Precision Piezo Z-probe Now available.

@Moriquendi

Ok sounds good to me. If you could stop making me want to buy more printer related things that would be great though

I've only really briefly looked into the piezo - I assume there will be some good mount ideas kicking around on the internet. Ill be probing onto glass if that matters?

Quote
Ed3D

Ok sounds good to me. If you could stop making me want to buy more printer related things that would be great though

But then you wouldn't be giving me any money...

Quote
Ed3D
I've only really briefly looked into the piezo - I assume there will be some good mount ideas kicking around on the internet. Ill be probing onto glass if that matters?

There are all sorts of mounts floating around but none so far for the threaded V6, however I'm about to start working on one. Glass is ideal for a piezo probe.

Idris

---

{Precision Piezo} Accurate, repeatable, versatile z-probe plus piezo discs, endstop cables, pt100, 50w heaters.

@Moriquendi
I have v1.2, it has the four 3-pins.

Forgive me if this is stuff you already know

With the v1.2 Orions the output is open-drain. What this means is that the output can be pulled low by the Orion but it cannot be pulled up. When the Orion is plugged into a controller and the controller has pullups enabled there is a large value resistor (20-100k) between the signal line and V+ so the signal line is pulled up to V+. When the Orion is triggered it shorts the sig pin to GND.

There are three possible causes for the problem you're seeing that I can think of.

1, You don't have pullups enabled on the pin that the Sig pin is connected to

2, You've somehow damaged the pin on your controller (ramps?)

3, You've damaged the T2 transistor on the Orion so it's always closed.

Can you measure the voltage of the pin on the controller without the Orion plugged in?

Idris

---

{Precision Piezo} Accurate, repeatable, versatile z-probe plus piezo discs, endstop cables, pt100, 50w heaters.

@ Moriquendi

How close does the disk need to be mounted to the hot end? Can I put it between the carriage and the Y axis or does it need to be right above the hot end?

Edit: Like this:



Edited 1 time(s). Last edit at 12/03/2018 02:13PM by Ed3D.

Irdis might have a slightly different answer, but...

In general, the fewer components in between the nozzle and the disc, the more accurate the sensor will be. Any joints or long pieces of material adds possible inaccuracy tot he probing point. What's most important is consistency, so the most direct path from the nozzle tip to pressure on the piezo disc is best. Anywhere that can flex or compress in between the nozzle tip and the piezo disc adds some uncertainty to the trigger point, which will translate into possible variance for the sensor system as a whole. Also, since the piezo more or less depends on movement, you want to decrease the amount of possible movement that the nozzle tip can have, as that adds a whole new possibility for error in the actual finished print. Therefore you want to restrict nozzle movement to one dimension (Parallel to the Z axis, ideally.) A very small amount of motion should be all that is required, less than 0l.1mm is adequate. I made 3 or 4 different types to test with, and they all work at less than 0.05mm deflection.

---

MBot3D Printer
MakerBot clone Kit from Amazon
Added heated bed.

Leadscrew self-built printer (in progress)
Duet Wifi, Precision Piezo parts

I'm kind of concerned about the hot end wobbling / not being as rigid if its clamped through the piezo. My other concerns is how the fillament guide would no longer mount to the hot end and so is pushing through the piezo too so extrusion / retraction will trigger the sensor? Maybe I'm misunderstanding on this one!

Is the Orion better / more rigid than the simple disk? Could I integrate the Orion into my print head and are there mount point dimensions / a solid model / a dimensioned drawing available?

Is it recommended to use the hot end sensor instead of sensors under the build plate? The Andromeda looks great and it would make my wiring and print head packaging a lot easier. Can you quote some standard deviations between the different models?

I think I'm set on a piezo right now I'm just wondering what the best config is. I'm using a coreXY motion type if that makes any difference - apologies for the barrage of questions!

Edited 4 time(s). Last edit at 12/03/2018 05:51PM by Ed3D.

Quote
Moriquendi
Forgive me if this is stuff you already know

With the v1.2 Orions the output is open-drain. What this means is that the output can be pulled low by the Orion but it cannot be pulled up. When the Orion is plugged into a controller and the controller has pullups enabled there is a large value resistor (20-100k) between the signal line and V+ so the signal line is pulled up to V+. When the Orion is triggered it shorts the sig pin to GND.

There are three possible causes for the problem you're seeing that I can think of.

1, You don't have pullups enabled on the pin that the Sig pin is connected to

2, You've somehow damaged the pin on your controller (ramps?)

3, You've damaged the T2 transistor on the Orion so it's always closed.

4. Intermittent error

5. User error

Hooked it up on the breadboard this morning. Power from a lab PSU and with a 4.7k pull up . Worked like a charm.
Hooked it into the printer. Worked like a charm.

Whatever I did yesterday was obviously wrong.
Thanks for the help

Edited 1 time(s). Last edit at 12/04/2018 03:01AM by bondus.

Quote
Ed3D
I'm kind of concerned about the hot end wobbling / not being as rigid if its clamped through the piezo. My other concerns is how the fillament guide would no longer mount to the hot end and so is pushing through the piezo too so extrusion / retraction will trigger the sensor? Maybe I'm misunderstanding on this one!

Is the Orion better / more rigid than the simple disk? Could I integrate the Orion into my print head and are there mount point dimensions / a solid model / a dimensioned drawing available?

Is it recommended to use the hot end sensor instead of sensors under the build plate? The Andromeda looks great and it would make my wiring and print head packaging a lot easier. Can you quote some standard deviations between the different models?

I think I'm set on a piezo right now I'm just wondering what the best config is. I'm using a coreXY motion type if that makes any difference - apologies for the barrage of questions!

I have had no problems with hotend wobble, it will be rigid both with orion and or with a drilled piezo.
It triggers like mad while printing, but that is not a problem. The controller will not care about it while you are printing.

I would say the Orion is nicer to work with, it's easier to get tuned properly and it sits well in place, and no extra board to mount.

For your corexy I would say a hotend sensor is easier than an underbed. I have underbed on my delta and an Orion on my i3 clone , now rebuilt to a Hypercube EVO.
Search on thingiverse and you will find many hotend mounts.

@Bondus,

Glad it's working for you now.

@Ed3D

Every controller I've used or even heard of ignores Z-probe triggers except when specifically probing. Triggers during printing have no effect.

The Orion is more rigid than a simple disk and you can integrate it into your print head .step files are available here on thingiverse However, I must warn you that designing your own mount is not as straightforward as it may appear, the rigidity of the clamp and mount are both very important as is the geometry of the parts that interface with the Orion PCB.

Hotend probing is better if you can use it, especially with a bed that moves in Z. Designing bed mounts that will support the weight of the bed and not trigger at the start of a probe move but then trigger with a low force from the nozzle is not trivial. It can be done but my experience is that hotend probing is easier.

It's very difficult to quote standard deviations for the Andromeda because everyones implementation is different. The Orion is self contained, we can test it on our printers and be reasonably confident that other people can achieve similar numbers, we can't do that with the Andromeda. I can tell you that just as the Orion had roughly twice the accuracy of the piezo disk models, the Andromeda has significantly improved performance over most bare disk implementations.

Idris

---

{Precision Piezo} Accurate, repeatable, versatile z-probe plus piezo discs, endstop cables, pt100, 50w heaters.

As an Andromeda user, I can verify Idris' statement.

The design for the mount is non-trivial, and has a HUGE effect on the ability of the sensor to do its job. And if you think that keeping the nozzle still with the probe attached is difficult, try keeping the BED stable and still. It can be done, and excellent results are possible.

I have gone with underbed sensing, but if I were to do it again, there's no way that would be my go-to solution. I'm very happy with my results now, but I'd most certainly go with hot-end probing if I were to design it again. In physical terms, it is much simpler to get consistent results from the hot end than it is from the print bed. The challenges of keeping things steady are the same either way, but there are a myriad of problems with under-bed sensing that you simply don't have to deal with when you put the sensor on the hot-end.

I believe that on a delta, the underbed sensing has some significant advantages over anything where the bed moves (in any dimension.) That said, there will ALWAYS be a difference in the sensitivity of an underbed probe based on the position of the hotend when it contacts the bed. This is entirely unavoidable. It can be minimised to the point where it doesn't make a difference large enough to hit the limits of the 3D printer, or at least the sliced GCode file, though it ultimately does limit the possible Z accuracy.

It is true that the bed must be stable for underbed sensors to work correctly but this stability is also needed even without a sensor. The traditional three point mounting allows the bed to move in the XY plane - constrained only by the friction of the adjusting nuts and often dubious screws to the build stage. I have previously used three sliding pillars between the build stage and the carriage - this works well but is obviously over-constrained if the build stage is out of true.

A printer that I am building at the moment will have a kinematic (Maxwell) mount for the build stage with piezos under the three ball components and nesting force provided by magnets set in repulsion. I hope to get the long striven for 1micron/1gram sensitivity along with a much more stable bed.

The sensitivity difference across the surface of the build stage with underbed sensors is not actually unavoidable. In fact, the method is inherently accurate if all else is equal: The sensors must all be equally sensitive and the stage design must be rigid and not have constraints other than the mountings to which the sensors are attached. On tests I have been able to indicate that the response is withing 10% across the whole build stage - including outside the triangle of sensors.

Mike

Quote
leadinglights
It is true that the bed must be stable for underbed sensors to work correctly but this stability is also needed even without a sensor. The traditional three point mounting allows the bed to move in the XY plane - constrained only by the friction of the adjusting nuts and often dubious screws to the build stage. I have previously used three sliding pillars between the build stage and the carriage - this works well but is obviously over-constrained if the build stage is out of true.

A printer that I am building at the moment will have a kinematic (Maxwell) mount for the build stage with piezos under the three ball components and nesting force provided by magnets set in repulsion. I hope to get the long striven for 1micron/1gram sensitivity along with a much more stable bed.

The sensitivity difference across the surface of the build stage with underbed sensors is not actually unavoidable. In fact, the method is inherently accurate if all else is equal: The sensors must all be equally sensitive and the stage design must be rigid and not have constraints other than the mountings to which the sensors are attached. On tests I have been able to indicate that the response is withing 10% across the whole build stage - including outside the triangle of sensors.

Mike

Explain the benefits of this set up that is far more complex, has shortcomings (and you describe them), costs more than just ONE sensor in the nozzle mount.
No benefit, not worth it.

---

"A comical prototype doesn't mean a dumb idea is possible" (Thunderf00t)

Quote
MKSA
Explain the benefits of this set up that is far more complex, has shortcomings (and you describe them), costs more than just ONE sensor in the nozzle mount.
No benefit, not worth it.

I would explain the benefits and even discuss the relative values of different approaches but your saying "No benefit, not worth it" shows that any such discussion itself would have no benefit and not be worth it. I have long since given up unproductive debates on this or any other forum.

Mike

Quote
leadinglights

Quote
MKSA
Explain the benefits of this set up that is far more complex, has shortcomings (and you describe them), costs more than just ONE sensor in the nozzle mount.
No benefit, not worth it.

I would explain the benefits and even discuss the relative values of different approaches but your saying "No benefit, not worth it" shows that any such discussion itself would have no benefit and not be worth it. I have long since given up unproductive debates on this or any other forum.

Mike

???
The meaning of my last sentence is: If there is no benefit, then it is not worth it !
So, explain the benefits because so far, you just raised issues. Issues that are known, even obvious BTW. Yet you plan to implement it ? Puzzling isn't ?

---

"A comical prototype doesn't mean a dumb idea is possible" (Thunderf00t)

Quote
MKSA

Quote
leadinglights

Quote
MKSA
Explain the benefits of this set up that is far more complex, has shortcomings (and you describe them), costs more than just ONE sensor in the nozzle mount.
No benefit, not worth it.

I would explain the benefits and even discuss the relative values of different approaches but your saying "No benefit, not worth it" shows that any such discussion itself would have no benefit and not be worth it. I have long since given up unproductive debates on this or any other forum.

Mike

???
The meaning of my last sentence is: If there is no benefit, then it is not worth it !
So, explain the benefits because so far, you just raised issues. Issues that are known, even obvious BTW. Yet you plan to implement it ? Puzzling isn't ?

While it may have been your intention to say that "If there is no benefit, then it's not worth it!", what you wrote can easily be understood as a statement "there is no benefit therefore it is not worth it", hence Mikes response.

Idris

---

{Precision Piezo} Accurate, repeatable, versatile z-probe plus piezo discs, endstop cables, pt100, 50w heaters.

Quote
MKSA
.................. Issues that are known, even obvious BTW. Yet you plan to implement it ? Puzzling isn't ?

Which Issues?

Mike.

BTW, in response to the only point you have made so far, Murata sensors can be had for under one UK pound each and Chinese (which may or may not be good) can easily be under 10 U.K. pence.

Quote
leadinglights
On tests I have been able to indicate that the response is withing 10% across the whole build stage - including outside the triangle of sensors.

I'm pretty sure that I said that you can get "close enough" but percentage isn't a measurement, unless you state what it is a percentage of. So what does that mean in real terms?

I've got my Andromeda sensors to less than 0.1mm consistency across the whole build stage, There are a few spots where if I were to drag a sheet of paper around the glass, it's probably a little tighter between nozzle and glass than it needs to be. I'm okay with this, as I always allow for some tolerance with my models, but this does mean that this adds to the potential dimensional accuracy. I'm willing to accept the amount of variation that I get from my 3D printer, but I would absolutely not accept those tolerances from a machine shop, for example.

Piezo sensors by their very nature measure force, and usually do so via movement. If you can make the build stage absolutely rigid so that any force exerted on it in any location were to be transmitted to a single piston, and absolutely constrained in the X/Y axes, then you could get perfect repeatability, but it's easier to make the bed REASONABLY rigid, and constrain the nozzle in the X/Y planes, while sensing force.

As with everything else in engineering, it's a trade-off.

I have acceptable results, which I accept aren't perfect, but they are able to produce results that work for me.

Hi SupraGuy,
Agreed it was a bit sloppy as it could have meant plus/minus 10% or within a 10% span, which was the meaning I intended. With one set of measurements I got a minimum of 21 grams and a maximum of 23 grams over the bed surface. It is actually quite difficult to get a pressure reading but I did it with a calibrated spring on the nozzle with a known starting position and speed to get a readout in volts over time. There are some background on this in the thread at [reprap.org] including some results in the first posting.

Agreed that piezoelectric sensors measure force; many people have assumend that they detect vibration. As far as the movement required to get a good output, there is likely to be many times more movement in the mechanism that applies pressure to a piezo disc than there is in the disc itself. I have mostly used discs with the pressure applied directly between the faces but even when they are being bent the movement can be very small.

Agreed also that good engineering will almost always need trade-offs and compromise: With care "acceptable" can be almost indistinguishabe from perfect.

Mike

p.s., I am still waiting for MKSA to enlighten me about the issues with underbed piezo sensors so that I can engineer the issues into a good compromise.

Quote
leadinglights


p.s., I am still waiting for MKSA to enlighten me about the issues with underbed piezo sensors so that I can engineer the issues into a good compromise.

You have been busy for more than two years with this, just reread yourself, others and myself (indeed, I realized all this has already been discussed) to find the issues.

Now, after all this time and all these experiments, you should be able to give the benefits of your choice compared to a hot end/carriage mounted unique sensor.

PS: Piezo are so good at detecting, generating vibration that it is their main usage especially the ones you are using.

Edited 4 time(s). Last edit at 12/05/2018 03:53PM by MKSA.

---

"A comical prototype doesn't mean a dumb idea is possible" (Thunderf00t)

Quote
MKSA
................................... Now, after all this time and all these experiments, you should be able to give the benefits of your choice compared to a hot end/carriage mounted unique sensor. .................................

All of the issues I found have been resolved to give very acceptable sensitivity, repeatibility, accuracy and reliability. This was done not by being certain that I already knew the answer but by trying to find solutions to each problem.
For my printers I prefer to use underbed piezo sensors. One of the main reasons (but not the only one) is that changing temperatures give rise to a voltage - it is easier to move the underbed sensors to a remote position than it is to do the same on a hotend or carriage mounted sensor if you want to heat the whole enclosure. Another reason will be seen in the photo at [reprap.org] How do you put hotend sensors on those hotends?

Mike

For me, the biggest disadvantage to the underbed sensors is the need to level the bed. by definition there, you can't have the mechanism completely constrain everything. Of course if you build everything to need no adjustment, then you could, but then if you did that, you'd have no need for a Z probe in the first place, all you'd need is to home Z, and you'd be done.

Any time you allow the bed to tilt, you have to accept varying forces across the bed surface, since pressure applied out of center will put tilting forces across the bed. This can be minimized by constraining the amount of allowable tilt, but it cannot be eliminated. (Again, if you could do so, you'd only need the single sensor anyway.)

Any compliance in the Z movement mechanism will result in a loss of sensitivity, and any compliance int he bed mounting system will result in a variation based on location. in addition, multiple sensors will always have a decrease in sensitivity outside of their mounting area, because downward pressure on the other side of a fulcrum will result in upward pressure at another sensor location. This is the fundamental compromise in the system. If we had 3 perfectly gimballed joints between mounting points and bed, we'd have the ultimate in levelling capability, but practically no ability to accurately detect pressure outside of the enclosed area. (This was what my first stab at the design acted like, which I'd label as "abject failure." My only good results were doing 3 point leveling with the probe points being basically right on top of the piezo sensors. and "acceptable" results only being within that triangle.) My oringinal setup used the 3 mounting points for the build platform, with the bed solidly mounted to the platform. The new one pushes the mounting locations out towards the corners of the bed, and senses pressure between bed and platform. Most of the bed is now inside of that square, though not all, but nothing is further outside of it than 1/2 of the square's length, which was what I determined will grant me acceptable results.

It is possible to build a system that doesn't need this, but the engineering requirements become much more difficult with the pressure transmission, and it rapidly becomes easier and cheaper to sense force at the nozzle. I would therefore advise anyone designing a 3D printer to explore that avenue first.

---

MBot3D Printer
MakerBot clone Kit from Amazon
Added heated bed.

Leadscrew self-built printer (in progress)
Duet Wifi, Precision Piezo parts

Quote
SupraGuy
.................................................. because downward pressure on the other side of a fulcrum will result in upward pressure at another sensor location. This is the fundamental compromise in the system. .....................................

The problem is that, despite being obvious it is not true. I don't have access to a drawing program so must rely on your visual imagination. If you take two sensors then applying a force above one of them will result in a charge developed by that force, call it 1 unit: If you place the force between the two sensors then each will generate 0.5 units of the same charge resulting in the same total: In the third case, having a force applied outside the two, say for example the same distance as the sensors are apart - again applying the same force will give a minus 1 unit charge on the one furthest from the force but a plus 2 units on the closest due to the mechanical advantage of the lever so (-1) + (+2) =1, again the same. There are many ways that this can be upset, having too short a time constant on the amplifier and mechanical flexing which means that the forces arrive at the sensors at different times is a prime candidate.

Edit. O.K. I have done a quick drawing to illustrate the point.



Mike

Edited 1 time(s). Last edit at 12/07/2018 10:10AM by leadinglights.

Now that I've properly installed and configured my orion on my HyperCube There is no turning back. All other automatic probing systems I've been using through the yeras are banned to an box of sort of fucntional but sort of poorly implemented solutions never to be used again.
So the next steep is to have a orizon on my Cyclops+ machine, What is the current situation there, it there anything I can do to help with the development of this?

I'm glad to hear you're enjoying it.

We have a couple of solutions for the Chimera and Cyclops though you will still need to do some work to integrate it into your specific printer.

Idris

---

{Precision Piezo} Accurate, repeatable, versatile z-probe plus piezo discs, endstop cables, pt100, 50w heaters.

Quote
leadinglights

[attachment 108496 PiezoPositions.jpg]

Mike

My problem with this is that it makes 2 assumptions, which aren't necessarily true.

1. That there is no friction in the system.

2. That the piezos measure force. this is not actually correct, they measure a rate of change in force, which is a different matter. In your third diagram, the piezo that sees 2F is also seeing it at 0.5v, whereas the one seeing -1F will see it at v, where v represents the velocity of the applied force. Because of the reduced velocity, it will read much lower. You are only looking at the force side of the equation and ignoring the rate of change.

The basic piezoelectric disc is not measuring a rate of change but a force. The thing that makes it look like it is measuring a rate of change is electrical leakage - mostly in the attached electronics. If the amplifier impedance was in the region of some hundreds of megohms then this would be quite apparent as the voltage developed across the disc would stay for many seconds. Another way of looking at it is that a potential applied to a piezoelectric disc will deform it and the deformation will last as long as the potential is maintained.

Mike

I'm really interested in trying this but I need to figure out how to get it to work with the Petsfang cooler (https://www.thingiverse.com/thing:2984806). I have a TronXY X5S with an E3D-V6 and the linked Petsfang cooler. I have a BLTouch Smart now but I really like the idea of the hot end being the probe, no X or Y offset and little to no Z offset.

I've looked around on Thingiverse but haven'e found a piezo mount that looks like it will work or an alternate cooler that looks as effective with a piezo mount. If anyone has any pointers I would appreciate it!

Hi there,

First off great sensor; I couldn't stand my BL Touch clone anymore.

Now I'm running into issues and need to pick your brains. My current printer (DIY) is running on a single leadscrew (to be upgraded to dual soon), and the issue is that when it probes the right side of the bed it goes lower into the bed before it triggers as there is less resistance (the carriage goes up since there is no leadscrew there). Now the best way to resolve that would be to increase the sensitivity, but I have to lower it quite a bit else it triggers early during the vertical probing move. Also here is my config:

; Z-Probe
M574 Z1 S2 ; Set endstops controlled by probe
M558 P5 R0.6 I1 H10 F250 ; Set Z probe type to effector and the dive height + speeds
G31 P500 X0 Y0 Z-0.35 ; Set Z probe trigger value, offset and trigger height
M557 X0:198 Y10:185 S30

Any suggestion on how I could solve that, at least until I can upgrade to a dual z?

@Arakhan,

What are your acceleration and jerk values during Z-probing?

Also, a couple of suggested changes to your M558:

Use P8 rather than P5, we don't need filtering.
H10 is much too high, I probe from 2mm

@tletourneau

I have a mount for the X5s, I just need to make a couple of tweaks and I'll post it to thingiverse, however I don't yet have a good part cooler for it.

Idris

---

{Precision Piezo} Accurate, repeatable, versatile z-probe plus piezo discs, endstop cables, pt100, 50w heaters.