Not very impressed with Smoothie

Jimustanguitar · Post by **Jimustanguitar** » Sun Apr 10, 2016 7:21 pm

I own two Smoothie Boards, am just now hooking the first one up, and wow what a mess. I'm kind of wishing that I didn't buy two of them before trying one first. Forgive the rant, but I felt like I needed to air my grievances in case it saved someone else the frustration later. Needless to say, a DUET board has been ordered.

Something about the board seems chemically difficult to solder to. I don't solder for 8 hours a day like some people do, but I have tacked pins and components onto hundreds of boards over the years, and the Smoothie just won't wick solder onto its own pads which is very frustrating.

Some of the pins that came presoldered on my board are tilted and don't align correctly for the LCD adapter. Arthur Wolf himself looked at this at last year's MRRF and kind of dismissed it as normal, and just bent the whole header straight again. Which had me worried that it would pull pads off the board and cause issues down the road...

You have to solder several header pins onto the board to make it work with the full graphic LCD board as well as the ethernet shield and voltage regulator. If you try to do anything else with SPI, like the thermocouple board, you have to modify the LCD adapter board to give you access to those pins. Granted, you can order the board in several different configurations, with and without pins already on it, but I was still surprised at how many times I had to uninstall the board to solder on one more headers or jumpers. I thought I was done at least 4 different times.

Using onboard power for the FETs (for heated beds and hotends, etc) requires either running extra power wires to connectors on a different side of the board, or soldering on jumpers that aren't included with the kit. The Rambo is this way as well, but it puts all of the power inlets in the same place and is much easier and more logical to do.

Speaking of power inlets, be very cautious that you have polarity correct. Don't go by the markings on the board, because there are two pads next to each other that cross, and the labels for one are the wrong polarity for the other. How stupid is this:
[img]http://i.imgur.com/l4xjghi.png?1[/img]

Their FETs don't have diodes across them for back-current from motors and fans so you have to solder on your own or else you void the warranty of the board.

The board's documentation exists, but it just seems difficult to navigate and semi-incomplete. For example, the documentation for the thermocouple board doesn't ever tell you where to connect it. It says to use an SPI channel, and then if you read their support tickets it tells you to use the LCD's SPI channel instead of the SD card's, but none of these are actually labeled or spelled out anywhere. You have to go digging into the pin diagram to reverse engineer which ones are SPI, and at that same depth in their documentation you'll also find that the LCD and a thermocouple don't play nicely together and there are several error reports related to this which have been opened by customers and closed repeatedly by the developers.

Here's what you can expect from trying to use the LCD and a thermocouple at the same time.
[img]http://i.imgur.com/SBAM7Dy.jpg[/img]
It takes 8 or 10 reboots to get back to where the screen works again, even if you unplug the thermocouple board or reload a different config file.

My own competency level probably plays into a lot of these frustrations, and I'm no expert, but as somebody who's not completely brand new to the game, yikes. I'd feel bad for a newcomer trying to figure this out for the first time. No fuses, either.

Am I the only one, or has the Smoothie underwhelmed a couple others as well?

Polygonhell · Post by **Polygonhell** » Sun Apr 10, 2016 7:42 pm

It's par for the course with Open Source stuff, things generally get to the point they are "good enough" for the major contributors to the project.
All the people who contribute to the project are familiar with it's quirks, so they never really get addressed.
No one likes doing documentation, so no one does. And the problems tend to be worse for less broadly adopted technologies, and those dependent on other Open Source components.

I use a lot of Open Source software in my day job, and despite the stuff I use being very broadly adopted, the documentation is crap and you have to be willing to look at the source code if you want to use it in anything even remotely complicated.

I find most of the reprap derived stuff to be basically the same way, both Hardware and Software.
It's not just a Smoothie thing, look at RAMPS and it's many derivatives, in an ideal world no one should ever have to flash firmware to make a configuration change.

Don't get me wrong I'm not knocking open source, it's just that it generally doesn't get "polished" for end users like commercial product where paying end users and competition drive development, in effect for open source the primary customers are the developers themselves, so development inherently has different focus.

Xenocrates · Post by **Xenocrates** » Sun Apr 10, 2016 9:06 pm

Don't own one myself (And feeling a little glad about it currently). But from an outside perspective, it doesn't impress either. So far on the forums I've seen two DOA, (one with the caps soldered on sucky, maybe it's that weird solder resistant board, other with FET stuck from the get go), and at least one blow a FET (And apparently the ADC for E.1 at the same time). It's not got great hardware. The developers are competent, but not very kind or polite. Documentation exists, but is sparse and peppered with a number of TODO's (At least the last time I looked at it).

But that sort of board design. Ugh. I'm no electrician (Or electronics tech). But a first (200, electronics rather than electrical fundamentals) level electronics class tells you pretty clearly. Fets, Solenoids, Relays, anything which can potentially generate a back-current, needs a diode. Fuses or circuit breakers are a basic protection that should be designed into any system, even a toaster, much less a delicate control board commanding hundreds of watts of power. And how hard would it have been to design a sane power connection. Sure, screw terminals add cost and take space. But even if you insisted on using solder pads rather than a connector or terminal, it should be relatively easy to connect the pads that need the same voltage (or if they are intentionally separated to allow you to have multiple voltages in, document them and move the pairs farther apart, without this cross pad bollocks). Headers that would potentially be used normally should be installed by default and accessible (Admittedly, even the RAMBO fails a little at this, as the LCD adapter partially obscures one of the analog EXT pins, but only one, and that could be solved by using a small radius on that corner (Which may be a side effect of the cutter more than anything).

Smoothie also seems to have a pile of M and G codes they use, which no one else does. They also have no problems changing what certain options mean, and pay little attention to making coherent families or series's of G and M codes. I dislike this. If I pick up a Haas manual from thirty years ago, the commands will do the same things now as they did them (G150 gained the ability to use more points in pocket definitions on newer controllers admittedly, but all the old commands run right), and there are logical sets of codes that are continuous or nearly so. G00, G01, G02, G03 for example, all define different varieties of interpolation. G12 and G13 are circular pocket cycles where the trailing digit matches the number of that direction of normal circular interpolation. G40, G41, G42, and G43 all do sane things that are related to each other (Guess what G40 does

Spoiler:

). There are some that are really out of sequence, some numbers that don't get used at all, and some codes, especially M codes, can mean different things on different machines and makes. M97 for instance is a local (within the main program but after a M30) subroutine call on Haas machines, but no other manufacturer includes that feature that I know of. It's a general problem with printing. But Smoothies' devs seem more willing to mess around with that than most (At least to me). Considering the praise heaped on them for having sane maintainable coding in the firmware, and attention being paid to reserving G and M codes permanently and not changing what a given option does, it's like they made their part nice and easy to work with, while making life for guys like me, who often will write code by hand living hell because they didn't have as much foresight or care about it. It may just be that it's due to all the different forks, such as 626 pilots, that I've even noticed it and it would be an issue for me.

{Rant}

Spoiler:

But then, this whole thing is just opinion and anecdote.
I'm sure there are people who are perfectly happy with their smoothie boards. And I'm sure there are a number of people for whom my criticisms are just someone who doesn't want to understand enthusiast progress yelling at people to get off their lawn. And there are certainly people with more understanding than I have, both of the industrial side, and the enthusiast side of this, who think that the two are fine working totally differently.

JFettig · Post by **JFettig** » Mon Apr 11, 2016 8:35 am

I'm also absolutely disappointed I bought one as well. Its just sitting on the shelf now. I'm done with it. Arthurs attitude really killed it for me, anytime I reported a bug he blamed it on me or something I was doing, especially when it came to Simplify3d. The board can't handle the g-code, even though its kind of ridiculous, any other board on the market has no problem with it. Everything else that didn't work right, it was my problem, like using Windows OS vs Mac or linux.... The motion control on the board cuts corners, messes with extrusion rates, etc. I never could get good prints out of it so I gave up. I bought a RADDS board for a Arduino Due and have been very happy with the results. I would have considered a duet if it had pololu sockets on it. I also didn't want to be stuck with a touch screen interface.
I might pull it out when I get my laser built, but I don't know how long I'll use it based on everything I've experienced so far.

Holy1 · Post by **Holy1** » Mon Apr 11, 2016 9:07 am

JFettig wrote:I'm also absolutely disappointed I bought one as well. Its just sitting on the shelf now. I'm done with it. Arthurs attitude really killed it for me, anytime I reported a bug he blamed it on me or something I was doing, especially when it came to Simplify3d. The board can't handle the g-code, even though its kind of ridiculous, any other board on the market has no problem with it. Everything else that didn't work right, it was my problem, like using Windows OS vs Mac or linux.... The motion control on the board cuts corners, messes with extrusion rates, etc. I never could get good prints out of it so I gave up. I bought a RADDS board for a Arduino Due and have been very happy with the results. I would have considered a duet if it had pololu sockets on it. I also didn't want to be stuck with a touch screen interface.
I might pull it out when I get my laser built, but I don't know how long I'll use it based on everything I've experienced so far.

I have tried to buy a Radds but there are none to be found

JFettig · Post by **JFettig** » Mon Apr 11, 2016 9:30 am

They appear to be out of stock.

Holy1 · Post by **Holy1** » Mon Apr 11, 2016 10:02 am

JFettig wrote:They appear to be out of stock.

Indeed. I have sent several emails but either didn't get a response or just got shoulder shrugging. No one knows if anymore will be made.

Post by **Eaglezsoar** » Mon Apr 11, 2016 2:14 pm

Jim, when I get a difficult to solder board I usually use a small amount of rosin in paste form and apply a light coat of it where the soldering will be done. This cleans the area and makes the solder "stickable"
The solder has a rosin core but some of the stuff I have seen from "other countries" seem to require a little more rosin.

http://www.amazon.com/Rosin-Paste-Flux- ... paste+flux

http://www.amazon.com/s/ref=nb_sb_noss_ ... acid+brush

Post by **Eaglezsoar** » Mon Apr 11, 2016 2:17 pm

Are the MKS Sbase cards any good? Just looking for opinions. I have two of them (v1.2) that I am thinking about putting up for sale.

JFettig · Post by **JFettig** » Mon Apr 11, 2016 3:42 pm

Any of the clones will have the same firmware issues I experienced. They do appear to be better laid out than the original(fucstercluck IMO).

I'd guess the solder issue is lead free solder which melts at higher temp than normal solder. I didn't have any issues soldering stuff on mine.

Nylocke · Post by **Nylocke** » Mon Apr 11, 2016 4:16 pm

Ive had problems with soldering too, 60/40 tin lead solder

Glacian22 · Post by **Glacian22** » Mon Apr 11, 2016 4:38 pm

Eaglezsoar wrote:Are the MKS Sbase cards any good? Just looking for opinions. I have two of them (v1.2) that I am thinking about putting up for sale.

I've been running an MKS Sbase with no issues.

U.S. Water Rockets · Post by **U.S. Water Rockets** » Mon Apr 11, 2016 4:56 pm

JFettig wrote:I'd guess the solder issue is lead free solder which melts at higher temp than normal solder. I didn't have any issues soldering stuff on mine.

I was going to suggest the same thing, but since others reported using Tin/Lead solder and having issues, then I'd suggest that the PCB itself may bit have thermal reliefs around pads or large copper areas around mounting points or on inner layers. Laying a board out to dissipate a lot of heat means it will suck a low wattage soldering iron down below the melting point of solder and may prevent the solder from ever reaching the correct temperature.

JFettig · Post by **JFettig** » Tue Apr 12, 2016 8:29 am

What I'm saying is that if the existing solder is lead free, it'll be harder to melt. I used tin/lead when I soldered on mine. I have a quality solder station that had no problem. Cheap irons don't do well.

Nylocke · Post by **Nylocke** » Tue Apr 12, 2016 11:08 am

This is with my HAKKO iron. Its not a super high wattage or high thermal mass but its not a shitty $5 shack iron or anything.

KAS · Post by **KAS** » Tue Apr 12, 2016 2:05 pm

I have a couple china spools that claim they have lead. You can definitely tell the difference with the cheaper ones.

626Pilot · Post by **626Pilot** » Wed Aug 03, 2016 7:40 am

I could never desolder broken components from a Smoothieboard, and my solder station goes up to 550 degrees.

I might be a little biased, since I've been working with the firmware since 2014, but I like it better than the Duet. The board design might be "weird," but the hot end power connector isn't as dinky as it is on the Duet. (Like, it makes me wonder if I'm going to twist it off the board when I tighten down the screws.) A recent blog entry indicates that they have done a pretty big refactor, and are getting ready to merge it into the main release once it's tested well enough. One of the things they're doing is continuous step generation, like the Duet has. That is claimed to fix the S3D bug.

I've had my Smoothies on ethernet before, but the SD card write speed is so terrible that I don't bother with that any more. I have OctoPrint set up on a Raspberry Pi 3, and it's marvelous. I have a touchscreen LCD, but usually I don't bother with it as the web interface has a skin that works perfectly on tablets and phones. It will also film a time lapse if you give it a camera.

The upcoming Smoothieboard 2 is what I'm really waiting for. It drives more steppers, which is what I need to start using a Diamond hot end. It also has a lot more RAM, an FPGA(!!!), an Edison slot, and probably other things I'm forgetting.

bot · Post by **bot** » Wed Aug 03, 2016 10:58 pm

Give the DuetWifi a shot, it should have all the improvements you're after (beefier hotend connector) and the ability for tons of motors. I'd like to see what smoothe 2 does, but the Duet seems much more reliable and well-thought out. It's a delivered, ready to use package whereas the smoothie has been a hackers dream and a users nightmare.

626Pilot · Post by **626Pilot** » Thu Aug 04, 2016 12:38 am

bot wrote:Give the DuetWifi a shot, it should have all the improvements you're after (beefier hotend connector) and the ability for tons of motors. I'd like to see what smoothe 2 does, but the Duet seems much more reliable and well-thought out. It's a delivered, ready to use package whereas the smoothie has been a hackers dream and a users nightmare.

I'm already used to the Smoothie way of doing things, and trying the Duet last year left me with the impression that the delta calibration is less advanced than my own. dc42's algorithm is certainly good, but when I tried it, depth-mapped Z correction was still in its infancy (read: undocumented and maybe or maybe not working yet). Correcting the geometry by itself didn't get me the flat 1st layer I needed, whereas my Smoothie firmware modification does. I also don't like that the coordinates have to be in a separate file, when they could just be generated on the fly. And, mine probes more points, so that when their Z correction is working (which maybe it already is), mine will still have finer resolution (less bumpy) than theirs.

Finally, as a developer, Smoothie is WAAAAAAAY easier to work on than Duet. Smoothie has its source in a very nicely nested directory structure. Duet is basically a pile of spaghetti in comparison, making finding the location of something you need more difficult. Many things are mashed into a single file, which in Smoothie would be split out to five or ten different files, in a directory tree that makes it obvious that that's where you'd look for whatever you need. Duet's organization is the result of developing using the weird Arduino IDE, which demands a directory structure very unusual for a C++ application. It encourages (practically demands) the placement of all source files in a single directory, which makes sense if you only want to write something that blinks an LED and nothing else, but really complicates things when you want to write a big application. Libraries are supposed to all have their own separate, single-directory trees outside (not within) the application's directory. I've lost many hours, and said many bad words, while dealing with that IDE.

What I'd really like to see is a Raspberry Pi driver shield, so that we could walk away from these barely-modern microcontrollers for good. Smoothie running on a Pi would be magnificent. You could run OctoPrint on the board, and file uploads of 20MB would be gobbled up in a second, and you'd have ONE BOARD that does everything.

Nylocke · Post by **Nylocke** » Thu Aug 04, 2016 8:56 am

DC has been writing/building his branch in Eclipse (https://github.com/dc42/RepRapFirmware/ ... ctions.txt). I'm not sure how good the organization is, but I feel like the Duet is the true all in one controller, especially with the release of the new DuetWifi.

If it would convince you to try it out I could talk to DC about me reorganizing the firmware into something more user-friendly for people who aren't the primary dev, and I would communicate with you to make sure you're happy with it too.

dc42 · Post by **dc42** » Thu Aug 04, 2016 7:00 pm

626pilot, if your Duet has the small bed heater connector then it must be the older version 0.6 board. The 0.8.5 board has a much larger connector, same as the Duet WiFi.

I guess you haven't looked at the RepRapFirmware source tree for a while. It isn't as flat as it used to be and I am gradually refactoring it into smaller modules. But there is still some way to go before it is structured as I would like.

Nylocke · Post by **Nylocke** » Thu Aug 04, 2016 9:54 pm

Would you like some help? I'm not as experienced as you but I'm fairly adept with different dev tools and languages.

bot · Post by **bot** » Thu Aug 04, 2016 11:43 pm

dc42 wrote:626pilot, if your Duet has the small bed heater connector then it must be the older version 0.6 board. The 0.8.5 board has a much larger connector, same as the Duet WiFi.

I guess you haven't looked at the RepRapFirmware source tree for a while. It isn't as flat as it used to be and I am gradually refactoring it into smaller modules. But there is still some way to go before it is structured as I would like.

Were there not some early 0.8.5 boards with small black connectors? I have some fairly small connectors on my 0.8.5, and from what I understand the DuetWifi uses the green type? Am I mistaken?

[edit] Nevermind, you said bed heater. I thought we were talking about the hotend connectors. I think that was the problem 626 had, the hotend connectors.

dc42 · Post by **dc42** » Fri Aug 05, 2016 2:20 am

My mistake, 626pilot was indeed talking about the hot end heater connector.

I've never had any worries with the hot end heater terminal blocks on the 0.8.5 being strong enough. You can always hold the terminal block as you tighten the screw if you are worried. However, along with other users, I have found it difficult to get the hot end heater wires into the terminal block unless I can see the entry holes, because they are so small. It's probably OK if you use ferrules and have the correct crimp tool to crush them into a square shape; but most people (including me) don't. The larger terminal blocks on the Duet WiFi make it a lot easier.

626Pilot · Post by **626Pilot** » Tue Aug 09, 2016 5:19 am

Nylocke wrote:If it would convince you to try it out I could talk to DC about me reorganizing the firmware into something more user-friendly for people who aren't the primary dev, and I would communicate with you to make sure you're happy with it too.

As I see it, his firmware is already good enough that he doesn't need my help, at least not in any problem domain I have recent experience with. If it was my project, I would add more probing points and finalize the depth map-based Z correction, if it hasn't already been. Might be a fun side-project, but I have serious deadlines to worry about these days.

dc42 wrote:I guess you haven't looked at the RepRapFirmware source tree for a while. It isn't as flat as it used to be and I am gradually refactoring it into smaller modules. But there is still some way to go before it is structured as I would like.

That is good to hear. It seems there is some cross-pollenation going on. Your algorithm is being implemented on Smoothie, and apparently they're doing a big refactor that includes something like your instantaneous step generation. And PanelDue support, of course.

Welcome to the SeeMeCNC Community

Not very impressed with Smoothie

Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie

Re: Not very impressed with Smoothie