Integrating a BCL-AMP capacitive sensor to LinuxCNC

I'd be willing to test this out and provide some data. I currently use the BCL-AMP setup with QTPlasmac and while it works, it could definitely be faster in its tracking. The sensor outputs the nonlinear signal which is also noisy. After filtering (hardware and software), it works, however I feel as though it is not sensitive enough now due to being averaged and filtered so much that any abrupt changes in height will cause the nozzle to touch.

A cleaner, linear response signal could be the ticket, as I have fiddled with the PID parameters quite a bit as well as my Z axis velocities and accelerations and it is still not really where I'd like it to be.

Let me know how I can be of help.

Uthayne is your man. He is very knowledgeable in this area.

Uthayne wrote:

Let me know how I can be of help.

 

Hello, thank you for your willingness to help.
As I mentioned, I’m not proficient in LinuxCNC, but I have great respect for it.I’d like you to let me know what you need, and I’ll do my best to adapt the product to your requirements and send it to you for testing. I think we should start a new thread to avoid cluttering the current one.I’m very committed to keeping the topic open and accessible to everyone without any secrets.Criticism, in particular, has always been, is, and will continue to be motivating for me.

ps. to introduce myself I present my child:
My Proma-Elektronika company

A cleaner, linear response signal could be the ticket, as I have fiddled with the PID parameters quite a bit as well as my Z axis velocities and accelerations and it is still not really where I'd like it to be.

Have you considered linearizing the output (with lincurve) before feeding the linearized distance to the
PID loop?

PCW wrote:

A cleaner, linear response signal could be the ticket, as I have fiddled with the PID parameters quite a bit as well as my Z axis velocities and accelerations and it is still not really where I'd like it to be.

Have you considered linearizing the output (with lincurve) before feeding the linearized distance to the
PID loop?

This has already been done.

Have you seen linear graphs?

This has its drawbacks though.

Non-linear characteristics are very good because they have higher resolution/accuracy for useful focus ranges and lower for a larger distance which is not very important but you are right, it is more difficult to handle for a standard PID.

The sensor can be switched between linear and non-linear characteristics as needed and desired.

PCW wrote:

A cleaner, linear response signal could be the ticket, as I have fiddled with the PID parameters quite a bit as well as my Z axis velocities and accelerations and it is still not really where I'd like it to be.

Have you considered linearizing the output (with lincurve) before feeding the linearized distance to the
PID loop?
 

 

PCW: That is not something that I had thought to try. I'll have to give that some thought and see how it performs. Thanks for the tip


prokopcio, I'd recommend starting a new thread in this subforum so we can discuss your design. Looks promising!