An engineer's observations and a question or two

nhf

#1

Just got my GF going with venting etc. Decided to use balsa wood for my first print of the Founder’s Ruler. Didn’t see balsa in the wood selections but found a vernier and chose that. Everything was going well until the final logo step - where the balsa ignited. At this point I noted three things.

  1. The big red button marked STOP doesn’t exist. Everyone knows powerful dangerous machines have the big red stop button. My automotive training kicked in, you don’t open the hood on an engine fire. I was fortunate the balsa only smoldered. Further opening the lid during a material fire would expose you to the flames and fumes.

  2. The exhaust fan doesn’t run much past the ending of the cut. It really should run long enough to exchange the air in the compartment once or twice. If something really noxious was cut, opening the lid would not be pleasant.

  3. The WiFi setup doesn’t work if the WiFi has a hidden SSID. I know WiFi security isn’t secure but I really prefer my SSID hidden.

Questions:

Q1) Would it be difficult to program the big button to act as an emergency stop button?
Q2) How about running the fan long enough to exchange the air a few times?
Q2) Does the WiFi in the GF support 5 GHz band?


#2

No, but wouldn’t be as safe as opening the door for cutting power to the laser. The button requires the microprocessor to interpret the push and perform the cancel and there’s always the chance it is unresponsive due to a bug. I think opening the door or disconnecting the pro external interrupt should shut off power to the laser regardless of the microcontroller state.

You’re getting enough volume of fresh air pumped into the machine by the fans that opening the door isn’t going to fan flames more than they already are.

I agree. It’s a common request to run the fan longer or on demand in the UI. Even when things go well there is often smoke trapped under the material that rises when the material is removed.


#3

You lift the lid for a fraction of a second, about half an inch, and then let it fall back again. It kills the print. No smoke escapes.

I’ve killed many a print doing it. :slightly_smiling_face:


#4

The WiFi module only operates in the 2.4 GHz band.

The WiFi can be configured to connect to AP’s with hidden SSID’s, but it is not officially supported, requires potentially voiding your warranty, and discussion of how to accomplish it in this part of the forum is verboten.


#5

As Jules pointed out, just cracking the lid for a split second breaks the safety interlock, kills the print instantly and the head retreats to the upper left corner, while the fans continue to run. It’s quite effective. :+1:
A kill switch would leave the head parked over the fire. Bad juju for the lens, camera and other systems in the head. :fire::heavy_dollar_sign:
Didn’t the air assist blow the flame out?

There is a big difference between an engine fire and a candle flame - and since the atmosphere in the machine is fully oxygenated by the intake and exhaust fans there is no danger of back flash. I have only seen one instance of a flame I couldn’t snuff with my thumb. In that situation dropping even a bone dry rag on it would extinguish it.
My personal fire management plan consists of a damp rag, a squirt bottle, a fire extinguisher and 911 - in that order.

In the case of ignition, the primary goal is to put the fire out. You have to open the lid to do that, or watch your laser burn. :grimacing: In my opinion, a little smoke in the room from a class A fire is the least of my worries.

There is a list of materials that warrant close attention because of the ease of ignition. Paper products, with corrugated cardboard at the top of that list - to which I just added balsa wood. Thanks for the heads-up! :sunglasses:


#6

Both of those are also read by the microprocessor.


#7

Did a few gas system designs over the years and even though an explosion hazard is not apparent here, a couple of the core safety items from this could carry over.

1: CPU’s initiate a controlled stop but not an E-Stop. Simple reality since computers can and do get into endless loop mode. A hard wire E-Stop, remote from CPU control, is the only safe guard past this.

2: Purges (before an ignition attempt) are required to move (at least) the equivalent amount of air that would fill the container five times. This was quite long in a few cases, (with huge fire boxes), but no industrial insurance company would certify a system with less. Anticipation of CFM changes and other factors made it reasonable to extend this past the five times margin when applicable, so as to not embed a permanent wall to any expansions in the future. Gives the control circuit an inherent ability to survive future expansions.

Can we do this?
PRO has an E-Stop ability with the shunt loop on the breaker. For those not inclined, a lid lift works.
As far as purge times (after ignition in our case) I would have to defer to the glowforge team to see if they considered how much air is moved over the run-on time period. Letting us extend beyond this time for our own peace of mind would be a big plus as far as customer support goes. It does not bother me at all to have the exhaust running to attempt a capture of remaining fumes while I open the lid and remove materials.

Should we do this?
As mentioned. If a fire is directly under the laser head, my immediate priority would be to put out the fire to prevent damage. Laser head retreating on a lid lift is ideal. Totally agree with PrintToLaser here. Rag, Spray, Extinguisher, 911.
As far as the purging goes, as long as the minimum suggested by the engineering team is locked in, an extension is entirely up to an individuals ability to ignore instant gratification.

Seems my two cents turned into half a buck. Sorry for the length.


#8

We’re not calling them prints, are we? :slightly_frowning_face:


#9

I am. You can call them whatever you want though. :wink:


#10

At 200 cfm and a 10-sec post operation run time (which is approximately what mine runs after the job is complete), isn’t that running through a little over 3 cubic feet of air per second? Withn interval volume of a little over 3 cubic feet (not considering equipment inside). Meaning the 10 second post op run time is moving about 10x the case volume?


#11

I think I’ll call them burnings… or Steve. They look like a Steve.


#12

Steve works. :smile:


#13

There’s no reason it has to do that. Pressing the button could be the same as clicking cancel in the app.


#14

By your definition, it is a controlled stop, not an E-Stop.


#15

No, it doesn’t have to, but by my understanding of a ‘kill’ switch, it’s the same result as pulling the plug.
An ‘Abort’ switch would be better, bit we already have that with the safety interlocks.


#16

Another engineer’s observation - fire safety and what to do in case of a fire is located in the user manual on page 3:


#17

Not entirely true.

The safety mechanism is actually done in dedicated hardware logic circuits, independent of the microprocessor. (I have a detailed schematic diagram that I have worked out, but I’m hesitant to post it on my wiki as I am not really interested in helping people defeat important safety systems.)

If the microprocessor managed to lock up, and the watchdog timer didn’t trip, lifting the lid would break the logic state and interrupt the signals that fire the laser.

Lifting the lid is a safe and sure way to halt the laser, second only to pulling the plug.

The button is part of that circuit, but the logic contains a latching mechanism that the button doesn’t interrupt. So making the button a ‘true’ e-stop would require microprocessor cycles, and rely on the microprocessor to recognize it. This may not actually be desirable, especially if you are trying to stop it because of a microprocessor problem.

Which, is why I suspect GF hasn’t, and won’t, implement the button as safety stop.

EDITING TO ADD: The interlock on the Pro models is also part of that logic, and opening that circuit does effectively the same thing as lifting the lid.


#18

Thanks for all the support. I would observe, that lacking a supervisory system, moving the cutting head, or “Steve’s head”, requires the microprocessor. One should include the human factor in the emergency loop, in this case the human factor pressed the big button a couple of times hoping for the process to stop. I will retrain the human to lift the lid a little if this happens again.

I am disappointed with the WiFi capabilities, they seem a little quaint for a gizmo that looks that good.


#19

If you are running a filter even if the machine shuts down, the filter will continue until you shut it off. That does bring to question if the Glowforge filter has its own air moving device built in. just having the filters seems to not justify the price or time to produce, so I assume it also does.


#20

I think the filter has both intake and exhaust fans. It also communicates with the GF by IR I think.