E3D Revo PZ Probe

[ChicagoKeri]

It's been a while!   Been busy with stuff.

But through it all, the Voron 2.4 has been working well.  However, I've never really liked Voron TAP all that much, though it does actually work fairly well across a variety of build surfaces.  

To be fair to TAP, it has worked better than anything else I've tried.   The MellowFly CNC TAP has been a great experience.... it works consistently well, with only a little maintenance needed after quite a long period of use.

But what if there was a way to achieve the same result for probing without the weight and rigidity characteristics of TAP?   Klicky and various touch type probes are already well developed but E3D has just come out with a new and exciting (what could go wrong?) Piezo electric probe built into a Revo Voron Coldside, The Revo PZ Probe Voron.  They've been out of stock for quite some time but are now available.

So, I ordered one and it arrived in a week or so.   The kit consists of a special Coldside Heatsink with an integral Piezo sensor, a dedicated circuit board, a 1.5m long harness for the circuit board, and inexplicably, harnesses for the heater and thermistor wiring.  There are no Hotend or Nozzle parts included, nor any instructions. There's a custom X-carriage to download and print with mounting holes for the circuit board and that's about it if your systems already set up for a Revo, as mine is.

The following are my initial impressions:

So, off with the toolhead, off with the (Mellow CNC) TAP, on with the X-carriage, the circuit board and the Revo PZ heatsink.  Some changes to printer.cfg and ready to go, right?

Maybe not so much.

E3D's online instructions are a bit sparse but there is an instructional video on YouTube from E3D.  Eventually, I got everything installed and configured, though the road was a bit bumpy.

Bump #1 - The Revo PZ is a 3.3 to 5 volt ONLY part. So, using my MellowFly SB2040 CANBUS toolhead's 24v probe connector, gpio25,  was out of the question. Fortunately, there is a 5 volt endstop port right next to it, gpio28. But, plugging the Revo PZ into this port crashed the toolhead!  Oh. It seems MellowFly built a board with a 5 volt port that cannot tolerate 5 volts on the sense pin, which can only tolerate 3.3 volts.  It seems that the Revo PZ's board output is apparently NOT "Open Collector" and toggles between 5 volts and 0 volts when triggered.  Now, where was that BAT85 diode again?   OK, it works with a BAT85 diode installed on the sense wire, (band towards PZ board), kind of what is needed when using certain inductive probes.  Not E3D's fault, but a heads-up would have been nice. An included BAT85 would have been nicer yet.

Bump #2- The Revo PZ is quite touchy and sensitive, so even though E3D's recommended printer.cfg changes reduce the likelihood of false triggers during probing,  they can still happen. For me, Quad Gantry Leveling would cause false triggers.  There is a button on the PZ control board to choose between sensitivity settings, but I found that dropping the QGL speed to 80 stopped the false triggers.  Advice is given on this, but it isn't that easy to find. 

Bump #3- the consistency is not as good as traditional TAP.  The initial testing of PROBE_ACCURACY is all over the place. It might return a great result of 0.005 range/0,001 standard deviation or as bad as 0.03 range/0.010 standard deviation.  As nothing was altered between subsequent tests, it is likely the characteristics of the piezo sensor.  Perhaps changing the sensitivity setting on the Revo PZ control board might help. For now, I changed QGL's tolerance and retries settings.

Other stuff that isn't exactly bumps... 

1- The Piezo sensor has a thin and narrow flat flex cable that leads to the Revo PZ control board. It is relatively short and quite fragile.  Short is good for signal integrity and not getting tangled and flopping about, but is isn't much fun to work around and impossible to fix if damaged.   2- The wiring for the PZ control board is a pain because E3D decided to make the board very small but yet offer different interfaces, requiring 8 wires. This necessitates a TINY JST SH connector with 1.0mm pin spacing. this is very fiddly to work with and difficult to crimp terminals for adequately.  It would have been nice if two different boards were made available.... endstop type and data type, each with only the relevant wires. 3- Piezoelectric sensors only produce output when motion is occurring, which means the output will only be a pulse of a specific duration, probably set by the PZ control board.  If for whatever reason the pulse is not read by Klipper, the Z drive(s) will merrily continue to try and stuff the nozzle through the bed.

 

And now the good news:   The toolhead is now indeed extremely rigid and roughly 35 grams lighter than it was with TAP.   This has given me my best-yet  SHAPER_CALIBRATE charts!

 

All in all, I like it.  I think that it will be quite a challenge to get it working as consistently as Mellow CNC TAP but I want to stick with nozzle touch probing and this is the only way I've found so far that doesn't involve significant extra weight and / or loss of rigidity.  I would prefer some kind of strain gauge arrangement rather than a piezo sensor but PZ is what I have on hand so....

 

Happy Printing!

 

 

Edited March 17 by ChicagoKeri
minor correction

[sfxfreak]

I recently did a similar upgrade, although in my Crucible. 

I actually found the docs on e3d's website very rich and detailed (although the web site itself could get wonky if you open too many sections up at the same time). 

https://e3d-online.com/pages/e3d-support-pz-probe

The changing of the operating mode did wonders for me in terms of reliability and the speed at which I am now probing. 

I was used to a piezo workflow having used a Precision Piezo ages ago on a Hypercube, but I still find them very accurate and - as you wrote - having none of the requirements of a Tap. Not as fast as a Beacon, but amazing, especially in the small form factor of a Crucible. 

The one thing I didn't do before getting mine was research on the connectors. I don't want to cut into the cables delivered with the PZ and had to temporarily route the sensor to the main board with the original loom, while waiting for a replacement from aliexpress. For those who might want to do the same - get an 8 pin SH1.0 connector - don't even think about crimping them tiny things yourself

It's a great set. I wish the hotend itself was capable of faster printing than what I am getting compared to a Dragon HF (not even the UHF), but it'll do for smaller printers. To be clear, I am using the High Temp heat block with a 60W heater (not the typical heater core), a PT1000 probe and an original e3d high flow nozzle.

[claudermilk]

Welcome to the forum! Thanks the insights.

[ChicagoKeri]

Update:

The Revo PZ Probe was not the source of inconsistent Quad Gantry Leveling!  It was merely the Z0 belt had loosened slightly... it seems my CNC Z belt tensioners have no means to positively lock the adjustment and the Loctite threadlocking compound didn't hold up on Z0. 

I did change the PZ Probe sensitivity from the default setting 0 (level 5) to setting 1 (level 10) which reduced false triggers. 

With all the belts evenly tensioned, QGL now completes much more quickly  It has even done a Zero Retries QGL once or twice.

So far, I'm liking the Revo PZ on Stealthburner.

 

PS. For anyone wondering about my earlier mention of the BAT85 diode;

Certain electronic inputs have a maximum voltage. My toolboard of choice, the Mellow SB2040, has an odd powered endstop connector, gpio28, which has a 5 volt supply, but the input of which can only tolerate 3.3v.  With many devices, an "open collector" topography is used, and the toolboard input sees either a 0 volt "grounded" signal or a "floating" open circuit similar to just disconnecting the input wire.   With many other devices, including the Revo PZ Probe a "source or sink" topography is used, so the toolboard input sees either a 0 volt "grounded" signal or the full voltage of the power supply to the device, in this case 5 volts.   Where this causes problems, one can use a diode, preferably a BAT85, to isolate the toolboard input from the 5 volts "source" signal, effectively providing a "floating" signal thus simulating an "open collector" topography. 

Happy Printing!

[ChicagoKeri]

One thing I noticed yesterday when retrofitting the Voron 2.4 with Revo PZ to a Printhead and X Carriage to parts printed from PPS-CF10;

The Revo PZ Probe coldside / heatsink is not very electrically conductive. That is to say, as is my custom, I apply an electrically conductive coating to the Printhead parts so as to ensure any static charges that may build up during printing can dissipate at the nozzle.   

Despite my best efforts, there is no consistent ground path from the Nozzle / Heater / Sensor Hotside to the well grounded / earthed Revo PZ mounting bolts.   

I get inconsistent readings, not entirely open-circuited but varying from about 22MΩ (Mega Ohms) to charging up like a capacitor.  This will likely suffice to bleed off large static charges.  Apparently either the anodizing of the metal parts or the piezo element itself form an insulating layer between the top and bottom of the heatsink.