Category: Science

Laser cutter controllers generally set the tube current (and, thus, beam power) through a digital PWM signal to the HV power supply. Confusingly, the same power supply input terminal can receive an analog signal controlling the output current. Both signals have the same 0 to 5 V range.

I have yet to see a PWM frequency spec for any HV laser power supply, although surely there must be one. The specs for the Cloudray power supply on my shelf seem typical:

I have no spec sheet for the replacement power supply OMTech sent, which is now installed in the laser and is measured below. I believe all similar HV laser power supplies, regardless of the nominal brand, are essentially the same inside and will have similar, if not identical, behavior.

Controllers descending from the GRBL codebase default to a PWM frequency of 1 kHz, a legacy of using the PWM output for spindle motor speed control. GRBL’s Smoothieware descendant has a configuration option for the period in microsecond steps with (I think) a default of 20 µs = 50 kHz. Ruida controllers similar to the (Ryxon) KT332N in my OMTech laser (seem to) default to 20 kHz frequency:

The laser frequency is used to set the pulse frequency of the control signal used by the laser. The glass tube is generally set to about 20KHZ

KT332N Manual, p 55

Knowing how a dozen measurements outweigh a thousand opinions, I recorded the power supply output current as a function of PWM frequency. The test setup is the same as for the original series of current measurements, with oscilloscope traces arranged thusly:

• 1 unused (yellow)
• 2 L-ON laser enable, low active (magenta)
• 3 PWM signal (cyan)
• 4 tube current – 10 mA/div (green)

I set the KT332N controller for a 200 ms pulse when poking the front-panel button, which is long enough to show any interesting behavior, and changed the PWM using its awkward controller interface. LightBurn provides access to the “vendor settings” which include the PWM frequency, which I set as needed:

So, we begin by varying the PWM frequency with a constant 50% PWM …

The default 20 kHz:

The upper half of the scope screen shows the entire 200 ms pulse, with the small slice near the middle appearing zoomed across the bottom half. The readout just above the buttons along the bottom gives the measured PWM percentage and frequency. The green trace shows the tube current is about 12 mA, half of the power supply’s maximum 25-ish mA.

The Tek current amplifier has plenty of thermal drift that I have not attempted to compensate, so always eyeball the average current with respect to the baseline around the pulse in the upper half of the screen.

No trace of the 20 kHz PWM signal appears in the tube current, which runs at a constant 12-ish mA for the duration of the 200 ms pulse.

Increasing the PWM frequency to 100 kHz (!) produces no change, although I cranked up the zoom timebase to better show the PWM pulses:

Reducing the PWM frequency to 10 kHz produces very small ripples in the output current corresponding to the PWM cycle:

At 5 kHz the tube current becomes sinusoidal, with an average around the same 12 mA produced at higher frequencies:

The sine wave current is about 90° out of phase with the square wave PWM, although much of that must come from delay through the entire power supply, rather than just an RC low-pass filter.

At 2 kHz the tube current takes on a decidedly lumpy look:

At 1 kHz there’s definitely something odd, perhaps a resonance, going on inside the supply, although the average current remains 12 mA:

At 500 Hz the PWM is slow enough that the tube current resembles the output of an integrator, rather than a filter:

At 100 Hz, the digital PWM signal is so far below the filter cutoff that it’s behaving as an analog input, with the tube current ramping between minimum and maximum:

The current has regular full-on glitches halfway through the “off” part of the PWM signal, so running at absurdly low PWM frequencies does not prevent them. Also note that the PWM signal does not control the current at the same speed as the L-ON enable signal, due to the low-pass filter rolling off the transitions.

Now, holding the PWM frequency constant at (the absurdly low) 100 Hz and varying the % PWM duty cycle …

At 30% PWM, the output current becomes triangular due to the low-pass filter:

At 99% PWM, the output stays at the power supply’s 24 mA maximum output, with small downward ramps marking the 1% off times:

Some observations for this HV power supply, which seems typical of similar supplies sporting other “brand names”:

• A PWM frequency below 10 kHz introduces output current variations due to the power supply interpreting the PWM waveform as a somewhat analog input, rather than a purely digital signal. This effect increases as the frequency decreases.
• An Arduino-speed digital PWM near 1 kHz will be interpreted as an analog signal, with the tube current varying significantly around the PWM signal’s average analog value. It does not control the current in an on-off digital manner.
• Due to the effect of the low-pass filter, the PWM signal cannot switch the tube current between “full off” and “full on” at any frequency. The current will always follow a ramp with a slope controlled by the filter rolloff, so low PWM inputs will have low peak currents.

I must switch to the controller’s analog output …

The LightBurn forums have many despairing posts from folks with CO₂ lasers sprinkling random dots all over their engravings:

• Weird dots all over my engraving
• Dots appear all around engraving
• Random dots when engraving
• Just search for engraving dots to find many more

Well, as it turns out, engraving lots of small test patterns on scrap acrylic and peering at the results revealed the same problem:

The test patterns were engraved at various power levels, which was the whole point of the exercise: I was looking at the current waveforms, rather than the acrylic. Despite that, the result should be solid blocks with no speckles in between, which is not quite what happened.

For reference, the test pattern:

An early hint came from a trace captured while looking at an entire scan line across the pattern:

See that isolated spike left of center, where the L-ON signal (magenta trace) is high? That shouldn’t be possible.

Setting the scope to trigger when the L-ON signal is high (= laser power supply disabled) and the tube current is more than a few milliamps (= laser beam active) captures those errant dots.

Sometimes a spurious pulse happens just after L-ON goes high to disable the HV output:

The X axis stepper DIR signal (yellow trace) shows the laser was scanning right-to-left, so the glitch will be just to the left of the 2 mm block in the pattern. In point of fact, it’s about ¾ of the way down the right-hand column:

A closer look shows a distinct circular pit at the end of the line:

The two left-to-right lines bracketing that line also show how the high-intensity pulses affect the laser beam startup intensity during a scan line.

Sometimes the glitches happen quite some time after the laser turns off:

Sometimes they’re in the middle of what should be a blank space:

The glitches are not always full-scale events. The two nearly invisible pulses just to the right of the block (bottom green trace) make the smaller dots you can see on the targets:

As far as I can tell, spurious dots happen most often with current levels around 20% PWM, less at 10% PWM, and rarely above 30% PWM. I think it has something to do with the chaotic spikes that the power supply produces at lower currents, instead of the relatively stable outputs for higher currents.

Although these measurements are for the replacement HV supply I got when the original supply failed, I saw similar chaotic waveforms with a Cloudray HV supply I bought as a backup. Given that other people have reported similar random dots with many other machines & power supplies, I think these scope traces show where the dots come from: all the power supplies behave the same way.

The only way to reduce the number of speckles is to use higher power, which will require higher scanning speeds to achieve similar results. Unfortunately, higher speeds give the power supply less settling time, so there may be no good answer.

I haven’t been able to find any “official” schematics for the HV laser power supplies shipped in typical lasers (there are many terminal wiring diagrams), so I have no idea how the L-ON signal controls the output current. Apparently the oscillating chaos inside the power supply occasionally punches through the output switch, which isn’t too surprising given the voltage and power levels in there.

If nothing else, the acrylic test pieces look pretty on the microscope positioner:

In the usual techie sort of way …

Just to see what the laser tube’s output looks like, I aimed a large photodiode toward the laser tube output:

That’s a venerable PIN-10AP photodiode minus its green human-eye filter, with an IR-pass / visible-block set of gel filters taped on the front to knock out everything except IR scattered from the laser’s snout. Nothing sits in the direct beamline.

The alert reader will kvetch about a CO₂ laser running at 10.6 µm, an order of magnitude off the right end of the photodiode response curve graphs, through stage filter gels not even pretending to have optical specs. Hey, stage light filters are utterly transparent to thermal IR and there’s plenty of invisible light to go around, so maybe this will work.

The coaxial cable trails off to the scope’s 1 MΩ input, so, although the photodiode does not operate in true zero-bias mode, I can at least look at its photocurrent driving a voltage into the scope input.

Surprisingly, the lashup kinda-sorta works well enough to show the laser’s light output tracking the tube’s current:

That’s a manual 20 ms pulse at 90% PWM, with the tube current at 20 mA/div. The oscillations at the start of the current pulse seem to excite the tube enough for the light output to stabilize when the real current comes along. I cannot tell if the exponential tail-off beyond the pulse is due to excited molecules cooling off in the laser tube or the poor photodiode recovering from Too. Much. Light. It. Burns.

The response is a little shakier at 50% PWM:

Dropping to 30% PWM requires more time to get up and running:

And 10% PWM looks downright awful:

Although the vertical scale for the photodiode trace doesn’t mean much, it’s obvious that the IR output matches the current input, right down to the littlest pulses. Sliding a bit of brass shimstock between the filter gels eliminates nearly all the photodiode output, so it’s not electrical noise. I think the long tail really shows the gases cooling off.

The alert reader will have noted the wee blip over there on the right, 21 ms after the start of the 20 ms long pulse and 4 ms after all those spikes shut off. Yup, the HV power supply can deliver a stray pulse when it’s not supposed to be enabled. More on that in a while.

With the laser cutter set up as before and the scope set up to calculate the time integral of the tube current, this happens:

The trigger is the Boolean AND of the top two traces:

• DIR signal = low = left-to-right X axis motion
• L-ON signal = low = laser power supply output enabled

The bottom trace is the laser tube current at 10 mA/div, which is, conveniently, also the scope vertical axis calibration, so you can read “amp” wherever you see “volt”. The four pulses correspond to a single scan line through the usual test pattern:

Scanning at 250 mm/s, each 1 mm block occupies 4 ms and the 2 mm block on the right is 8 ms long.

With all that in mind …

The white line in the scope screenshot is the time integral of the current, scaled at 50 µV·s/div and read as 50 µA·s/div due to the 10 mV/div = 10 mA/div equivalence. The integral is pretty much a straight line up and to the right during each pulse, showing that the power supply delivers a nearly constant average current despite the random-looking spikes in the shorter pulses, the oscillations at the start of the longer pulse, and the reasonably flat section after the current settles down.

Say it again: Totally did not expect that.

A closer look at the first pulse of a different line:

Today I Learned: The per-division scale of the white integral line is completely bogus in Zoom mode. Although the display still shows 50 µA·s/div (magenta text obscured near the middle), the integral line rises seven divisions. Some fiddling around showed there is no relation between the calibrations of the normal display and those in Zoom mode.

What is important: seeing the integral as pretty much a straight line with a reasonably constant slope, if you’re willing to ease over the bumps due to the current spikes. The slope of that line (yeah, the derivative of the integral, for a chunky definition of derivative) gives the average tube current.

Referring to the non-Zoomed trace, the integral rises by about 40 µA·s during each 4 ms pulse (and twice that for the 8 ms pulse), for an average current of 40 µA·s / 4 ms = 10 ma. Having previously established 100% PWM corresponds to 24 mA, 40% of 24 mA = 9.6 mA seems about as close as one might expect.

This looks less awful than I expected, fer shure.

More measurements are in order.

Having established that the RMS value of the huge current spikes at low PWM settings doesn’t amount to anything meaningful, I cranked the AM502 current amp gain to 10 mV/div, re-ran the tests for PWM values from 10% through 99%, and recorded the RMS value of a single line through a square of the same pattern:

Each square is 1 mm on a side and the pattern runs at 250 mm/s, so the laser will be enabled for 4 ms. For example, the test setup shows the result of a pass at 50% PWM:

The two cursors mark the duration of one block, with the laser current in the bottom trace starting off with the usual off-screen spikes, then settling down to a constant-ish 13-ish mA for the rest of the block. The 13.74 mA_RMS value (the AM502’s 10 mA/div matches the scope’s 10 mV/div, so you can read mV as mA) includes some part of those spikes (the higher gain clips the tips), but most of it comes from the stable-ish portion.

The whole measurement set as a slide show for your amusement:

When confronted with data points, plot them:

Huh.

I expected the line to pass through the origin, which it most certainly does not. One could make up a story about how the 30% and 40% PWM points are Close Enough to the line to sorta pull the bottom end over to the left a little, but even that doesn’t explain the known-to-be-weird results below 30% PWM.

A better story might be that 30-ish% PWM produces the minimum current required to fire the laser tube. Operating below that current works, in the sense that the laser produces a beam, but it’s out of spec. Running above that current eventually lets the power supply reach an agreement with the tube as to the operating point.

As before, those measurements do not account for the reasonably consistent results of scorching some cardboard:

Cardboard is not the best test medium and I now agree RMS isn’t the best measurement.

More study is indicated …

Laser power settings of 10, 20, and 30% obviously produce different results:

However, the scope traces for PWM values under about 25% all look pretty much like this:

Rather than a simple constant current source, the power supply produces very high amplitude current pulses for low PWM inputs, with no visible differences between any of the PWM values.

The scope can compute the RMS value of (a section of) the trace, so I aimed it at traces captured from the upper left block of this test pattern:

Because the pulses have such a high amplitude, I set the Tek AM502 current amp at 100 mA/div to capture the entire pulse. Measuring a part of the trace without a signal gives the baseline noise level:

The scope display is 10 mV/div, so 1 mV_RMS (close enough to the 894.4 µV reported just above the bottom label row) means 10 mA_RMS of noise. Given that 100% PWM corresponds to about 25 mA (DC-ish during the pulse), the RMS numbers may not have any significant figures.

A slide show of the results so you can page through them:

The RMS value comes from the trace between the A and B cursors.

Extracting the numbers:

• 0% PWM → 1 mV → 10 mA_RMS
• 10% → 2.3 mV → 23 mA
• 20% → 2.0 mV → 20 mA
• 30% → 3.0 mV → 30 mA
• 40% → 2.3 mV → 23 mA

Which says I’m measuring either too much of the wrong thing or not enough of the right thing: there may be no baby in this particular bathwater.