There’s two different effects here, they’d be different charts. I guess I need to find this RS video that this chart came from.
Response time (bandwidth) is mentioned in the Cloudray MYJG power supply specs: “Response Time <=1ms”. I do not know what the tube itself would do if you drove the tube’s high voltage high and low faster. Semiconductor tech has no switching transistor that can control 28KV, it has to control a flyback transformer. I don’t have a way to make a faster drive to try.
RF-CO2 has a response time of 90us.
Response time is “ok if I give the drive a high logic or low level, how long does it take the beam’s actual power to come out and reach 100% or go back to 0%?”
This requires some highly specialized equipment to measure, but we can do it with the laser. This exercise ALSO illustrates what the laser machine’s process actually needs.
Say you’ve got a 0.2mm spot size. When the laser is moving, this is making dashes. We’d have a hard time seeing if a dash actually turns on and off if the spatial period of the burn is less than 0.2mm, although it does show up. If we have a 1500mm/s laser doing rastering, then that spot size is a time period of 133us.
If you raster at 1500mm/s and have a 0.2mm (or 133us) wide black pixel with two white pixels on either size, then a 100W RF-CO2 output will ramp up to hit 100W in 90us, sustain 100W for 43us, then the Ruida’s command to fire stops, and it takes 90us for the output to decay to zero. Since this is symmetric and the ramp is basically a straight slope, the pulse has a net energy of 13.3mJ (a joule is a watt-second) which is exactly as commanded.
However, the pulse’s energy is lagging slightly, and unclear in the 90us ramp area- but again, this isn’t really apparent when it’s less than a spot size in spatial terms.
If there’s only one white pixel then another black one, the fire command to the supply goes low for 133us. But the beam output takes 90us to decay to zero, then there’s a 43us period where there’s no command and no output, then the command happens and ramping begins.
The ability to see this scales with raster speed, and spot size. If I start engraving aluminum and change to a 1" lens, that’s half the spot size and now we need a response twice as fast to be able to discern dashes at 1500mm/s
The DC-excited tube is 1ms on/off response, so no way can it make 0.2mm dashes at 1500mm/s=133us. It’s barely gotten into ramping up power before it’s commanded off, and then ramps down and won’t get to 0 power during the white pixel either, so white and black pixels are both basically grey of only slightly different shades.
Another way to see this: forget raster. The machine’s PWM applies all the time. So just make a line long enough for it to get to the specified speed (1500mm/s) in the middle. Set Power to 25% for that layer, and then add PWM Override=1KHz, a 1.5mm spatial period and 0.375 mm on-time/1.125mm off. The RF-CO2 is essentially instant on/off on this scale, you get clear dashes of 100% power.
The DC-excited tube’s output cannot go full on/off inside this period and instead makes a ~25% power output the whole time, with “bumps” of power going up and down with a 1.5mm spatial period. It’s just a weak burn line.
There’s two different effects here, they’d be different charts. I guess I need to find this RS video that this chart came from.
Response time (bandwidth) is mentioned in the Cloudray MYJG power supply specs: “Response Time <=1ms”. I do not know what the tube itself would do if you drove the tube’s high voltage high and low faster. Semiconductor tech has no switching transistor that can control 28KV, it has to control a flyback transformer. I don’t have a way to make a faster drive to try.
RF-CO2 has a response time of 90us.
Response time is “ok if I give the drive a high logic or low level, how long does it take the beam’s actual power to come out and reach 100% or go back to 0%?”
This requires some highly specialized equipment to measure, but we can do it with the laser. This exercise ALSO illustrates what the laser machine’s process actually needs.
Say you’ve got a 0.2mm spot size. When the laser is moving, this is making dashes. We’d have a hard time seeing if a dash actually turns on and off if the spatial period of the burn is less than 0.2mm, although it does show up. If we have a 1500mm/s laser doing rastering, then that spot size is a time period of 133us.
If you raster at 1500mm/s and have a 0.2mm (or 133us) wide black pixel with two white pixels on either size, then a 100W RF-CO2 output will ramp up to hit 100W in 90us, sustain 100W for 43us, then the Ruida’s command to fire stops, and it takes 90us for the output to decay to zero. Since this is symmetric and the ramp is basically a straight slope, the pulse has a net energy of 13.3mJ (a joule is a watt-second) which is exactly as commanded.
However, the pulse’s energy is lagging slightly, and unclear in the 90us ramp area- but again, this isn’t really apparent when it’s less than a spot size in spatial terms.
If there’s only one white pixel then another black one, the fire command to the supply goes low for 133us. But the beam output takes 90us to decay to zero, then there’s a 43us period where there’s no command and no output, then the command happens and ramping begins.
The ability to see this scales with raster speed, and spot size. If I start engraving aluminum and change to a 1" lens, that’s half the spot size and now we need a response twice as fast to be able to discern dashes at 1500mm/s
The DC-excited tube is 1ms on/off response, so no way can it make 0.2mm dashes at 1500mm/s=133us. It’s barely gotten into ramping up power before it’s commanded off, and then ramps down and won’t get to 0 power during the white pixel either, so white and black pixels are both basically grey of only slightly different shades.
Another way to see this: forget raster. The machine’s PWM applies all the time. So just make a line long enough for it to get to the specified speed (1500mm/s) in the middle. Set Power to 25% for that layer, and then add PWM Override=1KHz, a 1.5mm spatial period and 0.375 mm on-time/1.125mm off. The RF-CO2 is essentially instant on/off on this scale, you get clear dashes of 100% power.
The DC-excited tube’s output cannot go full on/off inside this period and instead makes a ~25% power output the whole time, with “bumps” of power going up and down with a 1.5mm spatial period. It’s just a weak burn line.
But a totally SECONDARY response graph is what happens at lesser power. Now, an RF tube is NOT complicated- if it has a 90us response and a linear rise/fall, it does that. A 100W tube at 50% duty 5.555KHz it will produce a sawtooth power output with an average power of 50% of 50W. And we can go to 5% duty and still get the same ramp up/down slopes. If instead we ask for 5% PWM at 20KHz, we get about 5% power on average. 5W on a 100W tube.
DC-excited is mighty unstable arc in this area. The tube actually drawing current (like a current meter on an oscilloscope), the tube’s side glow, and actual lasing are different things.
At 20% PWM 20KHz, the DC-excited tube is 17% average power output (I have a great Coherent meter!). At 10% PWM, it’s basically no output. So, you’re like, I can still get 5% output somewhere by finding the sweet spot, right? Not really. The DC tube’s really inconsistent, you might use 12.7% PWM and get 5% power at one moment, then find the tube output can also get stuck at 0% on this duty- or could leap to 10% power. Without a feedback circuit that senses laser output, we just can’t keep it stable in this range.
To be more colloquial, this is like having a manual transmission- and you’re not allowed to use the clutch. You want to go slow. 10mph, sure. But can you creep at 1 mph? Once we start turning the engine below 1000 rpm, it’s really shaky and can die easily. There’s nothing you can do with the throttle or brake to get a predictable power output at 300rpm- it might die, it might surge to a clean 1000 rpm, it might bounce around from 200rpm to 600rpm and shake the car, but you can’t keep it stable at 300 rpm. So it can’t creep at 1mph. But an EV can.
Sorry for the long-windedness, I’m thinking through this myself, and thinking how I might script a video explaining it, so I’m going through all the thoughts. Bottom line on the second point, a 100W DC-excited tube can scale from 100W to about 15W… then it’s sort of an exponential decay to 0W with only a small change in duty, and it’s so unstable and unpredictable we can’t use it. You could in theory find that 10.3% duty produces 5W one moment, but unstable- the next moment it could produce 10W or 0W.
