This repository aims to contain a collection of information on the power supply within the Glowforge Pro.
These power supplies frequently fail, and there is little documentation that can assist when trying to repair them.
Scott Wiederhold has done a remarkable job at trying to reverse engineer the GF in order to produce an Open Source replacement controller and his OpenGlow forum contains a lot of photos, teardowns and information.
I have been trying to repair a broken power supply myself, and wanted to combine a bunch of information from various posts, as well as what myself and others found.
Tip
Please contribute here if you can. Any additional or improved information will be greatly appreciated by many.
Caution
This information involves working with lethal voltages (400-22,000 volts) and high power laser radiation! It is intended only for people who are sufficiently experienced in working with high voltage electronics and understand the necessary precautions.
Warning
The information contained here is provided on a "best effort" basis and is not guaranteed to be accurate. Consider this information to be a starting point where you should then use your best judgement and double checks before relying on anything.
Some links to relevant information, mostly on OpenGlow:
- Teardown of power components
- High voltage troubleshooting
- Testing HV output
- Analysis/photos of GF boards
- OpenGlow schematic
The power supply is quite difficult to get out, as it wedged into a tight corner and attached using cables with very little slack.
Ken S has made two videos of the removal process which is very involved and will probably take an hour. At one point the machine is flipped upside down - take great care to ensure the gantry is secure as you dont want it crashing to one side during the flip.
A lot of basic teardown information can be found in this post from Scott.
The power supply is a custom unit which provides several voltages required by the machine:
- 3.3v for logic and control board.
- 12v for various accessories (pump, fans, motors, etc)
- 40v for higher power components (more fans, peltier perhaps?)
- 12-22kV for the laser tube.
It also has some control signals which communicate over the ribbon cable. The 12v, 40v, and HV outputs are only enabled when the right control signals are set.
It's possible to control the PSU when outside the machine, per this post
Remove the screws attaching the top lid (where the ribbon cable and HV red wire are).
Remove the panel on the side by undoing the 4 screws holding it on.
In that panel you can access two cable bundles between the top and bottom boards. Unplug these from the bottom board.
Remove the top cover and attached board. Take care with the red and white wires. The red HV wire has a grommet that splits apart when the cable tie is removed.
The top board is screwed into the lid with 8 Phillips head screws. You will need to use a driver with a thin shaft to remove the center one as there is not much space. Be careful not to damage or crack the core of the flyback transformer.
These are the connectors on the PSU. You could create extension cables in order to run the machine with the PSU outside of it for testing.
Molex mini-fit jr 6-way is the large 6-pin connector to the daughterboard Digikey link
TE Connectivity Economy 2.5 4 way is the smaller one next to it Digikey link mating connector
Molex 30-way 1mm ribbon cable note top on on side bottom on the other Digikey link
This is the pinout of the ribbon cable:
The long red wire carries the high voltage to the laser tube. This is a common cause of laser failures, with the insulation failing and the power arcing through to the gantry earth.
There is loads of information online, just search for "glowforge red wire".
The high voltage transformer can fail. It seems a common failure mode is for the components in the potted section to burn out. The currently only know fix for this is to replace the transformer from a donor unit (the transformer is a custom unit and cannot be purchased separately).
It might be possible to repair the failed parts, but it is risky and would require cutting the parts out of the potting , replacing them, and re-potting it. A vacuum chamber would be needed at minimum to remove air bubbles which would otherwise enable arc-over internally.
[TODO]
[TODO]
TODO: Describe HV principle, operation and parts.
The driver daughterboard has some reverse engineering attempts documented here
Using an appropriate dummy load and sending the right control signals, the various outputs on the supply can be enabled for testing.
Scott has a post covering this procedure.
Setting the various signals on the ribbon cable can be tricky. I designed a breakout board which brings these connections into a regular header socket, and also acts as a passthrough to the control board for verify operation when installed.
Designs and Gerbers are available here.
The secondary windings on the flyback can become shorted. This will result in no or very weak output to the tube, and potential failure of the output Mosfets.
It's possible to test for such a short with a ring test. A ring test works by applying a weak square wave to a resonant circuit around the flyback core. On a good flyback, the circuit will "ring" and oscillate with a gradual dampening. If there is a shorted turn, this will absorb all the energy and the ring will diminish rapidly, much like a ringing bell being touched.
This test can be easily done using just a scope, a capacitor (~0.5-2uF), and a length of thin wire. It's not necessary to remove the flyback from the PCB.
The diagram below shows the testing setup. It consists of:
- A scope with a signal output, 1kHz at 1 volt (often the "cal" output)
- A capacitor (non-polarised, ideally film) of between 0.5 and 2uF
- 4-5 turns of insulated wire around the core. Thin, as there's not much space.
- Connect it all in parallel - the flyback does not need to be removed from the PCB
- Set the scope timebase to 200uS/div, and the vertical to around 2-10mV/div
A flyback with no shorted turns will look something like the left picture, whereas if it looks more like the right picture, its almost certainly got shorted turns and the flyback will need replacing.
TODO: Add info on extending cables outside the GF machine.