i think you are on the right track attaching to the bottom side of the glass bed.  However, a NTC 3950 already encapsulated in a cartridge style case may serve as a better option.  A tiny bit of thermal compound or thermal glue with a layer of Kapton tape over to hold it all in place.

I don not think it will be very easy to get enough surface area of the bare NTC 3950 for a reliable reading on a consistent basis to print as the bed moves up and down on the Z-Axis.

At the speed and intervals that the bed moves at it will not have any impact on the thermistor.

I don not think it will be very easy to get enough surface area of the bare NTC 3950 for a reliable reading on a consistent basis to print as the bed moves up and down on the Z-Axis.

if you are using the stock sensor it is probably your issue. Unless there have changes made to Marlin and or Repetier there is no table in them for that oddball thermistor that XYZ used. You need to use the same type of standard 100k NTC thermistor that the hotend uses. Meaning you will have to figure out a way to mount it. I dremeled a grove out in the bed base material so the thermistor could be mounted under the heater.

Thanks for the reply but it's clear you didn't read most of what I typed up.     I am NOT using the stock thermistor, I didnt mention that at all.  Im using 100k NTC 3950 thermistors.  Same on both the extruder and the hot bed.

   Anyway I figured out the problem.    When I replaced the factory Davinci thermistor and installed the nee 100k glass thermistor, it was sandwiched in-between the glass/heated element and the factory fiberglass insulation mat.   The difference in heat from the fiberglass mat underneath it and the heating element above it created almost a 20 C difference. Which explains my variances posted in the chart above.       Excuse the fingerprint smudges from the thermal compound on the top of the tape in the pictures,  but I added a blob of thermal compound around the thermistor glass bead and used Kapton tape to thermally insulate the thermistor directly to the heating element.

Just like that the problem is fixed!    The temperature read outs are spot on now!      Works great!   

Struggling to get the bed heater temperature to read correctly on the Ramps 1.4..    I bought brand new 100k NTC 3950 thermistors on Amazon (since I used old spare ones and couldn't figure out what they were).   I have it programmed in Repetier 0.92.9 correctly as "100k NTC 3950".      It still shows the temperatures way way off.  Like this:
60c displayed on the LCD is = 105c measured by a laser thermometer
55c displayed on the LCD is = 82c measured by a laser thermometer
50c displayed on the LCD is =65c measured by a laser thermometer
45c displayed on the LCD is  = 54c measured by a laser thermometer
40c displayed on the LCD is = 46c measured by a laser thermometer

   The printer originally had a ferrite core (small black magnetic clamp) which was clamped on the heater wires and the thermistor wires. 

When the NEW thermistor wire is in the ferrite core with the heater wires,, the readings are insane, like 900c.   When I take the wire out of the ferrite core it settles down to about 20c which is room temperature in that room. So its causing some variations in the readings for sure.

Any ideas?   I'm at a loss.

If you are using the stock sensor it is probably your issue. Unless there have changes made to Marlin and or Repetier there is no table in them for that oddball thermistor that XYZ used. You need to use the same type of standard 100k NTC thermistor that the hotend uses. Meaning you will have to figure out a way to mount it. I dremeled a grove out in the bed base material so the thermistor could be mounted under the heater.

   The printer originally had a ferrite core (small black magnetic clamp) which was clamped on the heater wires and the thermistor wires. 

When the NEW thermistor wire is in the ferrite core with the heater wires,, the readings are insane, like 900c.   When I take the wire out of the ferrite core it settles down to about 20c which is room temperature in that room. So its causing some variations in the readings for sure.

Any ideas?   I'm at a loss.