Piezo discs for bed leveling - Fallacies and truths.

I have followed bed leveling methods almost since I made my first 3D printer more than 7 years ago. I tried many different sensors over the years since but have concentrated on nozzle contact detection with piezo discs since I first saw Njål Brekke’s video on YouTube more than four years ago [www.youtube.com]

Looking through several forums I have been astonished at how often the same tired old falsehoods about piezo z probes are repeated. In the hopes of correcting those I have listed the most common ones and my take on it – gathered from 4 years of experiments involving hundreds of piezo discs, four printers and trials with many different designs of sensors.

The most common falacies and half truths:

1. Piezo discs used in Z probes are microphones that detect the sound of nozzle contact – Generally untrue
2. Piezo discs need to be bent to produce an output – Not true
3. Piezo discs only produce an output with change in pressure – Not true
4. The characteristic frequency of a piezo disc is important – It is completely irrelevant
5. The nozzle needs to contact the print bed at speed to produce a usable output – Not true
6. The nozzle needs to contact the bed with force to produce a usable output – Not true
7. Piezo discs are temperature sensitive and lose output at higher temperatures - True but can be worked around with good engineering.
8. Piezo discs can give false signals from external noise - True but can be worked around with good engineering.

Taking these one at a time:-

1. Piezo discs have been used as microphones by injecting an acoustic signal through the nozzle and detecting it with a second piezo disc in the bed. This method has a lot of promise but does not seem to have gone any further than a Hackady article by Moritz Walter [hackaday.com] I have not been able to find any example where a piezo disc is used to detect the sound of the nozzle contacting the bed.
2. Piezo discs do not need to be bent to produce an output. Direct pressure on the surface of a piezo disc will produce a healthy output: Bending the piezo disc will need less pressure but has more compliance and hence is less rigid. Although very implementation dependant, bending the disc will give about 2.5 times the output but will move about 2.5 times more than a piezo disc under simple direct pressure.
3. Piezo discs consist of a non-conductive piezoelectric material between two conductive surfaces. When the piezo material is stressed a voltage will appear across the conductive surfaces and only internal and external leakage will cause this to decay. If the decay is slow enough then the output will represent the applied pressure – or bend in the case where the disc is bent.
4. The characteristic frequency of a piezo disc is the frequency at which it will resonate if held by the edge with nothing but air contacting the rest of the disc. In any form of sensor there is necesarilly something touching the disc so the characteristic frequency is moot.
5. No great speed is needed for a piezo disc to respond to nozzle contact. Almost the first recording I got of a piezo disc being used for nozzle contact detection gave an output of 8V at only 1mm per second [reprap.org] Presently I am developing a system where I am using a contact speed of 0.5mm per second.
6. The nozzle does not need to contact the bed with a great deal of force. Even at 20 grams of force there is no detectable witness marks left on Kapton tape or blue tape even when examined with a microscope. Reliable detection of nozzle contact is possible with as little as 2 grams of force.
7. Some piezo discs do lose almost all sensitivity by about 60°C but Murata ones are good to about 80°C if they are pre-conditioned by soaking for several hours at 80°C. Sensitivity will drop and then rise again.
8. Piezo discs are somewhat more subject to false triggering from mechanical noise than other sensors. The time of greatest concern is at the end of and X and/or Y movement directly before the Zpop (Z probe operation) when the printer may be shaking from the deceleration at the end of the movement. The workaround for this is to introduce a short delay between the end of the XY movement and the start of the probe operation.
   Other causes may be too violent acceleration on the downward movement at the beginning of the Zpop or even just general external mechanical or electrical noise.

I am presently working on a new piezo signal conditioning circuit that looks for the clean straight line rise in voltage which is a characteristic of a nozzle being driven down by a lead screw. I will make the programming and schematics available in the public domain once it reaches Beta testing.


Mike

Edited 1 time(s). Last edit at 08/10/2019 11:44AM by leadinglights.

Nice write-up Mike!
I have a universal kit and a set of andromeda's to boot.. So I have faith.
I've just to get my lazy butt to install them!

Cheers!

I have been doing some trials on a piezo conditioner for Z probing and have some results that may be useful to others. The particular test was probing at 5mm per second, something I have not done before as I find 1mm/sec or 2mm/sec quite sufficient. The plot below shows a distinct hump which I have not observed before, this hump is caused by the nozzle bouncing on the bed.



The significance of the bounce is that a trigger becomes unreliable if it occurs anywhere near the hump - will it show contact on the first bit or the second. Although the error here would be less than 10µm, it should be realized that this printer is set up to be as rigid and predictable as I can get it - in some commercial, kit and self-built printers, this may easily be an order of magnitude worse, particularly at probing speeds more than 5mm/sec.

Mike