Page 2 of 3

Re: Impressions: Duet vs. Smoothie

Posted: Tue Oct 27, 2015 1:31 pm
by mhackney
The "draw" for Duet and RepRapFirmware are 2 fold:

1) the delta kinematics calculate EVERY point of movement, not line segments. This is my #1, #2 and #3 reason for trying Duet. RRF is the only firmware that does this.
2) although 626Pilot's delta calibration is great, it is hampered by the limited memory space on Smoothie. I can't use the same firmware image to calibrate AND run ethernet and my panel. Duet RRF delta calibration is native and calibrates in less than 30 seconds on a single cycle of probed points. It's so fast and good that I have added the code to the header of all my gcode files so it does a quick calibration then prints.

Setting up a dev environment on smoothie was a breeze (or should I say, smooth). RRF is a total pain in the but. I also agree with the M500 but, I struggled so much with USB mounting and lack of ethernet based file transfer that its a wash. Making a local config change and then uploading via the web server or ftp is quick. And, since I calibrate on every print run now, that minimizes the issue.

Re: Impressions: Duet vs. Smoothie

Posted: Tue Oct 27, 2015 3:52 pm
by bot
eeeeh no. Smoothie is a well-maintained code base, but it's not as amazing as they'd have you believe. It's no better than porting repetier to duet. Repetier is already ported for radds and due, which would make it easy to port to duet I think.

The main advantage at this point of duet is the delta algorithm and its ease of use. It is not exactly like other firmwares, so it takes a bit of getting used to. Once it's familiar, it's very convenient to operate.

Re: Impressions: Duet vs. Smoothie

Posted: Tue Oct 27, 2015 10:14 pm
by plasma
mhackney wrote:The "draw" for Duet and RepRapFirmware are 2 fold:

1) the delta kinematics calculate EVERY point of movement, not line segments. This is my #1, #2 and #3 reason for trying Duet. RRF is the only firmware that does this.

I would love to see a comparison in print quality between smoothieboard and a duet. If possible.

Re: Impressions: Duet vs. Smoothie

Posted: Tue Oct 27, 2015 10:55 pm
by 626Pilot
bot wrote:eeeeh no. Smoothie is a well-maintained code base, but it's not as amazing as they'd have you believe. It's no better than porting repetier to duet.
I think the delta kinematics are similar, in terms of using segmented moves. However, that's where the similarities end. Their delta calibration system uses a bunch of hard-wired logic and never produced good results for me, and many who use it never get a good calibration. My calibration system uses fuzzy logic instead of hard-coded rules, and it works very well.

There is also the code quality, which in Repetier is not very good:

Code: Select all

#if NUM_EXTRUDER>5
    ,{
        5,EXT5_X_OFFSET,EXT5_Y_OFFSET,EXT5_STEPS_PER_MM,EXT5_ENABLE_PIN,EXT5_ENABLE_ON,
        EXT5_MAX_FEEDRATE,EXT5_MAX_ACCELERATION,EXT5_MAX_START_FEEDRATE,0,EXT5_WATCHPERIOD
        ,EXT5_WAIT_RETRACT_TEMP,EXT5_WAIT_RETRACT_UNITS
#ifdef USE_ADVANCE
#ifdef ENABLE_QUADRATIC_ADVANCE
        ,EXT5_ADVANCE_K
#endif
        ,EXT5_ADVANCE_L,EXT5_ADVANCE_BACKLASH_STEPS
#endif
        ,{
            5,EXT5_TEMPSENSOR_TYPE,EXT5_SENSOR_INDEX,0,0,0,0,0,EXT5_HEAT_MANAGER
#ifdef TEMP_PID
            ,0,EXT5_PID_INTEGRAL_DRIVE_MAX,EXT5_PID_INTEGRAL_DRIVE_MIN,EXT5_PID_P,EXT5_PID_I,EXT5_PID_D,EXT5_PID_MAX,0,0,0,{0,0,0,0}
#endif
        ,0}
        ,ext5_select_cmd,ext5_deselect_cmd,EXT5_EXTRUDER_COOLER_SPEED,0
    }
#endif
};
This kind of rat's nest can be found all over the Repetier source, as can magic numbers like "#if DRIVE_SYSTEM==3" - not descriptive, and if anyone wants to change that out for an enumerated value, it'd take a lot of effort. It's like a less-uncouth version of Marlin. Cleaner motion control, as I recall, but the code quality is what I'd expect from a college student. Smoothie's organization is light years beyond this.

Re: Impressions: Duet vs. Smoothie

Posted: Wed Oct 28, 2015 2:32 pm
by RMLynch
Forgive me if this sounds proper daft...

Would it be possible and/or relatively easy to locate and swap out the memory module(s) on the SmoothieBoard to something much greater?

Or would this be a nightmare to implement?

For those interested, 626Pilot specifically, here is a good read on the SmoothieBoard 2/Pro 2: http://smoothieware.org/blog:13" onclick="window.open(this.href);return false;

Re: Impressions: Duet vs. Smoothie

Posted: Wed Oct 28, 2015 3:35 pm
by WZ9V
I could be wrong but I believe it requires swapping the entire main chip since the memory is onchip. So unless you could find one in exactly the same package with exactly the same pin out and exactly the same internal devices and exactly the same mapping and with more memory, it would take porting the whole thing to a new chip aka Smoothie2.
RMLynch wrote:Forgive me if this sounds proper daft...

Would it be possible and/or relatively easy to locate and swap out the memory module(s) on the SmoothieBoard to something much greater?

Or would this be a nightmare to implement?

For those interested, 626Pilot specifically, here is a good read on the SmoothieBoard 2/Pro 2: http://smoothieware.org/blog:13" onclick="window.open(this.href);return false;

Re: Impressions: Duet vs. Smoothie

Posted: Wed Oct 28, 2015 3:57 pm
by RMLynch
WZ9V wrote:I could be wrong but I believe it requires swapping the entire main chip since the memory is onchip. So unless you could find one in exactly the same package with exactly the same pin out and exactly the same internal devices and exactly the same mapping and with more memory, it would take porting the whole thing to a new chip aka Smoothie2.
RMLynch wrote:For those interested, 626Pilot specifically, here is a good read on the SmoothieBoard 2/Pro 2: http://smoothieware.org/blog:13" onclick="window.open(this.href);return false;
Yeah I wondered about that but wasn't sure. After posting it, the thought popped into my head.

Thanks!

Re: Impressions: Duet vs. Smoothie

Posted: Thu Oct 29, 2015 1:16 am
by Nylocke
626Pilot wrote:but the code quality is what I'd expect from a college student.
Careful with them words, some of us here are college students :P

Re: Impressions: Duet vs. Smoothie

Posted: Fri Oct 30, 2015 11:31 pm
by 626Pilot
More memory could be added to a Smoothieboard v1. You can just buy RAM chips. This guy made a MOD/S3M player that runs on an ATMega644 (like Arduino, but normally runs at 20MHz and can be overclocked like crazy):

[youtube]http://www.youtube.com/watch?v=j6ijbexoq-M[/youtube]
I don't know how easy it would be to wire it in because I don't know whether the right microcontroller pins are brought out to any headers. It would require changing some of the code - it would have to know how to access the RAM.

On the other hand, if there's an LPC chip that has exactly the same everything but with more RAM, you could hot air the existing controller off the board and drop in the replacement. I don't know whether any such chip exists.

Anyway, I don't think someone is going to go to all that trouble just to run my firmware without disabling their LCD.

Re: Impressions: Duet vs. Smoothie

Posted: Sat Nov 07, 2015 4:37 am
by 626Pilot
I'm finding the upload speed highly variable. Sometimes it goes around 100KB/sec, but tonight it's more like 30KB/sec. I rebooted it and it's still the same. When I use FTP, it seems like it lets it upload a chunk of data, pauses, lets it upload another chunk, pauses, etc.

Re: Impressions: Duet vs. Smoothie

Posted: Sat Nov 07, 2015 9:57 pm
by bot
Are you on wireless? With a wired connection I get a consistent 250kB/s.

Re: Impressions: Duet vs. Smoothie

Posted: Mon Nov 09, 2015 8:42 pm
by 626Pilot
No, I got a long Cat-5 cable.

Re: Impressions: Duet vs. Smoothie

Posted: Mon Nov 09, 2015 9:09 pm
by bot
Are you connected directly to a computer Ethernet port or do you have a switch/router in between? Or even a hub? There could be several factors at play. Typically this isn't an issue any longer, but back in the old days cat5 cables could be either straight-through or cross-over. For connecting two clients (computers) together, the cable had to be a crossover rather than a straight-through.

Also, there could be a conflict of protocols setup for the ethernet port. I would try with a switch or at least a hub in-between to see if that changes anything -- if you aren't already running it in such a way.

Re: Impressions: Duet vs. Smoothie

Posted: Fri Nov 13, 2015 5:21 pm
by 626Pilot
I'm running off a gig-E switch that has no trouble talking to other machines. Because I can see the FTP client accepting data pretty fast and then pausing, I think it's waiting for SD writes to finish.

The Duet and Smoothie teams will make a lot of how some SD cards are much faster than others, but even a "crappy" SD card will gobble down megabytes of data every second if I put it in a $7 USB reader from Amazon, so I don't see the logic in blaming the card. I think it's down to this: the people who write device drivers for Linux and Windows know something that the Duet and Smoothie teams don't, because the same card that takes 30KB/sec of data on my Duet will eat a ten-meg file in a fraction of a second if I use the reader.

Also, I tried doing a wide print on my Duet with the dc42 firmware and I can confirm that at least on my printer, it DOES NOT provide full-bed printing. Using the stock print height, the nozzle digs into the surface over on the right side of the bed. If I change the bed height so that the right side prints OK, it will dig into the far left side of the bed. This is at least partly because dc42's firmware doesn't have Z correction. It seems that for the time being, if I want full-bed prints, I'll have to stick with my Smoothie fork, and just not have three nozzles.

I have some replacement Trick Trucks, and a set of 300mm arms, on order from TrickLaser. Maybe that will help the Duet calibration.

Re: Impressions: Duet vs. Smoothie

Posted: Fri Nov 13, 2015 5:47 pm
by bot
Are you speaking about the auto-calibration not allowing you to print over the full bed? I'm going to be attempting to use his autocal routine with manually measured probe points, to rule out inconsistent probing (and because I don't feel like hooking up a probe :P). I wanna see how it works.

Re: Impressions: Duet vs. Smoothie

Posted: Fri Nov 13, 2015 11:47 pm
by 626Pilot
His auto-cal doesn't allow me to print across the full bed.

Re: Impressions: Duet vs. Smoothie

Posted: Sat Nov 14, 2015 12:53 am
by bot
That seems weird. It was easy enough for me to calibrate the thing manually to use the whole bed. Did you try manual calibration?

Re: Impressions: Duet vs. Smoothie

Posted: Thu Nov 26, 2015 5:00 pm
by bot
Hey 626, do you happen to have the gcode coordinates for the grid you were using in dc42 autocal? I'm testing the autocal out by manually probing and inserting values, but don't feel like generating the extra probe points for the larger bed. If you have them lying around in the bed.g file, care to share them?

Re: Impressions: Duet vs. Smoothie

Posted: Thu Nov 26, 2015 5:16 pm
by mhackney
I posted a python script in the other thread that will generate them for you and the example has Rostock coordinates.

Re: Impressions: Duet vs. Smoothie

Posted: Thu Nov 26, 2015 5:25 pm
by bot
Excellent, thank you!

Re: Impressions: Duet vs. Smoothie

Posted: Sat Nov 28, 2015 12:34 pm
by JFettig
For someone who doesn't have much programming experience - how easy is a duet to set up? How well does it work as far as calibration goes? Can you adjust tower offsets? I can't imagine its any more difficult than setting up Repetier firmware on a rambo?

Are there any other panels that work? I'm not a fan of touch screens

Can the web interface do everything you would want to do with it? I personally like using an encoder knob for setting z-height rather than that weird crosshair thing that is in repetier host and pronterface.

Re: Impressions: Duet vs. Smoothie

Posted: Sat Nov 28, 2015 2:55 pm
by bot
The only kind of experience related to programming that might help would be knowing how G-Codes work. All the configuration is done with gcodes in a file that is on the SD card that is loaded on bootup. It's not as simple to use as the EEprom with repetier host, but it's decent. Which tower offsets do you mean? If you mean endstops, then yes you can adjust those. And tower position angles, but not individual tower radiuses (at least not yet).

The web interface does largely have everything you'd need (and more!) compared to the paneldue. The one thing I've encountered, though, is an unreliable connection with the web interface. I can't say it's any fault of the web interface application per se, but I often lose the connection, or the connection becomes very laggy, or I simply cannot connect to the web interface -- the address does not resolve. It mostly works, but I like having the paneldue as a hard-wired backup that allows me to control the printer in the event that the web interface stops working. The problems seem mostly to happen if the computer is left unattended for a while with a print going, and the screensaver has turned on or hard disks turned off. I've had some luck by keeping the web interface application in focus and preventing screensavers or anything from engaging.

The touch screen on the paneldue is actually kinda finnicky (at least with the stock 4.3" one from t3p3) The buttons are small and close together, and sometimes the touch will randomly hit a button nowhere near where you pressed. It kinda just takes some getting used to. For the price, I think it's definitely worth having as a backup control method.

Re: Impressions: Duet vs. Smoothie

Posted: Sat Nov 28, 2015 10:07 pm
by JFettig
Thanks for the reply Bot,

By tower offsets - I mean the angular offset that they are off of each other, 90, 210, 330 degrees is default for repetier.

I assume that the reliability of the web connection does not effect printing?

The problems you're experiencing with the paneldue are exactly why I don't like touch screens :)

As far as I can tell, its the only display that is supported?

Re: Impressions: Duet vs. Smoothie

Posted: Sat Nov 28, 2015 11:59 pm
by bot
I believe it is the only display that is currently supported. You can adjust the angular position of the towers around the circle. They assume a normal configuration and you can adjust +/- some degrees (+ moving the tower anti clockwise).

The reliability of the web interface does not affect printing. Though, sometimes, the web interface can not connect until the duet is rebooted -- which WOULD affect the print. This is why the paneldue is a must for me.

Re: Impressions: Duet vs. Smoothie

Posted: Sun Nov 29, 2015 8:22 am
by 3D-Print
The PanelDues is not available to order, and per Think3DPrint3D, will not probably be available until Jan 2016. There is a redesign by David Crocker in the works.