[rusEfi] 1991 Honda CRX B16A Swap #32

[AndreyB]

my intuition is to use the 16 for timing and the 1 tooth for tdc detection, is this the correct approach here?

Sounds like a legit next step.

Yes a pre-warm would help with first run, but a step-by-step approach is
a) get good tachometer - maybe with removed spark plugs if still researching/confirming signal or weak battery
b) confirm TDC location with a timing gun and injection disabled

and only after that pre-warm and add fuel into the picture

[ZHoob2004]

Didn't manage an engine start today, it seems I'm still missing a few things.

1. I forgot to install a jumper for the ignition driver...

2. I think I need a custom trigger written up just like the accord 24+1 or 24+4 wheels, but as 16+1 or 16+4 (both would be cool, call them Honda OBD0?)

3. I didn't have a 10uf cap on hand for the VR filter so I substituted a 1uf to see if it works. For future reference for everyone: it did not.

I'll be placing an order for the parts I substituted so I can get the filter working as intended and see what that does, as well as completing the jumper setup. I also used 4K7 resistors everywhere a 5K was called for because I didn't want to wait for them. I'll be adding them to my order, but I think I'll try it as is for a while because I can't imagine that making much of a difference. Maybe I'll regret this one too.

Sensors all work fine as far as I can tell, but I'm unable to confirm the map sensor at this time (I think it uses the same one as the accord?). I'll post my sensor curves eventually. I think I lifted them all from a megasquirt-pnp setup guide if you want to look yourself.

Here's the best sniffer capture I could get. PC6 should be looking at the 4 tooth wheel and PA5 the 16 tooth, but is being jumbled by the poor filtering. I may have them switched.

[kb1gtt]

Looks like the 5k's are all on the VR circuit, these should be fine as a 4.7k. Keep in mind that you probably do not need R111 or R112. If those are installed I would expect some problems with low RPM signals. It appears you have cranking signals, so you are probably good here.

Thanks for the note about the 10uF. I kind of expected the C1003 at 220uF would have been enough bulk storage. The 10uF is just because that's what the datasheet suggests. I guess the impedance's across the board must limiting it such that it needs more localized bulk storage.

[AndreyB]

2. I think I need a custom trigger written up just like the accord 24+1 or 24+4 wheels, but as 16+1 or 16+4 (both would be cool, call them Honda OBD0?)

24+1 and 16+1 are technically not needed as custom triggers any more since these could be made as a cobination of a 24/0 or 16/0 skipped wheel trigger plus cam/vvt position logic for phase detection, see http://rusefi.com/wiki/index.php?title=Manual:Software:VVT

If needed I can add 16/4 later if it would be needed once we have clear signal.

Just to make sure I understand you right, the picture does not show 16+1 or 16+4 yet, is that still the noise you are fighting?

[ZHoob2004]

The picture is supposed to be a 16+4, with16 being the top. The 4 tooth wheel looks very clear to me, and I believe is the exact same wheel used in many other Hondas. The 16 tooth is the being jumbled by noise so I'm still waiting on parts to try and fix that.

[ZHoob2004]

Ok I'm back. I got the capacitors and haven't had a chance to install them since then. My alternator went out, my radiator sprung a leak and my front ball joints failed. Oh, and my other car is out of commission for a cylinder head job.

Anyway, got the caps in place and it works less than before. I can't get any consistent trigger signal anymore and I don't know what's wrong. I also seem to have lost my tps functionality to the ecu as it is always showing 66.6% position, even though calibration works fine.

So far, temperature sensors work, 3/4 injectors work, spark works, iacv works. Only thing lacking is ego, 1 injector and my misbehaving trigger, so what do I do?

Trigger resistors are all 4.7 or 5k (I swapped the 4 lower ones for 5k), negative sides are pulled to ground and standard capacitors are in place, what am I doing wrong? No capture because they're nonsensical.

20170205_193540.jpg

20170205_193615.jpg

I do have an oscilloscope but it's an analog techtronix 465, however much that helps here.

[AndreyB]

I wish I've documented http://rusefi.com/forum/viewtopic.php?f=3&t=621 better But that was a honda with VR sensors in the distributor should be pretty close to your setup?

Do you have R111 and R112 installed or not installed? Do you want to try removing one of them to check if that would make a difference?

[ZHoob2004]

I tried with them both installed and uninstalled, I've got pretty similar results in either case. Uninstalled I get rpm signal briefly, but don't get the trigger counts I should be getting. Leds blink pretty frequently while cranking. No plugs in for smoothness.

[AndreyB]

Can you post a screenshot of your "trigger setup" screen"? "use front only" is theoretically the way to go with VR.

(paranoia mode) is there continuity to ground? pictures show 0R resistors but not much solder.

(crazy mode) flip sensor wires just to cross this off the list?

[ZHoob2004]

Don't have a screenshot right now, but I don't have rising edge checked because that seems to lock me out of the console, requiring a reset from my pc. There should be continuity, soldered them with paste and they pulled into place just fine. I try to be conservative with my solder so it doesn't look messy. I'm hesitant to flip the lines because this matches all the factory pinouts and I had a signal before I upsized the cap.

[AndreyB]

I had a signal before I upsized the cap.

sounds like a plan to down-size the cap?

[ZHoob2004]

I guess so. I guess I'm mostly curious about how you had your accord set up if you know any more about that. I imagine you did everything according to how the manual has it, except substituting the tdc sensor (4 tooth) for the cam position.

[AndreyB]

That was too many years ago, zero recollection But if I would be doing something super not standard I would totally write it down on the forum.

[kb1gtt]

negative sides are pulled to ground

I believe you mean the negative side of the VR input. Can that connection to GND be removed? Typically the negative terminal has a weak pull to 2.5V, then when ever the positive side crosses the 2.5V is when your MCU pin changes state. I'm not sure if it will work properly when the negative is connected to GND. I seem to recall it should, but I don't know for sure.

Analog scopes are just fine for this. But remember your straps on your probes are connected to the GND pin on your plug, and they are both connected to the GND plug which means they are both connected to each other.

[kb1gtt]

Oh, also I suspect the larger cap is causing problems, because when negative is connected to GND, it need to pass GND by some mV before it will trigger. the larger cap is likely absorbing the energy, as it should be doing, and is preventing the voltage from going to far negative. I think this will all start to play nice once the VR negative is removed from the GND conductor.

[ZHoob2004]

Ok, I populated the 0r to ground because I saw that done on a number of other boards with vr triggers. Do you think just getting rid of those will set me on the right track or do I need to pull the signal to 2.5v myself? I'll probably bust out the scope tonight and make sure I have a decent signal from the distributor so I know the ecu is at fault.

[kb1gtt]

The 2.5V is automatic. You do not need to do anyting, just connect the VR wire to the VR input. The pull ups and pull down's are for hall inputs only.

[AndreyB]

The pull ups and pull down's are for hall inputs only.

Not according to http://rusefi.com/wiki/index.php?title=Manual:Hardware:Frankenso:VR_input So far we were pulling down to GND more often than not.

[kb1gtt]

I do not under stand this. I see this in the schematic. Is there a forum thread where it was determined to be wired like this?

[AndreyB]

My honda accord rusefi#5 link couple of msgs above.

My dodge neon was running like that

[ZHoob2004]

I believe my sensors to be the exact same ones as used in the accord, just with fewer teeth on the wheel. Honda guys have been swapping distributors and ecus for decades now without problems, so I don't think the oem wiring is suspect either.

Current plan is to scope the sensor by itself (no ecu) and verify I have a clean signal. Then I'll add the ecu in and probe it while cranking and see how it looks. Finally I'll probe the output of the decoder chip and compare that to see where we're losing it.

I just remembered that the trigger led is often solid when the motor isn't cranking, is that normal, or should it be off due to the vr not generating a voltage? Would my ground jumper cause that?

[AndreyB]

Which exact led?

You have two channels of the same nature i would expect both on or both off while not spinning.

[ZHoob2004]

It seems to vary. I'm not being very helpful right now because I'm at work so I'll have to make more detailed observations when I'm home. I think when I first reset the unit both are off, then it depends where the wheels land after cranking I think.

[kb1gtt]

I don't see much for details about how we came to the sensor configuration on the Neon and Honda. Looks like the Neon was hall, with no shielding, so I expect this to use those jumpers. But the 93 Honda and 95 Honda are VR. The 93 Honda has shielding which matches my expectations. The shield is grounded on the engine, and your wires are probably twisted pair inside. If there is noise on the signal, it's probably caused by a faulty shield ground. However the 95 Honda, the shield is only on one of the wires, not both wires. So I expect the un-shielded wire may want to be grounded, but would probably be OK with just being connected to the + VR lead. The + VR lead is the one that floats to 2.5V.

So I wonder, for this engine, is the VR a set of twisted pair in side a shielded cable, or is the shielding only on one of the wires to the VR? Also can you confirm the shielding has continuity with GND, and where is it grounded? It may be grounded at the engine, or it may be grounded at the ECU. It may be required that you connect the shield GND at the ECU, or it may not be required at the ECU.

Do you have a schematic for your particular engine? Even if it's a pizza box schematic, that would be better than nothing. it's hard for me to help diagnose when I'm blind.

[ZHoob2004]

I've yet to see any official documentation on shielding for the wires (I'll check my service manual), but some searching turned up a picture where it looks like all 3 pairs of vr wires are bundled into the same shield. And as a reminder, I'm using all stock wiring with the addition of a jumper harness to change the ecu connectors, so the shield should still be connected to wherever it needs without my intervention.

Anyway, for more exciting news, it seems to be working (not the engine, the trigger). I removed the W1001 and W1002 jumpers, and got to work scoping the sensors. I'll attach the scope traces but they didn't end up being that important (good background info though). Scope traces at the distributor and the ecu ends both showed textbook vr outputs with peak-to-peak voltages of about 15v and fairly clear edges, so I went on to reconnect the frankesno and get cranking again. I found I would get an rpm spike on starting cranking, which would then drop to 0rpm. The solid trigger LEDs seem to have gone away with the jumpers, so I was basically out of ideas. Eventually, while agonizing over the trigger settings in TunerStudio, I realized my real mistake. Can you see it?

2017-02-06-224809_396x699_scrot.png

It occurred to me that my "crank" trigger is a wheel in my distributor, and my distributor is turned by my camshaft which, long story short, spins at half the rate of my crank. Rather than having 16 teeth, my crank trigger effectively has 8. Upon fixing this setting, and undoing all my other misguided efforts, I found I have a proper cranking signal and decent captures from the sniffer.

I got a bit overzealous and decided to try and start the motor (simultaneous fuel, only 3/4 injectors working :p), but got a backfire out of the motor and decided I had better wait until I have a properly constructed base tune before trying that again. Hopefully I'll have the motor idling on rusefi by the end of the week. Below find attached a bunch of scope traces and captures for future reference. 2v/5ms per division, 1x probe for all traces. Noise on channel A may be due to a loose terminal (why some are on B)

20170206_203652.jpg

20170206_210032.jpg

20170206_211345.jpg

20170206_205713.jpg

20170206_212046.jpg

2017-02-06_21_50_43rpm_0_maf_NaN.png

[kb1gtt]

This is good news. Your 91 might be the same wiring as the 93. The 93 shows the VR signals as individually shielded, then multiple VR's are then combined in a bundle which is then again shielded. So my guess is that yours is either the same, or close to the same. Also on this 93 schematic it shows that the grounding point for the shielding is also connected to a black-red wire that goes to ECU pin A26. If that harness pin on your shows continuity to GND, I would connect it to GND in the ECU.

Also if the signal is flipped in the ECU you might need to flip the + and - VR leads in the Frankenso.

[ZHoob2004]

When installing the jumpers I used the pinout of the factory harness and made sure all the grounds are connected, so if it connected to the ecu, it's connected to frankenso.

[ZHoob2004]

Out of curiosity, what do you think the highest sustained current I could pull out of the high side drivers is? I'm looking at how to hook up the vtec solenoid so I need to provide 12v from the ecu. Estimates online predict 0.5A for the solenoid, but I don't know if that's peak or hold current.

[AndreyB]

http://rusefi.com/forum/viewtopic.php?f=4&t=450 is not right for VTEC, http://rusefi.com/forum/viewtopic.php?f=4&t=892 is (actually that thread is a mess - there are two little boards there, both high side switches)

[ZHoob2004]

I'd like to keep everything internal if I can, could I find a dip package high side switch and slap it into the prototyping area? Do you think the board would provide enough of a heat sink if I'm only pushing 0.5A? Another option is I could just throw a mosfet on and use one of the low side drivers to make my own switch.