BTT Manta8P + Rpi CM4?

[justin White]

I'm about to start building a PnP. I've been searching for info on hardware setups and whatnot. Looks like OpenPnP can use Marlin. I'm not well versed in Marlin/3D printer firmware but I'm going to build a 3d printer soon as well and I like the all-in-one form facter of this setup for small machines. Is there any reason why this would not work with OpenPnP?

I've seen some discussion about using an Rpi with OpenPnP and best I can gather people were worried about RAM. The 2 CM4s that I've had laying around for a couple years are 2gb RAM models. Does anyone have actual experience as to the RAM requirements for an OpenPNP system and how I might expect it to run on an Rpi?

[Fernando Corona]

I had the same desire with my first PnP project of an all in one screen and PC combo like a tablet of some sorts.

Currently looking at trying out a SBC setup like the lattepanda or a gaming handheld PC like the GPD WIN4. Still researching at the moment. If this is interesting to the group I can post videos about testing it out. Currently designing a small form factor PnP I can try it out on.

For the controller I use Duet3d boards. I'm good at programming but much rather work on the hardware side of things. Duet boards are plug & play for me and can get me running in under an hour or so.  I've already used a duet2wifi and duet3 6xd both running latest RepRapFirmware. 

On Thu, Mar 16, 2023, 11:16 AM justin White <blaz...@gmail.com> wrote:

I'm about to start building a PnP. I've been searching for info on hardware setups and whatnot. Looks like OpenPnP can use Marlin. I'm not well versed in Marlin/3D printer firmware but I'm going to build a 3d printer soon as well and I like the all-in-one form facter of this setup for small machines. Is there any reason why this would not work with OpenPnP?

I've seen some discussion about using an Rpi with OpenPnP and best I can gather people were worried about RAM. The 2 CM4s that I've had laying around for a couple years are 2gb RAM models. Does anyone have actual experience as to the RAM requirements for an OpenPNP system and how I might expect it to run on an Rpi?

--
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/8a1b0a52-454e-491b-824a-374f49ef9781n%40googlegroups.com.

[justin White]

After looking at it a bit The CM4 on the Manta probably isn't quite going to cut it due to the Manta only exposing the SOC's native USB port. The Rpi4 SOC has a single PCIe lane which is how the Rpi4 gets it's USB3 port. Still Curious if the Rpi4 has enough guts otherwise to run OpenPnP with 2 cameras.

I have some custom machine control projects that also deal with machine vision, for those I currently use an Odroid H3. Since I always have a couple on hand I will likely use one of those at least initially, I suggest you look into them, they should be more than capable for OpenPnP. RAM is SODIMM and there are SATA, NVME, and EMMC storage options.

If something like an Rpi doesn't have enough grunt I will stick with the H3.

[Fernando Corona]

Thank you, I will do so. 

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/a17d1af8-5b77-4ca9-b1eb-df5b827be99cn%40googlegroups.com.

[mark maker]

Hi Justin,

Marlin is fine with reasonably new firmware version and proper configuration with A B C axes, and not extruders.

https://github.com/openpnp/openpnp/wiki/Motion-Controller-Firmwares#marlin-20

Don't know much about Raspi. While OpenPnP itself is Java and thus portable, there are native libraries at least for OpenCV and the OpenPnpCaptureCamera plus the all-in-one-installer that need to be compiled for the platform. I believe some users did it, but AFAIK you can't expect to simply install it from download.

https://github.com/openpnp/openpnp/issues/1437

https://github.com/openpnp/openpnp-capture/pull/47

Maybe that is changing now that Jason is modernizing the platform support.

https://github.com/openpnp/openpnp/pull/1534

Recently, some computer vision routines crunching on the pixel level were coded in Java, notably the self-tuning DetectCircularSymmetry stage that is now used in many stock vision pipelines. While it is surprising how fast the Java just-in-time compiler can be, it is still not the same as native, potentially vector optimized or even GPU run code. And while the Raspi is amazing given its price, it still can't compete with PCs, even (recent) Mini PCs. Plus you already mentioned RAM, which is important for Java to run the Garbage Collector at ease. OpenPnP is running through a lot of RAM, throwing around uncompressed HD or higher images from two camera streams, and copying them many times with absolutely no concern for reusing buffers. So if you care about speed, a Raspi might not be the right option. But then again, speed is not everything. 😉

_Mark

On 15.03.2023 16:44, justin White wrote:

I'm about to start building a PnP. I've been searching for info on hardware setups and whatnot. Looks like OpenPnP can use Marlin. I'm not well versed in Marlin/3D printer firmware but I'm going to build a 3d printer soon as well and I like the all-in-one form facter of this setup for small machines. Is there any reason why this would not work with OpenPnP?

I've seen some discussion about using an Rpi with OpenPnP and best I can gather people were worried about RAM. The 2 CM4s that I've had laying around for a couple years are 2gb RAM models. Does anyone have actual experience as to the RAM requirements for an OpenPNP system and how I might expect it to run on an Rpi?

[justin White]

Probably more trouble than it's worth ATM considering I have several of these I can use

https://www.hardkernel.com/shop/odroid-h3-plus/

As I mentioned I use those in a LinuxCNC machine vision project I do that uses much higher resolution cameras than I'd bother with on a PnP. I do use openCV but no special image processing, and take the frames raw. Uses barely any RAM which is why I thought maybe a Pi could handle this but I suppose it's not apples to apples and I don't need any extra headaches trying to get it running.

You received this message because you are subscribed to a topic in the Google Groups "OpenPnP" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/openpnp/69b1ba7zPK8/unsubscribe.
To unsubscribe from this group and all its topics, send an email to openpnp+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/4576fc5e-2fb8-912f-18a7-8cd6bb1a539b%40makr.zone.

[Fernando Corona]

After looking into the Lattepanda 3 delta I think I'm going to buy one to try OpenPnP using their touch screen as well. Also want to try making a pocket cad station with it for lite cad work on the go. 

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CA%2BQ02MPmZf0YAAK%3DB%2Bho0j0frqxiOSjU7w7Q_avahF2XtAfhmQ%40mail.gmail.com.

[Jarosław Karwik]

We have started buying these ones ( as replacement for Jetson TX2 as Nvidia stopped supporting it - we use two units as boat computers in my line of work) 

https://www.eglobaltech.cn/ViewDetails392.aspx  ( Mini PC intel Celeron J6412 )

I was nicely surprised that Celeron ( with Ubuntu 22.04) works that fast. And the price is good - barbone ( no RAM nor SSD) is 145$ including shipment from China. Usually take 8 days to get it. 

There is also a bit more powerful version with J6413

I plan to put one of these on my CHM36VA ( needs to be outside - it is fanless which should not be enclosed)