Skydog

[dan]

I wouldn’t recommend flashing your firmware unless you really trust the firmware provider and you have no other options.

Like I said, flashing your firmware is an escape hatch, not the main exit…

Submarine Escape Demonstration

[soldiercoleman]

Actually the fact that Glowforge listened to the concerns of the customers is why I decided to order. I don’t think I would have without the wonderful attentiveness that @dan and the @staff show the community.

Maybe I’ll think about playing with the firmware when I have my bigger and better Glowforge 2.

[rpegg]

I assumed he meant that a Raspberry Pi, Arduino or custom board would need to be added as the controller. But yes an added motion controller is necessary. A local PC would need to provide user interface and higher level processing to convert the design to G-Code or something similar that the controller board would understand.

[henryhbk]

The glowforge does not have a motion controller onboard

How does it move then? It may not have motion planning, but it must have stepper control of some kind inside. I can’t imagine it solely has an h-bridge being controlled over the internet. Stepper control etc must be local, otherwise network latency would cause total havoc…

[jdodds]

@dan_berry

If memory serves, there was some talk about the GPL-ed GF maybe being able to run GRBL code. I’m aware of some CNC projects like Chilipeppr ( http://chilipeppr.com/ ) that provides a web-based GUI to mills, which typically runs on something like a Raspberry Pi and sending GRBL code to it. Even then, it isn’t a design tool, it’s more of a control panel.

@karaelena

That senerio would not work. The glowforge does not have a motion controller onboard.

So something else would have take the CAD convert that to CAM. Then convert that CAM to a waveform. Then upload/stream that to the glowforge.

And there you have it in a nutshell, 99% of the GF owner’s eyes have just glazed over and they are thinking, “but, all I wanted to do was make some cool drink coaster with a picture of my cat on them. . .”

Reality is, if GF cloud service disappears, best case scenario is that 10,000 users have a very large brick sitting on the kitchen counter for a very long period of time. Whaaattttt?!!!

Some entrepreneurial coder and hardware types would have to either: a) band together and do a traditional CNC g-code solution requiring at a minimum a new controller board which would give them a shot at the 1% of the GF owners that are tech savvy enough to get into some full on CAD/CAM software OR, b) to get that shot at the 99%, they would have to do what the GF team is doing now for an interface except that it would be local not cloud based as who in there right mind would buy into a cloud solution if the one they just had vaporized into thin air.

This is a good discussion to get into, not because of any lack of confidence in the GF team, but rather a better “eye opener” understanding of what happens if the cloud solution fails to materialize for the long term.

I’ve said it before and I will say it again. . . Personally, I have 100% confidence in this team and this project.

[karaelena]

Traditional motion controllers (GRBL, TinyG, Smoothie, Marlin, LinuxCNC (or MachineKit) & Mach3 all have the ability to output stepgen pulses required to move the steppers (specifically to tell the stepper drivers which direction and how long). The glowforge does not have this. The Glowforge ‘cloud’ create these and outputs a waveform in which the glowforge ‘plays’ locally. So think of a 10 channel MP3 (each channel is some form of I/O - i.e. one channel for X Axis Step, another channel for X Axis direction, another channel for PWM for the Laser, another channel for the exhaust fan etc…)

What you described is the stepper driver. Not the thing that tells the stepper drive what to do.

This was all discussed here

We’ll release a GPL-licensed firmware for Glowforge

Bingo. We do both CAM and motion planning in the cloud. We just send the machine the actual waveforms to be fed to the motors (X, Y, each fan, focus, etc), laser, chiller, etc. There’s a bit of decompression and then the uP sprays the results at the right I/O lines. Not being constrained by g-code opens some pretty cool doors in terms of what’s possible… much of that will be behind the scenes, though.

[ihermit2]

Yeah, not an easy task to create a cloud replacement. For people with out eyes glazed over, I seem to recall the cloud computes, not only a motion plan for moving the laser around to wave forms but also a cooling plan which I suspect are waveforms for fan speeds,in the pro the power supplied to the peltier, and possibly pump speed if that is not fixed.

[rpegg]

I was at a Mini-Maker Faire on Sunday. It’s always interesting to notice that 80% of the booths are not developers of new products or tools but technical folks that were demonstrating their personal design for controlling a homemade robot, automating an antique teletype, building an insanely complex device for moving a marble from Point-A to Point-B. Why? Why not? They tinker because it’s fun. They build because it hasn’t been done before. Even if Glowforge continues to operate forever, I would bet someone in the maker community will attempt to provide that local control. Heck, I would try it myself if I had more time and a little less sense. Might never make it to a finished product but the design and geek level code will show up somewhere.

[marmak3261]

I’ll be on board with this. I did install DD-WRT in a Linksys router once upon a time. It was one of my first ventures into Linux. This type of open source software model for existing hardware would be a similar project. It will require some new skills, but we have a great user base that will make it happen, should need arise. Fascinating about the Skydog. Kudos to Dan for naming the elephant in the room here.

[dan_berry]

@karaelena I wasn’t aware they were taking ALL driver control out of the equation… but by doing that they’ve only radically simplified the stepper motion control, they haven’t eliminated it completely. A better quote from Dan would have been:

@dan

Funny Glowforge genesis story: I was going back and forth with Mark, our CTO, between closed-loop (servos, which I was advocating) and open-loop (steppers). He made a better case for steppers. Finally I said, “Well hell, if you want to go open loop, you don’t even need local control at all. Just put the damn motion controller in the server.” I didn’t actually realize what I’d suggested until Mark stared at me like I was from Mars for a while and it dawned on both of us just how powerful that could be.

Chilipeppr relies on two processors: a TinyG (or Atmel 328 running Grbl) for the G-Code interpretation, motion planning and stepper control… and a RPi for the user-friendly web-based front-end.

Skipping to using a (insert real-time OS embedded PC here, because timing is critical) doesn’t have to be responsive to end-users except through a BBB (blinking blue button) for the most part. There’s no reason such a setup couldn’t, say, translate G-code into motion control and then process the output as well once it’s done… IF GF were to flop. But I’m optimistic that’s a big ‘if’.

[caribis2]

As we have no glowforge in our basement/garage/workshop/bedroom, we need to speculate about something. And speculating about this is a fun mental exercise. However, as they sold a s-ton of glowforges,and assuming they deliver what was promised, there is virtually nothing to actually worry about. Even if Glowforge management makes an extra concerted effort to screw up the business end of things, the user base is already plenty large enough that someone would buy the company’s assets.

Now, ten, twenty years down the road, when we have nanobot based fabricators, this could be an issue. But that’s the problem with old stuff.

[rpegg]

That’s actually the only thing that gives me any concern. If Glowforge is too successful then some larger totally profit driven company will buy them out, or more importantly the technology/patents. What is the incentive or legal requirement for a new owner to continue support for the early sales? PowerCloud Systems didn’t shut down Skydog. Comcast bought them out for the technology and then shut down support for Skydog. Now I fully expect to totally fry my several year old Glowforge and move on to an improved product before any of that will happen but it’s still something to think about.

[caribis2]

Being really successful and selling your company would not exactly run contrary to the start-up philosophy. But the question is what are the motives of the buyer? Would a buyer start selling annual subscription services to the cloud interface? I would not be surprised, but my response was to DIY’ing software and board controllers to keep tens of thousands of abandoned glowforges functional because the company couldn’t make a go of it. Discussing the unrealized fears of Glowforge being too successful - there is a topic ripe to drive Dan nuts.

Companies certainly get shut down all the time by buyers. I just don’t see that as likely in this space: to small and fragmented. But who knows, companies do crazy things. What I do suspect is that after pre-selling $28 million in 30 days, Universal, Trotec, etc… are trying to figure out what way is up and someone at at least one power tool manufacturer and maybe a company like Danaher is running numbers.

[PrintToLaser]

Of course the motivation for the acquisition would be profit, but care would be needed to develop a model that would not alienate the existing customers.

I have no idea of the resources required to develop the cloud based control, but a small dedicated crew (with money) managed to accomplish it.
I think a disenchanted customer base of 10,000 could be motivated to replicate and support it.

We nave the backbone of that organization right here, populated with compassionate craftspeople who count engineers and legal talent among their numbers.

A salute to @jdodds for throwing this issue on the table and bringing it into clear focus, but like him, I have faith in the founders vision.
I doubt any one of the them would be content to see greed sink its teeth into their baby and rag-doll it to death.
My perspective may be naive and overly optimistic, but I prefer that to worry and pessimism.

pessimistic - being prematurely disappointed in the future.

[marmak3261]

This thread is sobering. I’m like @rpegg’s philosophy: laser like there is no tomorrow and hope for the best!

[hroger1030]

Good to know that this backup plan is in place. People who prepare for the worst with backup plans tend to succeed, and so the fact that you have considered this possibility speaks well of you.

Long live GF!

[takitus]

In order to run a glowforge without the glowforge site itself, it would be possible to make a simple premade kit for people to use that included a raspberry pi interceptor that would act as a wireless access point which would redirect the glowforge to whatever site we put together to provide the necessary waveforms. Encoding movement into waveforms is something that is pretty commonly done, so it wont be the major hurdle here.

How they are encoding/decoding the waveforms would be the first concern I have. The encryption of the waveforms could be a stumbling block if they are using a key we dont have access to. This could cause us to have to modify the firmware. If its a standard encryption/compression scheme we can just implement that and move on to the next phase which is determining how homing, power levels, focusing, and any other small nuances of the glowforge work. Once that is figured out we have the base for using software that exists on the market currently to provide standard laser cutter functionality.

Implementing support for the cameras would be a whole separate ballpark. We dont even know what spectrums their cameras are working with yet, much less how they are interpreting that, so reverse engineering that would be quite a large piece to bite off. I think getting the standard functionality out the door first would be enough to at least keep people from rioting in the streets.

[ihermit2]

I would hope they are using https or TLS and rejecting any site with an invalid/inappropriate/self signed certificate. Could make an intercept very tricky.

[rpegg]

Had to stop myself from responding with a long technical back and forth. Keeping it at a high level… I’m hoping that some simpler subset of the cloud control could be reverse engineered from understanding the firmware and ported by the community to run on a standard PC. Otherwise we would still be dependent on external site control. Of course the promised firmware would need to be source code with at least inline documentation and not just a digital file to have a starting point.

[dan_berry]

I’m just going to throw this out there for discussion sake… I look forward to reading the criticisms.