Precision Piezo Z-probe Now available.

I usually use -0.1 probe offset as technically the probe triggers below the bed. We are used to probes which trigger above the bed and have positive offsets.

The std deviation suggests your probing setup is close to optimal so for tighter levelling its probably worth looking at mechanics of the machine.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Well you are absolutely right with offset-direction. Think I need to reduce - not to increase!
When I had Z-Offset 0.0 it did not perfectly stick. But maybe, I over-compensated. Will try tonight and report back (right now I'm not at home)

Good to hear my probing setup is close to optimal - even with aboslutely no bolting together.
Mechanics might be the problem - it is a Delta with 1000mm vertical 2020 extrusions and 400mm 2020/2040 horizontal extrusions - rigidity is not bad. - But I fear making it more rigid will be hard. Already have some extra-corners between horizontal and vertical extrusions.


OK, ok.... I see. I have to take some pictures tonight

It's probably not the frame that needs making more rigid it's usually carriages, arms effector, hotend.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Well. Yes that might be the problem.

Carriages are with 4 POM-Rollers directly on 2020 extrusions, no linear rails.
Arms are 6mm OD / 4mm ID Aluminium-Tubes with cheap chineese Traxxas-Like ball-joints
Effector is printed on my kossel mini - which until now is only calibrated by hand - so maybe some deviation from here.
Hotend-Mount - printed too, but extremely tight - nearly could not mount it because it was so tight - don't think it might come from here.
In addition - because of the 1000mm height - there are nearly 2000mm long GT2-Belts - maybe even some stretch from them.


So my conclusion for now is: With the available parts my std. deviation & calibration is good, and I will maybe not be able to get any better without putting much more money into this printer.
Thanks a lot - for your answer - will save me some time searching for precision which might not be achiveable.

Will do some more tests, but as the first results are really good, I think I will put underbed piezos to my kossel mini too.

Edited 2 time(s). Last edit at 08/09/2017 04:46AM by Dancer.

Glad it working out okay. I think the road to delta printer perfection is insanity - this at least explains things.

NEW STOCK

We are also selling the piezo kit PCB in the store now, to buy on its own.

Ready made module will still be a week or two we are awaiting PCB's from new supplier.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Quote
DjDemonD
I usually use -0.1 probe offset as technically the probe triggers below the bed. We are used to probes which trigger above the bed and have positive offsets.

The std deviation suggests your probing setup is close to optimal so for tighter levelling its probably worth looking at mechanics of the machine.

After quite a lot of testing tonight (it's nearly 23:00):
std.deviation whole bed 0.026
Z-Offset +0.3 --> I'm using Marlin latest bugfix.

Too late for some Pictures today - will do some tomorrow, but for a start my openscad of the piezo-holder as attachment.
-) Cups bolted to top of extrusions
-) Piezos just loosely put inside the Cups brass side down
-) Insert loosely put on top of piezos - cut out is for solder-joints and wires
-) printbed bolted to inserts.

Sounds like you are making progress. 0.02mm deviation is not bad.

Keep us posted.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Here are my Pictures - Hope they are useful for somebody.

Preliminary probing results of my underbed mount piezo's:
Sensitivity set to where moving carriages do not light LED.
(Note I hadn't adjusted config.g for new bed height)

Moving probe cold at center (moves around outer portion of bed between probes):

Bed probe heights: -1.671 -1.659 -1.659 -1.659 -1.659 -1.659 -1.659 -1.659 -1.659 -1.659 -1.659 -1.659 -1.659 -1.659 -1.659 -1.659 -1.659 -1.659 -1.659 -1.659 -1.659 -1.665 -1.659 -1.659 -1.665, mean -1.660, deviation from mean 0.003

Stationary probe cold at center:

Bed probe heights: -1.665 -1.665 -1.665 -1.665 -1.665 -1.665 -1.665 -1.659 -1.659 -1.659 -1.665 -1.665 -1.665 -1.659 -1.663 -1.659 -1.659 -1.659 -1.659 -1.665 -1.659 -1.659 -1.663 -1.665 -1.665 -1.659 -1.659 -1.659 -1.665 -1.665 -1.665 -1.665 -1.665 -1.659 -1.659 -1.659, mean -1.662, deviation from mean 0.003

Moving probe bed 60C hotend 230C at center (moves around outer portion of bed between probes):

Bed probe heights: -1.552 -1.552 -1.552 -1.552 -1.552 -1.559 -1.552 -1.552 -1.559 -1.552 -1.552 -1.552 -1.552 -1.552 -1.559 -1.552 -1.552 -1.552 -1.552 -1.559 -1.559 -1.559 -1.559 -1.559 -1.552, mean -1.554, deviation from mean 0.003

Stationary probe bed 60C hotend 230C at center:

Bed probe heights: -1.565 -1.559 -1.559 -1.552 -1.559 -1.559 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552 -1.552, mean -1.553, deviation from mean 0.003

After those I did the auto level twice and changed config.g I went from -0.15 Z offset to -0.05 Z offset so quite a bit more sensitive. However I have the same problem when not at center as I did with the FSR's which I was hoping the added sensitivity would help alleviate. So some redesign is in order.

Here is the G code I ran to test multiple points twice with the results added at the end of the lines:

G30 P0 X-86.60 Y-50.00 Z-99999 S-1; X tower 0.120 0.120
G30 P0 X0.00 Y-100.00 Z-99999 S-1 ; between X-Y towers -0.038 -0.032
G30 P0 X86.60 Y-50.00 Z-99999 S-1 ; Y tower 0.108 0.102
G30 P0 X86.60 Y50.00 Z-99999 S-1 ; between Y-Z towers -0.052 -0.046
G30 P0 X0.00 Y100.00 Z-99999 S-1 ; Z tower 0.088 0.076
G30 P0 X-86.60 Y50.00 Z-99999 S-1 ; between Z-X towers -0.044 -0.038
G30 P0 X-43.30 Y-25.00 Z-99999 S-1; X tower 0.012 0.006
G30 P0 X0.00 Y-50.00 Z-99999 S-1 ; between X-Y towers 0.005 0.011
G30 P0 X43.30 Y-25.00 Z-99999 S-1 ; Y tower 0.062 0.062
G30 P0 X43.30 Y25.00 Z-99999 S-1 ; between Y-Z towers 0.018 0.016
G30 P0 X0.00 Y50.00 Z-99999 S-1 ; Z tower 0.012 0.012
G30 P0 X-43.30 Y25.00 Z-99999 S-1 ; between Z-X towers 0.019 0.019
G30 P0 X0 Y0 Z-99999 S-1 ; center 0.022 0.022

Hi I hope your are pleased with it those are good results. I'm jealous my corexy underbed setup is a no go, but the micro delta one is awesome.

Plus for added sensitivity I would only worry about carriages that move when probing and have sensors carried on them, so for delta it's unimportant, for corexy with moving bed it's the z axis that needs to move smoothly (when probing at least) x and y can be as rough as nails as they stop when probing.

Even moving y bed on an i3 isn't a problem as long as you pause before probing (see above for various methods/firmware mods).

Frame resonance is also a consideration even if sensors aren't moving with the carriage.

Edited 5 time(s). Last edit at 08/10/2017 03:06PM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Oh, I'm very pleased with the single point probing results. I need to change my design to eliminate the normal teeter totter effect of probing outside the triangle of the sensors without a rigidly mounted bed. With the FSR's I went to only probing in a line from center to a sensor. I was hoping the increased sensitivity would let me probe outside the triangle, which it helped alot but it's still there. I will design some bed holders that keep a spring pressure down on the bed at the sensors and see what that does. The other option is to do as Micheal Hackney did and move the sensors out on a triangle shaped frame where the triangle formed by the sensors encompasses the bed, but I don't have the room without going outside the frame.

That's a good idea, I don't know if we've convinced Michael to try the piezo system, he said he was getting 0.02mm delta deviation with fsrs, I think we can blow that away my kossel XL is down in the 0.004 area now.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

No luck with eliminating the usual problems with underbed sensors (teeter totter and variable sensitivity). I tried supporting the bottom of the piezo's solidly since I designed the mounts with a recess under them and tried spring loaded pressure from above the bed. Lessened the effects but didn't get rid of them. I went back to the recess and turned the sensitivity up. Like your video I can blow on the bed and trigger them. Had to back off just a bit and had to move to position then put a 500ms dwell before the actual probe to eliminate triggering while probing. After adjusting the H values in Duet autocalibrate I get .003 to .002 autocalibrate deviation. Then I leave it up to bed height map to correct for the "usual problems". The map looks really bad with that high of a sensitivity, but the results are great.



Looks to me like the effector mounted sensor(s) would be simpler to implement and far more accurate sans the bed map.

(Yep, I know that's why they were implemented, but I am one of those that have to hit my thumb with a hammer to know it hurts)

Edited 1 time(s). Last edit at 08/12/2017 08:11PM by Alexander Mundy.

I have not had teeter totter or instability on my underbed piezos except during the very first trials of the concept. Possible contributors include the sliding bearing seen to the left of the amplifier. This bearing is a close but not tight fit on a stub which takes out X and Y movements. Another possible contributor is that the load point for the bed adjusting spring suspension is taken out as close as possible to the LM8UU bearings - despite the carriage only being a 1.5mm sheet of bare FR4 glass fiber composite this is a very rigid setup.



(Drawing on the first entry of [forums.reprap.org])

I have not had any feedback from anybody who used the same system except Suggy but I rather think that the difficulties he was experiencing lay in other aspects of his piezo setup

Mike

Hi Alexander,

I think if you were going to eliminate this then Michael Hackneys frame idea as you mentioned before might be the best plan. I think I notice it less on my micro delta as the aluminum plate is 120mm but the bed printable area is only 100mm so I'm not probing the edges. I did used to see the same effect with fsrs.

Have you tried clamps on the mounts rather than just seats to rest the bed in?

Otherwise go to hotend probing and rigidly bolt the bed in.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Well it works great with bed compensation, so I'm pleased. I did try holding the bed down tight against the piezo's with various amounts of force from spring wire, that helped but didn't alleviate it and the sensitivity started dropping off at a certain point.

Quote
DjDemonD
Sounds like you are making progress. 0.02mm deviation is not bad.

Keep us posted.

std. dev on my Delta Mini with M48: 0.0134
One Microstep is 0.0125mm

I think this is the end of getting better with RAMPS and 1/16 microstepping.

Yes I'd say thats probably your optimum. Mike feels that the limiting factor is the controllers (all of them) they take around 1ms to register a trigger so even with 1/256th microstepping I don't get better than 0.01mm range with 0.007mm std deviation. Mainly as you cannot probe really slow or it wont trigger.

When we are talking about cartesian machines and a first layer is maybe 200-350microns having the bed levelled/compensated to 30 microns is going to be very helpful. You can argue for more precision with a delta if using the probing to calibrate (so long as you have very accurate endstops also).

If you want to test your printer's absolute accuracy mechanically then set normally closed/(inverted) for z-min. Use a multimeter to check there is no leakage of current to your hotend (from a dodgy heater cartridge or thermistor, measure from PSU GND to the hotend on volts ensure its 0v with machine powered and hotend heater on) attach the GND/- wire from the z-min endstop connector to the hotend and the SIG wire to a metal plate placed and secure on the bed. Now you have an electrical contact-contact probe (nice clean nozzle for this one ) and do the M48 repeatability test. The accuracy/deviation you get now is what your machine is capable of mechanically. The difference is the contribution of the piezo probe. If they are the same then the probe is performing as well as your machine is able to mechanically. Its a good way to control for the machine's capabilities when testing.Try lowering your probing speed to a very low value, see what difference it makes.

In fact it works so well if there was a (cheap - this is the problem) conductive PEI sheet available (or PC possibly) then I would consider developing a print surface probe kit, using electrical contact to trigger, the accuracy especially with very slow probing speed (as no force needs to be imparted to trigger) would be very high indeed.

Edited 4 time(s). Last edit at 08/15/2017 10:48AM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

It's also worth saying I see three types of repeatbility:
1) single point probing repeatedly, this is mainly about the probe
2) moving around and probing this is where the printers mechanical accuracy starts to factor in.
3) repeatability between printing sessions this is where deployable probes can be much less repeatable than fixed types depending on their design. Not really a problem if bed levelling or calibrating before each printing run but actual nozzle to bed gap will vary, which is undesirable.

Edited 2 time(s). Last edit at 08/15/2017 01:23PM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

For the duet wifi can I use the same connections as my DC42 IR probe then ? I'm connected at +3.3V, GND and Z_probe_in then. I could use Z endstop pins also I guess but is it possible to get a analog output from the PCB so I can set the force needed to trigger in duet?

For the titan mount is there any way of not gluing in the sensor? Where can I find the spec of the 27mm and 20mm if I want to grab a few more from my eletronics dealer (if I have to glue them in).

[duet3d.com]

Also how sensitive is the sensor? Does it need a lot of wiggle room to give a signal or can I design a rigid mount for it and rely on sensitivity? I'm going to use it with titan aero extruder.

Edited 1 time(s). Last edit at 08/17/2017 01:20PM by minim.

Hi yes use these same connections as for dc42 IR probe. You can use analogue or digital P1 or p5 under M558.

You need to bond the piezo for it to work well.

Buy more Piezos from. Rs components see above in this thread or the other piezo "research" piezo thread by leadinglights for part no.s, theyre murata brand with leads.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

What sort of machine are you using?

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

I use a Flashforge Creator pro. I designed my own mount for titan aero on it and was hoping to be able to test the piezo to see if I need more kits If it works good I want this on my corexy also.

I doubt you'll be able to just glue the piezo to your mount as pictured (I could be wrong, the sensitivity can be mindblowing, I've had one triggering reliably on a test stand by dropping a hair on it). This is still pretty new tech, lots of paths to be explored and the piezos themselves are cheap.

If you look here you'll find a titan bracket that I designed for a 20mm piezo that has been tested (though not with an aero). In my experience the trick seems to be controlling flex within the piezo and minimising extraneous movement.

Idris

---

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

I redesigned it a bit. Would this do the job?



Edit:

Actually I didn't dare that little to fled it. Changed the design slightly to give it some more strenght.



Edited 2 time(s). Last edit at 08/17/2017 04:38PM by minim.

Looks feasible. Let us know how you get on.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Just curious, is there any wiki or similar that shows setup of the sensor and adjusting the two pots? On the precisionpiezo website it was just a notice about going into another thread on this forum but wen I went into that it was a long discussion about tuning some other type of sensor and I didn't find anything related to this.

There are detailed instructions. See here [www.precisionpiezo.co.uk] and here [www.precisionpiezo.co.uk]

The first link on each page is the instructions.

Hope this helps.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

The new hotend mount printed through the night and I tested it now this morning. In the start it triggered with me blowing on the unit. Mounted it up and tested a few Z homings tapping it with my finger and it worked fine. Tested again with 400mm/min dive speed onto the bed and it pushed the hotend 1-2mm up (flex was in the mount) before it decided to trigger. Figured I had to increase sensitivity so I turned the VR1 pot (VR2 was tuned initially) up, tuned VR2 again and did a new test with the same thing happening. So I kept redoing this with small steps untill I suddenly got no detection at all. Most likely I turned the VR1 pot up so resistance skyrocketed and went down again. So then I hooked up my oscilloscope to check that the sensor was responding and it does respond, but maybe not enough? Did I break the board while adjusting or is the sensor not giving high enough signal? The peaks in the picture is me tapping the hotend slightly.

Edit: uploaded new pictures as I had the wrong channel on The one with 50mV scale is me tapping it gently. The one with 1Volt scale is moving the mount ALOT. I see that with the same sensor they manage to get 8-10V peak in another thread but no chance getting that here. Is it my piezo that is broken or what? Now I can't get it triggering properly at all so something is off




Edited 2 time(s). Last edit at 08/18/2017 06:16AM by minim.

Could be that a tap is a sudden knock but when homing it's more of a very light touch and it's not detecting a knock, don't know why it wouldn't trigger with pressure. This was happening to me, my mounts (under bed) are hollow by .5mm and getting flex and not pressure