Plugging in to Wanhao Duplicator 7

[Steve Hooker]

Hi,

I have a 3d DLP resin printer that, I think, needs this accuracy.

I've started a thread on the Facebook Duplicator 7 support group about this NZS. It looks like it'll take away the layer errors I and others are experiencing. There's 3,000+ of us there. Most of us are numnuts when it comes to modding, but many of us are prepared to dive in and follow along ;-)

I'm only able to guess that it would be possible to install the NZS inside the D7 box, where the Nema17 stepper resides, and that the NZS screen could be placed outside but those buttons...I don't know how to get to them. Could they be accessed over PuTTY which is how we access the Raspberry Pi, where NanoDlp lives anyway?

The D7 uses as it's mainboard an ATMEGA2560 Arduino, so it should be fairly easy to put the driver into it. Certainly, we can flash the firmware with the Arduino IDE.

I think I'm still a little confused as to the setup. I'm just awaiting some of the other, cleverer, modders in the Facebook group to post their thoughts, but I thought it may be good to link people here in too.

Be happy for ever.

Talk to you in a bit.

Steve

[misfittech]

The LCD and buttons are not needed in normal operation. The LCD and buttons are mainly for setting up the unit (microstepping, max current, etc), however this all can be done through USB and/or hardware TTL serial which can be connected to an Arduino if needed. 

The unit mounts to the back of stepper motor and takes up around 3/4" of space on the back of the stepper motor. 

Please let me know if you have any more questions.

Trampas

www.misfittech.net

[Steve Hooker]

Hi Trampas,

I have installed and wired up my NZS, uploaded the software, ready to go. But I cannot connect with PuTTY. I've researched here as best I can.

Maybe when I installed the drivers, the Misfit Tech net failed and I'm using a Microsoft driver from 2007, perhaps this is wrong? I've screen shot as many windows as I can think, perhaps you can see something?

Or maybe I should press one of the four buttons on the bottom of the NZS? 

Anyway, I've futzed around some, retraced my steps, reinstalled, restarted, the usual try, try again, but still at the error in PuTTY and cannot seemingly go further.

Of course, I may not have wired up the NZS correctly from the Wanhao board. But I have followed good instructions from the wise older hands on the Facebook user group. I have not used the 'enable' pin - yet. As I know it has something to do with an error, but don't know enough about this, and from your instructions, this seems to be optional.

Any clues?

Be happy for ever,

Steve

[Steve Hooker]

Hi Trampas, 

I see that the only thing I haven't tried is the unsigned driver restart on Windows 10, as per your hint:

Drivers for the USB serial port are here. On some windows system since the driver is not signed you might need to do the same steps as outlined here to install the driver.

 

I'll try that -- soonest.  

Back to the day job (for a few days).

Steve

--
=

Steve Hooker

Gwendda Fine Arts

Shropshire's famous

the world's only

Witch Photographer

Witches, fairies, ghosts and giants 

Now also dragons

http://www.witchphotographer.com

http://www.facebook.com/gwendda

http://www.twitter.com/gwendda

Mobile: +44 (0) 7903 723 298

+44 (0) 1952 400 420

[Steve Hooker]

OK, I'm in. Two things:

1. As suspected it was the Misfit driver that wasn't working, I had to restart with driver signature enforcement disabled. Reinstalling the driver worked, then. 
   
2. My USB cable was plugged into my hub. Once plugged into the main motherboard on my machine, I could get in and I used your recommendation Tera Term.

Now, I'm trying to configure the direction and steps per millimeter as although it's moving, it's not moving quite right -- it was before I added NZS.  

[Steve Hooker]

The steps per millimeter is in the Wanhao's main board's EEPROM. I don't know why it was cleared but re-adding M92 Z3200 and M500 to save it to the EEPROM again has cured the wong distance moved. Back to normal.

Changing the direction was easy:

:>dirpin 1

dirpin CCW(1)

So, effectivly, all I have done is :>calibrate, which give me a table of figures and:

:>testcal

Max error is 0.060 degrees

Then, to move in the right direction:

:>dirpin 1

dirpin CCW(1)

Still much to discover. 

Do I need to set :>feedback?

Should I set up spid, vpid, ppid?

What can I do with the enable pin? Would this be linked to the errorlimit? I'm thinking that if the print, in some ways, fails, the error limit would be breached, then the enable pin is triggered, the print stops and in NanoDlp the 'Print Stop Gcode' is run and the gantry rises to the top and the finish print sound is made.

[misfittech]

The feedback is set for the sPID mode from factory.  This a simple hybrid PID mode, what it will do is use PID parameters but never let current go below the 'hold current'.  This mode works well for most people and is easier to tune. The PPID is harder to tune, and will be quieter. 

You can play with the SPID parameters, I found that somewhere around P=0.6 D=0.01 I=0.01 worked well for me but your results may vary.  The general rule I use is that the I term determines the maximum static error you want. Then the P sets how quickly it will move when step occurs, too high and it oscillates. Then the D term helps improve stability, up to a point. 

Most people do not use error pin, it can be setup to tell you if there is an error, but the PID controller will do everything in the motor's power to prevent errors.  So the error pin is really only set if a you failed to home and hit end stop or other major error that rarely happens. One OF the things that can be done with error pin is to lower the drive current on the Smart Stepper and then use error pin to detect end stops.  This however requires some firmware changes to controller that has not been done. 

 

The enable pin can be used for enable signal from control board, when disabled the motors are free to move.  Again some people use this, but most don't. 

Trampas

[Steve Hooker]

Cheers Trampas, please forgive this newbie, I cannot get the syntax right for setting the spid values.

So, upon your advice, I'll ignore the error pin. No worries with keeping things simple for this stupid.

So, now everything is moving well with default values and calibrated to my Wantai Nema17, it's as if the NZS is not there. So, I have an existential crisis. I bought this, to absolutely ensure that my prints are the best they can possibly be, after reading this ringing endorsement. As he says, "each layer is exactly, precisely, repeatably the same height with a deviation so small I'm not sure how to calculate it." "Yeah, yeah," thought I, "that's what I want."

But, to be honest, my layers are already pretty good, I have a THK KR26 linear actuator, whose numbers for repeatability and backlash are jaw-dropping. But the chap in the Tony Stark link has an LCD, I do not, cannot, not really, as my Nema17 is hidden away inside the Duplicators 7's box. Now, I'm sure that this NZS is making each layer that much more perfect, each time it lifts, waits, drops and is exposed once more, those movements are going to be super accurate. Let me quote again from 'Tony Stark.'

The LCD constantly displays the current axis position during a print (in relative rotational degrees from "0") as well as the current positioning "error". I print at 24 microns layer height, and hover at less than 1 degree of rotational error. On my axis one complete layer (24 microns) requires precisely 1.42 degrees of rotation. All of that adds up to mean that no matter what, my layers are never more than half a layer off.

And further.

 As silly as it sounds, it takes a lot of the wondering out of a print when you can't see the underside of the build plate to just watch the error counter NOT rise.

But, Trampas, buddy... How can I show this to someone else? Should I get an LCD and wire it to the outside of my box as a comforter? Am I able to monitor such through the USB cable via TeraTerm? Of course, you can say, "just believe, Steve. Believe. Trust in the power of the NZS" :-)

I'll be pointing to this from the Duplicator 7 Facebook page, where I expect others to follow my footsteps in the snow. So apologies for my lyrical texts.

[misfittech]

try "spid 0.6 0.01 0.01"  hence include all three values on one line with spaces between them. 

You can wire the LCD external if you like, just run some jumper wires for the 4 pins.  

As far as demoing the power of the NZS, when printing grab the head and move it while printing. 

Also another thing you should do is increase your speeds and acceleration. With open loop steppers the speed and acceleration is set conservative to prevent missing steps, the NZS closed loop control prevents this. 

So power on you machine and grab the axis with the NZS and try to make it miss a step on the motor. Then try the same thing without the NZS.  After doing that try doing tests with faster print speeds, increase the speed without the NZS to point your prints fail, then install the NZS.  Other advantages are lower heat on motor, and quieter motors. 

Of course the NZS will make sure motor position is accurate, however if you have no problems without the NZS it can not fix a problem you don't have. 

Trampas

[Steve Hooker]

Excellent. I may get an extra LCD, for that warm fuzzy feeling, I guess.

I have seen videos of that grab/release trick on FDM printers. Not something I'd like to do with a vat full of sticky resin. But, there is a similar trick on resin printers. Ordinarily one cannot see the beginning of a print since it's dipped in resin and hidden in the solid, opaque vat, so people interrupt the regular lift of around 6mm, to a much higher lift to inspect the progress, maybe 50mm. But, it never goes back down to exactly the right layer height, leaving a distinct line, in essence, an errored layer to pay for one's impatience. With NZS, it would be perfect. A good trick to those in the know.

As for speed, I'll test this, but am reticent of speed increases as the UV exposed, cured resin actually sticks to the clear, thin FEP sheet at the bottom of the vat. There is an audible 'twang' or 'pop' as it is wrenched off the FEP sheet. Gentle and slow speeds are vital here to ensure that maybe a wee 15µm freshly cooked layer isn't damaged. Accuracy is the aspiration. More perfect layers and yet more perfect. Can never be too perfect. Add as much perfect as I can :-)

Thanks Trampas. You're the man.

[Steve Hooker]

Hi Trampas,

Everything is too tall. The smaller I go the bigger the error. 

Here's a photo of a test model printed with NZS and without. Both are at 5µm. Ther settings are exactly the same, I'd expect both to be exactly the same height.

My NZS settings are the default, save for spid and the microsteps. As in a screen grab above.

I have the steps per mm at 16 microsteps x a 0.9° stepper x 2mm pitch on my linear actuator = M92 Z3200 and all is good and accurate without the NZS but much taller with NSZ. So, I wonder if somehow NZS is confusing itself with microsteps? In between your reply, I'll try setting NZS' microsteps to 1. But I'm stabbing around in the dark and each 5µm print, with 2000 layers, takes 19 hours, so any pointers are most welcome dude ;-)

• At 10µm layer height, the NZS is actually printing at 11.44µm
  
• At 5µm NZS is printing at 7.03µm
  

Be happy for ever,

Steve

[Steve Hooker]

Oh, I should add, that when I ask my D7 to move "G1 Z180 F400" it does indeed move 180mm, with and without the NZS. Truly all seems well in the terminal.

[misfittech]

What is the height of the part you printed verses the expected height?

What was you previous stepper driver? 

The NZS could have a problem but given the number that have been shipped and installed I kind of doubt it.  I know the DVR8825 stepper drivers has a lots of problems with microstepping. 

Finally can you provide the belt pitch and gearing details as well as you controller's steps/mm?

Thanks

Trampas

[Steve Hooker]

In the photograph, my part is 10mm high, it's the regular 3dSLA_test model. https://www.thingiverse.com/thing:1219068. But, I've been having problems with other my models too.

Without the NSZ, using the Wanhao Duplicator 7's internal 1.4 Ramps board with the inbuilt A4982 stepper I measure 10.08mm high. 

Bypassing the internal stepper and using the NSZ I measure 15.20mm (or 7.6µm layer height). Both are printed at 5µm that's 2,000 layers.

I've added a picture of the Nema, just so you have an idea in your mind. Above, behind my FDM printed support, is a THK KR26 linear actuator.

As said, I have a 0.9° stepper so that's 400 steps per revolution. I've added the default 16 microsteps to NanoDlp and the ballscrew has a 2mm pitch. All adds up to 3,200 steps per mm. So, I add M92 Z3200 to the terminal (and all is well without the NZS and naturally, I'd expect models to print correctly with the NZS with exactly these same settings.)

So, if you're zeroing in on microsteps, should I try without any? That would be a drag since I want to print small layers. But, if I can test without microstepping and arrive at perfect hights in my models that would be a step in the right direction. If you could suggest some settings for me to attempt, I'll gladly try them.

[misfittech]

Was the NZS calibrated with motor disconnected from the ball screw? 

The positional accuracy off the NZS should be better than 0.11 degrees if calibrated correctly. Given that your error is 50% (which is huge) I am guessing it is not calibrated correctly.  A bad calibration can still be accurate on large move as it more a function of full rotations than fractional rotations. 

Is it possible to use feeler gauges and move the system small amounts? For example if you move the motor 90 degrees this should move 0.5mm is this the case? Also looking at the LCD to note the error is helpful, as 1 degree error is 5.5um on your system. So if the motor error is low and it is not 0.5mm then it most likely calibration error.   

Again with your current error a 0.5mm movement would be ~0.75mm which should be easy to determine with a feeler gauge. 

[Steve Hooker]

No, the calibration was WITH the ballscrew attached. I'll have to detach everything to do a calibration WITHOUT the ballscrew. Not a problem, but will take some time.

I don't have feeler gauges. I'll go get one from the shops now, if I can't I'll order online, again this will take some time.

Also, your additional LCD arrived earlier this week in the post. I haven't attached it yet, as it needs to be wired outside the D7's case. Not a prob... But will take... blah blah blah.

Great Trampas, you've given me something to go at. I'll be back soonest.

[misfittech]

https://www.amazon.com/Hotop-Blades-Feeler-Imperial-Measuring/dp/B06XHXJG31/ref=sr_1_2_sspa?ie=UTF8&qid=1516284296&sr=8-2-spons&keywords=feeler+gauge&psc=1

Also when you use the feeler gauges check the play in the system. You will have trouble getting more accurate than your play in the system... 

Thanks

Trampas

[Steve Hooker]

Sorry, Trampas, what do you mean by 'play in the system?'

[misfittech]

So if you can move the carriage system, ie flex it by hand. This is "play" in the system.  It is impossible to design something completely rigid and have it still move. So the trick is to remove as much play in the system as your budget will allow.  This is why a good CNC mill may only have a 12x12x7 working area and weigh a few tons.  That is it is designed such that if you grab the end mill it will not move more than .0001"  

So imagine a FDM printer where a blob of plastic comes out and makes a bump on the print surface. If the mechanical design flexes when the nozzle goes back over the bump then this flex is due to play in the system. This flex could happen when you change direction of the nozzle, and hence it can cause accuracy issues. 

Trampas

 

[Steve Hooker]

Gottcha! No, no play here. This, sir, is a THK linear actuator. No wobble, backlash so small it's insignificant. Repeatability that's so good you could set your atomic watch by. This is supposed to be rigid.

OK, I'm lush with hyperbole. Probably due to my naivety but from what I've read these actuators are the best you can get.

Remember, I only move on one axis. This is why I've been printing at 5µm layer heights and have been challenged to try 2µm. This photo is a close up of an ancient stoneage monolith, with some 17th-century graffiti and a match head for scale.

Will be back once I have a feeler gauge.

Cheers Trampas.

[Steve Hooker]

Hi Trampas,

Apologies for the delay. I bust my NZS, snapped the USB port and my soldering skills knackered it. Now, I have a new one from you but before I get the above height problem I have two other issues:

It's not going the right direction

This is weird since all was well with the other NZS first time, it went in the correct direction. I've tried reversing the wiring, every way I can. But it keeps going in the wrong direction! Maddening. I swap from 1, 2, 3, 4 to 2, 1, 3, 4 and from 2, 1, 3, 4. to 2, 1, 4, 3 and I've tried 1, 2, 3, 4. 

I read in motorwiring that, "the reverse or forward wiring of a motor is detected on first power up after factory reset." But, I rewire, factoryreset, calibrate and it's still going the wrong way. (I really, really want to avoid #define INVERT_Z_DIR in Repetier.)

I also read that motorwiring is for experts only. Naturally, I've tried it but the 0.9° Wantai Nema 17 keeps revolving non-stop with a slight pause on each revolution. It seems the only way to stop is to reflash the NZS's firmware.

As I said, I don't get why it insists on going in the wrong direction, as when I attach the now broken one, that one goes the right way. I'd use that NZS but unfortunately, I cannot connect it to the USB port to calibrate.

My LCD seems dead

I bought it separate from you. But when I plug it onto the board - nothing. Is there a trick with the three buttons that I'm missing? I see your YouTube video, but I'm getting nothing. Could it be dead-on-arrival? Is there something I should switch on in the firmware?

Arg! All is well now! 

My LCD finally lit up! I see now that it needs to be attached before a power-up. Simply attaching it while the power is on and it looks dead. Within the LCD menu, I spy a *CW and *CCW something and lo, my motor is going the correct direction!

I'll leave all of the above for others to find if they too are numbnuts, like me. <Bashes head on desk>

Now to try the layer heights and my feeler gauge... I have calibrated with the motor outside the printer, not attached to anything. Be back in a bit.

[Steve Hooker]

Ah! I see that my busted NZS isn't too busted after all. 

When I try to connect it to my computer via the USB I get an error "unknown usb device descriptor request failed." This was my bad soldering, I'll guess, as I took it to a phone repair shop and they did it under a microscope, but still the same error.

I can use the LCD to calibrate it. Haha!

Fantastic, as I've bought and modded another D7. I just need to order another LCD from you ;-)

[misfittech]

The firmware will detect if the LCD is plugged in when it is first powered up, if not it will not use the LCD.  The reason for this is that the Arduino LCD driver is poorly written and if the LCD is not connected it will wait forever for the LCD to return some data that never happens, I have been going through and trying to put time outs in where I can to fix these issues but for now to be safe the firmware detects if LCD is connected on power up. 

You appeared to have found the rotation direction setting in the firmware.  The encoder is designed such that we always know which direction is clockwise or counter clockwise.  On power up it will detect the motor wiring and always calibrate with the motor turning clockwise.  With the menu option you can select which way you want the motor to turn based on the direction pin, this is great as if you have dual motors for Y axis you just need to set one for clockwise and one for counterclockwise.  For most people this is much easier than rewiring motors... 

The USB ports are tough to solder. I also tried to find a robust mechanical connector with through hole pins to try and make it as tough as possible, however they can still break.  If you send the board back I will be happy to resolder the connector for you. 

Thanks

Trampas

[Steve Hooker]

Hi Trampas, 

The new NZS is working well and my heights are spot on. You were right, that I should have calibrated the Nema outside the machine, not attached to anything. In fact, I just held the Nema motor in my hand, as I set a calibration off in my hand from the LCD.

Here is the machine. Inside and out.

Unfortunately, the busted one, with the repaired USB port... I guess I got some solder flux under the first button and cannot get to the bulk of menus, specifically the CW/CCW menu. I can calibrate and observe the 'diagnostic menu' and that is all. It's knackered. I'd like to change the microsteps, too. But can't access the menu nor do anything over the busted USB port.

So, I'll take you up on your kind offer and write to you privately. I had heard you were a decent chap and in return for the karma, I'll swap you a picture of mine. Pick one, in large size, add the link to your choice, in a reply here, for the Google Juice, and I'll post one to you, gratis.

Steve Hooker

Gwendda Fine Arts

Shropshire's famous

the world's only

Witch Photographer

Witches, fairies, ghosts and giants 

Now also dragons

http://www.witchphotographer.com/

On the 'diagnostic menu. What does it tell me?

OK. I'm cool with revolutions per minute ;-) I used to have a 78rpm record player. I understand this. But why 'simp?'

'Err,' though? I see it flicker when my printer is holding, say between 0.06-0.07 or 0.12-0.13. And rise when my single axis moves 5mm during a print, ever so slowly and flicker between 0.07-0.40. When it's moving at its fastest and furthest - 180mm, at the end of a print, still pretty slow, it may rise to 0.9.

My machine prints at 3,200 steps per millimeter (0.9° stepper x 16 microsteps, x 2mm pitch). I know these err numbers are pretty small from the resultant prints - all good. But I'd like to know what it means. Could you educate me?

The Kdeg leaves me somewhat mystified too. I thought there were only 360° allowed ;-) I get that it's a positional readout, and I can power down and restart the printer when it's at its home position, to zero the display to 0deg, but I did notice, I think, after, that the display was showing a negative figure, somehow, which didn't make sense.

BTW: A print process is: home, move to the correct layer height at 5µ, stay there for 20 seconds for an exposure, move, very slowly up 5mm to pop the layer off the FEP clear sheet, pause for 0.1 second and back down, less slow, to the next 5µ layer. At the end of the print it shoots (still pretty slow) up to the top of the linear actuator for print removal.

[Steve Hooker]

Hi Trampas,

Thank you, it arrived quickly - two days ago, to my door, without me having to go to the depot. 

I installed it yesterday morning. But something is wrong. I've tried it after 15 minutes, two hours, six, now nearly 24 hours. I've read this thread and tried shorting everything I can see with the two points of a scissors. All in vain.

I don't know what happened - how I got here. I uploaded the sketch, I can't recall if the printer was also powered up. Then, I did have the printer powered up and connected via the USB to my computer. I wanted to get the LCD working, so needed to connect it this way then that way, each time I naturally toggled the power to the printer to see if I had the LCD correctly attached via my extension, but when it was I saw that it reported this error and I needed to reboot. Now, I know I should have only needed to attach the LCD  extension to the board, this way or that, but I reattached the LCD, too - this way or that. I guess I did this quickly, without much thought as to if the board was powered or not. And that's as far as I've been able to get ;-(

I've taken it away from the printer. Reinstalled the sketch. Tried with the points of the scissors. Pressed buttons. Left it nearly 24 hours... Have I screwed the AS5047D by futzing with the LCD?

[Steve Hooker]

Try the latest firmware in git hub. The encoder changed on the board, so it needs 0.30 firmware or newer.

Ah! Thank you again. Indeed I was using an older version of the firmware (though it was fine on the other boards).

Anyhow... All is well, I'm calibrating and putting my printer back together.

[Steve Hooker]

Hi Trampas,

Sorry, not been back in touch. Can't get it to work properly. 

I can calibrate and everything else with USB and the LCD, but cannot get it to move with the printer. I've sanity checked by swapping printer but the same results there, whilst the good NZS with the CW 0.25 works fine on both printers. I suspect then that there's something with the updated software that's screwed things. Indeed I think it's the earthing through the screws...

First, I have to confess, I'm no electrician. But, in my defence, I've wired these NZS so many times, I'm pretty sure I've not connected anything different (and my sanity checks confirm that I am indeed sane ;-). I should say, that the settings are exactly the same on both NZS boards, the motor = 1500, hold = 500, microstepping = 16.

It did move once via the printer's board. But just the once. It may have been me jiggling the wires thinking there was a loose connection. And calibration checking was giving strange results, like -15, 9. The numbers seemed erratic. The other weirdness I saw was when I touched the stp screw on the NZS with a screwdriver, the motor would turn, spontaneously. 

But now I've reconnected everything after the sanity check swap over, there's still no movement out of the board and touching the stp with a screwdriver does nothing. Now, and also the calibration check is giving quite low numbers like 0.2 and seems more consistent. Though I'd expect 0.6 or 0.7, as with the other board.

So, reading through your updated notes, and the fact that there's earthing through the screws, mine seem insulated from the board with plastic washers, I wonder if this is my problem?

I hope the above will help you diagnose my issue, or if you could suggest something to try.