Precision Piezo Z-probe Now available.

It works very well, typically you wont heat the nozzle to full print temp to avoid oozing, I'd suggest 120-160c and at those temps I've not seen any issues with PEI.

Idris

---

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

Any reason why the Andromeda couldn't be adapted for a nozzle probe? I have a lifting dual head design which doesn't have clearance for the Orion (and I don't want to space things further apart to clear). Would it be possible to drill out the central hole in the Andromeda to allow a 4mm PTFE tube to pass?

Lifting Dual Head

-Wes

I bought an Andromeda with a view to integrating it into my D-Bot Titan Aero set-up. I haven't worked it out yet but there is definitely room to *carefully* drill out the 3mm center hole to 4 or even 5mm.

4mm OD tube? That shouldn't be a problem, there should be clearance for up to a 7mm hole if you're careful however, I don't know how much strength would remain if you were to remove that amount of material.

Idris

---

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

I was wondering, could you guys also put up for sale the brass insert for the bowden collet that comes with the Orion Groovemount kit?

Yup, I've been meaning to for a while. Unfortunately, E3D won't do a wholesale price for me so I'll have to charge a small premium over the E3D price to cover my costs but it shouldn't be much and should be covered by the savings in shipping.

I've considered stocking the threaded V6 heatsink to go with the new Orion Threaded heatsink mount, anybody interested in that?

Idris

---

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

Oh, is that the same brass insert from the bowden to groovemount adapter as well as the one that's in the hotend heatsink as well?

Yes, it's this one.

Idris

---

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

I recently bought the Orion kit and have it installed on my printer, using an E3DV6 hotend. It properly senses a sideways tap on the hotend very easily, but pressure on the nozzle itself doesn't register unless a lot of force is applied. I have tried tweaking the sensitivity, but all that does is cause it to trigger as the hotend is moved up and down (this is a delta printer). I'm guessing that the mount isn't allowing the board to flex enough. I'm afraid to loosen the "sandwich" too much as that will cause the hotend to wobble. What is usually the cause of this behavior? Thanks.

Quote
MKSA
Looking at your schematic, I found it quite strange although I understand the principle.
A 3X Opamp for the piezo signal, its Vout fed to a comparator whose triggering level is determined by a R ladder adjusted by the microcontroller.
Now, the front end Opamp is "strange", and why is its output connected to a 2.5 Vref via a 1M resistor ? Regarding the R ladder. How many possible values ?, just pulled down or a combination of up and down ?. Why not a classical R ladder (R-2R, weighted R) or even a Digipot ?

Digipots are a lot more expensive than 5x resistors, and the DAC output on arduino's isn't all that reliable I've been led to believe. The circuit acts as a coarse digipot by either disconnecting the digital pins at junctions of the resistor ladder (by setting the pinmode to "INPUT", or pulls them to ground by setting the pin to "OUTPUT" and digitalWrite(LOW). So I've got 5 values to adjust to, which seems to be good enough for all of the piezo sensors I've tried thus far.

Essentially what this is doing is adjusting the reference voltage for the comparator portion of the circuit to within a range that it will filter out normal printer vibrations, and still pick up the piezo bump. Having the 1M resistor with the VRef allows the charge at the inverting input to slowly stabilize after a bump. The piezo is basically acting like a capacitor in this circuit, either charging or discharging based on the mechanical force. With the 1M resistor tied to that pin ensures that the recovery time of the input takes a little while, so you don't end up with a mess of extremely fast signals all coming through. It's got a very fast rise time, but a very slow fall time. The comparator is also inverting this signal, and drawing a pulled up interrupt pin on the Atmega88 to ground. Having it's "virtual ground" on the inverting input of the comparator be at midrail (~2.5v) allows the voltage to sit at a level that can be more easily adjusted for the noninverting input. It can go up, or down that way, rather than only up, and account for unforeseen environmental causes of voltage change.

Quote
Moriquendi
@Pyroball,

It looks very interesting, can you give an explanation of how it works? Specifically how the software and hardware interact?

Idris

See my reply above for an explanation of the electrical characteristics, but it should essentially work as a drop-in replacement for the Precision Piezo version, just with better filtering and self-calibration (no more fiddling with 3mm pots!)

I was chatting back and forth with DJDemonD on my original FFC cable chain idea a year and change ago. I'd be happy to get on a discord chat with you and go through the circuit and code to explain

Edited 1 time(s). Last edit at 01/11/2019 08:52PM by pyr0ball.

Automatic tuning of piezo sensors is a sensible direction for further progress. I have been working on a method of detecting the nozzle contact by looking at the morphology (shape) of the contact pulse in the hopes that it would eliminate adjustment. Present status in that pulse shape detection works very well but is on the back burner until I can find a user friendly way of doing the adjustment. The piezo element itself is rarely a problem, compliance in the piezo mounting and coupling from the nozzle to the piezo are much more important - eliminating flexibility in parts between the nozzle and the piezo element or flexibility in the mounting of the element to the extruder or carriage are much more important.

Mike

@ Holmes4,

You should be able to loosen the four assembly screws half a turn without any wobble issues, the mount is designed to preload the PCB which will keep the nozzle solid. On the other hand, if loosening the screws half a turn does cause the nozzle to become loose then it's likely you didn't have the screws tight enough in the first place, there shouldn't be any gap between the top piece, the PCB and the bottom piece.

@Pyr0ball,

That sounds great, I've never used Discord before, I've signed up but I don't know how to use it, perhaps you could PM me your account name and a time?

@Leadinglights,

I've been trying to look with more detail at the pulse shape and behaviour of the piezo disks but I'm running into problems with the oscilloscope probe loading the piezo and distorting the signal, have you seen this? I'd appreciate your advice on how to avoid it.

I've also been looking at ways of automating or removing the need to tune the circuit but what with providing support and manufacturing, plus life itself, I've had limited time.

Idris

---

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

@Moriquendi

If you'd prefer something else, that's fine too, but I know discord has screen share capabilities along with voice chat which makes things easier

As to your oscilloscope loading the piezo output, that's cus the probe doesn't have as high an impedance as an amp, and the output current of the piezo is tiny so it just gets vdrooped to nothing when probed in-circuit directly.

This behaviour can be mitigated by using another amp as a buffer. (if you want the exact same signal in/out).

Attach a probe wire to the part of the piezo circuit you're attempting to probe, and the other end goes to the noninverting input of your "probe amp". Use the topology in the below schematic, and attach your oscilloscope probe to the amp output:

@Moriquendi
You could use a voltage follower as pyr0ball suggests but I would recommend a 10k resistor in series with the non-inverting input. I think almost all modern op-amps have protection diodes on the input but enough of a charge could be transferred by a sharp blow to give problems. The alternative is to use a X10 or even X100 probe on your scope. Having said that, even with a 1MΩ probe the contact upstroke should not be distorted too much - the recovery after contact will be distorted but its shape is of little interest.

Mike.

I would try to give more information but I am typing this on a Chromebook perched on my knee in a lurching cruise ship several hundred miles north of the Arctic circle - just one of the difficulties of life

You poor man, what a terrible situation to find yourself in...

I tried using the probe on x10 but I still found that it caused issues with triggering. Specifically I'm trying to look at the signal generated by the Orion during actual probing on an active printer. It's possible that I was seeing not just loading issues but also injecting noise by extending soldering long wires to the piezo disks. I'll knock up a four channel voltage follower that can be mounted on the Orion and see what I can see.

Discord will work fine, I'm just not familiar with it is all.

Idris

---

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

Question regarding the Orion in a Bondtech BMG Extruder;

Normally, with just the E3d V6 in the BMG, you remove the bowden coupling from the E3D and insert a length of PTFE tube in the E3D so that 5mm is protruding (see the 30 second mark of this video)

Since the Orion is being inserted into the BMG, is the procedure the same? Remove the bowden coupling from the Orion and insert the PTFE tube so that 5mm protrudes?

I believe so yes though I've not used a BMG, you might need to adjust the length of PTFE that protrudes slightly.

Idris

---

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

Quote
pyr0ball

Quote
MKSA
Looking at your schematic, I found it quite strange although I understand the principle.
A 3X Opamp for the piezo signal, its Vout fed to a comparator whose triggering level is determined by a R ladder adjusted by the microcontroller.
Now, the front end Opamp is "strange", and why is its output connected to a 2.5 Vref via a 1M resistor ? Regarding the R ladder. How many possible values ?, just pulled down or a combination of up and down ?. Why not a classical R ladder (R-2R, weighted R) or even a Digipot ?

Digipots are a lot more expensive than 5x resistors, and the DAC output on arduino's isn't all that reliable I've been led to believe. The circuit acts as a coarse digipot by either disconnecting the digital pins at junctions of the resistor ladder (by setting the pinmode to "INPUT", or pulls them to ground by setting the pin to "OUTPUT" and digitalWrite(LOW). So I've got 5 values to adjust to, which seems to be good enough for all of the piezo sensors I've tried thus far.

Essentially what this is doing is adjusting the reference voltage for the comparator portion of the circuit to within a range that it will filter out normal printer vibrations, and still pick up the piezo bump. Having the 1M resistor with the VRef allows the charge at the inverting input to slowly stabilize after a bump. The piezo is basically acting like a capacitor in this circuit, either charging or discharging based on the mechanical force. With the 1M resistor tied to that pin ensures that the recovery time of the input takes a little while, It's got a very fast rise time, but a very slow fall time. The comparator is also inverting this signal, and drawing a pulled up interrupt pin on the Atmega88 to ground. Having it's "virtual ground" on the inverting input of the comparator be at midrail (~2.5v) allows the voltage to sit at a level that can be more easily adjusted for the noninverting input. It can go up, or down that way, rather than only up, and account for unforeseen environmental causes of voltage change.

What I thought as for the intended mode of operation but:
- Based on the schematic, the 3X amp as drawn is wrong, its IN- is connected to ground ! Schematic error ?
- The 2.5V Vref with a 1M will have no real effect as it is connect to the Opamp OUT. I can't see how it could: "so you don't end up with a mess of extremely fast signals all coming through.". So , is the schematic correct ?
- Using R-2R ladder would have given you a lot more possibilities at no extra cost. Used in DAC for decades.

Yet there are other ways, better ways, for ex. DC42 in his smart effector just feed the sensor signal after some conditioning to the ADC of the processor and do all the this work in software.
To "upgrade, correct" hardware in the field is quite difficult isn't ?

Edited 2 time(s). Last edit at 01/13/2019 11:54AM by MKSA.

---

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

We designed a piezo-based Smart Effector at the same time as the strain gauge version that we put into production. We didn't use an op amp, just a couple of resistors and a TVR diode to condition and limit the signal, then we fed it straight into one of the analog inputs of an attiny microcontroller. The attiny has an optional amplifier before the ADC, with gain up to x20 AFAIR. Which means that if you use the internal 1.1V reference, you can get a full scale sensitivity of around 55mV and a resolution of around 55uV. So no need for an op amp IMO.

The production Smart Effector has a digitally programmed sensitivity, and we would have used the same scheme had we put the piezo version into production instead.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote
MKSA
What I thought as for the intended mode of operation but:
- Based on the schematic, the 3X amp as drawn is wrong, its IN- is connected to ground ! Schematic error ?
- The 2.5V Vref with a 1M will have no real effect as it is connect to the Opamp OUT. I can't see how it could: "so you don't end up with a mess of extremely fast signals all coming through.". So , is the schematic correct ?
- Using R-2R ladder would have given you a lot more possibilities at no extra cost. Used in DAC for decades.

Yet there are other ways, better ways, for ex. DC42 in his smart effector just feed the sensor signal after some conditioning to the ADC of the processor and do all the this work in software.
To "upgrade, correct" hardware in the field is quite difficult isn't ?

Yep that was an error in the schematic I fixed before sending out for fabrication. Forgot to update the screenshot.

As for the 2.5vref, yes it's does effect the first amp in addition to that comparator because of the feedback resistor's lower value to the follower "pull-up". I may have some incorrect assumptions here as I'm still learning, but my understandings and observation of this circuit show utilising the voltage reference follower in this way gives the circuit a very specific characteristic: fast rise and slow recovery which prevents the circuit from sending a bunch of signals in a short time.

ADC signal processing may be better able to distinguish the waveforms of a bump and filter through software, but it's slow compared to the method I'm using (interrupt pins) which, since Marlin now comes with an interrupt pin function for endstops, means that it can get even more accurate z-offset measurements due to close to zero lag.

The field adjustments are something I plan to implement. I'll be setting up profiles for different sized piezo disks, mounting schemes, assumed mechanical stress factors, etc, by adjusting the upper and lower threshold that the circuit can calibrate itself for.

I'll not familiar with an R-2R ladder so I'll have to do some research on that

Edit: you know it's funny, I started out with the resistor ladder in parallel thinking I could figure out how to get 8 values out of 3 digital pins (3bit) and I just couldn't wrap my head around the math for it. Adding the other resistors in series was what I was missing, cus then I don't have to worry about figuring out individual values for each section of the ladder.

Thanks for that little gem! I'll have to poke around the circuit and see what works

Edited 2 time(s). Last edit at 01/14/2019 03:44AM by pyr0ball.

Quote
pyr0ball
ADC signal processing may be better able to distinguish the waveforms of a bump and filter through software, but it's slow compared to the method I'm using (interrupt pins)...

Done properly, it's not slow enough to matter. Using a probing speed of 20mm/sec (which is quite fast), if the code takes 100us to take an ADC reading and do the calculations, the jitter introduced would only be 2um.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote
dc42
Done properly, it's not slow enough to matter. Using a probing speed of 20mm/sec (which is quite fast), if the code takes 100us to take an ADC reading and do the calculations, the jitter introduced would only be 2um.

Hey I just realized I recognized your username from the thread over on soliforum for porting Repetier/RRFW over to the DaVinci Jr.

Small world!

Anyhow, I've got a GitHub up with all of the source files up for perusal. I'm open to making modifications to the circuit as well.

The board revision 0.1 that's up there is using an ADC input pin I believe and I've got a few of those boards already fabbed. I could run a kynar wire and bypass the amp to do some testing if you'd be able to provide some working code.

How large is the flash for a program like that? The footprint I'm using currently is the 28vqfn package which I can get up to the Atmega328 dropped on. Currently I'm going with the atmega88 as it's a move cheaper chip and my application size is only 5.6k

Edited 1 time(s). Last edit at 01/14/2019 01:44PM by pyr0ball.

The chip we use on the Smart Effector is the attiny44a. It's inexpensive, available in a small QFN package, and more than powerful enough for this job. The Smart Effector code uses less than 2K of flash (out of 8K) and 107 bytes of RAM (out of 256) plus stack. The code is at [github.com]. The code for a piezo might be even simpler because you won't need to allow for DC drift, unless your piezos generate significant DC when subjected to a temperature gradient.

I don't have any working code for a piezo sensor. We dropped development of it when we found out how difficult it was to drill piezos, given that the strain gauge version was already working and didn't pose any manufacturing difficulties. I guess if we were trying it now we'd use 3 small piezo discs around the hot end mounting hole instead of 1 drilled piezo, with each one connected to a separate ADC input.

AFAIR the atmega328p doesn't include the amplifier before the ADC.

Edited 1 time(s). Last edit at 01/14/2019 03:23PM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

A while ago I tried my own take on dc42s smart effector. Although it has never been used in hot blood it gave a good signal and the effector was very rigid despite being made of 1.6mm FR4 PCB material. Details on [reprap.org]
This may well benefit from having a self tuning conditioning circuit built onto the same board.

Mike

Edited 2 time(s). Last edit at 01/14/2019 06:07PM by leadinglights.

@dc42

Well the current code I'm using handles dc drift very well. I'm currently building a new branch of my project utilising an R-2R ladder. It should give me 16 values rather than the 5 I started with, so thanks for that tip! I'll see if I have any other spare I/O I might be able to use to get higher resolution out of it.

Before I go that route I'm gonna try setting up a pwm dac and see how that fares.

If I can get the pwm dac to work, then maybe I can use the ladder to adjust the gain on the first amp and get two different points of auto-tuning

got a weird problem I hafta ask, the answer seems obvious but I wanna ask to confirm before I do it. So basically it's been working well until today, when I get sudden G29 probe failures at the same exact two points (out of the 36 doing a 6 x 6 it's a huge 500 x 500 mm bed). So I checked and oddly enough once in awhile at those two points, the damn thing suddenly loses power, I thought it was a wire issue so I tested my wires, and they have continuity, so I decided to push the plug in more despite it looking like it's in pretty much all the way and it sorta works fine now?

I'm wondering if I should change the wires nevertheless (it's a 2meter continuous wire) or if it's the F6 board having issues (unlikely but I am liking this board less and less and really tempted to just go for my Duet Wifi). Any idea or to confirm what is causing this weird "no power" (like it just suddenly goes dark no led whatsoever) on just the same exact two probe points (aka the first 2 probe points nears the front right of the machine) on my machine? Thanks

Quote
iamthebest22
got a weird problem I hafta ask, the answer seems obvious but I wanna ask to confirm before I do it. So basically it's been working well until today, when I get sudden G29 probe failures at the same exact two points (out of the 36 doing a 6 x 6 it's a huge 500 x 500 mm bed). So I checked and oddly enough once in awhile at those two points, the damn thing suddenly loses power, I thought it was a wire issue so I tested my wires, and they have continuity, so I decided to push the plug in more despite it looking like it's in pretty much all the way and it sorta works fine now?

I'm wondering if I should change the wires nevertheless (it's a 2meter continuous wire) or if it's the F6 board having issues (unlikely but I am liking this board less and less and really tempted to just go for my Duet Wifi). Any idea or to confirm what is causing this weird "no power" (like it just suddenly goes dark no led whatsoever) on just the same exact two probe points (aka the first 2 probe points nears the front right of the machine) on my machine? Thanks

Some printers have in their 3D print space, strange areas where weird things happen not unlike in the famous Bermuda Triangle.

---

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

sounds to me like a broken wire inside the insulation. When moving to this exact position the wire is bent so it does not have contact anymore. moving on to the next point it is bent another way, contacting again.

I once had this Problem in the wire to the heater cartridge on the hotend. Had to search some weeks until I noticed, because Temperature was swinging when moving. Did PID tunings, changed Fan duct, put insulation on hotend - nothing helped. One day it was broken that much, i got thermal runaway. Scratched my head some eventings until I noticed it was always on front right where there was maximum bending on the wires...

Edited 1 time(s). Last edit at 01/17/2019 04:08AM by Dancer.

Likely places to look are wires around bends as Dancer mentioned but also poor contact between the sockets and plugs, "dry" soldered joints on PCBs, bad crimping of the socket or plug pins to cables. Failing the above it may be down to the Bermuda Triangle or the gods of missfortune - Murphy (USA) or Sod (UK)

Mike