@dc42 Tyvm for the response. I will do that with a scope this week. A subsequent bed probe with the second effector had a single point with similar error and coordinates, so I did a full teardown and found a likely contributor. The Seemecnc Rostock MAX series use two idlers at the base and the inner idler is hidden. Previously I had inspected the belts, motor pulley, and bottom idler but not the inner one. The combination of hairspray on a glass bed(with a sheet of paper covering the base of the belts but not the entire tower so some mist would float up) over a couple years had captured small particles of belt compound and filament dust and formed a couple beads the consistency of rubber cement between the smooth side of the belt and inner idler. Depending on speed and coordinate travel, this bead would change shape from a smooth, <0.2mm ellipse smooshed against the idler that didn't show errors to a 1-2mm hemisphere bead that would. Each of the 3 tower inner idlers had this bead, in different shapes, and removing them reduced(but not eliminated) the bed probing errors. This sticky bead was sequestering common signs of wear so the other sections looked pristine.

@mikeabuilder The lower frequency peaks in the 20's-30's are typically the ones with the very visible ringing of features and a greater concern. <25Hz are often loose mechanics. I would read the "Technical Details" of the Klipper description @
https://www.klipper3d.org/Resonance_Compensation.html

You can add that +- ShaperFreq chart to your excel doc to see which freq to use. Note the Duet also offers ZVDDD with shaper duration 2/Freq.

Short example, ZVD/DD/DDD are +- 15%, you have peaks between about 25.5 and 34. A shaper freq for M593 of 30 would cover the ranges of 25.5 (30*.85) through 34.5 (30*1.15).

@cadu Checked the effector under the scope - solder looked good except for fiber microparticle debris and some surface oxidation that was causing the initial inconsistent probe readings. Reinstalled the effector reporting errors and it performed as expected, with 3 45-point mesh probes completing without "inconsistent probe reading" error and without the double parabolic shape.

G29
45 points probed, min error -0.165, max error 0.116, mean 0.001, deviation 0.057, with all 3 sets within 0.006 of those readings

My best estimate atm is that the increased speeds, acceleration, and short high-g movements in the input shaping tests after updating the firmware created the hidden hairspray/microparticle bead on the hidden idler pulley. The parabolic bed shape completing without reporting errors was probably the intersection of probing coordinates and the belt bumping over the bead. Prints didn't show the error due to the longer movements flattening the bead.

@dc42 Tyvm for the response. I will do that with a scope this week. A subsequent bed probe with the second effector had a single point with similar error and coordinates, so I did a full teardown and found a likely contributor. The Seemecnc Rostock MAX series use two idlers at the base and the inner idler is hidden. Previously I had inspected the belts, motor pulley, and bottom idler but not the inner one. The combination of hairspray on a glass bed(with a sheet of paper covering the base of the belts but not the entire tower so some mist would float up) over a couple years had captured small particles of belt compound and filament dust and formed a couple beads the consistency of rubber cement between the smooth side of the belt and inner idler. Depending on speed and coordinate travel, this bead would change shape from a smooth, <0.2mm ellipse smooshed against the idler that didn't show errors to a 1-2mm hemisphere bead that would. Each of the 3 tower inner idlers had this bead, in different shapes, and removing them reduced(but not eliminated) the bed probing errors. This sticky bead was sequestering common signs of wear so the other sections looked pristine.

Hello, I'm experiencing some odd behaviour with my SE300(seemecnc custom version of of the Smart Effector) w/ genuine E3d V6 after updating firmware on a Duet2 Wifi from RRF2 to RRF3.4 and adding an accelerometer for input shaping(removed after gathering data).

Calibration appears to work, but bed Mesh consistently draws a bed mesh with opposing parabolic shapes, sometimes with max error 0.5, other times "7-9!" on a glass bed that wouldn't physically make that shape. Adding shims/clamps changes the values but not the overall shape.

If I print without using a bed mesh, it prints correctly except for some relatively minor bed height leveling issues fixed by manual baby stepping.

I'm not sure how to go about trouble shooting this. I have tried changing the probe type from P5 to P8, with trigger values from 50-150.

Additional possibly helpful info.
A couple months back(on rrf2) there was an error midprint that caused the nozzle to drive into the bed through the part after printing correctly 30min. That loosened things and I retightened the belts and found an endstop screw loose causing geometry errors and resecured all of them with thread locker. I also saw that the bed mesh was showing a single point near near infinity and the web control was going into page refresh error loops so I reloaded the SD from a backup.
This worked well since but when I was investigating the latest issue I saw that a motor mount was also loose, able to a wiggle few degrees but rarely due to the tight belts. After tightening that, rechecking belts and the other mounts, the double parabolic bed mesh still remained. Pulley mounts are flat ground with double grub screws. Thank you for any suggestions.

m115
FIRMWARE_NAME: RepRapFirmware for Duet 2 WiFi/Ethernet FIRMWARE_VERSION: 3.4.0rc2 ELECTRONICS: Duet WiFi 1.02 or later FIRMWARE_DATE: 2022-02-22 17:06:56

M558 P8 S0.05 A2 R0.4 C"zprobe.in+zprobe.mod" H30 F1200 ;
G31 P100 X0 Y0 Z-0.2

; Drives
M569 P0 S1 ; physical drive 0 goes forwards
M569 P1 S0 ; physical drive 1 goes backwards
M569 P2 S1 ; physical drive 2 goes forwards
M569 P3 S1 ; physical drive 3 goes forwards
M584 X0 Y1 Z2 E3 ; set drive mapping
M350 X16 Y16 Z16 E16 I1 ; configure microstepping with interpolation
M92 X80.00 Y80.00 Z80.00 E400.00 ; set steps per mm
M566 X499.80 Y499.80 Z499.80 E1998.00 ; set maximum instantaneous speed changes (mm/min)
M203 X15000.00 Y15000.00 Z15000.00 E15000.00 ; set maximum speeds (mm/min)
M201 X2400.00 Y2400.00 Z2400.00 E1000.00 ; set accelerations (mm/s^2)
M906 X1300 Y1300 Z1300 E1200 I50 ; set motor currents (mA) and motor idle factor in per cent
M84 S30

update I swapped to a spare SE300 with a Dragon hotend and it completed the mesh correctly on the second try. The first try reported Z probe readings not consistent. 25 probing points. The config file used was the same, with the same wiring whip and no rearrangement of the bed or control board that would have moved other wires and changed the electronic noise.

I would like to get the original working but don't know where to go from here.

update2 On the faulty effector I used a multimeter, Fluke 87V, on the right side of R11 and left sides of R13 and 14.
R11->R13 = 72.4 Ohm, R11-R14 = 68-72 Ohm with some fluctuation if my hand wasn't steady. Wiping some dust off appears to have steadied the R11-R14 probe reading.