Piezo20 probe and piezo kit now available
-
Lykle in most ways the nozzle wobble is lateral and comes from the clamp resting against the piezo which is compliant - and the 2 screw fixing system is the line about which it pivots, the triggering force is vertical and is the whole hot end plus clamp pushing up into the piezo. This is why I am keen to test the design with the four rods to slide the clamp up and down on. Or as you said use a version with 4 screws (like the mk I but smaller). Its no problem for the nozzle to move upwards by 0.15mm, its equivalent to having a sprung bed (but only very slightly sprung), its the lateral movement which can cause issues with squashed down first layers which are uneven when the nozzle passes over them on the second layer and curling edges of overhangs etc…
Sakey - If I might be so bold to suggest a strategy start with my mk5 design and try it then tell me where you changed it, if this is feasible. It seems silly to design from the ground up when my version works albeit needs refining to reduce lateral nozzle movement without eliminating vertical (upwards) nozzle movement.
-
Not taking anything away from DJ's work, was just trying to mount it into onto a Rostock with a machined mount to hold the hotend already (it seems to lets it slide up and down and then is clamped via 3 points at the top already which is seemed a good platform to start with for a piezo. I wanted to use this with as little modification as possible. My temp spacer is just a nut with a cut out in . Ill post a picture and stl when I re-assemble and get it working somewhat reliably.
-
That makes sense, it will be nice if these can be fitted between two existing parts without much change but my experience so far shows they need to be able to flex on probing contact, by an amount (around 0.15mm if my z resulting offset is anything to go by) to trigger clearly above the noise they generate.
-
Hmm, if the hot end can slide up and down in the clamp, you might be able to avoid sideways wobble.
Only trouble I see here is tolerances. You have to get it exactly right so that there is no sideways play but just enough space to slide up and down. That will be very hard to achieve with printed parts. So it will have to be post processed.But I like the idea of a sliding hot end, A little like the Auto lift hot ends but then for calibration.
-
Well unfortunately the rods Doug sent to me have not appeared, so lost in the post might be the issue, but I have now got some 4mm steel rods (wood screws shanks cut down with a dremel). I am printing a version now with holes for these rods. I intend to bond/heatpress the rods into the top piece, and then hand file the clamp piece so the rods fit and slide. This might be difficult due to the fact that the clamp will have to be tightened around the hot end exactly the right amount or these holes won't line up.
-
Okay so this latest version with rods to support the assembly is working extremely well. Almost no nozzle-wobble and calibration deviation at 0.02mm. It seems to be the optimum between sensitivity and nozzle wobble.
If you intend to make something like this, and are not planning to design your own (or if you are making your own, consider downloading this one and perhaps copy the salient features to your design - to save you time) this is the version to make. I am now calling this the beta version.
I have yet to test it with acetal or PTFE rods instead of steel but 4mm smooth steel rods work.
It's now on Thingiverse - http://www.thingiverse.com/thing:2069480
-
DJ I'll put some more in the post tomorrow mate
-
Glad to see it is working so well.
It was a pleasure working on this, even tho it was just a small part. -
That's way too kind Doug, the post is a constant source of irritation.
This version does seem quite solid. I've done one to fit Lykle's (congrats on getting the nimble on sale ordered mine already) effector and the other generic one to bolt onto underside of an effector or carriage.
I might make one up for the corexy (and a new x carriage to fit it to) be nice to try it on a ramps setup for the benefit of the majority who haven't got a duetwifi.
Also along the same lines thinking if I can make it fit a clone hotend bowden with the much larger push fit bowden couplers, might be tricky can't cut a 12mm hole in the Piezo…
-
I am really impressed by your works guys!!! great to see such projects.
-
Thanks on behalf of all who are contributing, it's been a very encouraging collaboration. 3d printing needs a z probe that's cheap, accurate and reproducible, that minimises the effects of tilt because it has no offsets.
-
Dj
Another package on it's way you should get it tomorrow sent it first class handed in at the PO desk
Doug
-
Awesome, much obliged. I've printed another one with the rod supported version on the kossel Xl and came out looking awesome and a mount for my corexy so hopefully I can use those rods and test the setup on ramps with marlin. I'm aware it's all been very DuetWifi so far. Rods have turned up so thank you very much.
-
Okay so made another one this time with 4mm wide 15mm long acetal rods (thanks Doug). Works fine nozzle is quite firm, sensor triggers normally.
Have attached it to my corexy (ramps/mega/Marlin RCbugfix 31-01-07 on). The usual messing around with Marlin and uploading firmware. However it will home nicely and entering a z offset of 0.2mm means the nozzle is nicely positioned just above the bed at z=0 on the lcd. It will do grid levelling if auto_bed_levelling_bilinear is enabled.
I have added instructions for configuration of Marlin on ramps/mega or compatible in the thingiverse listing http://www.thingiverse.com/thing:2069480
Interestingly Marlin allows the M48 probe repeatability test the result is:
Recv: Mean: -2.492500 Min: -2.510 Max: -2.475 Range: 0.035
Recv: Standard Deviation: 0.009014So hitting a standard deviation of 10 probes at just below 0.01mm repeatability.
-
Been chatting with Moriquendi about making it safer in terms of a wire break. Currently you'll just get a head crash.
Current thinking is that putting two Piezo transducers back to back into the unit (might need slight adjustments to the top part), would enable redundancy as well as potentially stiffening the system. The second Piezo has to be wired in parallel but with black and red reversed so they don't cancel one another out. If one Piezo is disconnected the other will still function. However you won't know it's failed potentially unless it begins triggering differently,which it might not given the way the comparator works in the signal board.
Any thoughts?
-
First of all, VERY COOL idea. I presently have FSRs on my bed. I'm following this thread with lots of interest.
One thought on redundancy, using your idea of back to back transducers… Maybe set it up such that both sensors are handled separately with signal conditioning, then the outputs are logically "AND"ed together. Then, if both don't trigger as expected (maybe after an RC time constant??), the ANDed output stays triggered (or is somehow latched triggered), which will signal an issue to the Duet's firmware.
Or similarly, have a second output from the signal conditioning board that would change states if the outputs don't behave as expected (i.e. last time, only one triggered). This signal could be monitored by the Duet firmware to identify a fault condition.
In either of these cases, you would probably have to match the gain and offset of the signal conditioning/comparator for each sensor such that they trigger at the same time. Or get really slick about it and let a tiny PIC automatically configure gain/offset for each side and also have it drive the triggered output with a buffered signal.
Hope that all makes sense. I don't know if what I suggest is feasible, but I thought I'd throw it out there.
-
One option my be to test the Z probe before doing the first probing move. I would expect a short downwards movement that is reversed rapidly to trigger the probe.
-
Toddimus - these are great ideas, some of it slightly over my head (not an electronics native), but I'll leave Moriquendi to comment as he made the boards. Suffice it to say some of our discussions earlier were about a new version of the board he is considering, which can support up to 3 piezo's as they are being used as under-bed sensors in one system, but which could also support a twin piezo in this design and perform some safety checking. Auto setup etc.. would be a great feature at this early stage, but if it ever becomes a product then I'd test a wide range of available transducers and pre-set each unit to function with very minimal adjustment by the end user.
David - I advise in the thingiverse listing to manually press up on the hotend at the start of a printing session and observe the led changing, and (on duet) to check the analogue sensor value is normal when the sensor is open, mine reads 216 and rises above 600-700 when given a firm upward press. I probe with 1/3 motor current so head crashes if they happen as I'm tweaking it all the time, are fairly undramatic. But yes a macro or something to test the probe automatically would be a nice addition.
-
I also considered that whilst it would offer nothing more for dual nozzle machines in terms of getting both nozzles at the same height, for dual x carriage machines one sensor per carriage would mean the two nozzles could be compensated in z, to exactly the correct z-height before printing.
What I also like with this approach combined with the very convenient nature of autocalibration, as we have it on Duet, is that a nozzle change requires no manual config adjustments, just swap it, autocalibrate and print.
-
So, the boards that are in the wild at the moment will accept any reasonable number of piezo elements wired in parallel and give a single output if any are triggered. With these boards it is not possible to differentiate between one element triggering and multiple elements triggering. I call this a single channel board.
I have a small number of three channel boards, three inputs that can be triggered separately and three independent outputs. These were designed to explore the use of piezo endstops, it's all well and good having a highly accurate and repeatable zprobe but if your endstops aren't repeatable you're no better off.
I like the idea of testing the probe by jerking the hot end, it's the simplest solution though it might make tuning more difficult with different hotends. Also I foresee problems with non-delta printers that either don't move the head in the z-axis or that cannot move it fast enough, x-y translation might work but increasing the rigidity of the hot end mount has been half the challenge so far.
With two independent channels you could wire two outputs to the controller and have the firmware look for simultaneous triggers and report a fault both aren't very close together. This could have other applications not just for other probe technologies (all wires can break, we just have a particularly delicate setup at the moment) but perhaps for measuring the offset between two nozzles as DjDemonD suggests.
Having the board detect a failed element could be done though I'm not sure it's the best solution, I'll have a think about how it might be done.
Moriquendi
