Honda Insight Forum banner

Sorting out a Citrus

10K views 151 replies 20 participants last post by  kamesama980 
#1 · (Edited)
My Citrus tends to stumble and lack power in lean burn when it is under increasing load. When it exits lean burn, there is a gentle burst of acceleration. I hope to uncover why. Items I'll be looking at:

ItemStatusNotes
EGR valveReplaced with measured good used valve, then with new valve.The original was terribly intermittent!!! And, don't reuse gaskets. Use a new gasket. The kind stamped out of a thin metal sheet may leak.
EGR plateCleanedWas plugged!!!
EGR passagesTBD
Spark plug indexingCorrect plugs for engine
Spark plug ageReplacedNo change in high load lean burn performance with new plugs
Spark plug coil performanceNew coils installed, no difference, old coils back in.No sign of rust or delamination.
Still worth testing. Corrosion possible in spring wires inside boot. Measured current draw of primary and found all three the same. New ignition coils made no difference.
LAF sensorReplacedThis is pretty much something that has to be done if the LAF sensor is aged. It is a very important sensor.
Manifold pressureNo smoke leaks except very small amount at EGR valve bearing (old and brand new valves)OBD and trim values suggested there might be a small leak. Smoke tester revealed a very small leak at the EGR valve sleeve bearing but this may be normal.
Fuel injectorsCleanedABSOLUTELY IMPORTANT. Low flow rate in an injector can lead to loss of catalytic converter. NOTE: Non-OEM projectors like Ultra-Power may have different flow pattern. Insight flow pattern appears to aim to normal and VTEC ports and have been designed to balance lean burn and high RPM use. Stick with OEM here. Unsure if age impacts atomization even if cleaned - this should be confirmed. The old injectors definitely show a flow imbalance. Do not reuse the seat seals or vacuum leak, possible imbalance. A hack is needed to keep dirt out above the seal if the cup type is not available (appear no longer available.) Needs more study, later
Fuel pump pressureChecked goodWell over 60 PSI
Fuel rail pressure regulatorReplacedOriginal unit replaced with one from another Insight before pressure testing; no difference. Replaced unit shows 41 PSI at tip at different vacuums.
New unit shows 43.5 PSI. No apparent difference in performance.
Engine DepositsAfter 7000 miles of Sunoco 93 and Techron, removed about 20% of deposits. May need to perform full cleaning.In past work, cleaning piston heads seemed??? to help but lasted only a few hundred miles after which deposits reformed. Not sure if there is anything to see here. We know that deposits raise cylinder pressure a little. Wonder if they affect boundary area flow in a way that creates turbulence and may modify mixing. Really need to borescope one of the few well-performing low mileage Insights to gauge this. Wondering if turbulence caused by disruption of the boundary layer by deposits is causing poor mixing in the cylinder and leading into misfires or partial combustion that could be limiting power or possibly introducing knock. Just a WAG and as likely to be wrong as right.
Valve lashGoodThis does affect performance and should probably be checked every 30K miles or so. If out of whack, tuning this can make for a smoother idle.
Leakdown testGoodLeakdown test showed excellent results in two cylinders and good results in third. Leakage in center cylinder might be due to a few visible scores likely caused by abrasive entering and getting stuck between piston head and wall before dislodging.
BorescopeUsing one with side cameras to be able to see the valves and cylinder topSince cleaning the injectors and doing some long road trips with 93 octane Sunoco, the cylinders appear to be getting cleaner. But beyond a certain point cleaning appears to have slowed.
Valve guidesTBDInclined not to worry about this with the good leakdown test.
Exhaust SystemBeing replacedCatalytic converters melted likely due to injector imbalanced from low flow rate due to varnish etc.
Compression / worn ringsGood. Measured 200/195/205Leakdown test suggests these are not a concern.
ECM / control loop issuesA02 and 505 ECUs behave similarlyA 4.88 Hz signal permeates readings of sensors, and they are also modulated by fuel injector current. This may be normal but it might align with jerkiness. Theory that it might be modulating the AFR readings. Needs more investigation. Examination of a well sorted car is desirable.
Knock sensorAppears to work properly; broke during replacement. Knock sound and knock retard scan values are closely correlated.In taking a longer look at OBD real time parameters, knock retard was observed. It is greatest during deep lean burn and correlates closely to knock sound in engine and drop in performance during lean burn.


Have I missed anything?

I won't look at all of these areas if I find the root cause before I get to them all!!!

I have already looked at several of these areas, but have decided that I need better data. OBD2 is too slow, so I plan to measure a number of sensors in real time, with many samples per cycle of whatever I'm measuring. This includes
  • Manifold air pressure
  • IAC valve position and changes
  • EGR valve position
  • LAF sensor (A/F ratio)
  • Injector pulse width
  • Changes in pulse width
  • Timing advance
  • Crank sensor
  • Throttle position
  • Camshaft sensor
  • Instantaneous changes in RPM
  • Correlation between instantaneous RPM change and instantaneous fuel injection change (deviations may imply misfire)
  • AFR imbalance as detected through LAF sensor (this will be a challenge!)
  • Some OBD2 parameters I can't measure, such as fuel trims, knock retard, etc
Arriving are some new tools such as an in-car fuel injector cleaner and parts to build the "black box" that will read the values above. I hope to be able to do A/B tests in many cases, such as with different EGR valves, different LAF sensors, and more.

Any ideas for measurements are welcome. This will probably start in earnest between Christmas and New Years.
 
See less See more
Discussion starter · #3 ·
@Gasoline Fumes, then hopefully this exercise will help both of us. What specifically happens under load in lean burn?
 
  • Like
Reactions: Gasoline Fumes
Discussion starter · #4 ·
First part of this project is to set up a high speed data logger for key engine parameters.

Wish you could buy this from, say, Adafruit: "ECM/MCM Breakout Board. For DIY high rate data logging or full electric conversions."
Product Blue Rectangle Gas Electric blue
 
Discussion starter · #5 ·
This is going to take a while. I need to learn kicad and find a PCB house.
Blue Rectangle Art Pattern Gas
 
  • Like
Reactions: trickle
Discussion starter · #7 ·
Bucking or "herky-jerky," as it's been called here in the past. I'll be cruising at maybe 45-50 MPH at 125 MPG, come to a slight hill, give it a little gas to maintain speed, and it'll drop to about 75 MPG and start bucking. Sometimes really badly. Giving it more gas will drop it down to about 50 MPG and it'll run smooth again and begin accelerating. I can't maintain LB on level ground at highway speeds either.
Yep. I suspect that this is underdamped response to various inputs. I hope that examining air-fuel ratio, manifold air pressure, EGR position, injector pulse width, and maybe some other stuff (timing?) will tell a story. It might be nice to have a measurement of fuel rail pressure and some other stuff.

More progress soldering:
Rectangle Circuit component Art Font Audio equipment
 
  • Like
Reactions: minor4326
Discussion starter · #9 ·
@trickle Very interesting! It would be interesting if detonation is detected. I should be able to see this. Might actually be able to quantify the benefits of removing carbon and different grades of fuel!

Still making progress.on the wiring. This doesn't get old. Only 90 wires left!
Rectangle Textile Font Circuit component Electric blue
 
Discussion starter · #14 ·
Umm, why not just use a scan tool?
Great question! I do own an advanced scanner that can read and clear the ECM, IMA, EPS, airbag etc systems. Newer cars use CAN for OBD but the G1 does not. The data rate is 10400 bits/sec or about 1000 bytes/sec but in fact most of that is overhead and one can only get less than 10 messages per sec (IIRC). In data stream modes, if you select all possible data, the sample rate drops to one sample every three or four seconds per item.

This will allow me to log 16 sensors for short periods at over 100 mega than samples/second, but 1000 samples/sec will be more than enough for most sensors. It will let me see what the ECU sees and how it responds, using charts. I'll share these once I have this running.
 
Discussion starter · #15 ·
Could you add more to the thread title? A more descriptive title might be helpful in the future if this thread ends up with some good solutions.
It's a great idea, I'm not sure yet where this will go... Keep that thought.
 
  • Like
Reactions: Gasoline Fumes
Discussion starter · #16 ·
Only 65 more wires to go (on this board)...
Rectangle Textile Font Technology Electric blue
 
Discussion starter · #18 ·
I've thought about connecting a knock sensor to the ECM that's isolated from the engine to see what the timing is without knock. And carefully listen to how much it knocks!
I have a recording somewhere taken by connecting my knock sensor to a computer's mic input. It is just a piezo element tuned to a specific frequency with a piece of metal.

Which reminds me: a few weeks back I bought some small piezo sensors and some small magnets with the idea that I would tape them into various parts of the engine. Not sure if I will be able to do anything with that.

Was also thinking about adding an accelerometer to detect bucking but that also might not be too useful. The human butt is really good at detecting bucking among a bunch of road vibrations. Not so good at seeing it in a waveform.

it'll all be worth it when you turn the key and it works
Yes, because it is supposed to do nothing at all! I am checking connections twice so that I'm not naughty, but nice to my ECM.

Only 42 connections left!!!!!
Circuit component Rectangle Electric blue Computer hardware Font

(getting more familiar with kicad is definitely increasing in priority...)
 
  • Like
Reactions: trickle
Discussion starter · #20 ·
sean are you familiar with torque app?
Yes. There are some threads about programming the AFR PID with a formula for a lean burn display.

30 more wires to go!
 
Discussion starter · #23 ·
Harder. It was getting pretty mindless and I was thinking of all the other things I wanted to do (the ones I need to do lock themselves out of my mind automagically) so I called another member on the phone and chatted while I did the rest... so they are DONE!

Now to test them. Murphy and I are good friends so I can't wait to find what mistakes he has laid for me.

You can see where I changed techniques.
Product Rectangle Font Electric blue Electrical wiring
 
Discussion starter · #25 ·
powered through it tho, looks nice too.
Not any more:
Circuit component Rectangle Art Font Technology


I need to lengthen four or five wires and add about 25 to get 100% connector coverage. This will then let me examine every signal in the MCM and ECM. Considering creating a markdown file for each ECU describing each signal with sognal screenshots from the logic analyzer/oscilloscope but given how long it took to get this far, I am concerned... Maybe I can fill it out as needed.
 
  • Like
Reactions: trickle
Discussion starter · #27 · (Edited)
hmm.. do you have sacrificial test modules? how many different ones are there?
There are almost as many (or more) BCM, MCM, and ECM versions than model years of G1 Insights. But I got my printed circuit board connector from a 1998 Civic ECM. However, you can purchase some of these ECU connectors and extra pins from Mouser and Digikey (TE Connectivity AMP .040 and .070 Multilock series.)
 
Discussion starter · #29 ·
I will start with a Saleae Logic Pro 16 which has sixteen channels of logic or oscilloscope functionality (thank you Saleae for the maker's discount!) The oscilloscope is limited to +/-10V but will tolerate up to +/- 25V. Most (all?) of the analog signals are below 5V so the 10V limit on a 12V signal is not a concern. Some outputs and switching signals operate at battery voltage; these are effectively "on" or "off" so these will be treated as logic unless there is a reason otherwise.

I made a list of the signals I want to look at first and was unable to keep it below sixteen!

For long term data collection I am looking at some combination of an AVR128DAxx part, a Teensy 4.1, and a Raspberry Pi 4, for different reasons:
  • The AVR128DAxx can handle 5V and may become the primary ADC for some circuits. The 28 pin version is a DIP so it is easily breadboarded and has 10 ADCs. The 32, 48 and 64 pin versions have tons of analog inputs and can be found on inexpensive breakout boards from the person who ported them to Arduino. So, this may be the default for interfacing to analog pins. On the minus side, it tops out at 25 MHz.
  • The Teensy 4.1 can handle 3.3V but it is programmable with the Arduino environment and uses 600 MHz processor. The support from the manufacturer is excellent, with customized libraries available open source. The Teensy 4.1 has eight serial ports so it is a candidate for handling the RS-422/485 connections. I'm not sure about its ADC capabilities but the 3.3V ceiling is a limitation. It employs high speed USB so between that and the 600 MHz processor, getting collected data to another system is not going to be the bottleneck.
  • The Raspberry Pi 4's primary use is for data logging. Offloading the logging to the Pi avoids timing issues when an Arduino attempts to write to an SD card. With a USB3 interface, vast amounts of fast storage are an option.
I can easily collect volumes of data that will only take up space. I may need to find a happy medium where I am collecting enough information to spot trends or diagnose a problem but have a rolling log that overwrites the oldest data when storage reaches capacity. It may also be useful to collect baselines of data in certain configurations, and perhaps with some common failures simulated to have as a future reference.
 
Discussion starter · #30 ·
Wires added to harnesses to enable observation of any ECM or MCM signal.
Slope Electric blue Pattern Art Visual arts
 
  • Like
Reactions: trickle
Discussion starter · #31 · (Edited)
An important thing to note is that one can't just connect to any of these signals without risk of adversely affecting car operation. For example, the "ground" for each driver might be a ground reference for DC (unchanging) signals but active signals can put currents on the ground that overlap the signal being measured. Module designers take extra steps to prevent the noise on one subcircuit from influencing readings on another subcircuit. This includes measures beyond the scope of this post. But it means that if a measurement circuit is connected improperly, it can unexpectedly introduce noise into the signal being read by the module and change it.

It's also possible that the device hooked up to a sensor loads that sensor, thereby changing it before it is read by the module. This also can cause adverse behavior. Finally, it is possible to hook up something wrong and break the module.

In other words, the breakout harness made for this project is necessary but far from sufficient. For some of these signals, great care will need to be taken to ensure that the instrument measuring the signal does not change the signal. Therefore, every signal will need to be assessed as to the most appropriate way to capture it without modifying it to the point where the tapped signal causes the car to behave in a meaningfully different way.

So, this will not be a matter of simply wiring a bunch of these taps to an Arduino. In some cases it may require significant accommodations to ensure that the signal being measured is not altered by the measurement circuit.

The largest problem is perhaps that there is no access to the power and ground connections in the module that is necessary to minimize noise. This hopefully underscores that this is not intended for permanent installation but only as a diagnostic tool to identify issues when they arise or confirm continued proper operation during a "tune-up".
 
Discussion starter · #32 ·
My newest Insight diagnostic tool, ready for its first live test:
Circuit component Hardware programmer Font Passive circuit component Computer hardware
 
Discussion starter · #34 ·
IT WORKS!!!!
Computer Personal computer Audio equipment Gadget Computer keyboard


First capture:
Rectangle Font Slope Parallel Screenshot

So much stuff to see here!
  • MAP - waveform shows changes in manifold air pressure WITHIN a revolution!
  • (not visible) EVAP - shows low duty cycle...
  • IACV - another potential clue to vacuum leaks at idle
  • NEP - not expected! who consumes this?
  • CKP - I wonder if this can be used to detect misfires?
  • TDC1 - not sure what this sensor does, or if a lobe is supposed to be missing?
  • TDC2 - kinda hard to check my connections - what is this used for?
  • INJ1 - I didn't have enough inputs to capture all injectors. But wow - injection almost a full revolution before ignition!
  • Ignition 1 - 3 - you can clearly see the 1-3-2 timing pattern.
This is only sixteen of the signals. It doesn't include the LAF sensor either. So much to discover! I can't wait until I use it to examine MCM signals! But next is to study these and decide what else is needed.

I designed this to accept plug-in modules which can be used for specific analyses. I think the first module should be a fueling and lean burn analyzer. This will require access to accurate AFR information. I hope to be able to SEE what seems to be misfires and what changes when certain things are done to the car to identify the cause.
 
Discussion starter · #35 ·
that seemed pretty fast even though i can't imagine how much work went into it. does it have a name yet?
"Modular Advanced Diagnostics" or MAD because you'd have to be mad to build something which requires soldering 208 wires and deals with connectors with 104 pins??? It took MUCH longer to build than I thought it would, and building plug-in boards and Python scripts to capture and interpret data will also take a long time.

One of the tasks ahead is to operate it next to my OBD scanner in live data stream capture mode, so that I can convert voltages read by this tool into values provided via OBD2.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top