This guy seems to have lost his house and a pet because his printer

This guy seems to have lost his house and a pet because his printer caught fire. Take appropriate precautions - you are using high current, high temperature, high complexity, home built machines

http://www.soliforum.com/post/57749/#p57749

Use wires that can take the current you are pushing through them. Make sure wires to moving parts are not going to internally snap due to stress creating a conductor bottle next that causes huge heat concentrations. Have a smoke alarm above your machine - they are cheap, even free sometimes. Have a fire extinguisher that is appropriate for electrical fires to hand. (Not water!). Ensure your mains power has appropriate RCDs installed.

This is cheap, easy, basic safety. It’s worth it.

I’ve seen one scourched printer, luckily fire didn’t get far, but wood printers make me nervous for this reason (mines all aluminium). Even the plastic parts can light really easily, just take a lighter to the end of some filament to see!

This is a real shame, I’m ordering a fire extinguisher on Amazon now, despite them (CO2 ones) not being particularly cheap!

I expect Sanjay has seen pictures of my melted e3d, for everyone else - what happens when the thermistor goes wrong is the heat block melts and the Nozzle drops out (with molten metal on it) the print will continue with a molten heat block until it ends. The heatsink and fan do such a good job you can even re use the heat break when you get a new heat block and nozzle.

Marlin has an option for a second thermistor. Reifsnyder (J-Head) tested this. If the temperature difference between the 2 thermistors is to high the hotend is switched off.
2 sensors are ok but do not protect against a shorted FET.
A thermal fuse is the best way to protect.

Honestly, I think all-metal construction is also a good way to protect. Nothing to catch on fire, house doesn’t burn down. Whereas things like the UltiMaker 1 could be more dangerous.

Problems:
Printer gets hot, printer starts on fire, Thermal fuses on digikey are rated at 240 trip temp MAX.
Solutions:
Place a thermal fuse in the heat break, at some point where the temp should be quite low, even a point where if it increased it would cause problems with the filament. (IE on the E3Dv6 near the PTFE tube, maybe a 200deg C fuse)
Circuit breaker/fuse on the DC input (should be there anyway)

As a community, we really should come together and work on a board that has the following features:
1: a connector where current rating doesn’t depend upon assemblers abilities (no screw blocks, these are a fire hazard if not properly cared for)
2: Trace thicknesses and and copper-weight that are big enough to support excessive currents
3: fuses,
4: Quality FETs
5: Standardize on a obtainable power supply.
6: Start using proper EMI/EMC techniques
7: Opensource peer review these things.

Note, i am not a layout expert, or a power expert, i am an RF/EMC engineer.

@ThantiK , the printer in this case was all-metal.

I would be interested to know where EMI impacts safety, printer being affected by outside sources?

Heated bed connectors are what scare me, a very large number of printers exceed the rating.

Standardised power supplies are hard too, there’s very little out there with the power to run most heated beds, especially when you consider price!

My best advice, all metal hot end with secure thermistor (like E3D 6 and something I saw in @ThantiK 'a photos earlier than that), plus a well constructed (NOT GEETECH) electronics board that has good connectors. @LulzBot have some excellent stuff. Finally the firmware needs to be less configurable and more sensible, Marlin has HEATER_SANITY_CHECK but I have no idea if I have it enabled or if it even works in current builds. There is no great reason to run bleeding edge firmware, we should be giving people stable binary releases of Marlin for their electronics, which means take all the printer settings out of the source code and put it in flash.

I really want to make an ARM based board and firmware and make proper use of DFU mode.

@Bracken_Dawson I really feel like there needs to be a peer review system. Before you make your production run. A group of people, have a hang out session. Look a the schematics for flaws (not missing features). Things like “Trace thickness is too thin”, “grounding in not adequate” and “you need some heat syncing”… Right now it is someone makes a board, sends it off to get assembled, then we all tear it apart when it starts being sold.

Peer review would help us reach a better board faster.

Peer review is important, but equally so is just getting lots of hours on boards in various configs and testing. All the best boards have been through several revisions now.

I work on Storage controllers for a living, I’ve seen some astonishingly bad flaws on boards pass multiple levels of review and only be found because I have over 200 of them in test.

We need a team of people working on and reviewing the design, even if only one does the main board. Then we need samples in the hands of this communities experts, the people who can work with a crap GEETECH board effectively, as @Thomas_Sanladerer pointed out in one of his videos. Then we will have something we can give to a total newbie.

@Bracken_Dawson The 1000s of hours of testing will end up happening in the community (yes not all by one person, but we will get there), the peer review is something that is not happening, but should. even if you only invite a few of the members. When i hear “John Doe” is the PCB designer, i cringe… You need some reviews, and by people who are knowledgeable and motivated to achieve the same goal.

Wires that can handle the current you pass through them. Yeah…I got an idea for that. Use a standard PC power supply and have all of the different wires from it actually power individual things instead of powering a board that then re-divides the power. Is it insane to think in such simple terms? Two sets of wires could be sent to power the heated bed. Each sets of wires on the heated bed would be its own complete circuit connected to a MOSFET and the power supply.

All very good suggestions, who wants to start a peer review committee? Preferably on http://reprap.org

@Camerin_hahn @Bracken_Dawson If the boards are designed by people who know what they are doing it will be much safer. Most (if not all) current reprap electronics are designed by people who are not experts. People who only have some basic electronics knowledge and can use eagle think they are an expert. Look at a few current electronics board and a 10min review can show some fatal flaws.
(Examples of bad boards are DUET, Melzi, sanguinololu)
The current generation printers have many single fault conditions that can cause overheating.

@Nils_Hitze The mentality of the board users needs to change. Now they want the cheapest board. I reviewed some boards in the past and give some comment on other boards.
The reaction was always the same.
1st, They told me I was wrong, until I proved I was right.
2nd, They then claimed I can’t be a big problem because it works.

The problem is that the users want the cheapest board and the sellers want to make profit.
Safety is not an issue for both until it goes wrong.

PC power supplies are messy and a waste of space, I tend to use a server power supply with a custom loom now.

I’d rather use a cheap 24V supply and half the bed current.

Those of us that are Electronics Designers, and work in the European Union, are too aware that if we design hardware that injures or kills someone we are personally liable (and will likely receive no support from our employers). With that in mind, safety is a high priority in any hardware I design and, frankly, the minimalist design ethic of RepRap-style electronics makes me very nervous for safety of users. Combine that with the trend of operating printers remotely or unattended and you have a potential disaster in the making.

http://www.raspberrypi.org/forums/viewtopic.php?t=55641&p=432092 - Smoke Detector on Raspberry Pi. We could actually integrate that into @OctoPrint really fast

@Nils_Hitze I actually just thought about something like that yesterday when I read that post. I wouldn’t integrate it into OctoPrint though but better yet either into OctoPi or even better into some little watchdog circuit (with info pushed to the user(s) though via all possible means)

jupp - all in for it

@Nils_Hitze Smoke detectors are only useful if you are present to react to them. Better monitoring and safety features need to be put into hardware and firmware to deal with potential and actual failure conditions.