Questions & answers:
Q: Is this setup on the pic OK?
A: Yes, the switches are mounted in right direction, electronically. However you should do something so they do not break off due to vibration. The IGBT might need some heatsinking too, depending on coil type and dwell configuration.
Q: Can HALL and TPS share +5V?
A: Yes. (MAP also)
Q: What parts do I need to get to install DD FlyBack on my V3.1?
A: 6.8V or 20V SMB footprint diode which was just added to rescue kit
Q: So I need to solder 6.8V or 20V SMB footprint diodes on places for diodes
A: NO !!!
The FlyBack SMB diodes that are on the schematic must always be there, there was never an option about that
mount 0V (=wire), 6.8V or 20V SMB zener (transient protection diode) on the places that are R2512 pads in the schematic: perfect footprint for mounting SMB diodes. Just follow GenBoard/Manual/DDFlyback
Q: how can I connect VR sensor. What goes to VR+ and what to VR- side of EDIS sensor?
A: VR- goes to GND. VR+ to trig1 (primary trigger) input. You need to short the onboard jumper (with max 10k) so the VR+ signal makes it's way to the LM1815 input. For more detailed answer ask on the relevant page. (maybe InputTrigger/AnalogSide)
A2 (same): The VR sensor should be connected with one side (vr+) to the trig1 input, the other 2 pins should be connected to GND (one is vr- signal, the other is cable shield) - DB
Thanks Dave
Q: For lm1815 circut I need to solder D14 (3.9V diode) and short SJ6 right? Anything else?
A: The components you need to mount are listed on the BuildProcedures/SectionThree page, please check schematic to be sure - DB
Q: I can't get lightbulbs connected instead of injectors to work. How can I test if hardware on v3.1 is ok? I have connected input trigger and I have rpm readout in MT but it just won't fire injectors (lightbulbs). My h[0] tables is h[0] = 01 02 04 08 00 00 00 00 and config.alternate = 03 and I am using injectors a,b,c and d so that shouldn't be the problem, and I have FETDRIVER_INVERTING in my make. I have constant 5V output on FET gates no matter how RPM or TPS change.
What can I check?
A:
-MAP signal (with MAP=0 pulsewidth can be zero or very low)
- check...I have 98 kPa on MT display
-measure the other PORTA pins as well (maybe you measure the wrong pins)
- check...I measured wright pins
-try to map h[0] to other pins (maybe you already ruined some pins of AVR)
- I have tried maping all pins and they all react the same...when I use mdh80 I allways get 0V on the avr pin that is defined first in h[0] table, is this OK???
-check the injector pulsewidth on LCD or in MegaTune
- check...it is changing with rpm and it is arroun 9 ms when rpm is 0
-try mdh.. see testing on GenBoard/Manual/DigitalOut
- check...When using mdh80 I have 0V on AVR pin (good!) that is first in h[0] table , but I don't have change on FET driver output it is 0V when input is 5V (good) and 0V when input is 0V (bad! Should be 12V). Does that mean that FET drivers are gone again?
-There is a good chance it's output is shorted. Did you check that with DVM when the chip was unsoldered? Shorted output could have been the reason for the previous smoke (it never smokes just because).
- check...it is not shorted, I have check all outputs on FET drivers and neither has connection to the gnd. Did the same on the FETs signal inputs with same result. Resistance betwen all 4 FET driver outputs and FETs are 1k.
-Also check FET driver supply, just to be sure.
- check...I have 4V on their pin 7
-This is the problem. It needs 6V minimum. Probably the diode that powers it from 14V is broken. Short the diode.)
- YES...:-) that was the problem I've tested all FETs with mdh and they all work :-) (I've even realized how mdh works..80, 90, A0, B0...:-P)
Q: Can I use diode that I've got from Marcell for D14 (needed for VR) on the place D6 (FET drivers suply)or I'll just short D6?
A: D6 type is originally LL4148 (small current SOD80 footprint diode) but a short will do just fine. Really, the diode was replaced by a short on v3.2 in fact. The diode was originally intended to protect the FETdrivers from a reversely applied supply; but we found that if the big transient-eater diode fails due to such an abuse, the board dies anyway.
Q: I have measured voltage at my V3.1 input and it was arround 10V, but in MT it says that it is 14V, then I tryed to power it with 12.5V, and in MT it says that it is arround 17V. Is this OK?
Besides the reading it causes no problems (changes dwell and injpwm duty somewhat).
Fix:
You have to follow GenBoard/InitialConfig, see batt_cal. The variable name and LCD text (also in global.h) is a very good hint that this is related to VBatt: \n
uint8_t batt_cal; // VBatt calibration appr.BC=default
Decrease batt_cal.
Q: (12.06.2004) I have tried to start the engine yesterday and I've find a new problem. I've done some testing before with light bulbs and everithing worked OK but wbo2 sonde was disconected. I've tested wbo2 and it worked OK, but yesterday when I connected everithing together there was a problem. With engine standing still rpm was reading 800-1200 rpm and injectors were working. When I disconected wbo2 everithing was OK, then I tried again with wbo2 connected and there was the same problem but then I disconected trigger and everithing was ok again. So there is a problem when wbo2 and trigger sensors are connected, what can I do to solve it?
A:
-It is most likely caused by bad grounding. Don't reconnect WBO2 probe to a running engine. (disconnect at idle should not cause problem, but better shut off engine).
- I didn't start the engine, didn't want until I fix the problem. So do you think that I didn't connect ground and analog ground the right way or I don't have good ground all together?
-I really don't want to guess how you did it. Check GenBoard/Manual/InitialTesting/VerThree on how it should be done. You can split GND-part to a separate page because of importance (so that part can be referred easily and more verbosity is allowed). If you're not sure of the wording (which is relatively simple: connect GND-s at the battery), make a drawing of your wiring, which shows the ground wires, I will review it.
-If the GND and GND5 are both properly connected with separate wires to the battery (-) you can cut the onboard connection between them (arch next to the FETdriver chips; this is only installed for safety: so bad install just causes noise instead of smoke). A cold WBO2 heater consumes 3..6A current and that can cause serious GND fluctuation without proper grounding.
- I've connected all gnd5 and gnd wires directly to the battery. OK. But problem remained.
- When I turn on the power vems react like the engine is running (wbo2 is on and fuel pump relay is on)
-So you have trigger noise.
- How can I get rid of it?
-Describe or draw trigger ground and trigger source and trigger setup in general.
- It is a VR sensor, trigger+ is connected directly to the EC36 pin 27 and trigger- is connected directly to the gnd on the battery. That is not the way to do it, trigger GND should at least have a separate wire to GND on the connector. Otherwise the small voltage drop over the GND wire feeding the logic inside the box will trigg the box. But at the first sign of trigger problems I suggest that you connect the trigger+ and trigger- at the throughholes available for this purpouse next to the VR chip's. By doing this you eliminate two sources of noise.
- I've just tried that (connected vr+ to JP7 and vr- to JP8) but the problem remained the same. It reacts like it is too sensitive and when I turn power on it gets some false signal and then it won't stop. Could it be that something arround LM1815 chip is damaged? Maybe something like D14 diode? How can I check this?
- Then I tried to turn my test trigger wheel and when the wheel stoped vems turnd of wbo2 and FP and then it was all OK no false rpm readouts or injector pulses. I turned power off and on again and the problem was back.
-This sounds like trigger noise in any case.
- And then one of IGBT's smoked out.
-Something is very wrong here. If it is a table-setup, the IGBT cannot smoke with lightbulbs (not coils) connected. If it is an engine-setup, IGBT cannot smoke, it just blows the 5A fuse that is used in the coil power (usually +14V side, separate 5A fuse for each coil). But if it's engine setup, you connected a test-trigger-wheel? (that alone should not hurt by itself, if the engine is otherwise stopped)
- It was engine setup but it had one 16A :-( (it is 5A now) fuse on +14V side of coil since EDIS coil has only one +14 input for both coils.
- Could it be maybe that IGBT was damaged last time when my FET driver smoked and now he coused troubles?
-No.
- I had wasted spark coil and spark plugs (grounded) connected all the time (can you elaborate? that is not valid operation) and they seemed to work OK but one side of coil seemd to work more frequently.
- It was wasted spark coil from EDIS connected with spark plugs that were connected to the battery ground with wires. I've connected lihgt bulbs now just to be shore and I still have the same problem.
-Make sure you have uptodate config on the project page, write firmware version (see GenBoard/UnderDevelopment/FirmwareChanges) and organize the page thematically (firmware, link to config.txt, trigger, power, etc..) not chronologically.
Q: Here is the problem: my vems V3.1 when it has connected trigger, wbo2 sonde and light bulbs instead of injectors and coils is geting false rpm info as soon as I turn ON the power. Bulbs are turning on and off and rpm readout is around 800-1200rpm. Then I turn trigger wheel to arround 300rpm and when wheel stops then everithing is ok: bulbs are off and rpm is 0. False rpm readout does not appear if trigger or WBO2 sonde is disconected.
- GND and GND5 connections: I have all wires meet at the battery ground, so every GND wire from EC-36 is going to batery gnd and every sensor gnd wire is going to batery gnd.I have cut onboard connection from GND to GND5. I have tried to connect VR+ directly from VR sensor to JP7 on board and VR- from sensor to JP8, but it didn't help.
- Trigger: I am using LM1815 chip so shorted SJ6 and SJ1 (down side of the board) and solderd D14 diode. I have tried VR sensors from OPEL, FIAT and FORD and result is the same all the time.
- WHAT is wrong? How can I check if everithing around LM1815 is OK, it feels like it is too sensitve? Please help guys I am runing out of ideas.
-This is a very strange problem indeed, SJ6 and SJ1 is correctly jumpered. By jumpering SJ3 to VCC you can increase the threshold value. But I feel that that would be a band aid in this situation, you must have a more serious problem with the trigger system. I suggest that you check that pin7 of U11 isn't shortcircuited to GND through C38 or R87. (Should be high ohm)
- Pin7 is not shortcircuited to GND I've measure resistance of 1 Mohm ---> sorry I've measure wrong value it is 1.1 kohm when sensor is not connected and 1Mohm when sensor is connected is this a problem???
-I haven't actually measured on an actual board, but the 1.6Mohm resistor makes me belive that anything below 1Mohm is wrong and 1.1k _will_ cause false triggs! I think that you found what's wrong with your board. Unsolder C38 and R87 and R87. Measure resistance after unsoldering each component. If resistance is high ohm (probably open circuit) with them unsoldered fit a new capacitor and unless you can easily source a 1-1.6Mohm resistor locally (not likey to be in rescue kit) measure that the R87 is ok (1.6Mohm) when unsoldered from the board. Then fit it again and check resistanse from pin 7 to gnd again. If it's still 1.1k with them unsoldered you should remove the chip and fit a new one.
- Jörgen I've just find out that I have crapy DVM + very low electronic knowledge. My DVM shows diferent values depending on scale I use so when I use 20Mohm scale it shows 1Mohm and with 2000ohm scale it shows 1,1kohm no mather if VR is connected or not :-(. I am really sorry for the wrong informations. But I have one good news: yesterday I just tried to touch field conected to pin 5 with wire conected to EC36-28 (+5V for sensors) and I didn't see any change with my problem but today I've soldered a bridge on SJ3 (to +5V field closer to pin1) and now everithig works PERFECTLY. I would just like to ask what is the downside of this kind of setup? Why did you call it a band aid?
I also suggest that you measure the AC voltage at pin 3 of U11,(measure between GND very close to the chip and pin 3). If you have more then a few mV at this pin you have a problem. Please write what you see here when there is oscillation and when there is no oscillation.
- I've measured arround 3mV between pin 3 and GND near chip when it is still and the same when it is geting false rpm and trigering light bulbs, but then I tried to measure voltage between pin 3 and GND rail that is going on EC36 pin5. I've measured 3mV when it is still, and 16mV when it is getting false rpm.
Measure the above ground-noise relative to VR- (or VR- relative to LM1815 GND)
I'm almost certain that your VR- is not connected properly to GND. (maybe to GND5? or VR- and GND meet near battery (bad) instead of near EC36 connector (good).
-But back to your problem there could be some small peaks that don't show up in the AC reading, it would help a lot if you could check the wave form with a scope. if there is noise C30 (which helps shape the input signal) may be broken, try replacing it if you don't have a scope 1nF is will give you a 16kHz filter frequency, you could easily have a bit more here for most applications. As an example my engine would run with a 33n capacitor, but with multitooth you are probably limited to 2.2n. Also check that R55 is around 10k, if it is shorted the filter isn't working. Fitting a 100n for C30 would be good for testing if you don't have scope, if it false triggs with that we can start looking elswhere (to low rpm limit to allow the engine to run with multitooth, but good for bench testing).
-If possible replace C39, it's the chip's supply cap. If that is broken there will be strange results. Also fit C79, it's one of the panic noise suppressor capacitors, it will improve the supply. BTW, only change one thing at a time! You will not know what fixed (or hid) the problem othewise.
-If we go back to the SJ3 jumper again, in the open setting the threshold voltage will be selfadjusting. When jumpered to VCC the treshold voltage will be 220mV, unless the voltage swing to 220mV the chip will ignore the next 0v crossing. If it is jumpered to gnd the threshold value will be 0v (+-25mV) and pretty much everything will trigg the chip. In the Open setting the threshold voltage will be around 80% of the peak value on Vin.
- I've tryed to short SJ3 so that pin 5 is shorted to VCC, but it didn't help. It definately should have eliminated te problem (knowing about the 16mV AC reading.)
This may seem strange but is power connected to pin25 on EC36-Yes
And if all coils injectors and WBO² are un-pluged will it still power up-Yes it will
Also check if the VR pickup has a short to ground. I see that you tested more then one so I guess not.-No it has 772 ohm resistance from VR+ to VR-
If you just jump jp12 to TRIGG2_CONN will it still see RPM at powerup? - I've connected jp12 and jp13 (is that right?) and it didn't help. was this with or without the VR connected to jp12,jp13 too? -It was without VR connected to jp12 and jp13, I only have one VR connected to tigg1 conection (EC36-27) and gnd.
also is this powered by a car battery or a powersupply? -it is powered by a car battery
I've soldered a bridge on SJ3 (to +5V field closer to pin1) and now everithig works PERFECTLY. I would just like to ask what is the downside of this kind of setup? Why did you call it a band aid?
Because connecting pin5 of LM1815 to VCC disables the adaptive hysteresis. It might work, but adaptive hysteresis is good for suppressing noise.
This is getting to the point that should not be started without scope.-sorry I dont have a scope
- diode is on the picture top-right corner. Probably 5V1 zener (but shouldn't matter too much in this case).
- 1-103 10k
- 3-103 10k (1 and 3 are input part)
- 2-753 75k
- 4-105 1M this is the peak detector together with the (220nF) cap next to it. Can be measured with DVM (DC 200 .. 2000 mV mode)
- 5-103 10k output pullup resistor
- examine the solders and reheat any that is suspicios. With the latest results, the most suspicious part is the peak-detector (part 4 on the picture, marked as "105") at the LM1815.
- use table-testing (not engine!) before the trigger is sorted (obiously)-I am using it
Tonci, why don't you notice my GND notes above that I keep stressing many times now. (do you know how to use history in wiki?)
Anyway, since this page is not properly formatted, I wrote cleaned-up checklist (GND as well) to GenBoard/Manual/InputTriggerHardWare
I've managed to turn ON THE ENGINE!!!:-)))...and it feels NICE!!! It will need a lot of work to set it up but that will be a joy :-))))
I've left pin5 on LM1815 conneted to VCC and didn't see any problems with that setup although I didn't want to rew it above 3000rpm since the engine is out of the car. What sort of problems I can expect with this setup?
Check the LM1815 datasheet. IIRC this sets relatively high fixed hysteresis (arming threshold) so most likely problem is "no triggerring" at very low RPM. If the VR signal amplitude is high enough, it might work even at lowest RPM on a cold winter morning cranking.
Well tonigth was very cold winter night and there was no problem with that so it migth be OK, but I'll look into it. I'll folow Jörgen's advices.
I didn't use shielded cable for VR sensor, but I plan to use one. Now I need to install engine in car....I am so happy :-))
Very good, but take care to progress in small steps. Don't carry away WOT enthusiasm. Especially with ignition at high power.
Don't worry I'll ask questions as allways.
j table has 4 digit hexa values
j table in tables.txt can have either
- 2 digit hexa values (like A3) - so old tables will work fine
- or 4 digit hexa values (like A32B) (effectively same as A3, 2B being the fractional part; fractional part ignored, except internally in the controller during the ve_learn process)
- or 0..255 decimal values (if numberformat=10 given)
The 4 digit hexa values are dumped by the firmware so the result of VE learning can be saved. The LSB is not particularly important, but provides some info on which loadsites did VE learning happen. When ve learning is on, and EGO correction is enabled and working (sensor warm enough), and other conditions are met (coolant warm enough, relatively slow change of MAP, RPM and ego-correction value), the 4 digit value changes slowly (the 4th digit weights 1/16 of 3d, that weighs 1/16 of 2nd, etc... similar to other numbersystems). The computer needs the precise values internally, but the 2nd two digits (lower byte) is not very interesting after that.
Ok I got it, thanks for the answer Marcell
Q:I have a problem with ...It shows 2700-2800rpm?! at crancking. I thought that something was wrong with my hardware but when I put back the old firmware from 2004-11-20 everithing worked ok. Trigger parameters in config file seemd OK in both firmwares...here is config from [latest STABLE1_0 firmware] and from [2004-11-20 firmware].
- Did anybody else had the same problems with latest STABLE1_0 firmware?
- no
- Is there something that I am missing?
There is a common mistake of
- using bad global.h for make mtt
- forgetting to upload config.mtt after reflashing
Almost sure that bad values made their way to config.rpmk[] due to improper configuration after reflashing.
therefore any report must publish the output of mcd and mct commands (saved to mcd.txt and mct.txt files for example). Otherwise firmware version and config could be enough to reproduce the problem unless the above typical mistake is made.
I've just downloaded firmware 2004-11-24_v12 and I have the same problem. I have made config.mtt and tables.mtt in the file where is my latest firmware so I think that I am using right global.h. Here are my outputs of mcd and mct comands: [mcd.txt], [mct.txt].
The funny thing is that everithing is working OK with old firmware (that I downloaded 2004-11-20), even started the engine.
Is there some value in config that is doing this?