@Chriss:
I thought I would revive this thread with some useful information. I stumbled on this thread after realizing that my IR probe started giving me weird results on my FR4/G10 build sheet. (I found this interesting article that shows how IR can find voids and defects in G10, https://irinfo.org/11-01-2015-hazouri/)
Based on my recent tests all at room temperature, FR4/G10 causes some issues with the IR probe. I have a spring steel build sheet with 3M 468mp adhesive topped with a .020" G10/FR4 sheet. I can confirm that I get readings from the IR probe that are not correct on clean FR4 (cleaned with 99% isopropanol and no coatings and with a lightly sanded finish on the FR4)
To confirm my suspicion, I performed mesh compensation probing, then picked 2 spots on the map showing a high next to a low. The two spots are 40mm apart in the x-direction with the IR probe measured difference of almost 0.2mm. Disabled mesh compensation (not necessary for the G30) but necessary for the dial indicator measurements) probed G30 S-1 got a reading from the IR probe of 0.005mm of what the mesh compensation point previously measured (it showed consistency in measurements of the IR probe). Then I took an analog dial indicator attached to my x extruder stepper motor and zeroed at the first point and dragged the dial indicator to the other point. The dial indicator showed a worst-case of <0.01mm difference at most between the two points. This included sweeping the area around each point with the dial indicator in case my offset on my IR probe was not perfect. This was all done with power to the stepper motors to ensure repeatable measurement points.
To further confirm, I placed one continuous piece of masking tape over the two spots and remeasured with the IR probe and the Dial indicator. With the tape in place, I measured <0.01mm difference in height between the two spots with both the dial indicator and the IR probe.
Also in regards to your plot of the PEI bed, I believe PEI is transparent to IR so you would be measuring whatever is below the PEI.
I just painted the FR4 build sheet (previously was showing a lot of height variation) with high temp paint. (I hope it sticks well when removing parts (after curing the paint)). After painting, I get the nice flat bed measurements I was expecting on my printer. If the high temp paint works for printing, I will try to post up the paint I used. I will likely ditch the FR4 and paint the spring steel sheets for my custom built mag bed. It will depend on how well the paint holds up to being printed on. At least if I paint the spring steel, I can cure the paint to a higher temp which might help it hold up to the plastic extrusion temps better.
As a side note: I use Airwolf Bed adhesion coatings on the fiberglass with really good luck with their ABS, PC and PLA coatings. I have had mixed results with the nylon coating, but it has been a really long time since I tried printing on it.
Also, I found I have to run mesh compensation mapping with the IR-Probe when the bed and chamber are cool and at a constant temperature. I found there is a ton of drift as the sensor heats up above a hot bed.