An update on my repaired EGR valve. I noticed some herky-jerky in lean burn. This could be many things but since the EGR valve is now easy to test, I retested it.
Compare to post #19, made after the initial repair. Note that there are small jumps as the valve opens and closes, particularly near the bottom of travel, where the EGR spends most of its time. Also note the steep jump where the valve "pops open".
Before anyone concludes that repairing an EGR is not effective long term (and it probably is not), two theories:
1) Residual graphite, which I used to lubricate the valve when it was open, has gotten onto the potentiometer.
2) The fix really is not that great and the potentiometer is failing again.
Could it be that the wiper doesn't have enough contact pressure now?
Assuming the valve isn't leaking, as long as the sensor voltage has settled near 1 volt when the valve stays closed for a few seconds the output voltage shouldn't cause herky-jerky since in use the valve closes and the sensor voltage should be settled long before the car attempts to enter lean burn.
Of course the sensor voltage jumping around while the valve is moving (as your graphs clearly show) when not closed is a good indication of a damaged pot.
It seems that in driving the car the voltage deviations from a linear curve wouldn't affect lean burn since the valve only opens when not in lean burn. Your graph indicates that the slight resistance changes with the valve closed might get the ECU's attention and it does something other than move the valve (perhaps it changes the AFR). When my car had herky-jerky it felt like the engine lost power for an instant.
On my car the sensor voltage usually sits at 1.01 as per the OBDIIc&c. Sometimes it is 0.99 until the next opening/closing event. But it doesn't vary from whichever number it settles on until after the next "open" command. I've never seen it change once it settles. And if you are driving at a steady speed the valve is closed for quite awhile before lean burn is initiated.
The next step is to open the valve, and blow it out then clean it thoroughly, and then reassemble it.
The service manual says to use carburetor cleaner on the valve itself.
I looked very closely at the valve and valve seat with a magnifying glass but couldn't really tell if it had been leaking or not. The valve contact width is very small. Maybe 0.025" ?
Before that, perhaps I should see if herky-jerky coincides with EGR movement, using an accelerometer, a really smooth road (where???, but fortunately the acceleration is back and forth so road vibrations could be filtered out) and measuring the pulse width going into the EGR, the voltage from the potentiometer, and perhaps at the same time, the AF sensor, injector PWM and timing, to see if those are factors.
Just as a check, you might want to check the total resistance of the sensor pot. My defective valve and the newer Accord valve are both ~4.43k ohms.
I would first eliminate the EGR valve affecting herky-jerky by temporarily placing a piece of thin metal shim stock under the EGR gasket then install the valve to absolutely make sure no exhaust is getting into the intake system through the valve and then check for herky-jerky. Leave the EGR connected so the ECU doesn't have a fit.
If the car settles down in lean burn you have better proof that it is only the valve causing the problem and not other things. And it might be proof that the valve is leaky when closed.
The next thing may be to check the crankcase ventilation valve that it isn't stuck open or closed.
A test that does not require the alternator would be to see if the EGR valve is sensitive to vibration. From the data, it appears that it takes a measurable reduction in voltage before the valve starts closing, which suggests internal binding. If the engine is on, its vibrations might cause it to close or open more erratically. So perhaps that needs to be a new test: set the valve to a certain position, let it stay there with the engine on, see if it moves. Repeat and plot.
Hit valve with a small hammer while it's on the bench??? As an old timer machinist once told me, "Don't force it, ....... use a bigger hammer."
When I bench tested my defective valve I tapped on it with a short metal rod before reading the voltage after each change in drive voltage.
If the 1st or 2nd tap did cause a change, the voltage change was very small. After the first few taps the output voltage didn't change with more taps.
I also pulsed my bad valve ON/OFF at ~1.5 times per second and it was really clapping shut from a little opening each time. The sensor voltage as shown on a Tektronix scope just settled to the closed value with no problem. Yet that valve had bad herky-jerky.
Here is the valve after about a month and 1600 miles: