Did you miss that part at the beginning where I clearly laid out that it was my interpretation of what was going on?
Poor little horseyā¦
Will never forget @marmak3261ās quoteā¦about feeding France for 3 months with all the dead horses that have been beaten on this forum.
Well thatās obviously false. If it were true, people wouldnāt be asking for it to be an option.
Regarding accuracy in placement, there is a helpful feature that I think is very clever for this system of visual placement.
The ānudgeā feature where you can bump the placement with the arrow keys. The distance of the ābumpā diminishes with each increase in zoom level, allowing tighter incremental control.
Iām not posting this in argument for visual placement (which I happen to love), just to bring attention to another tool at your disposal.
I have seen, and am encouraged by the improvements that magically appear in the machineās operation. From little things like the dimming lights when the lid opens, to others including extended fan operation after the job. It used to take an extended amount of time for the zoom, either in or out to respond. That is instant now.
Gratefully, the company decided to begin shipping with the software still in beta to cut our wait, and from all evidence the software team is actively attacking their task to improve user experience.
The company graciously provided this platform for our input because it is valuable to them in the productās development - so keep on providing your input for the sake of a better glowforge!
True, mechanical limit switches break if slammed against, optical limit switches would solve that problem. Mechanical ones could have been made easy to replace by the user. It aināt rocket science.
If anything they should have installed them to avoid head banging, and could have been used as a fallback if camera alignment fails. Like if it canāt find the head after trying to find it for 60 seconds, run to what it thinks is 0,0, and let the limit switches take up the slack.
Are you kidding? There isnāt a can of dog food unfilled in the US as a result of the horses being whacked here
Pretty soon weāre going to be up to our earballs with decaying horses, nowhere left to put them!
Yes, definitely a major disruption and break between horse maintenance and care and public values regarding the role horses play in the chain of being. It is not good at the moment for many horses. Our sentiment has changed but the reality of horse flesh hasnāt, (except as a metaphor for certain topics on the Glowforge community forum.)
Perhaps you are reading the word ācapableā as not including the idea of āpotentialā, which I would imagine @chris1 is really saying but repeated the word capable rather that saying potential. Thatās my take.
I agree. It is not capable at the moment to do numeric positioning. It seems that the potential for building in that capability for the GUFI is fairly strong and I think there are a lot of folks on the forum who see that from potential to capability shouldnāt be that difficult from a programming and firmware point of view.
Defining difficulty, complexity and implications is a slippery thing. Again, why Glowforge has settled on this feature set for the time being and have not been able to add features at regular intervals would be pretty neat to know. I can only imagine, and that is without any real world engineering experience.
I just had to say, this is now my new favorite BADH animation.
Itās insulting for you to insinuate that Dan could believe something like that.
Iāll be sure to apologize to @dan if I ever actually meet him. I suspect heāll survive the bungling attempt at explanation on my part.
He seems to be too professional to get mad at people for putting words in his mouth, so I agree that youāre almost surely in the clear.
Were you thinking about this?
Here, as a final reason that they decided not to use limit switches, Dan says that his laser cutter (which was imported by āHurricane Lasersā, I believe) had a loose limit switch screw which caused a fair amount of damage during a homing routine. Of course, this is a screw failure, not a limit switch failure, and itās not like the Glowforge doesnāt have screws in it.
Bonus snipit at t=16m49s : whatās in the Glowforge head?
Exactly.
Pretty sure this topic has been beaten to death. But I am going to drop in my 2 cents anyway. Thou I can respect the engineering challenge of not using limit switches- I really donāt understand the decision on not using them. But it is what it is. The cards have been dealt and at this point, there is no backsies. Since the bed loads in without a hard mount point, creating a register on it is a non-starter.
Iāve milled out risers that give the bed a better positive ālockā then the stock mount points in the chassis. But downfall you lose z-height and even with the bed static phantom āvirtualā 0,0 optically is not as dead nuts as I would like.
With that said, My wife would give two #$!@ about all of that. All she cares about is, āIs the Pewāing going to happen where I place the image?ā Which really I all I care about. She can walk up to it and use it. Vs. āI tried to go to 0,0 and it went over there?ā and me asking āOh did you goto G54 or G55?ā ā(door slams)ā
Fair enough.Using the head camera certainly can do amazing things when it works. I do see the benefits of the Faster. Better. Cheaper. design strategy, and I am glad they used this approach for most things. I think that until they work all the head alignment bugs out, not including limit switches has only accomplished one of those goals, Cheaper. Faster? No. They have had 2 years and havenāt worked out the bugs. Better? Acceptable alignment is 1/4 inch off. I have no doubts it will get better, but right now it aint.
He had a bad experience with a single limit switch. Once.
Iāll probably keep beating this horse. Its fun, and someday it may walk again.
Iād say there are a whole slew of people who donāt give a ratās ass about limit switches, zero stops and numerical positioning, saw the promo video, noted that you use a camera image to drag and drop a design and will live blissfully ignorant of the potential of such features. Who are we to tell them that that the cake is a lie.!
And those will be the same peor who get their glowforge, try the camera placement; find out it sucks and then come here and post about it in support. Or just give it a bad review I suppose.
Or perhaps be the lucky ones who get a production unit like mine and find out that itās just a dream to use.
Limits switches donāt break or come loose in a properly designed system. I have never broken one and the only time I have had them come loose is on early RepRap designs where they were held in place by P clamps that could slip.
I always use the button type because they are more accurate than the lever types and arrange it mechanically that the axis hits a stop before it can destroy the switch. With a low force drive like NEMA17 with belts there isnāt enough force to break the switch, so the body or its mounting bracket can act as the stop. On a milling machine with ball screws it needs something more substantial to stop the axis. If the switches are wired up normally closed and are the button type and fastened securely it is very hard to imagine them giving a false negative.
On the other hand how many times have we seen GF banging against the end stops and today even the front door. The camera method seems far less reliable. And even if saves a few Dollars per machine, how much engineering time and support time has it cost and will continue to cost until they make it reliable? Will they ever get that back?
And I donāt understand Danās claim the camera can watch the head and correct it on the fly. We have seen how long it takes to grab a shot of the head, upload it to the cloud, analyse it and send motion data back to do the next stage of homing. I canāt see how it can do that during a job. And if it is blindly executing stepper motor waveforms sent from the cloud does it really even know where the head should be to any accuracy? Steppers lag or lead when accelerating and decelerating.
The makers of my washing machine must think that lever switches arenāt reliable for sensing if the lid is lifted. They use two double-throw switches, wired in such a way that if they donāt both change state when the lid is opened, they dump power through the main fuse and kill the machine.