Keeping a constant distance to a curved surface


Thread Starter

Suttipan Limanond


We are designing an adhesive dispensing system which requires the glue head to be at a constant distance from the surface where glue will be deposited. I am thinking of using a range sensor but wonder if this is really necessary. Further, I am not quite sure how to use this measurement anyway since most commercial motion control boards normally take the feedback signal from the motor encoder.

So I would appreciate if anyone could tell me if range sensor is really needed in this application. And, if so, how should I proceed?

Thank you very much

An application we made we used a ultrasonic distance sensor with an analog output. With this analog output we changed the distance of the arm which needed to be as constant as possible to the object.
The data from the sensor needs to placed in a control which will change the position of a arm.

In your application I would probably search for a good analog sensor, a PLC and an position controler like the servo controler for a the needed servo motor. There are servo controler which can be programmed so a PLC controler would not be needed.

Good luck, Sisko Bos
Application engineer
[email protected]
We have had very good luck in this type of application using a hybrid control of the vertical axis. There is the standard position control from an encoder, necessary if nothing else but to bring the axis into approximate range to start the "flying height" mode, and to take it out at the end. Then there is the "height" mode running off a range sensor (capacitive or ultrasonic usually) into an A/D converter on the control board. To minimize the transition problems, we control the height as a separate software axis using the A/D value as feedback. It does not have its own axis, but its output can be used as an electronic gearing input into the standard positioning vertical axis. The position loop is still active, but its setpoint in this mode is continually modified by the height loop.

We have used this successfully in many laser auto-focusing applications that I believe run much faster than what you would be doing with a dispensing application.

In any case, I don't see how you can do this without some kind of height sensor and some method of closing the loop.

Curt Wilson
Delta Tau Data Systems

Cameron Anderson

Suttipan, your application is easy. Use a Banner U-Guage ultrasonic or the L-Guage laser with a analog output and bring it into IDC's H3501(Analog Position Controller) and an Electric Cylinder to maintain uniform spacing between the glue head and the surface by following the analog output of a distance sensor. Go to IDC ( for the controls and actuator and to Banner ( for the sensors.

Cameron Anderson
Motion Control Specialist
This has been done with machine tools. It is called cutter radius compensation. The idea is to keep the servo path a constant distance from the part while using a round edged cutter. The cutter can do straight lines or arcs while maintaining the servo path away from the work
by the distance of the cutter radius. Since the part path was programmed into the controller, the radius compensator is an offset to that path which is accomplished by using a secondary
linear/circular interpolator with the cutter radius as the radius of the interpolation.
Is this an automated machine where the glue path is well defined?
If so, this technique is viable.
If there is not a well defined path, then you will probably have to sense distance as you suggested.

Thomas B. Bullock, President
Bull's Eye Marketing Incorporated
Industrial Controls Consulting Division
104 S. Main Street, Suite 320
Fond du Lac, WI 54935
PH: 920: 929-6544
FX: 920: 929-9344
E-mail: [email protected]

Michael Griffin

I have a little bit of experience with dispensing adhesives in beads (although I am not an expert in this). I haven't done anything with a spray system.
In cases where you are dispensing a bead to a known surface in a simple pattern, it is fairly common to simply control the height with a cam. The stepper (or servo) simply drives the nozzle at a constant speed but the cam controls the height. A stepper (or servo) drive is used to give a known and constant speed to get acceptable process capability (glue weight).

In applications where glue is being dispensed in a complex pattern which is not suited to a cam, or where the same machine must follow many different patterns with various products, all the systems I have seen used
some sort of small robot. Some robot manufacturers (e.g. Adept) even have special glue dispensing packages or configurations. In this case the robot manufacturer is offering you an off the shelf dispensing solution. I have seen similar systems offered by companies which make glue dispensing equipment. Some of these robots are of the simple X-Y gantry type.

In both cases I have only seen systems in which the nozzle followed a fixed path while the part is securely fixtured. It is possible the robot systems many offer following the surface as an option, but I am not sure on this point.

If you are trying to do this with a motion controller, I suspect you will need a rather sophisticated system which offers this as a built in option (some sort of electronic camming with feedback) to get acceptable

Michael Griffin
London, Ont. Canada
[email protected]

Davis Gentry

I can think of two ways to do this - either the range sensor, as you indicated, or if you know the profile of the piece to which you are applying the adhesive you can progam that profile, with an appropriate offset, as a motion program.

Our (Delta Tau) motion control boards will take a
feedback from an MLDT, or from an analog feedback, or a number of other sensors as well as standard resolvers or encoders.

Davis Gentry
Applications Engineer
Delta Tau Data System
The servo axis 'Z' must maintain constant distance from a surface.
The surface will be in motion, and the axis 'Z' must adapt to constantly changing part height.

The feedback from the height transducer is a secondary feedback.
The servo loop itself would be a typical motion loop
(encoder feedback on the back of the motor typically)
The position command or target position, comes from the height sensor.
The servo controller changes the command positon as referenced to the servo slide to 'adapts' the height of axis 'Z' on-the-fly.

As you probably recognize, it is better to not have the sensor near the glue, if you can avoid it.

Option one is do do exactly as described above.

Option two is to record the positions manually jogging the 'Z' height at many locations.
This teaches the points for the height locations. Although not as fancy, it is more practical on simple parts.

A third option is to use a sensor to generate the profile, without applying the glue. This task records the 'Z' height profile into memory. It can be played back from memory at the required application speed. (I tend to prefer this to keep the sensor away from the adhesive)

Some things to consider, effecting control system and/or sensor selection
Sensor type and location
Update time and part pitch (slope)
What is you height tolerance?
Surface finish
Remember sensor is most likely offset from application point. This can present challenges.

I have worked with several motion controller that support this function. You have several choices on controls, depending on your requirements.

What are your cost objectives, and programming background ?

Let me know if you have any additional questions

Dave Kane
[email protected]

Guy H. Looney

I've done a lot of applications w/ dispensing adhesive. I've successfully used analog feedback for height adjustment as your question illustrates. I've also used the setup for flow control to compensation for bubbles developing in the line. I've been reading the responses, & I've got a few comments:

1) The guys at Delta Tau are correct as usual. Based on personal experience, Galil, Optomised Control (now Baldor), ORMEC, & Yaskawa could do
this (I'm sure there are more, I've just haven't used them). Program the motion based on the "typical" pattern. Next wire in the device w/ analog feedback (there are a blue million of them......from a simple LVDT to a super sensitive sensor from Keyance) that gives you the resolution & responsiveness you need. Finally set up the controller to calculate the
difference between the zero point ("typical" pattern) & the offset point (actual pattern). As long as the system can calculate the corrections fast enough you'll be fine.
Quick example: Line speed of 10" / second & repeatability requirement of 0.004" (4 thousandths). I found an analog sensor w/ a
resolution of 0.000001" (1 micron) at 0.1 msec update. It's extremely important to realize that the resolution of the sensor is directly
proportional to the update rate (slower update rates allow for higher resolutions). The line speed of 10" / second tells me that a 1msec update
rate allows me to "see" the path every 0.001" (1 thousandth). The resolution of the sensor is more than I need, but the update rate is just as
important as the resolution. One last thing to consider. The color of the object you're detecting will affect the resolution / repeatability of the sensor. The darker the object, the worse it gets.

2) I would not recommend a PLC for this application unless the line speed is extremely slow. PLC's would typically take the analog feedback into a standard I/O rack & then have to communicate the value to the motion control card. This would take several milliseconds. A dedicated motion control card could process this information & update the position at the microsecond level. As seen in my example above, update rate is extremely important in this application.

3) Electronic camming would only be useful if the pattern was constant. Electronic cams (which are very cool by the way) are designed to work like mechanical cams. I've yet to see a mechanical cam that can compensate for changing heights. Let me back up for a second; electronic cams could actually compensate for changing heights by placing an offset value with each point in the cam table & using the analog value to set the offset value. However, this would probably take longer than simply correcting a
general path. Either way, more & more motion controllers are incorporating all the features that would be needed. You could pick up a 3 axis
controller for less than $3,000 & accomplish about anything you'd need.

4) As far as a specialized robot goes, I've done all my applications with ball screw & belt tables (some off-the-shelf, some customer built from
components). The customer mounted them in x, x-y, x-y-z, & x-y-z-theta formations. Sure, Adept & other such companies make some neat stuff, but
it's overkill for a lot of applications. By the way, I've done the applications w/ steppers, servos, & a combination of both.


Guy H. Looney
Motion Control Engineer

A.C.E. Systems, LLC
170 Medearis Drive
Old Hickory, TN 37138
email: [email protected]
work: (615) 754-2378
fax: (425) 944-5017
cell: (615) 330-0044

Suttipan Limanond


I just want to thank every one who kindly responded to my questions, especially those who replied at length. As a result, I now start talking with a few motion control and robot suppliers whose products are capable of `loop-over-loop' control. And the project seems to be heading in the right direction. Thanks to all the experts on this list.

Suttipan Limanond.
Just contact Creative Automation.
They have Automatic glue Dispensing equipment that contain special software to dispense glue in almost any situation.
Ask for Jack Engel.
Best regards,
Just contact Creative Automation.
They have Automatic glue Dispensing equipment that contain special software to dispense glue in almost any situation.
Ask for Jack Engel.
Best regards, 800-688-6220