----- Original Message ----- From: "Jeff Oberholtzer" To: Sent: Tuesday, August 22, 2006 2:04 PM Subject: [dsm-ecu] Re: Knock Board I actually took a scope home from work over the weekend, hoping to get to the bottom of this. What I found was: 1) The 1990 tach pulses are definitely inverted from the 1991+ pulses. The 1990 tach driver pulls the output to +12 volts when the coil fires, whereas the 1991+ tach circuit in the transistor inverts that, so that it pulls to GROUND when the coil gets fired. 2) I was wrong about some 1990 ECU's using the same 2-diode input circuit on pin #109. That was a Colt ECU! Doh! It had the 4 vertical boards, like a 1990 DSM, but the tach circuit of a 1991+ DSM. 3) The 1990 ECU tach circuit is made to only accept +12 volt pulses. The ground-going pulses from my GVR-4 never made it past the first stage. 4) But here's the REALLY weird part: the tach pulses to the knock board were there ALL the time, no matter whether the ECU was receiving tach pulses or not! So, I can only assume that the muting pulses are being generated by the ECU from the CAS or ignition pulses somehow, although the knock chip has a separate CAS input. What this means is that I was wrong, the knock will not be effected evidently by using the wrong ECU in your car. But, it also means that whatever the tach pulses are needed for in the ECU is NOT receiving any tach signals at all when you use the wrong ECU. Although I'm sort of back to square one, the good news is that I've also developed a cheap circuit that will convert the 1990 tach pulses to be 1991+ ECU compatable, so that the 1990 car's tach will still work, and the ECU will see the correct pulses, for whatever it is using them for. Jeff O. --- In dsm-ecu@yahoogroups.com, "spoolamatic" wrote: > > After the tread about the blanking pulse for knock ,I finally got a > chance to check it out and had a ecu I was fixing so I hooked up > enough goodies to it so it would work and this is what I found going > to the KB . Not sure if any body has ever put a scope on it but this > is what I found . > > The trace in the file I put up has the top trace on the #54 ignition > transistor wire and the bottom one is on the 4th pin on outer edge > next to case going down to the main pc.(#5 on bottom KB). This pulse > was the same duration (946us)for all rpms at least to about 4500 rpm . > (couldn't turn cas any faster). > > Pins 1 & 2 on top left in pic was a inverted CAS signal with #1 being > a little rounded off maybe some capacitor coupling or something . > They changed very little with rpm's but pin 1 was low for about 1.5ms > after the coils should have fired ,this pulse (not pictured) would > change some with rpm but not much . > Pin #7 from left on bottom (#5 going to main pc) had the same signal > as the transistor (#54)signal . > > All this was done with wire 109 (tach) not hooked up and when it was > I didn't find the pulse going to the knock board .Not sure wich is > the real blanking pulse but one of them I would think . > > Just would be nice to know if any of the timing of the pulses are > code related . > ----- Original Message ----- From: "Todd Day" To: Sent: Wednesday, August 23, 2006 4:14 AM Subject: Re: [dsm-ecu] Re: Knock Board > > >> After messing with both types I found if 109 in hooked to ground or >> hooked to power or left unhooked no error code would show up .But if you >> only feed it with half the number of pulses it would normally see it would >> show code #44 with in a few seconds so it must be used for error codes only >> as I though earlier . This is one reason swapping ecu's wont show a error . >> > > There is a hardware flip-flop in the ECU that the code samples at > 75degBTDC. Since it is a flip-flop, if the coils are firing every time, the > code will see a train like 10101010. It does check for 11111 or 00000 and > considers those cases okay. I'm guessing because the ECU itself can turn > off coils during downhill engine braking. > > And yes the knock only uses the ECU/CAS as far as I can tell also . It >> looked to me from the signals I saw that the knock was blanked for about 1ms >> after coil fire .But that would be apox 36 crank degrees at 6000 rpm so it >> migh be unblanking it instead of blanking it which would make more sence to >> me . Any thoughs on this ? >> > The crank angle sensor interrupt fires at slit to no-slit and no-slit to > slit transistions. > At no-slit to slit (75degBTDC), we approach the firing range. The code > checks for ignition failure (above), does a ton of calculations deciding > which coil to fire, when to start charging it, and when to fire it. At the > very end, the knock is sampled, knocksum is calculated, the knock gate bit > is set, and then oxy sensor feedback is done. > > At slit to no-slit (5degBTDC), we've already just fired (ie, here's your > lower bound on spark advance!), more coil energization calcs are done (which > I've never quite understood fully), the knock gate bit is cleared, overrev > is determined, air mass per cylinder is calculated, fuel amount is > calculated, and injectors are turned on. > > So knock is only listened to for 20deg starting at 5degBTDC and extending to > 15degATDC (aka 75degBTDC). > ----- Original Message ----- From: "Bill Sundahl" To: Sent: Wednesday, August 23, 2006 12:47 AM Subject: [dsm-ecu] Re: Knock Board > My favorite topic. > I want to point out my research that I dumped in the files section a > while ago, but then never announced (I never felt like it was quite done). > http://autos.groups.yahoo.com/group/dsm-ecu/files/1g/Hardware/KnockChip.doc > I apologize for missing this thread last month, I should tune in more often. > > Most of what you found here seems consistent with what I saw. > Your 946 us pulse is the blanking pulse, and as you found, is triggered > by the ignition pulse from the CPU. As far as I know, the only thing > here that is software related is the ignition pulse itself. > > Do I understand correctly that one of the signals (CAS?) went away > when you fed a pulse into pin 109? This is something that I never tried > as that pin did not appear to effect to the knock chip. Many of these > signals (including the tach 109) pass through IC4, which largely seems > to be a 5V to 12V level converter, but also has some logic. So something > could be going on in there that would effect the knock chip. > > Does sonic knock detection fascinate everyone here, or is it just me? :-) > -bill > >> >> Posted by: "spoolamatic" spoolamatic@yahoo.com >> >> spoolamatic >> >> >> Mon Aug 21, 2006 3:58 pm (PST) >> >> After the tread about the blanking pulse for knock ,I finally got a >> chance to check it out and had a ecu I was fixing so I hooked up >> enough goodies to it so it would work and this is what I found going >> to the KB . Not sure if any body has ever put a scope on it but this >> is what I found . >> >> The trace in the file I put up has the top trace on the #54 ignition >> transistor wire and the bottom one is on the 4th pin on outer edge >> next to case going down to the main pc.(#5 on bottom KB). This pulse >> was the same duration (946us)for all rpms at least to about 4500 rpm . >> (couldn't turn cas any faster). >> >> Pins 1 & 2 on top left in pic was a inverted CAS signal with #1 being >> a little rounded off maybe some capacitor coupling or something . >> They changed very little with rpm's but pin 1 was low for about 1.5ms >> after the coils should have fired ,this pulse (not pictured) would >> change some with rpm but not much . >> Pin #7 from left on bottom (#5 going to main pc) had the same signal >> as the transistor (#54)signal . >> >> All this was done with wire 109 (tach) not hooked up and when it was >> I didn't find the pulse going to the knock board .Not sure wich is >> the real blanking pulse but one of them I would think . >> >> Just would be nice to know if any of the timing of the pulses are >> code related . >> > > ----- Original Message ----- From: "spoolamatic" To: Sent: Wednesday, August 23, 2006 10:44 AM Subject: [dsm-ecu] Re: Knock Board --- In dsm-ecu@yahoogroups.com, "Todd Day" wrote: > > > > > After messing with both types I found if 109 in hooked to ground or > > hooked to power or left unhooked no error code would show up .But if you > > only feed it with half the number of pulses it would normally see it would > > show code #44 with in a few seconds so it must be used for error codes only > > as I though earlier . This is one reason swapping ecu's wont show a error . > > > > There is a hardware flip-flop in the ECU that the code samples at > 75degBTDC. Since it is a flip-flop, if the coils are firing every time, the > code will see a train like 10101010. It does check for 11111 or 00000 and > considers those cases okay. I'm guessing because the ECU itself can turn > off coils during downhill engine braking. > Wondered why it was this way . > And yes the knock only uses the ECU/CAS as far as I can tell also . It > > looked to me from the signals I saw that the knock was blanked for about 1ms > > after coil fire .But that would be apox 36 crank degrees at 6000 rpm so it > > migh be unblanking it instead of blanking it which would make more sence to > > me . Any thoughs on this ? > > > The crank angle sensor interrupt fires at slit to no-slit and no-slit to > slit transistions. > At no-slit to slit (75degBTDC), we approach the firing range. The code > checks for ignition failure (above), does a ton of calculations deciding > which coil to fire, when to start charging it, and when to fire it. At the > very end, the knock is sampled, knocksum is calculated, the knock gate bit > is set, and then oxy sensor feedback is done. > > At slit to no-slit (5degBTDC), we've already just fired (ie, here's your > lower bound on spark advance!), Not sure if you got that this pulse was the same at any rpm even if you just passed a slit and stopped .(dont remember if L to H or H - L ) I based the 36deg off of 6000 rpm =100rps = 1rev per .01sec = 360degrees per .01s = So .001 sec = 36deg . But this would make the pulse last for different crank degrees at different rpm's in contrast to your comment below so my presumtion must have been wrong . more coil energization calcs are done (which > I've never quite understood fully), the knock gate bit is cleared, overrev > is determined, air mass per cylinder is calculated, fuel amount is > calculated, and injectors are turned on. > > So knock is only listened to for 20deg starting at 5degBTDC and extending to > 15degATDC (aka 75degBTDC). > I guess you mean 75 degrees before bottom dead cenrter . But this makes me ask if at WOT or other say running at 20deg of timing it does not listen for knock for 15deg ? ----- Original Message ----- From: "Carl Morris" To: Sent: Wednesday, August 23, 2006 11:22 AM Subject: Re: [dsm-ecu] Re: Knock Board > spoolamatic wrote: > >> > So knock is only listened to for 20deg starting at 5degBTDC and >> extending to >> > 15degATDC (aka 75degBTDC). >> > >> I guess you mean 75 degrees before bottom dead cenrter . > > I think he was referring to 15degATDC being the next-cylinder-to-fire's > 75degBTDC. I was just thinking about how only two cylinders fire per > revolution and getting confused, but then I remembered the waste spark, > so I think this makes sense to me. > >> But this makes me ask if at WOT or other say running at 20deg of timing >> it does not listen for knock for 15deg ? > > Are the sound waves "ringing" the block for a bit of time after > the actual knock event? > > -- Carl >