Unofficial Prusa Mini Wait Thread  

Page 91 / 147
  RSS
burtronix
(@burtronix)
Trusted Member
Posted by: @derek

@burtronix

No, it shouldn't matter.  Gcode is gcode and it's only going as fast as the printer is, which is only a handful of lines per second.  1Ghz should be plenty, assuming your resources aren't bogged down with other tasks.  The size of the gcode also doesn't matter, provided it's not bigger than the remaining RAM(512MB-OS usage with the Zero)

The Pi zero is 1Ghz single core and the Pi4 is 1.5Ghz quad core.  On top of other features, the Pi 4 is WAAAAY superior to the Zero in every way.  More expensive, yes, but the better value is in the Pi4 considering the Zero hasn't seen a CPU update since 2015.

Where you'll see the main difference in performance(between Zero and other Rpi versions is with uploading files into Octoprint and giving the printer commands.  The faster the CPU, the better.  On paper, the Rpi4 is 6x faster than the Pi zero but it's actually a lot more in reality due to CPU architecture and the fact your OS is tieing up a ton of resources on the Rpi zero.

I'm not an expert with these things but hope this helps.

Maybe I'm misunderstanding how gcode works. For instance, if you're printing a hollow cube with just 1 inner & 1 outer perimeter, a single layer would just need just 8 (2x4) move instructions. But if the shape were hexagonal, a layer would need 16 (2x8). Is it not true that more complex shapes need more lines of gcode? Don't more lines of code take longer to both upload & process ? 

Whatever you find to do with your hands, do with all your might!...
Posted : 05/02/2020 8:38 pm
minos197
(@minos197)
Active Member

To be fair all the pi needs to do is read a file like by line and broadcast it over uart at 115200. If that was all the pi did, it wouldn't be a problem even for the zero.

What makes things risky is the Web ui and the video streaming.

The pi 4 is also not the ideal solution since it has crazy power requirements and non standard USB c implementation. Combined with a non ideal power supply, and a camera it will respond to a low voltage warning by powering off the USB power rail, killing the print.

If you are to buy a device for octoprint look at the pi 3A. It is smaller cheaper powerful enough and with less finicky power supply requirements

 

 

This post was modified 2 months ago 2 times by minos197
Posted : 05/02/2020 8:57 pm
munkisquisher
(@munkisquisher)
Eminent Member

@burtronix

yes, one line of gcode is an instruction "move this far in this direction" or "start squirting plastic" or "heat up to this temp" If you are printing a curve it's made up of lots of short line segments, the length of those segments which leads to the precision of the curve is adjustable in the slicer

What makes it harder to keep up at a higher precision is that if you are printing 1mm long segments of a curve, the printer gets to the point it needs the next instruction much faster than if it's printing a 5cm long line. So the pi needs to send instructions faster.

It's still a tiny amount of data to send compared to the theoretical limits of what usb can send or what the pi can process. It's just that with a single core pi zero, there's wifi, power management, and the whole octopi os and app that's competing for resources. It can bog it down for short periods even if it runs at 10% load most of the time. That would normally not be noticeable, but any pause in streaming the gcode makes your printer pause.

Does anyone know what kind of buffer there is in the printer for storing the gcode that's streamed to it? a decent buffer would hide any glitches in transmission.

This post was modified 2 months ago by munkisquisher
Posted : 05/02/2020 9:04 pm
burtronix liked
Derek
(@derek)
Eminent Member
Posted by: @burtronix
Posted by: @derek

@burtronix

No, it shouldn't matter.  Gcode is gcode and it's only going as fast as the printer is, which is only a handful of lines per second.  1Ghz should be plenty, assuming your resources aren't bogged down with other tasks.  The size of the gcode also doesn't matter, provided it's not bigger than the remaining RAM(512MB-OS usage with the Zero)

The Pi zero is 1Ghz single core and the Pi4 is 1.5Ghz quad core.  On top of other features, the Pi 4 is WAAAAY superior to the Zero in every way.  More expensive, yes, but the better value is in the Pi4 considering the Zero hasn't seen a CPU update since 2015.

Where you'll see the main difference in performance(between Zero and other Rpi versions is with uploading files into Octoprint and giving the printer commands.  The faster the CPU, the better.  On paper, the Rpi4 is 6x faster than the Pi zero but it's actually a lot more in reality due to CPU architecture and the fact your OS is tieing up a ton of resources on the Rpi zero.

I'm not an expert with these things but hope this helps.

Maybe I'm misunderstanding how gcode works. For instance, if you're printing a hollow cube with just 1 inner & 1 outer perimeter, a single layer would just need just 8 (2x4) move instructions. But if the shape were hexagonal, a layer would need 16 (2x8). Is it not true that more complex shapes need more lines of gcode? Don't more lines of code take longer to both upload & process ? 

Yes, the bigger the file is, the longer it takes to both upload and start the job.  Sorry, I thought you meant while the printer was printing if it would be too slow.  A better CPU will help with both uploading and starting the job but will affect starting the job more, because that entails processing the file and putting it into active memory.  After that it's on cruise control because it's only feeding commands one at a time.  So basically if you're using a Pi 3 or 4 for Octoprint over a Zero, you're paying for the convenience of not waiting several minutes for your job to start.  

Posted : 05/02/2020 9:04 pm
Derek
(@derek)
Eminent Member

@munkisquisher

Most printers have a config file in the firmware that defines the buffer size for usb.   You can edit it.  I believe it has to be a value that is a power of 2 (2, 4, 8, 16, 32, etc)

 

If your buffer is too small, the Pi Zero will definitely run into hangups and will intermittently stop.  This was the issue with the Raspberry Pi 1 many years ago and the fix was merely to increase your buffer to something higher like 32.  Same would apply with the Pi Zero.  The higher the buffer you set, the longer it takes to start the print(although it's only a few seconds in reality).  

Tbh, I've never had to mess with this on a Prusa, so I don't know how to edit it.  I haven't messed with a Pi Zero since I had a Printrbot.

This post was modified 2 months ago by Derek
Posted : 05/02/2020 9:26 pm
charles.h13
(@charles-h13)
Noble Member

@minos197

I disagree.  The PI 4 uploads the file faster and starts the print faster.  Every second is important in my busy life.  I have 4 printers on PI 3B and and 3 (soon to be 4) on PI 4.  I will convert them all to PI 4.  They are significantly faster at upload, connection to the printer, and starting the print.  I am not an expert, but it is noticeable. Connection with a PI 4 is near instant and the 3B can take 10-30 seconds.  That can add up over time.     

Posted : 05/02/2020 9:35 pm
burtronix liked
clemens.m
(@clemens-m)
Estimable Member

@ charles-h13

Is every printer owning his Raspberry Pi - on Octoprint instance running on one Raspberry Pi?

I was thinking about having 2 Octoprint instances running on one Raspberry Pi 4 (with 1GB RAM) - both having a USB-CAM for remote control. I read about in the OctoPrint forum, that this should in principle be possible.

Best regards,
Clemens Mödlin

This post was modified 2 months ago by clemens.m
Posted : 05/02/2020 9:44 pm
LH
 lh
(@lh)
Eminent Member
Posted by: @littleneutrino

@aualow

😯aww, another one after mine that has shipped 😛 it'll happen eventually lol.

I'm in the same boat as you, with orders after mine already shipping.  I keep having to remind myself that patience is a virtue, lol.

Posted : 06/02/2020 12:14 am
charles.h13
(@charles-h13)
Noble Member

@clemens-m

Each one has its own pi

Posted : 06/02/2020 12:33 am
Area51
(@area51)
Reputable Member

Nice that that Mini's are rolling out again. Eager to hear, what changes where made to the printer... As there is no official announcement from Prusa Research...

Take a look at my designs on Thingiverse or on PrusaPrinters ;)...
Posted : 06/02/2020 12:51 am
Derek
(@derek)
Eminent Member

Anyone have trouble flashing the firmware on initial boot?  Mine just resets...

Edit: Never mind.  Pushing the knob is NOT intuitive! 🤣 🤣 

This post was modified 2 months ago by Derek
Posted : 06/02/2020 2:02 am
brian_charlton
(@brian_charlton)
New Member

I know that the Prusa team are all very busy, but what would be very cool.....

Someone from Prusa to update and complete the wait thread google document with the verified delivered (received) minis.

I understand that the document is an online and community managed collaborative document. However, I believe that many people would not bother to update the received status.

Just a thought/hope/wish

This post was modified 2 months ago by brian_charlton
Posted : 06/02/2020 4:34 am
stevensmith
(@stevensmith)
Trusted Member

10/13 order - Shipping notice received today. Spreadsheet updated. Woo Hoo!

Posted : 06/02/2020 4:59 am
Texy
 texy
(@texy)
Reputable Member
Posted by: @charles-h13

I will post more on speed tonight.  I think it is actually faster.  I am going to print the same item with the same filament and confirm speed tonight.  

Hey Charles, how did you get on with the speed comparison between your mk3/s and the Mini?

Texy

Posted : 06/02/2020 9:41 am
burtronix liked
charles.h13
(@charles-h13)
Noble Member

@texy

Time for away from me.  I need to open a new spook of filament and pick a test print, I will do it tonight.  

Posted : 06/02/2020 10:23 am
burtronix liked
crawlerin
(@crawlerin)
Reputable Member
Posted by: @brian_charlton

I know that the Prusa team are all very busy, but what would be very cool.....

Someone from Prusa to update and complete the wait thread google document with the verified delivered (received) minis.

I understand that the document is an online and community managed collaborative document. However, I believe that many people would not bother to update the received status.

Just a thought/hope/wish

I don't think they can do that, it's their internal data, plus it's customer data. People share it on voluntary basis, giving consent by filling those lines. But company cannot do that for their customers. GDPR and privacy and stuff.

Posted : 06/02/2020 11:16 am
Texy
 texy
(@texy)
Reputable Member

Prusa are actually not happy about the unofficial spreadsheet as customers are complaining that there Mini didn't ship before others below them in the list. So, no, there will not be contributing to it. However at least the shipping table is really now a table and gives more info than it did previously.

Texy

Posted : 06/02/2020 11:20 am
joan.t
(@joantabb)
Veteran Member Moderator

@charles-h13

If you are opening new Spooks of  filament, this might be an ideal model

Thingiverse

https://www.thingiverse.com/thing:4421

🤣

Joan

This post was modified 2 months ago by joan.t
I try to make safe suggestions,You should understand the context and ensure you are happy that they are safe before attempting to apply my suggestions, what you do, is YOUR responsibility. Location Halifax UK...
Posted : 06/02/2020 11:27 am
burtronix liked
crawlerin
(@crawlerin)
Reputable Member
Posted by: @texy

Prusa are actually not happy about the unofficial spreadsheet as customers are complaining that there Mini didn't ship before others below them in the list. So, no, there will not be contributing to it. However at least the shipping table is really now a table and gives more info than it did previously.

Texy

They stated few times that numbers are assigned to orders, but it's not an indicator of shipment queue. Their ERP decides on that based on parameters we may not know (components, plastic parts inventory, destination, arrangement with couriers etc.). It's rather inconsiderate if people are bothering representatives based on this spreadsheet,we will never have full picture.

Posted : 06/02/2020 11:33 am
charles.h13
(@charles-h13)
Noble Member

@texy

We have to live with the wait so I guess they will have to love with the spreadsheet.  

Posted : 06/02/2020 11:50 am
Page 91 / 147
Share:

Please Login or Register