EEVblog #318 - Makerbot Replicator 3D Printer Unboxing & Review

[mikeselectricstuff]

I'm a bit surprised that there isn't some form of locator system to it. An optical head locating method would be a good idea. Maybe in the next gen. Then it could self-calibrate.

I was thinking that some sort of auto-sensing would be a good idea, especially as they don't seem to have any locknuts on the adjustment, and platform position is clearly a very critical parameter for build quality.
I was thinking some sort of simple mechanical sensing of touchdown.
Or maybe even thermal - move head slowly down til it detects a sudden increase in thermal load!

[EEVblog]

I'm a bit surprised that there isn't some form of locator system to it. An optical head locating method would be a good idea. Maybe in the next gen. Then it could self-calibrate.

Even with good cal, you still have the issue of no absolute position sensing.
I just had a fail on the example on the SD card (I found it, bottom of box), an it rattled the bot so hard I thought it was going to fall to pieces!
Obviously the positional accuracy got a bit out of wack...



Dave.

[firewalker]

Dave.

Modern Arts? 

Alexander.

[free_electron]

There is another reason this installer deployment is annoying. I know they broke python deliberatly between version 2 and 3 . So there you have conundrum 1... What to pick. I also know from experience that some libraries may work fine in 2.7.1 but do not work in 2.7.2 because they need patching. I have a legacy python system at work that usese numpy, scipy and some other toolboxes.
A new version comes out with something i want to use... Well i have to wait until the libs get ported. And ot don't happen at the same time. Sometimes it takes months. That's just .. Bovinebyproducts...

So that is why i am so grumpy. You use a system that is known to be picky. Please give me an exact duplicate of what was used to develop it. That way : no surprises..

The three unsigned drivers it wants to install are an arduino driver and two makerbot industries drivers. It also installs the two ftdi drivers, which are signed.
Afaik this thing uses ftdi chips to talk. So why do we need these special drivers ?
I know some arduinos use a home-brew usb serial converter using another atmel chip. If that thing were pure cdc class it would not need a driver, just an inf file. But alas, it aint't...
The point is : i don't want unsigned drivers. Not because they are not signed, but because they have not gone through WHCL... And that is asking for trouble.

Besides, on windows there is the winusb layer. You don't need drivers if you don't want to go through the hassle of passing WHCL. Winusb is part of the os. Just use that api to talk to the machine. Done. Problems all solved.

I went to the makerbot forum. Someone had a question runiing this on windows 8... They have a two page long instruction sheet on how to disable the requirement for signed drivers ( yes it can be done, but is an elaborate process ). That is just .... Wrong. Sign your drivers or move to winusb.

[Monkeh]

[rant mode on]
1) When is the damned broken sauce community going to employ turnkey installers ? Give me 1 file that contains everything that is needed. i don't wanna have to collect stuff from other locations, i don't wanna read documentation or have to resort to web for an answer. I don't want to figure out what versions of what modules i need and where to find them.
Put it all in 1 zip file and give me a setup.exe with 1 click. 

They're not going to do that. GPL forbids it. You may technically (at least to my understanding) not bundle GPL licensed software together with software that is not GPL licensed. This is part of the political aspect of the GPL, to put a burden on people who want to close their source code, much like the open source community feels that commercial vendors are.

Wrong. The limitations are on linking.

I know some arduinos use a home-brew usb serial converter using another atmel chip. If that thing were pure cdc class it would not need a driver, just an inf file. But alas, it aint't...

It is pure CDC and it does just come with a inf.

The point is : i don't want unsigned drivers. Not because they are not signed, but because they have not gone through WHCL... And that is asking for trouble.

Because Microsoft's slapdash procedures cover every hardware configuration and catch every possible flaw.

[Monkeh]

Anyone else notice the lack of strain relief on the control board connection? Sloppy..

I imagine they swapped to the external PSU so they don't have to go through all the certification for a mains power supply and can get a decently high voltage for their heaters etc. Why muck around with a crappy ATX PSU?

[bxs]

Great video, thanks Dave.

I don't have the money to buy it, but even if I had, they would not get it.

This thing should do self alignment, not that manual thing, get some sensor guys.

It should come with that thing mounted on, if is because of transport, hey this thing should have a rest position here is locked and same to transport.

Also think that at least they should put a support at rear to mount the PSU.

About the soft, I read complains about no download at front page, but look better it's there 

I'm a Linux guy so a got the replicatorg-0037-linux.tgz and tried to run and guess what? Yes, didn't work!

So I looked better, this soft is a mix of Java, Python, and some native binary sutff  a hell of a mess 

I have python 2 and 3 in my system, so I pointed "replicatorg" to phyton 2 but no luck, looked at terminal messages and guess what, fatal error in java, great... put java7 in a local dir and pointed "replicatorg" to use it, now it worked 

But Murphy's law again, I click to close the "replicatorg" and in terminal I still get things like "# A fatal error has been detected by the Java Runtime Environment:" at least is at end...

Well, the soft didn't impressed me a bit 

Finally, the quality of the exterior case that Dave printed out, maybe ok for the money, but still ugly.

[Monkeh]

What I don't get is why they chose for bloody Python again. Yeah, I mean sure... Open source, hacker community, portability and all that, but python is NOT the way to go in my opinion. It's not professional enough for a commercial product like this. Just develop a platform specific program for both windows, linux and mac and be done with it. And you can still open source the whole software.

Do you have the least idea how wide-spread perl, python, php, and other open-source languages and libraries are in commercial software? Go take a look at the bottom of the readmes for big software suites some time.

This one for Photoshop CS5 is revealing: http://www.adobe.com/products/eula/third_party/pdfs/cs5_products_readme_20100428.pdf

[Bored@Work]

Do you have the least idea how wide-spread perl, python, php,

Why, oh why, did you feel the need to mention the fucking P languages?

Perl: Unmaintainable. Perl's motto "There is more than one way to do it" in practice means everyone has his own way to shove some Perl up your ...

Python: A language by idiots for idiots. When was the last time you saw a non trivial Python program that didn't already throw a few error messages in your face at startup?

PHP: More security issues than you can shake a stick at and a library even more incoherent than the POSIX APIs.

Really, these three languages alone in their short time of existence have created more legacy code issues than probably all COBOL and FORTRAN code together.

[Monkeh]

Do you have the least idea how wide-spread perl, python, php,

Why, oh why, did you feel the need to mention the fucking P languages?

Perl: Unmaintainable. Perl's motto "There is more than one way to do it" in practice means everyone has his own way to shove some Perl up your ...

Python: A language by idiots for idiots. When was the last time you saw a non trivial Python program that didn't already throw a few error messages in your face at startup?

PHP: More security issues than you can shake a stick at and a library even more incoherent than the POSIX APIs.

Really, these three languages alone in their short time of existence have created more legacy code issues than probably all COBOL and FORTRAN code together.

Because despite above comments, the languages are out there in force in commercial products.

As to your comemnts about Python, every day. I rely on Python software on a daily basis and, because it's not written by idiots, it works fine.

[amvakar]

Do you have the least idea how wide-spread perl, python, php,

Why, oh why, did you feel the need to mention the fucking P languages?

Perl: Unmaintainable. Perl's motto "There is more than one way to do it" in practice means everyone has his own way to shove some Perl up your ...

Python: A language by idiots for idiots. When was the last time you saw a non trivial Python program that didn't already throw a few error messages in your face at startup?

PHP: More security issues than you can shake a stick at and a library even more incoherent than the POSIX APIs.

Really, these three languages alone in their short time of existence have created more legacy code issues than probably all COBOL and FORTRAN code together.

Because despite above comments, the languages are out there in force in commercial products.

As to your comemnts about Python, every day. I rely on Python software on a daily basis and, because it's not written by idiots, it works fine.

Those commercial products succeed when the end-user does not need to know nor care about the language used. As long as updating the runtime breaks things horribly because the open-source maintainers decided that stable syntaxes and APIs aren't any fun, that is quite difficult.

[Monkeh]

Do you have the least idea how wide-spread perl, python, php,

Why, oh why, did you feel the need to mention the fucking P languages?

Perl: Unmaintainable. Perl's motto "There is more than one way to do it" in practice means everyone has his own way to shove some Perl up your ...

Python: A language by idiots for idiots. When was the last time you saw a non trivial Python program that didn't already throw a few error messages in your face at startup?

PHP: More security issues than you can shake a stick at and a library even more incoherent than the POSIX APIs.

Really, these three languages alone in their short time of existence have created more legacy code issues than probably all COBOL and FORTRAN code together.

Because despite above comments, the languages are out there in force in commercial products.

As to your comemnts about Python, every day. I rely on Python software on a daily basis and, because it's not written by idiots, it works fine.

Those commercial products succeed when the end-user does not need to know nor care about the language used. As long as updating the runtime breaks things horribly because the open-source maintainers decided that stable syntaxes and APIs aren't any fun, that is quite difficult.

Because development and improvement are no fun, either. Welcome to the world of software development: Things move. You want things not to move? Don't move them for yourself.

[JoannaK]

Dave: Having this good print out of the box gives me good impression of Makerbots (one of the few).. I have seen so many bad prints (apparently bad build/calibrated systems?) made wth their machines that this one looks amzingly good in comparision.

Like you noticed that Frosting-output (all jiggly) first layer was due way too big gap between nozzle and the heated platform. The first layer shoudl actually Squeeze a little so that it'll stick well ..

I have no experience of ABS printing, but at least with PLA (and reprap) one can get nice clean part-bottoms (allmost glass-like smooth) when using heated paltform.

I have seen some automaticly adjusting Build platforms, but usually it adds a lot mechanic/electronics and it's not worth the effort. Once you'll get it leveled, the only real adjustment will be z-direction (parts tend to live due weather, heat etc)

Usually I try to avoid rafts (messy, waste of material etc), unless the part is so narrow that it won't stand up on it's own.. Another reason for raft is a base for supports with some complex prints.

[EEVblog]

I have killed my replicator with that spiral cube example on the supplied SD card.
Printed just fine before that as you saw, ran the example, it vibrated the shit of the Replicator, and now the negative Y axis skips and locks up big time.
No obvious physical issues with belts, tension etc
Took the wife to the lab to show it off and print some stuff only to have it fail.
Using the old thing-o-matic to get the job done.
Not happy!

Dave.

[ftransform]

perhaps there is a Chinese agent putting a virus so they could corner the market

upload a rar of the file :p

[Ed.Kloonk]

I have killed my replicator with that spiral cube example on the supplied SD card.
Printed just fine before that as you saw, ran the example, it vibrated the shit of the Replicator, and now the negative Y axis skips and locks up big time.
No obvious physical issues with belts, tension etc
Took the wife to the lab to show it off and print some stuff only to have it fail.
Using the old thing-o-matic to get the job done.
Not happy!

Dave.

That's a pisser mate. 

[johnmx]

This new version could be better, but IMHO it still is an expensive malfunction toy.

[bxs]

I have killed my replicator with that spiral cube example on the supplied SD card.
Printed just fine before that as you saw, ran the example, it vibrated the shit of the Replicator, and now the negative Y axis skips and locks up big time.
No obvious physical issues with belts, tension etc
Took the wife to the lab to show it off and print some stuff only to have it fail.
Using the old thing-o-matic to get the job done.
Not happy!

Dave.

It's hard to understand how a commercial product can have so many problems, for the problems it should still be in indoors development/testing, the thing is that the image of the company is on stake 

I just had a sort look at software and it really screams a bunch of software hacked together with some weak duct tape, fine the a home thing not for a commercial product.

They really need to use some of that 10M and invest in people to take care of the big problems and also the details...

[Bored@Work]

It's hard to understand how a commercial product can have so many problems,

Their customer base maybe doesn't mind. Being cool and all that for just $1749.

[jpelczar]

They really need to use some of that 10M and invest in people to take care of the big problems and also the details...

I'm guessing: two guys (tops), overflowed with tasks to do all development and testing. No matter how good the engineers/developers are, things won't work properly if they're given too much work in short time to do.

Who else does it apply to: big money project you get all the work to do ?

[firewalker]

I wouldn't expect something perfect for the given price.

Alexander.

[Bored@Work]

I wouldn't expect something perfect for the given price.

The given price is $1749. I would expect something good for that.

[Slothie]

Speaking as a professional software developer of 30 years standing who currently makes his living programming Python, I would say that people who denigrate python as not suitable for professional commercial software are speaking from a position of ignorance. Python is an extremely powerful language that is both fast and reliable when used by people who know how to program properly. Like any powerful tool, however, its not idiot proof because idiots are so good at what they do.
All the shortcomings thar Dave experienced while installing the software could have been avoided if the developers of "ReplicatorG" had gone to the trouble of using the "py2exe" tool to bundle their software with the correct version of python runtime and any python library dependancies. But they didn't and that is why Dave had the problems he did.

[EEVblog]

[t's hard to understand how a commercial product can have so many problems, for the problems it should still be in indoors development/testing, the thing is that the image of the company is on stake 
I just had a sort look at software and it really screams a bunch of software hacked together with some weak duct tape, fine the a home thing not for a commercial product.
They really need to use some of that 10M and invest in people to take care of the big problems and also the details...

I agree.
But I don't know how far they are ahead or behind the commercial competition, it could be that with all it's issues, the Makerbot is ahead.
But in either case they'd better be careful.
I've only had a brief play with a competing one and it seemed better on the software side, and on the hardware results side too, but that was only brief.
I've lost count of the number of people who've commented "what did you get a Makerbot for, they are crap, get XXXXXXXX" etc. So it seems they have to be careful about their rep. My failed example isn't helping with that I suspect.
If I paid $1800 for it, I'd be pretty pissed at this failure.
If I did something stupid to break it, then fine (although the software shouldn't let me), but running an example on the supplied SD card is inexcusable.

Dave.

[nitro2k01]

Dave, when I watched the video, I reacted to two things.
1) You took the platform adjustment process too lightly in my opinion. As far as I understand it, the purpose of the adjustment is that the distance between the platform and the head should be smaller than the height of filament track, so you get enough pressure for the first layer to stick to the platform. And then see what happened to your first print.
2) Then there's the rubber feet that you just threw to the said and said "who cares" when they didn't stick. They were likely there for a reason, obviously to absorb shocks. My guess is that's what killed it, maybe because all three platforms took a sharp turn at the same time and that created a shock which killed a stepper motor or made a dent in one of the gears or what have you.