Piezo20 probe and piezo kit now available
-
I am still occasionally seeing it register the tap at my Y tower a tad late - I can see the effector "tip" a bit. It's only at the Y tower, and I'm not sure what's causing it. I can't adjust the sensitivity without affecting the other probe points negatively, so I suspect it's something about either the bed there, or that tower / belt. Adjusting belt tension didn't help it. This is the tower that had a bad bearing, and I'd hoped that would fix it. It seemed to improve, but is not 100% resolved.
-
I was getting v poor probing at one point but only on the left side of the bed. I wondered if it was the bowden tube or one of the screws being looser than the others, turned out to be a loose electrical connection.
-
If I probe at 850mm/min (with jerk & acceleration reduced, though) I don't get the tipping at that point. below 500mm/min, it starts doing it at more points, all on that side. I suspect maybe effector tilt isn't giving me as clean of a tap on that side. (I have plans to have an aluminum effector milled custom, but haven't had time to properly design it)
-
Ok, first, don't tap that fast. Results got progressively worse.
Second, I found my issue, I think. I like repeatability as proof, so I'll do a lot of rounds of checks and see how consistent it is. When I'd tightened down the piezo probe down with the 3 screws on my hotend, I got one lower then the other two, so my hotend was sitting at just a slight angle. This was causing poor results on one part of the bed. I straightened it up and now (4 rounds done) I'm getting good probe points all around the bed. I'll report back if that turns out not to be the fix.
Edit: Just finished a height map. I should try slowing down the probing speed now that things are better. This was at 600mm/min. http://imgur.com/kk1uaGH
-
Could you post your repeatability figures if you captured them?
-
I didn't… And then I tweaked too far trying to improve more, and heard the ceramic crack on the piezo. Sensitivity definitely went down. I'll drill another today and swap it out.
-
I'm probing at 300 mm/min.
-
Perhaps my "simple mount" isn't good enough?
I drilled a new disc, it's still very sensitive going by the LED. But when I try to dial in R1 there's no room between "triggers on any printer move" and "doesn't trigger on tap".
The top nylon washer I'm using is 3.5mm in diameter, and that's all of the piezo it's touching - is that too far in, and preventing it from bending enough perhaps?
-
Do you mean it's 3.5mm difference between ID and OD?
-
Yes.
-
I think you may be running into the same problem DjDemonD had in the beginning. Looking at your effector it doesn't like you're controlling lateral movement of the hot end. DjDemonD had to add sliders to allow vertical movement of the hot end while preventing lateral movement causing triggering.
Idris
-
The Slot for the hotend does actually by design have the hotend groovemount on an E3D "drop down" a bit into the mount. If anything it might be a little too snug of a fit. That's what I was hoping would make this ideal.
How tight do you push down on the piezo? Just do it's held flush against the hotend, or do you flex the piezo some (in a "resting" state)
-
It can be preloaded a small amount, as long as further upward pressure will bend it more, we're only talking 0.1mm more. Its the rate of bend that generates the signal rather than the absolute amount.
-
Ok, back in business.
I think my piezo was grounding the brass plate against my hotend mount. Maybe. I trimmed a spot in it around each screw, to be sure it wouldn't make contact. I also went back to the piezo I got the best luck out of… don't think I actually damaged it. I'm using a P1 probe type with a low trigger value. Seems to be working at the moment, anyway.
I think drilling these without damage is going to be the challenging part, at least at first.
-
My repeatability testing is now underway. If you have suggestions on the methodology or ways to display the data, let me know and I'll adjust: https://docs.google.com/spreadsheets/d/1EwNdSNXZR_OhwChLFrTYqFDFcKjB-4z6OwGkocKgW7E/edit#gid=0
Currently I'm just moving to Z5 and doing a G30 s-1 50 times. I have the min, max, average, and stddev at the bottom.
Edit to note I'm powering off and doing a full delta calibration between each set.
-
Thats good data. When I had a piezo in a machine running marlin I did the M48 repeatability test a few times and came up with similar results.
Recv: Mean: -2.492500 Min: -2.510 Max: -2.475 Range: 0.035
Recv: Standard Deviation: 0.009014Not quite as good, but the hastily rigged carriage this machine uses is quite wobbly itself.
I've mainly been using my delta autocalibration deviation as the metric, which is, I accept, flawed as there are many other factors feeding into this.
-
My delta calibration leaves a bit to be desired still. Just working on the probe for now.
Calibrated 6 factors using 16 points, deviation before 0.055 after 0.054
I still have some work to do in solving an oddity I've talked to dc42 off and on about, to get that down. Having more accurate probing (and using the nozzle) should hopefully help that.
-
Started a listing as a repository of 20mm versions which are currently in alpha testing.
http://www.thingiverse.com/thing:2152709
Feel free to try/mod one, its a basic module at the moment for integration into where-ever you need it. Not fully tested but the module does trigger the signal conditioning board when tapped so I am assuming it will be basically functional.
Very small, no rods/pins you need signal board, and 6x20mm m3 bolts and 6xm3 nuts and a 20mm piezo with 5mm hole drilled into it. I put a drilling template on there too.
Please feedback if you make/test one I haven't really a suitable machine to test this properly. I will be putting a version up later today which fits into a titan extruder.
-
Could you post your repeatability figures if you captured them?
If it's of any interest, I just probed my bed 89 times and had the exact same reading every time, to four decimal places, 232 steps and 2.9000mm
Idris
-
Thanks Iris that's a very encouraging result!
