Page 1 of 2

1995 Honda Accord - auto, vtec (#5)

Posted: Thu Jul 17, 2014 3:01 am
by AndreyB
1995 4 cylinder SOHC, vtec, OBD-I, distributor
Auto (TCM separate from ECU)
honda_1995_ex.jpg
honda_1995_ex.jpg (102.73 KiB) Viewed 36267 times
The wheels a bit ridiculous, but whatever. Now I need to get plates & figure out the state inspection.

http://rusefi.com/wiki/index.php?title=Vehicle:Honda_Accord_1995

Re: 1995 Honda Accord - auto, vtec

Posted: Thu Jul 17, 2014 3:16 am
by AndreyB
Engine mount control solenoid?!. OMG.

Wow. That's definitely another use-case for http://rusefi.com/forum/viewtopic.php?f=5&t=612

Re: 1995 Honda Accord - auto, vtec

Posted: Thu Jul 17, 2014 6:02 am
by puff
may i offer you my congratulations on the new test mule? :D

Re: 1995 Honda Accord - auto, vtec

Posted: Thu Jul 17, 2014 9:42 am
by kb1gtt
Injector resistors? I see 3 VR sensors, what is the CYP sensor sensing? Is that Crap Your Pants?

Looks like the trans needs to know the TPS. Could it be Toilet Paper Sensor as a match for the CYP sensor :) I would suggest you measure the ECU's impedance and make sure your ECU has the same resistance. Other than that, this looks to be not an issue. I also see the ECT sensor. You should do the same thing with that,make sure your ECU offers the same resistance as the stock ECU.

I wonder why the trans cares about baro pressure. I also see that pressure seems to be reported by the ECU, but the ECU only has one MAP sensor, which is probably not Baro. So this could be a potential problem. You may need to take some measurements to see what this signal is doing. If you simply ignore this signal, I would wager a guess the trans will work but will probably not shift quite as desired.

I've got a solenoid motor mount on my daily driver. When the engine is below 1.7kRPM, it activates to make the mounts softer, which is supposed to make less vibration. I hate that thing as it's in the way when I do the oil. You have many of those motor mounts.

Any how, good news, looks like that auto trans is not a real issue. I wonder how well the pin out works with Frankenso.

Re: 1995 Honda Accord - auto, vtec

Posted: Thu Jul 17, 2014 2:25 pm
by puff
I bet trans needs baro to see the engine load to make decision wether to shift the gear or not...

Re: 1995 Honda Accord - auto, vtec

Posted: Thu Jul 17, 2014 2:35 pm
by blundar
Honda's injector supplier (Keihin) didn't have a saturated injector when this was made. Or they had a ton of the older P+H injectors left in inventory. In any case, the resistor box on the injectors serves as a current limiting job to allow the injector drivers (NPN Darlington arrays, 2 inj per array, made by Sanken) to control the injector with <1amp of current. No magic smoke lost.

Engine mount solenoid I know nothing about.

Vtec is an on/off. There is no PWM on this. On the factory ECUs, it is run as a SOURCE output not a SINK. It uses a moderately beefy integrated transistor thing (Allegro/Sanken SK5050 / SI5050 / SI5151) which has a high side switch with a monitor circuit to detect open circuit and short circuit conditions. Feedback provided on one pin of the 5 pin package. The solenoid itself is moderately beefy - it takes a little shy of 2Amps to drive it if I remember correctly. Vtec is NOT variable cam timing - it essentially changes to a different cam profile. The stock ECU has independent fuel and ign tables for vtec and non-vtec operation.

The distributor is an interesting thing... 3 sensors.
CKP - crank position. 24 teeth / cam rotation, evenly spaced
CYP - cam position, 1 "tooth" (looks like a snail shell) fires before #4 cylinder if I remember right
TDC - top dead center, 4 teeth fires at TDC for each cylinder.

It has been *years* (7? 8?) since I looked at the code for the distributor handling. All 3 sensors are used by the factory code. The TDC sensor is used mostly as a firing edge while cranking in order to get the car to start quickly. (Did you notice how fast it starts for a distributor car?) TDC is also involved as a sanity check. CKP pulses checked vs it, etc. These things use an ASIC to do most of the counting of pulses and stuff so a lot of the really nasty timer oriented stuffs is offloaded off the main MCU, which is an OKI 66K variant not found outside Honda ECUs. They're at least functional on stock code to about 10K RPM.

Coming out of the ECUs, Hondas are inverted dwell. Low output is charge on the coil, high output is idle coil. The ign output line is pulled high internally in the ECU by a (?)12k resistor if I remember right. The ignitor module (usually inside the distributor cap) inverts the polarity of the control signal at the coil itself.

I'm not sure about the ECT sensor bias, but I'll email you something.

Re: 1995 Honda Accord - auto, vtec

Posted: Tue Jul 22, 2014 3:10 pm
by AndreyB
blundar wrote:The distributor is an interesting thing... 3 sensors.
CKP - crank position. 24 teeth / cam rotation, evenly spaced
CYP - cam position, 1 "tooth" (looks like a snail shell) fires before #4 cylinder if I remember right
TDC - top dead center, 4 teeth fires at TDC for each cylinder.
That's what I got for TDC and CKP while cranking the engine:
Image
Not really evenly spaced. Do you think it's just the way the starter spins it?
Also eventually I would need CYP sensor for real sequential injection in the right cylinder. Ignition should be fine since there's a distributor.

Re: 1995 Honda Accord - auto, vtec

Posted: Tue Jul 22, 2014 5:53 pm
by puff
do you know what's the physical pattern of the upper one (at least how many teeth are there?)
if this inconsistency is caused by the way starter works - there might be a problem with starting a motor…

Re: 1995 Honda Accord - auto, vtec

Posted: Tue Jul 22, 2014 9:03 pm
by kb1gtt
I wonder if the starter slows down due to compression. Perhaps try pulling the plugs to prevent compression, then see if it becomes more consistent.

Re: 1995 Honda Accord - auto, vtec

Posted: Wed Jul 23, 2014 7:42 pm
by Nobody
Yes normal, it's the compression of engine slowing cranking speed, then once past TDC it will accelerate. 4 cylinder engines have fairly large torque peeks...

Re: 1995 Honda Accord - auto, vtec

Posted: Wed Jul 23, 2014 8:24 pm
by puff
what about v8? from my point of view this could pose some challenges for starting the engine. i guess it's definitely not that bad idea - to start the engine using the signal from dizzy and only after the engine is started (i.e crank sensor signal is consistent) - move to the trigger wheel signal.

Re: 1995 Honda Accord - auto, vtec

Posted: Sun Jul 27, 2014 7:39 pm
by Nobody
V8 is going to be smoother than the 4-cylinder example that was posted. The 4-cylinder has a compression cycle every 180* (degrees), while V8 every 90*, so less torque ripple. While cranking and before engine starts, I tend to look at compression cycle like a pneumatic spring.

If you want sequential fuel injection, you need to run off of crank wheel and use cam to determine if its an intake or exhaust cycle, this includes starting. Crank decoding is probably one of the most important things that need to be nailed down and given priority to (hardware and software). On a high performance engine less than 1* of advance can be the difference between knock or no knock. At 7500 RPM 1 degree is 22us, 1/10 of a degree 2.2us (0.0000022 second). Think this illustrates why this should have priority, also lets not forget about rounding errors at high RPM, where you want the most precision is where you have the least.

In the past I have chewed up 6 timer/counter hardware channels just on crank decoding. 2 of which were used to convert to angle domain… timer and modulus counter 0-3599 (this was a 2-stroke so 0-359.9*).

Re: 1995 Honda Accord - auto, vtec

Posted: Sun Jul 27, 2014 8:23 pm
by puff
Nobody, have you seen the russian's code for crank decoding? Do you think it's good enough for the purpose? BTW, what is the usual procedure to calculate the angle? say, you have a 36/1 trigger wheel, you get the time per 1 tooth, you divide this time into 10 equal pieces and hence you get the angle from the last tooth? does it compensate somehow for the inconsistency of tooth duration during various cycles? does it actually need to compensate for it?

Re: 1995 Honda Accord - auto, vtec

Posted: Sun Jul 27, 2014 11:33 pm
by Nobody
How you approach crank decoding is to a large extent is hardware dependent. How many timers/counters, at what resolution (16, 20, 24, 32 bit), the higher the resolution the better the granularity. Also on modes of operation, single/double capture, with/without de-bounce filter…

If you attempt angle then that should be done in hardware, because you would link other functions to it i.e. injectors and spark (ignition coils). On some MPUs you can have a common counter/timer buses to which you would link said functions.

Converting time to angle in hardware, with constant minor corrections can be a timed modulus approach… I updated critical timing twice per crank tooth, but again approach varies and very much MPU dependent. In simplest terms yes you would look at time per tooth and how many degrees that tooth is. You need to know if it is accelerating or decelerating and by what delta and factor that in with smoothing/filtering. This part can be somewhat complex to very complex. Usually you have a high speed timer (in nanoseconds) that increments modulus counter by 0.1 or 1 degree, this timer is constantly being updated and modulus counter corrected incase it is out of sync or error is too large. Ideally you would want at least 50 high speed timer ticks per modulus counter increment at max RPM.

If your engine is running well (no misfire) and all pistons are doing about equal work, then your torque ripple will be consistent and timing repeatable. But if missing then your accel/decel will vary throughout 720* (4-stroke) and decoding require more work/logic.

I really don’t have a good answer for you, as I think this is one of the largest variable in many ECUs. I’ve swapped out ECU to different unit on same engine with no modifications and had it make more power, just because it was more repeatable/accurate…

Re: 1995 Honda Accord - auto, vtec

Posted: Mon Jul 28, 2014 12:55 am
by AndreyB
Nobody wrote:How you approach crank decoding is to a large extent is hardware dependent. How many timers/counters, at what resolution (16, 20, 24, 32 bit), the higher the resolution the better the granularity. Also on modes of operation, single/double capture, with/without de-bounce filter…
* rusEfi is implemented based on the idea that with modern 100+ MHz microprocessors the relatively
* undemanding task of internal combustion engine control could be implemented in a high-level, processor-independent
* (to some extent) manner. Thus the key concepts of rusEfi: dependency on high-level hardware abstraction layer, software-based PWM etc.
There is a brief hi-level description of trigger decoding @ http://rusefi.com/docs/html/

Re: 1995 Honda Accord - auto, vtec

Posted: Mon Jul 28, 2014 3:07 am
by Nobody
russian wrote:
Nobody wrote:How you approach crank decoding is to a large extent is hardware dependent. How many timers/counters, at what resolution (16, 20, 24, 32 bit), the higher the resolution the better the granularity. Also on modes of operation, single/double capture, with/without de-bounce filter…
* rusEfi is implemented based on the idea that with modern 100+ MHz microprocessors the relatively
* undemanding task of internal combustion engine control could be implemented in a high-level, processor-independent
* (to some extent) manner. Thus the key concepts of rusEfi: dependency on high-level hardware abstraction layer, software-based PWM etc.
There is a brief hi-level description of trigger decoding @ http://rusefi.com/docs/html/
If I dragged this off topic then sorry…

Image

Here is what I was referring to as torque for V8. These torque peaks and valleys are dampened to some extent by mass of drivetrain/flywheel and springs in clutch or torque converter fluid coupling. Now picture a fouled spark plug that intermittently fires, there will be a timing shift/skew that will throw subsequent cylinders off if using a global timer for all scheduling (based on complete revolution). An internal combustion engine is not a constant speed device (deg/sec - angular velocity) but always accelerating and decelerating (worst at low RPM).

Understand your approach, but I guess I design for best possible accuracy. I wish OEMs would put encoder like on servo motors, just about every modern 32bit MPU can decode that in hardware… Nissan 300Z had that.

NOTE - Most if not all use crank as master reference and cam to determine intake or exhaust cycle. Some ECUs will run fine with loss of cam sensor, they use MAP as backup reference to determine intake of exhaust stroke.

EDIT ADDING MORE DETAIL BELOW

Image

Image

Image

Image

Re: 1995 Honda Accord - auto, vtec

Posted: Mon Jul 28, 2014 7:17 am
by rus084
Nobody wrote: Image
oh, very large ripple (2000%), this is probably bad ...
I thought that the 4-cylinder engine is balanced, and it turned out that you need at least 8 cylinders to the engine does not slow down sometimes

Re: 1995 Honda Accord - auto, vtec

Posted: Thu Jul 31, 2014 12:12 am
by AndreyB
One step forward: I've sniffed the 12V ignition module control signal

Image

Two steps backwards: the car does not start, looks like no spark, I cannot remove three of the four spark plugs! See http://honda-tech.com/showthread.php?t=3217244

Re: 1995 Honda Accord - auto, vtec

Posted: Thu Jul 31, 2014 6:29 am
by puff
were they welded in?
what could be the cause for no spark? no signal? wet plugs?
what are the first three graphs? the top one - cps? how many teeth are there?

Re: 1995 Honda Accord - auto, vtec

Posted: Thu Jul 31, 2014 6:52 am
by puff
how does the 'no spark' diagnosis explains that the engine finally starts? I'd say that sounds more like the lack of fuel...
probably it's silly, but i'd say that it's easier to check for spark of no-spark with a timing gun (with capacitance/induction sensor?) at least you'd know that the dizzy is delivering HV to each plug.

p.s. this diagram looks pretty much complicated Image
what are the yel/grn and yel/blck wires?

PPS channel 5 & 6 on your screenshot seem to be almost the same?

Re: 1995 Honda Accord - auto, vtec

Posted: Thu Jul 31, 2014 11:11 am
by AndreyB
puff wrote:were they welded in?
what could be the cause for no spark? no signal? wet plugs?
what are the first three graphs? the top one - cps? how many teeth are there?
This morning I've removed #4 plug and I am making good progress on #2 plug. I am not getting it - they look pretty fresh, why would they be stuck there? I guess I should apply some unit-seize.

So far the plugs I've removed are gapped right - they should be 0.039-0.041in and they are more or less that.

Could be weak/dead ignition coil. Could be weak/dead igniter.

A lot of info @ http://rusefi.com/wiki/index.php?title=Vehicle:Honda_Accord_1995 to answer some of your questions

Re: 1995 Honda Accord - auto, vtec

Posted: Thu Jul 31, 2014 12:32 pm
by AndreyB
The book has coil testing procedure (measuring two resistances) - test passed, the coil is fine. Changed two plugs - this did not help, will attach the other two plugs with more leverage later.

There is no ICM test procedure in the book. Should I just check that coil input gets some voltage while cranking? Would that be a test of the ICM? Is that safe?

Re: 1995 Honda Accord - auto, vtec

Posted: Thu Jul 31, 2014 3:59 pm
by Nobody
Given the year, it likely has engine bay fuel pressure regulator. If it does, pull off vacuum reference line and check for fuel, if any is present then diaphragm is torn (try this after cranking). Long cranks can often be fuel pressure related.

Re: 1995 Honda Accord - auto, vtec

Posted: Fri Aug 01, 2014 10:19 pm
by AndreyB
Welcome to my luck: it was the ignition control module I guess. $50 for a used dizzy and this POS starts now!

Re: 1995 Honda Accord - auto, vtec

Posted: Tue Aug 05, 2014 2:28 am
by AndreyB
That ~40Hz noise is gone if I disconnect the TCM plug A. The +5 ref line looks clean on the scope, this makes me ask if one of the TCM lines is a digital signal? Who knows... For now I'll just have the TCM plug A disconnected :)

http://rusefi.com/wiki/index.php?title=Vehicle:Honda_Accord_1995

Re: 1995 Honda Accord - auto, vtec

Posted: Thu Aug 07, 2014 12:12 pm
by AndreyB
[video][/video]

Re: 1995 Honda Accord - auto, vtec

Posted: Thu Aug 07, 2014 8:48 pm
by sturovo
Great work Andrey!

Re: 1995 Honda Accord - auto, vtec

Posted: Thu Aug 07, 2014 9:14 pm
by AndreyB
More like "great work Jared" - that's his design :)

Re: 1995 Honda Accord - auto, vtec

Posted: Thu Aug 07, 2014 10:06 pm
by kb1gtt
Thanks for the pat on the back. Lots of work put in by many people. It's a collaborative effort. Many people had a hand in making this stuff go.

Re: 1995 Honda Accord - auto, vtec

Posted: Fri Aug 08, 2014 6:30 am
by Tambralinga
This model engin F22 or H22 ?