Measuring VCC pin8 shows only 0.3V
Testing the v3.3 units purchased recently (#316, #317, #328 & #360) with this: http://www.mmcompsys.com/download_zone/6_sensors/SEN.pdf Magnetti Marelli sensor - which is a common sensor used by the likes of Omex and MM themselves. Its resistance was measured at ~620 Ohms.
It turned out that the testbench (not the factory testbench of course, but the testbench built by the customer) with missing GND (EC36pin26) connection ruined the P259 (TPIC6A259) chip in 2 of the ECMs. The P259 chips shorted the supply for the LM1815 (VR-input-processing) trigger chip. Removing the P259 fixed the problem. (There are 10 FETs installed anyway which provide enough outputs, so the P259 outputs usually not used for a 4-cyl anyway). Grounds must be connected (GND and GND5 wires), there has never been an exception to this rule for assembled controllers (unassembled bare boards were shipped with an GND-GND5 arch in 2004, to remind the user of the importance of ground wires: but the arch is cut during assembly), more on MembersPage/PhatBob/UserGuide.
- A scope trace shows a good AC signal ~(+/-10V) on VR InputTrigger
- measure the input signal mean value, in DC setting. Eg. if the signal just slighly goes under 0V (eg. minimum is -20mV) for some reason, that would explain all. A scope in AC mode does not show this.
- Voltage measured at 9.2VAC, 0VDC
- measure the input signal mean value, in DC setting. Eg. if the signal just slighly goes under 0V (eg. minimum is -20mV) for some reason, that would explain all. A scope in AC mode does not show this.
- same config (from MembersPage/JimW/ExampleConfig ).
- please issue the (Man) mdv, mdV, mcd and mct for each and save the result, and publish. Very important to see the config that was actually read back
- Results are here [Unit 317] [Unit 328] [Unit 360]
- please issue the (Man) mdv, mdV, mcd and mct for each and save the result, and publish. Very important to see the config that was actually read back
- all have been loaded with firmware v1.0.23 and the
- verify firmware integrity (eg. prog.pl without "w" option) for each
- All verified ok, no errors
- if you unclamp any
- check resistor ad LM1815 pin7 (first check that it exists, than maybe measure resistance, should be 100..180k)
- 360 measured as 180K
- with scope, measure on LM1815pin3 and LM1815pin12. DC mode is needed to see if signal goes below GND
- Pin 3 0.0->0.3v Pin 12 0.5->0.7v
- check LM1815pin5: (the 3-pad SMD jumper) isn't it connected (solderblob) to either GND or +5V ? It shouldn't, and it's unlikely to explain anyway.
- don't clamp it unless the problem has been identified and fixed
- check resistor ad LM1815 pin7 (first check that it exists, than maybe measure resistance, should be 100..180k)
- #316 (primary&secondary VR) and #328 (VR primary/Hall Secondary) read the engine speed perfectly
- #317 & #360 show no engine speed
- there has never been a problem with primary trigger with assembled controllers earlier (they get tested, with appr +-5V signal, lower than yours), so 2 out of 4 is extremely surprising, suggests another round of careful testing and reporting details.
- publish a few seconds of the serial-output after Man mdf01 mdkff when the trigger signal is applied. This reveals if the trigger gets input.
- Logs are here [Unit 317 output] [Unit 328 output] [Unit 360 output]
After Emil's suggestion I took a VR sensor with a 500 Ohm impedence and wired it to the EC36 plug on my bench. Testing on a known working unit (316)I set the primary trigger type to coil and put the sensor on a mains adaptor, it gave a value constant reading of around 1500rpm. One of the suspect units was then tested with the same settings and there was no reading (0000rpm)
We should examine the VR waveform. This depends on the toothwheel (not the sensor: 800 Ohm should not be a problem by itself). If the amplitude at the missing tooth is way too high (max difference between amplitude of any 2 teeth should be < 15%), it is possible that one ECM triggers properly and another not. Obviously, either ECM must trigger OK on a proper-waveform VR signal. Watch wheel-error, since that might be a good indication of this. Do you have a picture of the toothwheel ?
The logs above (for 317 and 360) show B6=0000 (watch the 2nd B6=.... not the first) which means there is absolutely no trigger detected during or before the capture-period (not just lost teeth because of tooth-amplitude).
Pictures of tooth wheel
This wheel works with Omex systems and two of the v3.3s