Project

General

Profile

Modifications for high speed PWM and laser control outputs

Added by HackerspaceFFM almost 5 years ago

Hi,
since quite a while we have converted our china K40-III laser cutter to Laos. As we killed the original laser tube and with it the original laser power supply as well, we first had to buy replacements. We selected not to buy the same open-PCB power supply that also needs this big ballast resitor, instead we went for a boxed one that looks more like a PC power supply. It seems that those boxed laser power supplies are pretty common and available for a big range of different laser tubes, starting from 40W to more than 200W and those are not requiring that ugly ballast resistor.

attachment:NewTubeWSupply.jpg!

In one manual of those power supplies I found a hint that the laser power can be adjusted either using an analog voltage or a PWM signal at about 20kHz. As our power supply is able to deliver a little bit more than the maximum of 18mA allowed for our CO2 laser tube, I wanted a circuit where I can limit the maximum current independant of the PWM signal provided by Laos. I also read a lot about signal integrety issues (= speed issues) and hooked up a scope as well: I can confirm that the standard opto-couplers that are used on the Laos board are too slow to provide good PWM signals at a rate of 21 kHz that are required by the power supply. So I've done and successfully tested the following modifications done to Laos that provide good high-speed opto-decoupled PWM and Laser trigger signals as well as hardware settable current limit. I like to share this circuit with all Laos users, hopefully that for the next Laos board revision this option will be part of the board:

attachment:LaserPowersupplyConnection.png!

I have build this mod on a small proto-board. You should change two resistors on the Laos board as well (from 330R to 220R) to provide a little more power to drive the LED in the optocoupler within specifications (checked this: also not an issue for the mbed). With this modification in place I can use the trimmer to set an absolute maximum laser power that is used for 100% PWM setting. I also checked the signal with a scope and can confirm that they are much much faster and even 50kHz or more should be not an issue with this circuit. Keep in mind that this circuit needs 5V at the side of the laser power supply to work - but most laser tube power supplies provide this. Don't use the 5V of the Laos here because this would render the galvanic isolation between Laos and laser tube supply useless!

attachment:HighSpeedPWMMod.jpg!

Have fun.


Replies (2)

RE: Modifications for high speed PWM and laser control outputs - Added by jaap almost 5 years ago

Why are you also using an external optocoupler for laserOn? The current optocoupler is ok for that, right?

There is also an option to export laserPWM on output J27-01, that you could use instead of piggybacking on the existing optocoupler (remove resistor 31).

RE: Modifications for high speed PWM and laser control outputs - Added by HackerspaceFFM almost 5 years ago

The optocouplers currently used in the LAOS design are damn slow - I saw considerable signal issues at even a few kHz. I was afraid that this could have an impact when engraving raster images at higher speed and DPI settings, so to be on the safe side I just fixed both signals. Those xx8x7 optocouplers are not designed for higher speed, that's why I have to go for a different one. The 6N137 is a pretty good available high speed optocoupler. There are versions with two channels in the same size like xx2630/xx2631 but those are a little harder to get, therefore I recommend 6N137 (albeit many high speed optocouplers like xx2601, xx2611, xx2630, xx2631 migth work as well).

My Laos kit arrived with no transistors for Q1/Q2 and as the optocouler has a socket it was just the easiest way for me to use this socket instead J27 (I want to avoid an additional transistor as this can also have negative effect on signal speed [a push/pull stage can turn of the LED in optocoupler more quickly than a single transistor like Q1]).

    (1-2/2)