $12, plus or minus. (Both might be plentiful in wrecking yards—I haven’t been to one in years so I don’t know that.) You have to:

1. Remove the knock sensor module from the memcal (if so equipped).
2. Supply 12v (ignition) and ground to the external module (terminals B & D respectively).
3. Connect the module’s output (terminal C) to ECM terminal B8 (on the 24 pin connector).
4. Connect the knock sensor output to module terminal E through a resistor (in series with the sensor). You might get away with a 5% 100k resistor (putting you within ~4%, not counting the sensor’s tolerance), or better yet, use a 1% 95.3k resistor (putting you within <1%). Or, you could get a 100k sensor and forget about the resistor—certainly not cost effective if you have to buy a new one. (I’m assuming you can use a series resistor here just as you can use a parallel resistor when putting a 100k sensor in a 3.9k system.)

Replacing a 165 with a 730 should be much easier, simply move the wire at ECM B7 to B8 (the 24 pin connector), and make sure there is no knock circuit on the memcal. You will have to first move the wire already on B8 (A/C On signal), and of course there are a lot of other wires that have to be moved but I’m just referring to the knock portion of the swap here.

Either way, the only problem I see is that there would be no ESC# signal the 730 is accustomed to. I seem to remember, though, reading there was a code change needed for this swap, maybe that’s what it’s for, perhaps it’s disabling error 43 (ESC Failure)—I’ll have to find it again.
--------------------------------------------------------------------------
EDIT: Found it!

RBob said, "The malf 43 parameters can be changed, or just tie a 3.9K resister from pin F9 to ground."
--------------------------------------------------------------------------

As for the resistor networks, they are only for limp home mode. They don’t even have to be there. The only memcal I bought came with nothing but a one-shot PROM mounted on it. You can see where they made a not so neat job of removing the resnets—and why they didn’t just erase and reprogram the stock EPROM beats the heck out of me. It worked fine without the resnets but I don’t know if they masked off any error codes that might have been affected. Judging from their sophistication I doubt they did.

As with the knock circuit, it seems to me that using any resnet besides one specifically calibrated to your specific setup would be pure luck if it was close to the factory cal. It seems to me any power change would require a new resnet cal. I’m thinking that a spare ECM might be an easier/cheaper, and better driving, backup. Of course you’d have to have it with you and swap it out—something I’d rather not do on my way to church, regardless of how simple—but frankly, I’ve never experienced limp home mode so I’m unclear as to why the concern.

It would be interesting if someone with a modified motor tried various resnets in their system then force limp home mode (at the moment, mine is stripped to the unibody except for the dash and front fenders).

jwscab

Well, it's been awhile, but I was finally able to sit down and spend 10 minutes looking at this.

I have verified that the resistor values appear to give VERY similar results to an AUJP memcal I have here, so thats good news.

I haven't looked into figuring out which resistors perform what scaling functions, but if I have time, I will start playing around with it. As for now,I have a bunch of other things to tend to. Couple that with a pain in the a$$ scope at these low frequencies(don't ask) that won't be for awhile.

John, did get a chance to look at those parts I sent?

I have about 5 'spare' boards here, if someone is really interested,shoot me a PM and I'll let you know what the deal is with them.

JP86SS

iTrader: (1)

Funny thing, I just took them out of the shipper today.
Havn't had time to tie my shoes lately. Another couple weeks and I'll be back to normal (I hope). Work and weather have kicked my butt this year. Spring is near so I got to get moving on some of my activities.

bald_noggin

iTrader: (1)

Hi guys,

Sorry to bring this one back up.

I'm just starting to get back into my car and I had come across this post on the Resnet. Just looking at the diagram layout of the resistor networks, shows that on the 16 pin chip, that pin 7 goes to 8. Is it not actually pin 7 to pin 9?

I have also been doing some quick readings on a AUJL (P/N 16150499) for a 89 350 TPI. It appears at this stage, to be nearly identical to a AUJP netres, with in 10 to 100 ohms and is probably identical to the AXCR displayed earlier. Probably because the resnet's have the same part number! I'll do some more precise checks and update the document.

I also have two V6 memcals here. One is for a 3.1 Beretta and an Aussie 3.8. I'll record these ones also just out of interest.

Cheers,
Col.

Cyclone17

I've duplicated the AUJP resnet and put on a programmable eprom that is fully functional in the 730 ecm on a 350" tuned port engine. I still have about five of them laying around unused, as well as about a dozen 730 ecm's. If anybodys interested drop me a line.

jwscab

bald_noggin,

the resistor circuits that are in this post are from a v6 memcal, so the pin 7 to 8 and pin 7 to 9 are one difference between a v8 and v6 netres.

drb930

Subscribed!

Saar

still looking for info here? i have a few 7727/16149396 MEMCALs lying around..