Converting CHMT-36VA to Duet 3 6HC control

805 views
Skip to first unread message

Jarosław Karwik

unread,
Jul 18, 2024, 2:06:13 AM7/18/24
to OpenPnP
Hi,

I have been converting my machine  CHMT-36VA to Duet 3 6HC  - after I removed original controller. I have planned to use my own, but due to selling the machine  I decided to go for widely known generic control.

There is one source of information:

But:
- Was it really done ? There are issues with the info as the machine is 6 axis - not as claimed 5 ( the missing one is driver for tape peelers wheels)
- Has anyone else used Duet for the conversion ? I will use I&S from OpenPnp, but any details specific for the machine are welcomed.  E.g.  driver current for X/Y/peel axes

 







Jarosław Karwik

unread,
Jul 19, 2024, 10:00:25 AM7/19/24
to OpenPnP
Just got Duet.

Under the hood - ready for conversion.


chm2.jpg

Few findings:
1)  Original stepper supply was 36V - it is too much for Duet 6HC ( 32V limit in my revision). I will replace all these supplies ( there is 36V, 12/24V and 24V) with single bigger 24V supply.
With Duet you would need two extra expansion boards with drivers ( which can take 48V as well as encoders)
2) Some connectors ( for motors ) fit directly :-)

SM

unread,
Jul 19, 2024, 3:39:38 PM7/19/24
to OpenPnP
>> 32V limit in my revision

it's a pity, because many small stepper motors are really fun at 36V and above :-)
Message has been deleted

Wayne Black

unread,
Jul 19, 2024, 4:06:52 PM7/19/24
to OpenPnP
Yes, it was really done by me. I'm pretty sure others have done it as well. Peelers are treated as "Actuators" not "Axis". My conversion was a full gut. I replaced the head, swapped the pumps for vacuum ejectors and replaced the closed loop drivers w 48V units. I may throw some unused 72V CP motors on from another project. Don't have a lot more info on it as I'm a bad note taker, but here's some pics I took at the time of doing it;


On Friday, July 19, 2024 at 1:03:01 PM UTC-7 Wayne Black wrote:
Yes, it was really done by myself. I'm pretty sure others have done it as well. Peelers are treated as "Actuators" not "Axis". My conversion was a full gut. I replaced the head, swapped the pumps for vacuum ejectors and replaced the closed loop drivers w 48V units. I may throw some unused 72V CP motors on from another project. Don't have a lot more info on it as Im a bad nto taker, but here's some pics I took at the time of doing it;


--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/3a63d376-aced-47bd-b538-25ccfa8dad1an%40googlegroups.com.


--
Wayne Black
Owner
Black Box Embedded, LLC

Jarosław Karwik

unread,
Jul 19, 2024, 4:20:10 PM7/19/24
to OpenPnP
Hi,

@Wayne, these pictures are behind protected access :-(
And you are right - peelers need to be configured as actuators ( with scripting)
Do you remember what current do you use on X/Y ?  There are two different motor sizes, unfortunately no type designations.

About 24V setup:
- I will just start with standard Duet. Original Chinese setup could not place reliably with 100%  speed ( vibrations) , I was using like 50%.
- If the 24V is really limitation - I will add two Duet 1HCL modules with 48V supply ( as this is needed only on X/Y). This is very easy, although gets a bit expensive.

I will test

Wayne Black

unread,
Jul 19, 2024, 4:30:56 PM7/19/24
to ope...@googlegroups.com
sorry, heres the pics. Probably not that helpful, but I know people want pics or dont think it happened. Anyrate, machine works fine but is currently been torn apart for the new head.

inside.jpg
electronics form factor.jpg

Jarosław Karwik

unread,
Jul 19, 2024, 4:41:20 PM7/19/24
to OpenPnP
Wow, you made a beast !!! And likely  very noisy with these ejectors....

Wayne Black

unread,
Jul 19, 2024, 6:01:12 PM7/19/24
to ope...@googlegroups.com
On the contrary it is much quieter than the original pump set up. Not to mention much faster pick ups and releases. 

Jarosław Karwik

unread,
Jul 20, 2024, 1:47:15 AM7/20/24
to OpenPnP
Small investigation:
- It uses SX7236M  old and simple stepper driver    https://www.semicon.sanken-ele.co.jp/sk_content/sx7232m_sx7236m_ds_jp.pdf
- I have checked reference voltages and sense resistors and got following peak current ( Itrip):
   X/Y/Peeler  - 1.4A
   Z - 0.46A
   U/V - 0.39A

The X/P motors are NEMA23 ( 56x56x77mm), Y is a bit smaller NEMA23 ( 56x56x52)



vespaman

unread,
Jul 20, 2024, 2:54:22 AM7/20/24
to OpenPnP
Thanks for sharing the current limit setup! 
This is useful for any driver upgrades (I have been thinking about upgrading for about a year, but still haven't - machine is working good, but I think a more modern driver would yield some extra speed, esp if going for higher voltages, like Wayne has. Also reduction of heat..

- Micael

dc42

unread,
Jul 22, 2024, 4:26:07 AM7/22/24
to OpenPnP
Is there a voltage adjustment pot on your 36V PSU? You might be able to turn it down to near 32V. If it's a genuine Duet then the 32V limit on the older Duet 6HC boards is quite conservative. The ICs and mosfets on that rail are all rated to at least 40V absolute maximum (the 5V regulator is 36V max recommended). The only other components on that rail that might not like a higher voltage are the electrolytic capacitors, which in principle could be rated at either 35V or 50V; however we've been using 50V capacitors for a long while.

Later Duet 6HC boards are rated to 50V VIN.

Jarosław Karwik

unread,
Jul 22, 2024, 4:33:40 AM7/22/24
to OpenPnP
Well, I do have such pot and genuine Duet.

But   as far as I read the specs and connection schematics I cannot supply stepper motors with 32V and outputs with 24V . And all peripherals are 24V originally.

PS. Any way to check if Duet is genuine ? ( I bought it locally. it all looks as genuine, but you never know...)

dc42

unread,
Jul 22, 2024, 9:01:11 AM7/22/24
to OpenPnP
if you post a photo of the top side of the Duet then I should be able to tell you if it's genuine. Our resellers are listed at https://www.duet3d.com/page/genuine-resellers.

Jarosław Karwik

unread,
Jul 22, 2024, 9:45:47 AM7/22/24
to OpenPnP
duet.jpg

Wayne Black

unread,
Jul 22, 2024, 11:18:03 AM7/22/24
to ope...@googlegroups.com
Just my opinion; I understand you already have the 6HC. I have several 6HCs and Mini 5+ in different machines. They both work great. If I was to do the CHMT conversion again I would go w a Mini 5 ethernet using the x2 on board motor breakouts to 48v closed loop drivers. The cost of x2 1xds needed to break out signals for the 6HC is nearly the cost of the Mini5. The Mini 5 arrangement would be so much more physically compact for a 48V CL on the XY. You want that 48V+ on the XY.

Just my opinion, good luck

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.

Jarosław Karwik

unread,
Jul 22, 2024, 3:13:54 PM7/22/24
to OpenPnP
Well,
I bought 6HC locally and strangely mini5 was only 25EUR cheaper. So I took the risk of using just 24V with 6HC.
If this is not enough - I will have to add 2x XD or 2x HCL . There is a lot of space in CHMT-36VA

BTW - What kind of speeds/acceleration do you have - when using 36V drivers ? I will be able soon to test drive it with 24V

Wayne Black

unread,
Jul 22, 2024, 4:12:42 PM7/22/24
to ope...@googlegroups.com
BTW - What kind of speeds/acceleration do you have - when using 36V drivers ? I will be able soon to test drive it with 24V

Objectively, not really sure. Subjectively, the XY rapid speeds are not much different but the sound is way quieter and the 600lb Arcflat table the machine sits on doesn't swing like a pendulum like it did with previous drivers. That is to say motion speed relatively the same, motion quality much better, again subjectively.

vespaman

unread,
Jul 23, 2024, 4:49:45 AM7/23/24
to OpenPnP
Hi Guys

BTW - What kind of speeds/acceleration do you have - when using 36V drivers ? I will be able soon to test drive it with 24V

Objectively, not really sure. Subjectively, the XY rapid speeds are not much different but the sound is way quieter and the 600lb Arcflat table the machine sits on doesn't swing like a pendulum like it did with previous drivers. That is to say motion speed relatively the same, motion quality much better, again subjectively.


Wayne, aren't you running more than 36V (48?)?
For reference;
In my "standard" machine, settings are X; 2100.000/s acceleration 17000.000, Jerk 10000000.000
Y; 1500/s, accelleration 5500.000, Jerk 0.000.
I have settled for these settings, I used to run at slightly more, but then I lost steps at one time, and had to slow it down a bit.
This is a 48VB, so maybe differs slightly with the 36, since there's more mass on Y, but linear rail for X.
(I always am running my machine at 100% speed).

The way to test if loosing steps, is to repeatedly press 100mm moves in the same direction, then back. I always see issues on deceleration first.

 - Micael

Jarosław Karwik

unread,
Jul 23, 2024, 9:38:37 AM7/23/24
to OpenPnP
Work in progress - one more day needed:
Zrzut ekranu z 2024-07-23 15-35-13.png

I have also printed from covers with fan & cable connectors:
Zrzut ekranu z 2024-07-23 15-35-34.png

Wayne Black

unread,
Jul 25, 2024, 10:47:29 AM7/25/24
to ope...@googlegroups.com
Hi Micael, sir speedy ;)

Yes, running 48V for XY and 24V everything else.

I will add that the Duet CAN expansion is a little quirky. You, or at least I, need to ensure that at power on the USB connecting Duet is unconnected and not connected  during mains power up. The reason is the CL XY drivers CW/CCW setting seems to reset out of sync with the Duet config direction if USB power keeps the Duet alive and mains power is removed from the CL drivers. This only happens on the X axis and I believe if I rewired the connector it would behave correctly regardless of the power on/usb connection sequence.




--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.

Jarosław Karwik

unread,
Jul 26, 2024, 2:48:58 AM7/26/24
to OpenPnP
Now all wired and briefly tested. Big credits to Wayne for his schematics :-).  It matched 99% ( led strip had reversed polarity - minor thing, just you have to be aware that such thing can happen if you do it on your machine)

Zrzut ekranu z 2024-07-26 08-39-02.png

Had to change cable arrangements - to avoid keeping 3m of USB cable inside ( I used to have keystone cubes)
Now inspired by favourite Icotek  ( https://www.icotek.com/ )

Zrzut ekranu z 2024-07-26 08-39-15.png

Also small modification for better and safer Z homing ( previously there was a notch - software had to look for it).
Now it is absolute mode

Zrzut ekranu z 2024-07-26 08-39-30.png

And final cabling report
Zrzut ekranu z 2024-07-26 08-47-41.png

Wayne Black

unread,
Jul 26, 2024, 10:24:20 AM7/26/24
to ope...@googlegroups.com
Cool, did you find the extra 2 conductors in the main cable to the head? I dont recall the total # of conductors in the cable to the head, but mine had 2 unused addl ones. Pretty handy if you want to add more lighting to the head and ditch the gantry LEDs.

Jarosław Karwik

unread,
Jul 26, 2024, 10:37:20 AM7/26/24
to OpenPnP
Yeah,

There are two unused cables ( yellow and white/blue) on the machine side.
It would require quite digging to get to them from head  side  - like cutting shrink tube at least
Zrzut ekranu z 2024-07-26 16-34-44.png

Wayne Black

unread,
Jul 26, 2024, 10:53:15 AM7/26/24
to ope...@googlegroups.com
Ah yes, I forget. I replaced/ing my head and replaced the head cable with a new DB9 cable. 

Another thing that come to mind is that when I replaced the controller I rerouted the Y limit so I could open the top plate fully. After that I had a ton of false limit triggers. I replaced the original non shielded wire with shielded and the problem is solved. Fyi the routing of that signal wire is super weird, it runs under the cable chain and then thru a hole in Y bearing stand, really funky. The shielded wire is considerably thicker so I couldn't use the original route and zip tied it to the cable chain bracket. No more false limits and the cable enters the machine with the rest of the cabling.

Jarosław Karwik

unread,
Jul 29, 2024, 3:11:40 AM7/29/24
to OpenPnP
Yeah, this one is unusual.
But my cable looks like shielded - even if the shield does not seem to be connected anywhere.
Zrzut ekranu z 2024-07-29 09-10-24.png

dc42

unread,
Jul 29, 2024, 4:10:43 AM7/29/24
to OpenPnP
@jaroslav that 6HC board looks genuine to me.

Endstop cables for Duet boards don't normally need to be shielded if they are normally-closed switches. OTOH if normally-open endstop switches are used and the endstop cables run close to stepper motor wires, they need to be shielded.

Jarosław Karwik

unread,
Jul 29, 2024, 4:20:41 AM7/29/24
to OpenPnP
Unfortunately CHMT uses NO mechanical switches - and with only relatively weak pull-up in Duet inputs they are susceptible to noise.
Message has been deleted

Jarosław Karwik

unread,
Aug 3, 2024, 2:37:14 PM8/3/24
to OpenPnP
This is how the switch looks like ( there one more like that under the head):
Zrzut ekranu z 2024-08-03 20-13-16.png

Can be rewired - or just flipped as it seems to be NO/NC switch.
But it is not really needed - I did not have false alarms during testing.

Now the initial speed tuning:
Zrzut ekranu z 2024-08-03 20-14-58.png

Basically - I had to go to approx. 2/3 of Wayne settings.
The bottleneck seems to be Y axis acceleration. Will retest it again at some point with vision fiducials

I was able to go through I&S - this is quite lengthy process. I will have to repeat it again once I had some issues on the way.
Here some comments about some les obvious steps and issues;
-  Is there a way to configure in Duet axis without limits ? This is needed for peeler axis. I have configured very wide range. Some other options might work - like using relative movement only for this axis
- I have initially configured the camshaft head incorectly . You need to put your real cam radius ( I&S will not ask about it). Also it was not obvieus that Z has to be configured for setps/deg units.
- I took Wayne's X/Y limits - but my machine is smaller . This caused crash during I&S backlash calibration ( when it runs to soft limits in iys procedure)
- I configured myself few things ( when going second time over freash configuration with  I&S) - but then some things are not checkled and e.g. regex for the driver were missing when I configured the driver manually nefore launchig I&S
- It is unclear for me if vision fiducials should be temporary fixture or something more permanent. I finally made semi permanent attacheemnt:

Zrzut ekranu z 2024-08-03 20-30-28.png

Now the machine is ready for final configuration session:

Zrzut ekranu z 2024-08-03 20-32-17.png

vespaman

unread,
Aug 4, 2024, 4:26:08 AM8/4/24
to OpenPnP
So, in relation to a standard machine, such as mine, how much faster/slower did you accomplish on X/Y? I.e. What where your preliminary results?
(And what settings do you use, Wayne?)

I'm asking, since it may tip me over to actually do the driver upgrade.. :-)


  - Micael

Wayne Black

unread,
Aug 4, 2024, 2:19:58 PM8/4/24
to ope...@googlegroups.com
Hi Micael!

Ive pushed that machine aside for the time being, so I'm not really sure on the settings. I know Ive tweaked the settings quite a bit over time. 

As I understand your main concern is speed, correct? I'm not sure how much you'd gain going to a different controller in regards to increasing kinematic speed. I can say for increasing overall process speed ditching the pumps in favor of ejectors will make a big difference. Or at least it did in my machine. With ejectors pickups and releases are seemingly instantaneous. The old +/- pump scheme suffered either extended duty cycle (firing pumps prior to Z travel) or extended dwell time to ensure pickups and releases. 

Of course ejectors require a compressor w adequate capacity. I use a 27Gal 'quiet' compressor in my garage and its perfect. I havnt calculated compressor cycle time, but I just do prototypes and send production fab out of house.

vespaman

unread,
Aug 4, 2024, 2:47:15 PM8/4/24
to OpenPnP
Hi Wayne,
Correct, speed is my no1 prio. But I am not thinking about changing controller, only (maybe) stepper drivers, to the one you are using (T60 IIRC), and increase Voltage on X/Y. Y is acceleration limited with the standard driver, as you probably know, so I might also want to change the motor if it is still acceleration limited with the T60. So this is why I was wondering what gain I could foresee. 
If @Jaroslaw has achieved 70% of your speed settings, I presume it is about on par with a standard machine. Or maybe I am misunderstanding... ?
Also, if I understand correctly, the T60 uses the closed loop feedback as primary reference, so micro steps should be slightly more precise (not that I need this currently).

I am running the vac pump 'always on', so I don't foresee any speed improvement there, however, I will change into ejectors + buffer reservoir as soon as I have moved my machine into a new location, since the sound of the pump makes me tired - I run my chmt a couple of full days in a row, when I produce my stuff. Currently I think my "breaking point" is about 1k (same) boards a year, if more, I'll probably outsource.

 - Micael

Jarosław Karwik

unread,
Aug 5, 2024, 3:31:13 AM8/5/24
to OpenPnP
Hi,

I was able to do more testing ( mainly going over I&S further and further)
I have found out that backlash compensation test is actually good way to check if machine speed settings are ok - as it goes over several scenarios with multiple moves and speeds.

For reliable operation I set up following parameters:
- X -  1000mm/s 10000mm/s^2
- Y -  800mm/s    3000mm/s^2

It is unfortunately much  lower then my original settings. But this works in very reliable way and backlash calibration reports 0.03/0.037 ( X/Y)

It does not look that bad - when jogging with full speed. If you wanted production speed - might be not enough,  but for small series and prototypes- should be OK.

dc42

unread,
Aug 5, 2024, 5:46:28 AM8/5/24
to OpenPnP
@jaroslaw if you know the data for your stepper motors you can calculate the approximate maximum speed before the motors start to lose torque. See https://www.reprapfirmware.org/emf.html.

Jarosław Karwik

unread,
Aug 5, 2024, 5:52:36 AM8/5/24
to OpenPnP
Unfortunately I do not know the type - they are "NO NAME" (all of the motors on the machine).
It would be guessing. But then I can simply make few experimental settings and launch backlash test to see where the breaking point is.

dc42

unread,
Aug 5, 2024, 6:02:16 AM8/5/24
to OpenPnP
Do you know what voltage the motor drivers used originally?

Jarosław Karwik

unread,
Aug 5, 2024, 6:06:45 AM8/5/24
to OpenPnP
Yes,

It was 36V for X,Y,W ( peeler) and 24V for all others ( Z,U(C1),V(C2)). I was able to check operating current ( ~ 1.4A for X/Y/W ) by measuring reference voltage on original drivers.

dc42

unread,
Aug 5, 2024, 10:14:26 AM8/5/24
to OpenPnP
It probably needed 36V on X and y to reach the speed it did. Are you getting about 2/3 of that speed?

Jarosław Karwik

unread,
Aug 5, 2024, 10:23:18 AM8/5/24
to OpenPnP
As I wrote before the top speed on original Chinese controller and software was not really usable - machine was shaking too much for my taste,  I usually set it up to about half the scale.
It is hard to tell what was the original speed - I do not  remember precisely, but It ws not shown in  any reasonable unit, more like percent of full speed. So the 2/3 is kind of very imprecise visual impression.

Currently, If I go to about 150% of my safe speed ( with some step loss) it starts shaking again - so I will tune it a bit, but I will not go any further.  I prefer slower, but more reliable operation as the machine is not target for production.

vespaman

unread,
Aug 5, 2024, 10:48:43 AM8/5/24
to OpenPnP
I am running stock (36V) hardware reliably with about double your speed & acceleration settings. This is about as fast as the stock chmt is running, but not as brutal since the managed acceleration from OpenPnP.
Can you not reuse your original 36v stepper driver, and connect it to the Duet (or is there no step/dir outputs on your Duet)?

IMHO Reliable is even more important for production - here you do not want any issues what so ever!

But then again, if you are satisfied with your current setup, that's good enough :-)


 - Micael

Jarosław Karwik

unread,
Aug 5, 2024, 12:16:07 PM8/5/24
to OpenPnP
Well,

There is still space for tuning - but unlikely I get over thees 2/3 of your speed.
My Duet 6HC does not have step/dir outputs - there is separate model just for that.  It is a pity - as there are enough free outputs for two axes on generic output pins, just that this is not supported.

Knowing all that I should have bought Duet 3 Mini 5+ - as it has two step/dir external connections just perfect for X/Y. Well, I can still buy FDCAN expansion card for step/dir or complete 48V controller.
I will test current setup a bit more before final conclusion.

Wayne Black

unread,
Aug 5, 2024, 1:17:38 PM8/5/24
to ope...@googlegroups.com
Hi Jarosław 
If you dont mind me asking, where are you located and what is the purpose of the machine? Hobby, education, business?

Jarosław Karwik

unread,
Aug 6, 2024, 2:17:08 AM8/6/24
to OpenPnP
Hi,

I am located in Gdańsk, Poland.

The machine purpose was a combination of business and hobby.
I do design and manufacture professional CNC controllers ( for these machines : https://www.felix-gluer.com/ ) as well as some control electronics for electrical boats.
Most of these are low volume ( like 30..100 per year), so I manufacture them myself. If it goes higher - I have contact to professional assembly company.

This machine assembled few batches of my CNC controllers ( with original Chinese software),. Then I used it as development platform for my 7 segment S-Curved  controller. It almost works now , just that I do not have much time for it.
I decided to modify the machine for feeders ( SIplace and Dima), but found out that it would be significant work. At this time I also noticed that 3D printed kits offer much more flexibility and nice cartridge-like cheap feeders, so I went for Pandaplacer ( but Bing machine was close ). I have unassembled kit waiting for its time.
As I sold  the original CHMT controller I have choosen Duet as replacements to make the machine operational to be sold - and  once ready, it will go to Denmark.

I decided to document here the conversion process - as it is a bit more demanding then just flashing Smoothie based firmware to original controller ( which is the most common case).

Wayne Black

unread,
Aug 6, 2024, 10:47:42 AM8/6/24
to ope...@googlegroups.com
Very cool.
I thought I recalled you we were going to make a custom controller for openpnp. I have similar issue that I only play w the chmt machine when I get a new design and I pull out the pnp and work on if for a while. 

Message has been deleted

Jarosław Karwik

unread,
Aug 9, 2024, 2:46:39 AM8/9/24
to OpenPnP
I got feeders working :-) - unfortunately for the price of the drag pin :-(

Zrzut ekranu z 2024-08-09 08-32-44.png

Without that the pin is activated immediately - even if it does not reach first feed location.
It was kind of surprise for me that the pin operation can be merged with move which should be executed before pin operation

Now I think there is only one thing left - poor light for the camera
On the left - base setup - looks bad especially with black test board.
Right - a bit of external light from my mobile phone.

Zrzut ekranu z 2024-08-09 08-26-14.pngZrzut ekranu z 2024-08-09 08-26-51.png

I will have to add small led ring to the camera - sonething like Jan did:

Wayne Black

unread,
Aug 9, 2024, 11:05:44 AM8/9/24
to ope...@googlegroups.com
Ha, well done :) Ive killed more than one drag pin. The cage lets go before any major damage to the head fortunately. 


mark maker

unread,
Aug 10, 2024, 1:45:21 PM8/10/24
to ope...@googlegroups.com

Are you aware of this?

https://github.com/openpnp/openpnp/wiki/Axis-Interlock-Actuator

You can create safety rules, so the machine cannot move in X/Y when the drag pin is down.

_Mark

Jarosław Karwik

unread,
Aug 11, 2024, 4:34:58 AM8/11/24
to OpenPnP
Hi,

I am aware of interlock - but I have not noticed before the range interlock for drag pin.
While really good one for testing - I am not sure I would configure one by one 29 drag feeders with interlock. Especially that it might have load the system much.
This would have to be more automatic - feeder implementation would have to provide these drag zones.

Hopefully now I understand how it works and no more broken pins..... ( well, I have ordered several spares)

Small testing with camera light:

Zrzut ekranu z 2024-08-11 10-24-43.pngZrzut ekranu z 2024-08-11 10-24-56.png

Works, but no ideal - you can still see led reflections - will have to try stronger diffusor ( LED is on 50% PWM)
I also need to adjust camera a bit - had to move it up and now I need to adjust lens again
Pictures with and without light

Zrzut ekranu z 2024-08-11 10-29-07.pngZrzut ekranu z 2024-08-11 10-32-43.png

How have you solved the camera light problem ? I will have to ask Jan about his solution - is there anyone who could share his proven solution ?

mark maker

unread,
Aug 11, 2024, 5:09:36 AM8/11/24
to ope...@googlegroups.com

While I wrote the code for the interlock, I have never personally configured such a drag feeder. But other users seemed happy. I don't see how you would configure one for each feeder. I assume dragging happens with slow speed, so one rule with a simple speed limit is sufficient. Assuming you normally have the speed set at a higher setting, it will prevent accidental jogging with the drag pin down.

_Mark

vespaman

unread,
Aug 11, 2024, 5:44:31 AM8/11/24
to OpenPnP

How have you solved the camera light problem ? I will have to ask Jan about his solution - is there anyone who could share his proven solution ?



I used the board Jan did, it works OK, and is inside the tube, the two wires are pulled through the small hole on the camera tube. I then 3d printed a diffuser, that can be found thingiverse (search for "charmhigh"). The diffuser is supposed to be press-fit into the bottom of the tube, but it may need tweaking, depending on your printer setttings and filament.

 - Micael

Jarosław Karwik

unread,
Aug 11, 2024, 6:01:11 AM8/11/24
to OpenPnP
You are right - thanks. Might be even possible to set "special" drag pin speed and reject anything else. Will experiment with that as this would be really foolproof method.

Wayne Black

unread,
Aug 11, 2024, 12:15:32 PM8/11/24
to OpenPnP
Looks like to you need to lower the lights to the just below the camera lens. The lower you can get the LEDs wo being within the fov of the camera the better. Lots of LEDs and lots of diffusion. Unfortunately, the oem camera placement will make getting the light lower for the clearance a problem and not having light 360 around the fov is not ideal. 

Jarosław Karwik

unread,
Aug 11, 2024, 12:23:35 PM8/11/24
to ope...@googlegroups.com

I know. That's why I decided to go for proven Jan's solution

Jarosław Karwik

unread,
Aug 12, 2024, 3:49:54 AM8/12/24
to OpenPnP
PCB ordered:
Zrzut ekranu z 2024-08-12 09-42-56.png

I had some side discussion with Mike - and this is worth sharing.
The ultimate solution would be coaxial lighting ( which is superior) .
Zrzut ekranu z 2024-08-12 09-46-24.png

I even have nice module on my shelf, but this would require a lot of extra work ,
as some parts would have to be metal. This would also elevate camera few centimeters up ( due to mechanical constrains).

Zrzut ekranu z 2024-08-12 09-42-09.png

Mike Menci

unread,
Aug 12, 2024, 5:01:47 AM8/12/24
to OpenPnP
Here i9s how I have it - on one side of the head - M12 lens from ELP very near Coax box-top glass:  
I had to go 62mm (bottom coax box ) to PCB distance due to DIMA Feeders - to avoid collision with feeders structure .... 
PnP_Head_4xNema8_2xNema14 _coaxLight.png
Mike

Mike Menci

unread,
Aug 12, 2024, 5:09:54 AM8/12/24
to OpenPnP
And this is how matchKeyence_ELP camera.SLDASM .png 
space I need   - ----
What is the size of your coax light-Box ?


Mike Menci

unread,
Aug 12, 2024, 5:20:24 AM8/12/24
to OpenPnP
And enclosed are images which I get from vision 
Desktop.rar

Jarosław Karwik

unread,
Aug 12, 2024, 7:47:21 AM8/12/24
to OpenPnP
I have 65x36x40mm coaxial box  ( COP-30W  module)
When I put it under my camera ... well... it looks nice - clean and sharp
I will compare it later on to the Jan's solution to see  the difference

Zrzut ekranu z 2024-08-12 13-40-47.png

Jarosław Karwik

unread,
Aug 17, 2024, 5:11:18 AM8/17/24
to OpenPnP
The result from Jan's ring and diffuser from vespaman (https://www.thingiverse.com/thing:6667386/comments)
The LEDs still can be seen ( here 33% PWM) , but the metal pads are sharp with good contrast.

Zrzut ekranu z 2024-08-17 10-52-59.png

I have tried different diffusers ( transparent/white/grey plastic) - best results with grey ( actually it is not pure grey - Vertigo)
They need 0.15mm layers to print properly

Zrzut ekranu z 2024-08-17 11-00-59.png

The  ring as such:
Zrzut ekranu z 2024-08-17 11-01-12.png

And final look:
Zrzut ekranu z 2024-08-17 11-00-40.png


Now recalibration ( as the camera was touched and slightly moved) and final testing

Mike Menci

unread,
Aug 17, 2024, 6:12:56 AM8/17/24
to OpenPnP
I would never print a diffusor for camera Led ring  - you can find appropriate plastic in Celling Led Lights from China (remove glass and use plastic sheets only)  >>>>: 
unnamed.png 
plastic sheet from led lights.jpg

Wayne Black

unread,
Aug 17, 2024, 11:21:55 AM8/17/24
to ope...@googlegroups.com
I found a square plastic iso alcohol bottle that made for a perfect diffuser. In finding that I tried lots of materials laying around.

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.

vespaman

unread,
Aug 17, 2024, 3:11:09 PM8/17/24
to OpenPnP
Yes, your diffuser  is similar to mine, I cannot actually make out the LED's on mine, but the position of the LED's will make some kind of reflection back. I have 30% and my white PETG has a lot of pigment, so that might explain the difference. You might want to try a thicker diffuser, if you are not satisfied with your current result.

Also, how far up you push the diffuser will make a difference; the lower, then better, but you will start to see corners fuzzy if you move too far away from the lens.

However, the center of the picture with the cross hairs, is the most important part (for fiducials), so I doubt you will have any issues even if you can make out the LED's. I bet the light is good enough on feeders. Maybe, if you want to use OCR, the visible LED's might be an issue.

 - Micael

Jarosław Karwik

unread,
Aug 19, 2024, 4:36:24 AM8/19/24
to OpenPnP
Well, I have tried Mike suggestion:
Zrzut ekranu z 2024-08-19 10-28-07.png

It is a bit better then simple 3D print, but the LED stil lingers there.

It just works differently - the original lamp does have leds in similiar way as monitors backlight .
Meaning they are directed toward the edge of the plastic - and there are 3 layers - white paper (kind of mirror), transparent plastic and white semi-transparent plastic.
This is actually good way to make such camera light - just that this particular machine does not have enough space for such design.

Anyway - even current design with my 3D print is enough for good operation - even if there are small led reflections ( the contrast is sufficient).

Oscar Illerup

unread,
Sep 25, 2024, 11:15:49 PM9/25/24
to OpenPnP

He 

e2cad892-1deb-4f63-a616-c40fa8b8dcf2.jpg
bc65deca-d9d9-4715-b6bb-74adebce5e03.jpg
I am now the lucky owner of the CHMT-36VA. After receiving the machine, I tested and optimized it, and with the initial setup as shipped, I was getting around 900-1000 CPH. I have now installed 2 1HCL boards and have them up and running in open-loop mode (so for now, the same speed).

I'm struggling a bit with the setup of the encoders. I have used an oscilloscope to identify the A+, B+, GND, and +5V pins, and have connected them to the encoder input of the 1HCL board, but I haven't had any luck getting a good signal from them. Does anyone here have experience getting the encoders up and running?

Best Oscar Illerup

Oscar Illerup

unread,
Sep 25, 2024, 11:15:51 PM9/25/24
to ope...@googlegroups.com
He 

e2cad892-1deb-4f63-a616-c40fa8b8dcf2.jpg
bc65deca-d9d9-4715-b6bb-74adebce5e03.jpg
I am now the lucky owner of the CHMT-36VA. After receiving the machine, I tested and optimized it, and with the initial setup as shipped, I was getting around 900-1000 CPH. I have now installed 2 1HCL boards and have them up and running in open-loop mode (so for now, the same speed).

I'm struggling a bit with the setup of the encoders. I have used an oscilloscope to identify the A+, B+, GND, and +5V pins, and have connected them to the encoder input of the 1HCL board, but I haven't had any luck getting a good signal from them. Does anyone here have experience getting the encoders up and running?
Best Oscar Illerup


--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.

dc42

unread,
Sep 26, 2024, 7:19:24 AM9/26/24
to OpenPnP
Oscar, can you post a link to the datasheet of your stepper motors?

Jarosław Karwik

unread,
Sep 26, 2024, 11:24:26 AM9/26/24
to OpenPnP
I got them running, but on my hardware board. They are differential - may not work otherwise ( although some do)


Zrzut ekranu z 2024-09-26 17-22-06.png

Zrzut ekranu z 2024-09-26 17-23-22.png
Reply all
Reply to author
Forward
0 new messages