speed sensor/fuel pump
#16
Rennlist Member
Sorry to be off topic. Bob, any concern about timing accuracy with only 4 references per revolution? That's only giving each cylinder its spark timing within 90 deg every cycle. Plus belt slop/hysteresis. It's obviously working no problem but I'm just curious from an armchair engineering perspective.
I personally think you can have too many reference readings and it can lose count especially the older ecu's, one of the reasons the sync is there so it can start counting again. Where my motor has trouble is at load speeds but that is primarily a mass issue as I have an aluminum fly wheel and a knifed crank which limits how much inertia mass I have. And on top of that I have a 4 puck clutch, so it can get interesting in the pits
#17
Rennlist Member
My concern would be spark timing angle rather than fuel. Especially in those low speed situations where the next engine speed update is still waiting up to 90 crank degrees plus has an inherent error from the belt stretching slightly (assuming that the timing table was tuned during steady states, no quick RPM changes). The Motec is really capable of course but it will only be as accurate as the signal it gets. Sorry if I come off as being critical, I'm in the process of building my 951's setup and I'm just interested in the pros and cons of different schemes. I agree that too many teeth are an issue, I went to a 36-1 on my red car because the ECU doesn't work well with the raw signal from the stock speed sensor... needs outside help for dividing the signal into smaller groups every sync. Not to mention the integer/division issue inherent to the number of teeth per sync vs the number of cylinders firing per cycle. Luckily a lot of the newer stuff is compatible because of the demand from Audi guys and improvements in processing speed versus cost/availability.
#18
Rennlist Member
My concern would be spark timing angle rather than fuel. Especially in those low speed situations where the next engine speed update is still waiting up to 90 crank degrees plus has an inherent error from the belt stretching slightly (assuming that the timing table was tuned during steady states, no quick RPM changes). The Motec is really capable of course but it will only be as accurate as the signal it gets. Sorry if I come off as being critical, I'm in the process of building my 951's setup and I'm just interested in the pros and cons of different schemes. I agree that too many teeth are an issue, I went to a 36-1 on my red car because the ECU doesn't work well with the raw signal from the stock speed sensor... needs outside help for dividing the signal into smaller groups every sync. Not to mention the integer/division issue inherent to the number of teeth per sync vs the number of cylinders firing per cycle. Luckily a lot of the newer stuff is compatible because of the demand from Audi guys and improvements in processing speed versus cost/availability.
One of the reasons I went with the set up I did is I did not want a my sensor sticking out on a bracket. A lot of the sensors key off of a toothed wheel that go in front or behind the engine pulleys, and the sensor is on a low mount bracket. I have seen these systems knocked out of alignment due to an off. My system is keyed off the cam gear and all my sensors are hard mounted to the cam gear cover which moves them from a low mount to a high mount position hopefully removing it from taking a hit during an off. I could easily double the magnets I have mounted in the cam gear but I have not seen the need to. The Motec is an amazing ecu that has a lot of capabilities I will never use, it also has the flexibility to adjust to different sensor configurations. The ecu requires a minimum of 1 reference signals per revolution and I am using 4. There is one sensor out there that combines the ref and the sync signal into one. Be interested to see how that one is set up.
#19
Rennlist Member
My concern would be spark timing angle rather than fuel. Especially in those low speed situations where the next engine speed update is still waiting up to 90 crank degrees plus has an inherent error from the belt stretching slightly (assuming that the timing table was tuned during steady states, no quick RPM changes). The Motec is really capable of course but it will only be as accurate as the signal it gets. Sorry if I come off as being critical, I'm in the process of building my 951's setup and I'm just interested in the pros and cons of different schemes. I agree that too many teeth are an issue, I went to a 36-1 on my red car because the ECU doesn't work well with the raw signal from the stock speed sensor... needs outside help for dividing the signal into smaller groups every sync. Not to mention the integer/division issue inherent to the number of teeth per sync vs the number of cylinders firing per cycle. Luckily a lot of the newer stuff is compatible because of the demand from Audi guys and improvements in processing speed versus cost/availability.
https://rennlist.com/forums/944-turb...ing-gauge.html
Last edited by Tom M'Guinn; 12-11-2017 at 02:59 PM.
#20
#21
Rennlist Member
Recreating it functionally with off the shelf parts doesn't seem like it would be that difficult. Filter out the negative cycles and anything over a certain voltage, and feed the filtered signal to a schmitt trigger to turn it into a square pulse. When the last S100 dies, maybe I'll give it a try
#22
Recreating it functionally with off the shelf parts doesn't seem like it would be that difficult. Filter out the negative cycles and anything over a certain voltage, and feed the filtered signal to a schmitt trigger to turn it into a square pulse. When the last S100 dies, maybe I'll give it a try
You have the right idea. Another way would be overdrive it into a receiver chip that would square up the edges and clip the peaks.
Have done that to turn sine into ECL clock before.
#23
Rennlist Member
It's on my someday list. If nothing else, it might be helpful for guys who want to use the stock sensors with a stand-alone, perhaps with some active processing that produces a 60-2 pulse since 132 teeth isn't the norm.
#24
Rennlist Member
Most higher end standalones support the 132+1 pattern, FYI. A lot of Audi builds have similar setups and the vendors have added support for this style of pickup.
#25
I had some high rpm trigger issues with VEMS and stock 132+1, Bosch sensors and Facet Sensors at different gaps would run solid, but Facet sensors worked for some reason. Guessing because of the high freq of the wave above 5000rpm and 132+1.
#26
Rennlist Member
At 6k rpms, that's about one tick every 76 microsecond or 13,200 ticks (peak to peak) per second. That's fast for sure, and "maybe" faster than some of the old school ECU's can handle, but VEMS literature says it uses a "modern RISC microprocessor" -- so in theory it should be able to handle 76uS ticks in its sleep. If it's running at a modest 1GHz, for example, it would be able to run bout 76,000 single-cycle instructions between each tick of a 132 tooth flywheel. You'd "think" that would be more than enough processing power to deal with 132 teeth, even at redline. VEMS advertises it handles 135 teeth flywheels too, further suggesting speed isn't the real issue. It's also worth pointing out the original DME runs a vintage 8051. The FRWilk schematics show it running on a 6MHz processor, which means it would only have about 456 clock cycles between ticks at 6k -- a testament to the Bosch engineers who programmed that box, back in the day when every bit of memory and every clock cycle counted. Triggering issues are more likely the filter/conditioning I'd guess. The stock speed/ref sensors produce a raw A/C signal that is spiky and dirty, and all that increases with RPMs. The S100 chip is designed for exactly these sensors, and does a great job producing a clean digital pulse from a very messy analog signal. Trigger circuits on after-market ECU's like VEMS are designed to handle a wide range of sensors/speeds, and my hunch is that it just isn't converting the raw sensor signals into digital pulses as cleanly/perfectly as the s100.