What to measure and what to think about when running secondary trigger VR on a 3.2 or 3.3
- is the correct solder jumper installed?
- SJX
- is the sectrig LM1815 pin7 adaptive threshold RC circuit soldered on fine ?
- check the schematic
- Don't forget to look for missing components, especially the 0805 resistor from LM1815pin7. 100 .. 180k is fine, or even a 0k short.
- What signal does the LM1815 output to the AVR on IC3?
- Should be a 0/5V square waveform, only 5V for a short 60usec after the input negative going edge
Make a report of your measurements. All pins.
Report reasonably please
- NEVER forget the version serialnumber history of the board
- did you make the sectrig part ? You likely forgot a part from the sectrig part. It involves only a few components, just check the sch and the components. What part connects to LM1815pin7 ? Something is in the air (with 1 pad) ?
- take pics of the LM1815 involved area
- Make a report of your measurements. All pins.
If you start to write these few things systematically, it's very likely the problem is revealed in no time. Rumbling on irc you waste much time, without result. This is very important. Reports without the most basic info, or to the wrong place is useless.
Emil's board - reply after the report requirements stressed on irc
v3.2, standard, Earlier trigged by primary trigger VR in a car.
- It seems my board works both with secondary=HALL and VR driven by soundcard.
- configs: http://kombi.ulkhyvlers.net/vems/audi/
- Emil Larsson made [log] ([same log more readable format].
- "SECTRIG 82 81 PRIMTRIG" and "SECTRIG 81 PRIMTRIG" are both fine. Any other value coming after sectrig is a mistrig.
- We made a [new 1.0.29] that is more graceful for bad or noisy secondary trigger, see [log]. For suspicious sectrig, not firing (0x78, 'x' means no sync, no fire) is just safer.
- verify: v3_firmware_1.0.29.zip (537kb) 2005/12/27 17h45 ;;; md5sum of vems.hex is: b2264bd1abd5ff910eb551b68f50c682 vems.hex length: 271811 date: Dec 28 00:21
- Latest problem: playing back slow trigger (100-200 rpm) with trigger_tooth > 0 wouldn't sync up.
- configuring trigger_tooth=00 with the same wav being played it syncronizes right away.
- while leds are flashing, i can change trigger_tooth back to 28(decimal) and they keep flashing
- 200 rpm syncronizes after a while sometimes. 150 rpm not.
- playing back 250 rpm or more seems to work with any trigger_tooth setting
- even tried 2000 rpm without any problem
Miska's board - onboard trigger setup is good now
- V3.3, with auditrigger setup (complete board)
- Tried board in another engine and it idles now
- if rpm is raised even a bit more, some sparks are lost. It's always same 2 cylinders that miss the sparks. Do you know which two ? What does "sparks are lost" mean ? It sounds like misfire ? Or you don't even see the LED flashing and suspect a firmware problem ? If so, is the wheel_error counter increasing ? The mdd04 log looks reasonable ?
- 5 leds are connected to ign channels, and last two in firing order are not flashing.
- A snippet from mdd04 log: first PPPPP\n for some time and when touching gas pedal and it changes to PPPPP\nPPPP\n\nPPPPP\nPPPP\n\nPPPPP\nPPPP\n\n
- in the sectrig there was a very little positive bump between real signals (~0.5v max or maybe even smaller)
- VR-cables have 2 signal wires + shield around. Shield is connected to ground pin in motronic connector and according to factory wiring chart it is connected to common grounding point in engine block and shared between every shielded cable
- measured with ancient scope near 10V peak voltage from secondary trigger and arond 2.5V peak from primary
- measured VR-sensor resistance was around 850 .. 900 ohms
- 10k pullup in sectrig
- if rpm is raised even a bit more, some sparks are lost. It's always same 2 cylinders that miss the sparks. Do you know which two ? What does "sparks are lost" mean ? It sounds like misfire ? Or you don't even see the LED flashing and suspect a firmware problem ? If so, is the wheel_error counter increasing ? The mdd04 log looks reasonable ?
Apparently, the 5cyl audi has more uneven fuel or air distribution than I thought ( if the lost spark is caused by those cyls leaning out - note that this is very low load, should not be dangerous). Possible causes:
- uneven air distribution between cylinders. Might be the case. Could be solved with individual cyl fuel trimming. Might be hard, if behavior heavily depends on RPM (in worst case, if a certain cyl gets less than average air at RPM1 and more than average at RPM2) it might be tougher.
- uneven fuel distribution, pressure drop in fuel rail. This should only show at very high power, so unlikely for this case
- injector flowrates can be different
- sometimes injector flowrates are within 1% for the fully open case, but deviate shockingly (>6..10%) for a short 1 .. 1.5 msec pulse because of differences in injector-opening speed. Seen the effect for siemens injector-sets. Power-flyback helps somewhat. However, unlikely for this case, because this is worst at idle, and gets better as injector pulsewidth is raised because of MAP or RPM increase.
- Old stuff below
- No any reasonable output signal on output pins (pin10, 12) when input signal is applied Note that truth table say that pin11 must be low for pin10 to reflect RC pulse.
- Measured 100mV level from pin 12
- stays at 0V constantly, except for some noise when voltage drops in pin14
- in comparison between primary and secondary, only difference is missing output
- Input signal
- Feeding 5V logic level signal through series cap
- Measured secondary trigger input signal 900mV square wave from pin3 (-200mV to +700mV)
- Measured primary trigger input signal 970mV square wave from pin 3 (-200mV to +770mV)
- All components ok (according to sch) in visual check
- VCC (pin 8), GND (pin 2) ok
- in Pin 14, raising voltage after positive going zero-cross, falls down to 0 at negative zero-cross
- in pin 7, voltage rises when input signal is applied.
- Measured 2V level from both primary and secondary triggers with applied signal
- Repeated the tests with output disconnected (SJ7 open) and 5.4k pullup to Vcc with same results
- Problem was solved by replacing lm1815 chip. There was still some problems, IC3 voltage was 3.88V constanty. I desoldered C103 and it seemed to help (solder blob under it)
- When testing in car I had similar problems (secondary and primary signals gets mixed, lots of false secondary triggers) that Magnus is describing below. With mdd04 logging applied, trigger pattern looked like SpSPSpSPSpppppSpPpppppSpPpSpP...
- Using appr. 0.5m shielded cable with 2 signal wires (signal and signal gnd) between motronic and EC36 connectors. Both VR signals have their own cable. Shielding is connected only to motronic connector end.
- Added logs from cranking:
- 10k pullup to 5V at crankhome VR input?
- secondary is pulled to 5V with 10k resistor
- logs (with R182 shorted) [mdd0c_02.txt] and [mdd0c_03.txt] shows that LM1815 input (LM1815pin3) and adaptive hyst (LM1815pin7) is OK (for this RPM at least ;-).
- On the other hand, every crankhome pulse (2.5th cyl) gets to sectrig, instead of every 2nd (which is every 5th cyl). According to visual & DVM checking hall connection on board should be ok. But anyway, masking every 2nd signal did not work. This was fixed: HALL didn't get supply.
- Mpeippo is definitely leading the quest with new [log] (and 05) log, now with HALL supply :)
- Looks beautiful
The trigger is now good at cranking, h[2] fire order fixed now, and it sounds like it wants to fire up
It starts but loses sync soon, mdd04 pattern looks like ..PPnPPPPPnPPPPPnPPPPnnPPPPnnPPPPnnPnnPPnPPPP
- probably noise is injected to secondary trigger signal causing false trigs? Very likely
- but how can ign noise be injected if a shielded VR cable is used ?
- Is the shiled connected on one side only ?
- maybe it's bad grounding ? review ground GND and GND5 on MembersPage/PhatBob/UserGuide and search for ground loops
- try to separate the ignition and the shielded VR cable physically
- remember that it's also possible that ign-noise is injected in the primary trigger VR so review
OLD problem - the 20k pullup on VR crankhome fixed this.
Analysis of [logfile] (R182 still onboard): primtrig events between sectrig-s:\nÿ1ÿ
It's obvious (last column is the total_sectrig % 270) that there is a periodicity of 270. But why are there 5 events (at positions 0, 83, 136, 203, 218) instead of 1?
- Jorgen has seen ABY flywheels with large counterweights welded that is seen by the sensor. Maybe that's what we've just captured ?
- first of all, try to pullup the crankhome VR signal to 5V with 4k7..100k pullup resistor. This can be done outside the ECM, without unclamping the controllel. 20k is good starting value, [ result at LM1815pin3]. Should start the engine to idle, we'll see if it gets worse at high RPM.
- the smaller (stronger) the pullup resistor the better, unless it biases so strong, that it prevents triggering at a cold winter morning lowRPM cranking. Eg. 20k pullup would bias the 2nd LM1815's input signal to appr. +227mV, while 100k pullup biases with only +47mV, assuming crankhome VR resistance=950 Ohm (measured by Emil M). If you can tell crankhome amplitude at certain RPM, we can calculate strongest pullup that still allows triggering at 80..100 RPM. The measurements were captured at 200 RPM, but some at the input, while others at LM1815pin (railed).
- Emil says (forgive him the bad place, irc): I dont know what RPM emil_m's [signal scopeshot] (which is, by the way, REVERSE polarity!!!) is from, but he has appr +-2.6 Volts (5.2V peak to peak) at cranking. Than at 40% RPM, it will still be over +-1V, so we can safely bump up sectrig VR to appr +900mV. That allows a strong 4k7 or 5k6 pullup to 5V.
- Is the ignition/injection interfering ? (though the pattern is a bit irregular for that)
- Check the ground (and GND5) rules on MembersPage/PhatBob/UserGuide and describe/draw precisely how your grounds are connected. Take pictures too
- check the HW and config InputTrigger/AudiTrigger
- is the sectrig polarity allright ?
- to increase noise immuninty, you can try to bump up the adaptive threshold capacitor from 220nF to 220nF+1uF or rather 220nF+2.2uF by soldering a cap in parallel with the existing capacitor, at pin7 of 2nd (sectrig) LM1815
- but it would be nice/proper to kill the source of the noise
- are you using shielded (2 core + shield) cable for sectrig with the shield connected only at the ECM end ?
![[ result at LM1815pin3]](http://linkopingsmotorsport.se/1029/pullup_vr.gif){kind=link}
![[signal scopeshot]](http://linkopingsmotorsport.se/1029/G4.gif){kind=link}
Magnus board (aka 944_Driver on msnordic)
- V3.3 Serial 351
- U12 and R162 soldered to board and a pull up resistor is added between U12.11 and VCC as shown in the LM1815 data sheet.
- Stock Porsche 944 engine with 2 crank sensors but no cam sync. (Motronic 1.1, working fine with the sensors)
The problem is that interference from the starter gear sensor is triggering the secondary LM1815 circuit too. Cranking on the starter with the plugs removed and I get about 15 Vpp from the starter gear (quite high for cranking speed) sensor and 1.5 Vpp from the crankhome sensor. The interference found in crankhome signal is about 200 mVpp. Shielded cables are used all the way from sensors to the board.
So you are sure, that
- the crankhome VR sensor (600 Ohm?) is connected
- try to use separate shielded cable for the crankhome VR, not the same as used for the cranksignal. So use 2 shielded cables, min 2 wires in each, 1 for GND and other for signal. Connect shield on ECM side only
- try to divide the 15V crank signal with eg. 10k/1k resistors at the ECM side, and see if the interference drops to 10 (it's at the board: we'll find out sg.) or aint drop (in the cable: in this case move the dividor circuit to the sensor end of the cable)
- Do you see other suspicious things, like serious ground or supply noise ?
- No, ground and supply looks fine.
- where is the location of interference ?
- Probably at the sensors, my guess is that the crank home sensor is picking up magnetic field from the starter gear too since the sensors are sitting rather close together.
I tried to decrease the starter gear senor output by applying a 1.5 kOhm resistor in parallel with it, but that didn't do much good. The interference got smaller but not much.
Is it possible to modify the circuit around the second LM1815 chip in such a way that the interference is ignored at all engine speeds? Just connecting the U12.5 pin to VCC will probably not fix it for all rpm I am afraid since the interference will get bigger with increasing rpm.
Does anyone know how Bosch solved the problem since it works fine with the stock motronic box?
It's very important to check that the secondary VR ("crank-home") is not inverted. The falling pulse-edge at the zero-crossing must be fast and sharp. This looks okay on the scope for me.
It's also possible that the signals mix in the magnetic field. If separate cables are used, and scope shows that primary is injected into secondary even with ECM removed, than this is the case. The 135-tooth can be reversed if necessary, so it can easily be subtracted from the secondary, with only just one properly sized resistor - but hard to size the proper value for this mixing resistor without a scope. (maybe also possible, looking at the mdd04 log output, and sweeping through possible values with a variable R: the midpoint of the R-range where the interference is removed is OK).
This is probably the problem. I found an old motronic (3.1?) schematic last weekend and it has some resistors in a network between the VR-sensor inputs. http://www.vems.hu/files/MembersPage/MagnusKarlstrom/Motronic/ML31AUS.pdf and http://www.vems.hu/files/MembersPage/MagnusKarlstrom/Motronic/ML31DUS.pdf I am rather sure the starter gear sensor is connected to pin nummer 8 and 27 in the system connector and crank home sensor to pin 25 and 26. That makes sens to me and matches with the pin out for my older box. The intressting part of the circuit can be found on the upper part of page 2. I wounder that kind of circuit 'S100' is? An older version of LM1815?
I have also taken apart an old motronic box from a volvo turbo (0 261 200 022) and the input circuit for the VR-sensor is not easy to understand. I have not been able to figure it out yet.
Magneti Marelli use a PI-filter on the VR-inputs. It is 18n capacitor from input to GND. Then 18K resistor in series with signal. And this followed by 18n to GND. The effect of this is a high pass filter. It will attenuate the signal at high frequencies. As it is used for a 6 tooth crank signal it will probably be appropriate to raise the cutoff frequency for higher number of teeth. Try substituting 18n with 3n3 or thereabout. It should be placed close to the connector or outside the box as there will be high frequency currents in the signal and GND paths between the filter and VR sensor.