Bed leveling probes
A summary of printer bed leveling probes for bed tramming, tilt compensation and flatness error correction.
Work in progress
Much of the knowledge base on 3D printers now revolves around which printers and components are the best purchases in terms of performance, price, or ease of use. This knowledge is useless when the manufacturers discontinue their existing models or go out of business altogether. Although many commercial probes are depicted and discussed below, the purpose of this summary is not to recommend or suggest any device or method, but to show how they work and the strengths and problems with each.
The probes shown below are all more or less usable and of sufficient accuracy for the 3D printers that they have been used on. There are problems with all of the Z-probes, sometimes nothing more than the need for a specific print surface, and sometimes unexpected gross inaccuracies on a printer that had previously been exceptionally accurate. With almost every problem noted in this summary, there has been some work to reduce or eliminate the problem.
Notes on terminology:
The term “Bed Leveling” will be used in this document as a general term to cover out of flatness correction as well as assisted manual bed adjustment and automatic bed tilt compensation. The term “Bed Tramming" is only used to cover the adjustment of the mounts of the bed to adjust the bed plane normal to the X and Y axes. The single word “Tramming” has been avoided as this implies setting all three axes mutually perpendicular. The term “Bed” will be be used instead of “Build Surface” or other terms in the interests of brevity. This document only applies to FFF 3D printers
_Contents
Manual Bed Tramming
The earliest, and still accepted way of tramming a 3D printer is by hand. To manually tram the printer bed, the nozzle is moved directly above one of the adjustment points on the bed; a piece of paper is placed on the bed and the nozzle is lowered until a slight resistance is felt when trying to move the paper. This point is set as zero on the Z axis in the software. The nozzle is moved above each of the other adjustment points and the height of the bed is adjusted so that the same resistance is felt with the nozzle at zero. Feeler gauges or pins of a known diameter are preferred by some.
When manually adjusting the tramming, it's important to note that the zero position may differ slightly from the true zero due to the thickness of the paper or tool used. For instance, with 80gsm paper or Post-it notes, this difference could be around 80µm to 110µm, and can often be overlooked. In cases where the gauge thickness is significant, as with a 2mm pin, the height of the nozzle tip above the bed during measurement is referred to as the Z-offset.
LED Bed tramming tool.
A tactile switch with a battery and LED is used instead of the paper test or a feeler gauge, An example is the Filament Friday E-Leveler 2 Using a bed tramming tool eliminates the subjective feel of the paper test but it is less suitable for generating a bed mesh.
Manually attached z probes.
A common early method of partially automating bed tramming was to temporarily attach a microswitch or tactile switch to the carriage of the printer. This method is seldom used now although a manually attached Z-probe is sold by Trianglelab https://trianglelab.net/products/z-probe-1 as the High Precision Z-probe This consists of a thin membrane switch secured to the printer nozzle by an elastomeric collar.
Manually deployed probes.
Z probes which are attached to the printer carriage but can be moved by hand between parked and deployed positions.
A simple yet effective manually deployed probe is based on an Allen key. In its parked position the tip of the Allen key is above the nozzle, while in the active (deployed) state, the key is below the nozzle and aligned with a microswitch.
Automatic bed probes
In automatic bed probing, the height of the bed surface is measured at multiple points under the control of a computer. Each measurement is typically taken by lowering a sensor from a fixed height above the bed until it detects the presence of the bed. The sensor can be non-contact, like capacitive, inductive, or optical sensors (referred to as proximity sensors), or alternatively, a mechanical probe can activate a switch or another sensor upon contact with the bed surface. The mechanical probe can be a discrete device, or a nozzle on the printer, in which case the switch or sensor can be located in the carriage to which the nozzle is attached, or underneath the printer bed.
In other probing methods, the bed is scanned by a sensor in a plane parallel to the bed, obtaining a measurement of the distance to the bed at multiple points along each scan line. By eliminating the need to move the probe up and down at each point, the mapping of the bed becomes significantly faster and less susceptible to errors. The sensors used for scanning can be capacitive or inductive types with an analog output, or time-of-flight devices such as LiDAR or acoustic probes.
Z-offset is the height of the nozzle from the bed when the sensor detects the presence of the bed. Most Z probing methods will have quite a large Z-offset while methods using nozzle contact will have very little, if any Z-offset. In sensors where the sensor output is a measure of the distance from the nozzle to the bed, this measurement will stop changing when the nozzle contacts the bed and this type of sensor can be used both as a scanning sensor and as a nozzle contact sensor.
Probes other than nozzle contact types will have additional offsets on the X and Y Axes. While this can be compensated for in software, with some geometries, such as Delta printers, the errors can be difficult to correct. In these cases, the errors can be somewhat reduced by positioning the probe as close as practical to directly under the nozzle.
Self-deployed probes.
Z probes that are deployed or retracted by using the normal X, Y & Z movements of the printer to move a probe from the parked position to the active position.
In the self-deployed probe depicted below, the probe is held in the parked position by two magnets but can be extended so that the tip of the actuator is below the nozzle by moving the carriage upward so that the upper tip of the actuator rod is pushed down by a stop on the frame of the printer.
With the actuator down a flag blocks an optical path and bed detection occurs when the actuator is lifted so that the optical path is open.
Retraction is effected by moving the carriage to a position where the nozzle can be moved below the bed level while the actuator is in contact with the bed.
A problem with self-deployed probes is that there are a large number of geometrically important positions, each to be defined with X, Y, and Z coordinates, and possibly speed: The position of the carriage before extending the carriage and the position to end the deployment movement; the positions at the beginning and end of the retracting movement need to ensure that the nozzle does not strike the plate and that the carriage does not move down past the point where the magnets are in contact.
Powered deployment probes.
Probes where electrical power is supplied to extend the probe beyond the nozzle.
Servo Probes
While the servo-deployed microswitch is simple and reliable, the most significant reason for their adoption is that such servos and the knowledge of how to use them were available in the maker community. The shortcomings of this arrangement are the mass of the parts and the mechanical play of the bearings.
Solenoid Probes
Dockable probes.
Z-probes in which either the whole probe or the effector part of the probe is only connected to the carriage or effector to probe and parked remotely during printing.Examples on GitHub are Klicky; Quickdraw; Euclid
Dockable probes typically use a microswitch and are accordingly acceptably accurate as well as being inexpensive. By removing the probe from the carriage they minimally contribute to carriage mass.
A disadvantage of dockable probes, as with self-deploying probes, is that they need defined positions for the probe to be deployed and stowed. As well as contributing to the complexity of the software setup they may restrict some part of the printable area.
Dockable Delta Probe.
A dockable probe using a piezoelectric sensor is shown below. In this case, only the actuator is moved between the effector and a parking platform.
This probe is used only to generate a bed mesh while using a contact force of less than 10 mN. The Z offset is measured by nozzle contact directly over a second piezoelectric sensor mounted under the bed and the contact force is again under 10mN.
This probe can be seen being deployed, stowed, and taking both bed mesh and Z-offset at YouTube
Proximity Z-probes
Proximity sensors are widely used in industry. Capacitive and inductive sensors are particularly popular and have been integrated as Z-probes in 3D printers without any modifications. On the other hand, optical proximity sensors used in industry have generally shown poor accuracy, leading to the development of purpose-built optical Z-probes for 3D printers. While ultrasonic sensors are available, they have not gained widespread adoption for 3D printer bed leveling.
When setting up a printer with proximity Z-probe, it is usually necessary to measure the Z-offset separately – often a manual procedure. However, if the sensor can measure the distance to the bed, it becomes practical to determine when the nozzle makes contact with the bed. This principle is used in the Ultimaker 3 Capacitive Probe and on the Beacon Contact Eddy Current Probe. Moreover, placing a proximity probe under or very close to the printer nozzle is impractical, which necessitates adding offsets for the X and Y axes to the settings for the printer.
Capacitive Probes.
Capacitive sensors, widely used as Z-probes on 3D printers, are a popular choice due to their compatibility with most bed surfaces. They are offered in various physical sizes and voltages including 5V versions. However, factors such as the humidity of the bed surface material, composition of the bed surface, presence of different materials like metal screws, and variations in temperature and supply voltage can impact the absolute accuracy of capacitive sensors.
Capacitive probe with nozzle contact detection.
The Ultimaker 3 incorporates a capacitive sensor to serve as a nozzle contact Z-probe. This sensor generates an output that is proportional to the inverse of the distance from the sensor to the bed. When the nozzle makes contact with the bed, the distance, and consequently the output, stops changing, signaling detection of the bed surface.
Find graphic
Inductive Z-probes
The inductive proximity sensor has a very acceptable accuracy and is available in similar physical sizes and supply voltages to capacitive sensors, they are a very common sensor in commercial 3D printers, most notably on Prusa printers. Inductive proximity sensors detect conductive objects by inducing a current into them and then detecting the induced field. While some inductive sensors have some thermal drift, others have built-in temperature correction.
Inductive proximity sensors are widely used in commercial 3D printers, notably in Prusa printers as well as being available for retrofitting or design. They offer a satisfactory level of accuracy and are available in comparable physical sizes and supply voltages to capacitive sensors. Inductive proximity sensors operate by inducing an oscillating magnetic field into conductive objects and then detecting the induced field. Because of this, they must be used with a conductive (presumably metal) bed While some inductive sensors may experience thermal drift, others are equipped with built-in temperature correction.
Inductive proximity sensors are commonly used in commercial 3D printers, notably in Prusa printers, and they can also be adapted to or designed into other printers. They provide a satisfactory level of precision and are available in similar physical dimensions and supply voltages as capacitive sensors. These sensors function by generating an oscillating magnetic field in conductive objects and then detecting the resulting field. For this reason, they should be used with a conductive (typically metal) bed. Although some inductive sensors may encounter thermal drift, others come with built-in temperature correction.
Eddy current probes.
Eddy current sensors differs from traditional inductive probes both in their construction and their operation. Unlike older models, they lack a ferromagnetic core and operate at a much higher excitation frequency, and, unlike traditional inductive sensors, the eddy current probe provides a proportional value for distance to the surface, allowing it to scan the surface at speed and map it, rather than providing a binary on/off signal at discrete points. GitHub, Duet3D Scanning Z Probe; GitHub, BD Sensor
The eddy current probe has the advantage of combining effective resolution with a fast response speed. These qualities make it suitable for use as a nozzle contact probe because the distance value stops changing when the nozzle touches the bed. This principle is used in the Beacon Contact Z-probe.
Optical Z probes
Differential IR probe
LiDAR probes
Bambu
Underbed sensors
Underbed switch
Among the earliest investigators on underbed sensors for bed tramming was Joseph Prusa who showed that a method as simple as having tactile switches under the corners of the bed. Joseph investigated a method of fitting tactile switches under the corners of the printer bed and a video of this can be seen at YouTube
Force Sensing Resistor
Force-sensing resistors were popularised by Johann Rocholl on Kossel Delta printers. The earliest implementations used three FSRs equally spaced around the periphery of the bed. Although simple, accurate, and reliable, this arrangement was not suitable for use with a heated bed and it became necessary to insulate the FSR elements from the heat of the bed.
Piezoelectric Disk.
With piezoelectric sensors, it is important to note that they respond to changes in force rather than to static force. In the context of bed mesh generation and Z-offset measurement, it is dynamic events that we are focusing on, and a static measurement is not required. By employing signal conditioning with a high-impedance amplifier that has a frequency response ranging from less than 1Hz to several hundred Hz, we can obtain data that indicates when the nozzle comes into contact with the bed. When well-designed, this data can even reveal the quality of the contact, such as whether the nozzle was clean or contaminated with plastic residue.
Piezoelectric force sensors are accurate, consistent, and dependable [1], though they are not ideal for installation within a printer's carriage. They can, however, be placed under the bed of a printer, but the use of underbed sensors presents its own set of issues.
Acoustic Sensors.
Transducers, typically piezoelectric ones, have been used to detect vibrations from contact. This detection can occur passively, where the noise generated by the nozzle contacting the bed is sensed, or actively, where mechanical vibrations are transferred from a transducer in the hotend and detected by a receiving transducer in the bed. One further approach involves using a single transducer to both generate the vibration and detect the nozzle contact. The intensity of the induced vibration decreases upon nozzle contact, similar to lightly touching a vibrating guitar string with a finger.
Acoustic transducers can be used as proximity sensors by sending out a pulse and detecting the returned echo.
Strain Gauge
Strain gauges are robust, accurate and sensitive and have an acceptably fast response when used for 3D printing Z probes. The output from a load cell is linear and minimally influenced by thermal considerations and the output can be stable over long periods. Unfortunately, the design and placement of sensors to reduce the effect of unwanted mechanical and thermal stresses is not trivial and the cost of load cells and conditioning circuitry are greater than other sensor technologies.
Strain gauges used in bed tramming have typically been either beam-type load cells or foil strain gauges bonded to a mechanical part of the carriage/effector or the bed supports. Beam load cells originally designed for utility weighing scales, along with the associated conditioning circuitry, have been adopted.
Nozzle contact sensors.
Electrical contact sensor.
The Lulzbot Mini uses electrical contact between the nozzle and washers mounted on the corners on the bed. While simple, this method does require the nozzle to be clean enough to ensure good electrical contact and does not allow areas of the bed other than the corners to be probed.
Voron Tap
As described on the GitHub page for the Voron Tap, "Tap is a nozzle-based z-probe for the V2 and Trident printer designs. The entire toolhead moves to trigger an optical switch." GitHub
The Voron Tap is exceptionally accurate and reliable, but with a nozzle contact pressure of 500 - 800 grams, it is only suitable for printers of rigid construction and having a robust print surface on the bed.
Piezoelectric Element
An early example of a piezoelectric element mounted to a printer carriage was made by DjDemonD RepRap Forum This evolved into the several nozzle probes designed and marketed by Precision Piezo including the Orion. A more recent piezoelectric nozzle contact probe is manufactured by E3D and sold as the PZ Probe
Strain Gauge
Duet3D produces the Smart Effector for Delta printers. This uses a 3-element strain gauge which is made from the copper of the printed circuit board.
Accelerometer Sensors
Accelerometers installed on the carriage or on the effector of a Delta printer have been utilized for bed tramming. An instance of this is the Rostock V3, and the image below displays the effector of this printer.
Clean detection of the deceleration of the nozzle upon contact with the bed needs a rigid system: Soft bed surfaces, contaminated nozzles, loose or elastic drive belts non-rigid printer frames can all contribute to an inaccurate or possibly failed detection. Existing accelerometer probes use a very high probe speed, typically 30mm per second.
Sensorless Z homing.
Some stepper motor drivers, notably those from Trinamic, [2] have a feature called "Stallguard" designed to detect the approach of a stall condition; a condition where the load on the motor may cause steps to be missed. This feature can detect the nozzle contacting the bed and is found on several printers such as the Original Prusa MINI+.
Sensorless homing is unsuitable for Delta printers or Cartesian printers with 2 or more steppers with independent drivers. Despite this, the Stallguard feature is useful on any printer where there is an appreciable risk of a nozzle contact going undetected.
