PDA

View Full Version : Wondering about upgrading Chinese x-axis from stepper to servo



john banks
02-23-2012, 6:46 PM
Would the Chinese laser tube respond quickly enough to make it worth attempting an upgrade to 80+ inches per second? Present high quality rastering is about a quarter of that.

This sort of kit caught my eye, partly whilst looking for what to replace my early failing x-axis stepper driver with...

http://www.motioncontrolproducts.co.uk/products/6/119/dsdlsm_servo_system/

The compromise option might be this:

http://www.motioncontrolproducts.com/drives/digital-stepper-drive-dmd556.php?cat=2

Michael Hunter
02-23-2012, 7:21 PM
From what I heave heard - no it won't respond quickly enough.
If you have to replace stuff anyway it might be worth a try though : you can always slow it down if it does not work properly at top speed.

Rich Harman
02-23-2012, 7:33 PM
I've been considering the same thing except I will change both axis when I do it.

You can get a GeckoDrive Stepper driver and stepper motor that is twice the power of the existing one. It would give you a resolution of .0005" and higher accelerations.

I'm thinking that the first best thing to do is to replace the belts with some high quality ones like these: http://shop.polybelt.com/3M-Pitch-Poly-Steel-Open-End_c95.htm

The belts stretching is the only thing I can think of that is causing my machine to be inconsistent.

If you want to go all out, switch to servos with linear encoder strips. Then it doesn't matter if the belts stretch.

Dan Hintz
02-23-2012, 8:11 PM
I question the ability of the Chinese power supplies and tubes to respond quickly enough to handle higher transit speeds...

Rodne Gold
02-23-2012, 10:14 PM
Apart from any other issues , The machines flying head is relatively heavy compared to others and the rail/bearing system is not optimised for high speed travel so you might not have the envisioned success. A stepper driver is $70 or so , a much cheaper option , you can most likely find a local solution if you cant wait from china. Might be you can get the existing driver repaired - can't see it using anything special inside.
No use making a Yugo a rolls , its still a Yugo....

john banks
02-24-2012, 3:29 AM
The failing board looks OK, and replacement like for like is cheap, but it got me thinking.

The x axis is much lighter than the y and would be cheaper to change but only help rastering.

Main question is laser and power supply response. When I increase speed from 400 to 800 mm/s there is a sudden deterioration of the vertical edges on rastered fonts but not regular enough for further backlash optimisation. However the edge is so crisp at 400mm/s I think the laser could go faster if the head could accurately as the initial limitation seems mechanical. Perhaps just better belts and higher spec motor driver could help a lot as you say Rich.

What material do you think would be good to try a 1000 DPI 50% dither pattern and still see every mark on the material when magnified?

Rodne Gold
02-24-2012, 3:42 AM
Anodised aluminium , clear acrylic , whatever. The issue is actually the beam size , you will be lucky to get 0.003" and thus just about any of our glass/RF tubed lasers cannot support more than 300 dpi , to get 1000dpi spot size has to be 0.001" which is not achievable

Uros Sovilj
02-24-2012, 5:13 AM
Yugo was an industrial horse of Jugoslavia, coutry that i live in before it fall apart.
I drive Yugo for several years and i t was a great car in that time:) you only need to replace machine few times a year:)

Rodne Gold
02-24-2012, 5:31 AM
Weren't some of the cars bodies made of compressed cardboard?

Uros Sovilj
02-24-2012, 5:33 AM
It was a TRABANT made in east Germany. I know for an incident when almost hole bodie was eaten by pig's:)

Dan Hintz
02-24-2012, 6:39 AM
The issue is actually the beam size , you will be lucky to get 0.003" and thus just about any of our glass/RF tubed lasers cannot support more than 300 dpi , to get 1000dpi spot size has to be 0.001" which is not achievable
With a standard 2" lens (and no collimator), 0.005" is about the physical limit of what's possible... on a Chinese system with a glass tube, I would expect more fringe effects, so 0.005" may be right at the bleeding edge, too.

you only need to replace machine few times a year:)


I know for an incident when almost hole bodie was eaten by pig's:)

You crack me up, Uros ;)

Uros Sovilj
02-24-2012, 8:57 AM
It's an urban legend but knowing a car wondering that could realy happened:)

john banks
02-24-2012, 5:41 PM
Thanks, based on the useful info above about spot size, I tested some (regular checkerboard type) 50% dither patterns on acrylic, and they look like solid (not dithered) at 300 DPI, a bit randomly less solid at 250 DPI and convincing dithered at 200 DPI, especially with a short focal length lens. However, when I speed up to 800mm/s the 50% pattern at 200 DPI looks fairly solid again like it did at 400mm/s at 300 DPI and above, although I've only tested with a long focal length lens so far. This does suggest that there would be no point going to 800mm/s even if you could control the axis movement more reliably because of the laser/power supply response time.

It also confirms why some dense dither patterns made at 500 or 1000 DPI don't have enough contrast with areas that have solid undithered rastering. Does it make sense though to render some graphics at this resolution or higher because whilst the spot size is a few times larger than each pixel, it could still make the edges sharper and less aliased because the time the laser is turned on can be controlled better with the higher frequency of information?

Oddly today without doing anything, the x axis stepper motor driver is working smoothly again, but a spare is on its way.

Great story Uros!

Rich Harman
02-24-2012, 8:43 PM
Were you doing the scanning both ways (X Swing) or unidirectional?

Rich Harman
02-25-2012, 12:40 AM
In about 6 weeks or so GeckoDrive will be releasing their G215 driver which does sub-micro stepping. I think I will wait for it to become available.

I also plan to switch to 25mm width steel reinforced belts.

john banks
02-25-2012, 3:44 AM
Bidirectional.

Rich Harman
02-25-2012, 4:18 AM
I think for the kind of comparison tests that you are doing, unidirectional would be more appropriate since the backlash will have a greater effect at higher speeds.

Rich Harman
02-26-2012, 3:53 AM
I performed a simple test today, engraved some text at 200, 400, 800 and 1,000mm/sec, at 20%, 40%, %80 and 100% power.

Out of the four, the 800 speed looked the best (brighter) but under very close examination you can see the scan lines not exactly lining up - almost certainly due to the belt vibrating. This was on my 1440 x 900 machine - long belt.

At speed 1,000 the stepper was skipping, I could not complete it. Even so, it was possible to tell that the laser was firing appropriately.

Some math;

At 300 dpi and 1,000mm/sec the laser must fire (or not fire) 11,811 times in one second. Put another way, it takes 85 microseconds to pass over each dot. A really coarse guess on my part allows 40 microseconds for the laser to fire, or to turn off. That is not a difficult task for electronics. The lost cost 80 mHz chips that I work with have a timing resolution of 12.5 nano seconds or about 1/300 of 40 microseconds.

So, electronics or software-wise, running much faster speeds should not be an issue. However, I do not know what the limitations may be in the laser tube itself as far as how long it takes to build up the charge in order to start lasing.

I am completely confident that 1,000mm/sec is practical and easily achievable with some simple upgrades to the motion components.

With that in mind allow me to reiterate that my primary reason for upgrading the motion components is for accuracy. Any speed increase is a bonus.

The numbers are 12mm high.

225218 225219

john banks
02-26-2012, 4:28 AM
I will repeat my 800mm/s 200 DPI 50% dither where every dot alternates but unidirectional and with 40mm lens and making sure I am not overpowering the acrylic. But even if this still produces an undithered result it would just mean we have to make our dither patterns coarser as most existing laser file prep does?

I don't like the edges of text at 800mm/s it sounds for the same reason where there is a randomness to the backlash requirement whether that is due to motion.

It would be nice to get the moderate sized tables to raster with higher quality vs speed.

john banks
02-26-2012, 4:39 AM
Thanks for the pics, mine looks similar at 800mm/s
Black acrylic? Solid not dithered? Unidirectional?

Rich Harman
02-26-2012, 5:12 AM
Black acrylic, solid, 63.5mm lens, unidirectional.

john banks
02-26-2012, 7:35 AM
225221
Top line is unidirectional, bottom bidirectional. 35 to 55% power as the speed increases. 0.1mm scan gap. Solid not dithered. 12mm high test. Black acrylic (Perspex), film removed. 5 PSI air assist through cone. 40mm lens, focused (tried defocusing and didn't help).

Excuse the distortion from macro image quite close. The bidirectional had a lot more dust to wipe off when it came out the machine.

The 400mm/s looks best, the higher speeds are atrocious.

Backlash settings are 0.2mm for 600mm/s and 0.3mm for 800mm/s. So this would apply an effective 0.13mm at 400mm/s and the bidirectional at this speed is good.

800mm/s with 75mm lens is too awful to show.

It is this splattering that limits our speed.

I would have thought if it was related to the laser then it would be twice as worse at 800mm/s than 400mm/s, but as you can see it is many times worse.

So would motion control make the 800mm/s look more like 400mm/s? We're not asking the laser or power supply to turn on and off quickly with solid blocks, we just need the time at which it turns on to be correct.

We also see this splatter effect on slate and wood at speeds over 400mm/s.

Don't get me wrong, I'm not at all unhappy, just trying to fine tune. 400mm/s is giving great results.

john banks
02-26-2012, 7:50 AM
Looking at this close up the unidirectional at high speed is sharp on one side of the vertical edges and splattered/dragged on the other but even at 800mm/s one edge is sharp suggesting good timing from the laser/power supply. On the bidirectional it splatters/drags in alternate directions on each scan line.

With air assist off it looks much better.

matthew knott
02-26-2012, 8:54 AM
out of interest, do these lasers have to run in a pulse condition all the time, i know co2's are happy running in a continues mode so why doest the tube just switch on for a period of engraving and off when a non engraved area? does a cw co2 not engrave well or do the tubes not like it? We have a glavo co2 and a epilog, the galvo is a modified yag (coherent gem tube) We just supply it with a TTL on signal and it seem to work fine. The epilog has the same type tube, but it seem to all run at a pulse frequency. Be interested in why this is?
Cheers

john banks
02-26-2012, 9:51 AM
I have tried:

Reduced or no air assist
Swapping stepper motor drivers with the y axis
Different length focal lenses
Fooling it into thinking it is running at 400mm/s but with half the step length and acceleration and speed values so it actually runs 800mm/s
Tightened up the x belt from slack in 8 stages
Tried with and without protective layer on the acrylic
Tried engraving with wider scan gap so we can see what is going on
More and less power

End result is that none of the above solve the issue at 800mm/s where it is perfect at 400m/s. Running unidirectional shows that often in the direction of travel the laser starts at the leading edge of a part of a character nicely and then will stop at the far end with a short gap and then a "splatter" of up to a mm, and then at the next leading edge it may then be off, but often will recover again. It seems that the positioning accuracy is not being lost through missed steps but stretching that temporarily and randomly upsets the rastering accuracy?

john banks
02-26-2012, 10:16 AM
Have also tried reducing (including dramatically) the take off speed and acceleration of the x-axis.

Rodne Gold
02-26-2012, 11:22 AM
Pulse rate of the C02 lasers used is 20k... 800 speed at your dpi might exceed this ? It can be changed but Shenui say don't?
Great experimentation tho.. You really working this thing over...

Rich Harman
02-26-2012, 1:56 PM
Interesting. I wonder if the splattering has something to do with your laser being 100 watts, I am not seeing the same issue.

Somewhere I read, and now I cannot find where, that the tube requires at least 20K.

john banks
02-26-2012, 2:23 PM
I changed to y unidirectional scanning and the splattering is dramatically reduced and it ran for a while before the stepper started missing steps at 800mm/s (on the x-axis I loosened the belt to accommodate this!), but even with the same acceleration settings on the x axis (x is higher by default presumably because of its lower mass), it has problems. It has a longer belt but Rich's belt is longer still with less of an issue.

Have also done a 200 DPI alternate pixel dither (so 100 cycles per inch) at 800mm/s and I can see there is dithering and you can see alternate spots (although the overall impression of density is much higher than 50%), using 40mm GaAs lens for reputedly smaller spot size, but it does appear that the laser can make some effort at rendering each pixel whilst moving at 6300 pixels per second.

So I'm getting more convinced again by the merits of upgrading the sloppiness in x-axis control. I don't think the massive degradation in quality between 400 and 800mm/s that I see on my x-axis is due to the laser as it would then show "errors" twice as big, and these are massive as though something mechanical is wrong, there are stray dots engraved a millimetre beyond where they should be, and a fraction of that is obvious and would be on the 400mm/s rastering.

Would be a shame to upgrade and still have sloppy belts though, so need to think about the options from a simple controller upgrade, through a controller and servo, with or without an encoder strip.

My stepper drivers are set at 4.45A peak. Everyone else the same?

My steppers say 57BYGH350C-03-32. http://imexcn.com/pdf/Stepping%20Motor/HR-57BYGH350.pdf shows a 57BYGH350C as rated current 5.2A. My supplied drivers will go up beyond 4.45A to 5.2A and 5.8A.

john banks
02-26-2012, 2:53 PM
So I turned it up to 5.2A and it improved nicely. I tried to tighten the x axis belt but it still starts to miss steps if I do. So I loosened it again. Then increased the stepper peak to 5.8A and it is now looking like Rich's 800mm/s engraving...

This is looking like an exceptionally good target for upgrading now.

"The rated current is what the motor is rated at. The peak current refers to the amount of current the driver outputs.Non-microstepping drivers
Peak Current = Rated Current
When using a driver that only does full stepping, the rated current is the same as the peak current. (Rated current = Peak Current).
Microstepping Drivers
Peak Current = 1.4 x Rated Current
When using a driver that is capable of doing microstepping (microstepping = 1/2, 1/4 stepping or more), the definition of peak current becomes 1.4 times the rated current. Microstepping drivers are made differently in order to maximize their ability to drive the stepper motor. Therefore, step motors can handle up to their rated current multiplied by 1.4. (Peak Current = 1.4 x Rated Current). This will not damage the motor because the power output is more or less the same."

The 3M660 is listed as a microstepping driver, therefore a sensible upper limit would seem to be 1.4 x 5.2 = 7.28A so maybe I'm not overdriving them at all at 5.8A?

Rich Harman
02-26-2012, 3:21 PM
Mine are at 3.2A and 3.8A. I forget which is which.

john banks
02-26-2012, 5:55 PM
I am running around in circles and vulnerable to placebo effects.

I thought the extra current to the stepper was helping but careful
A-B-A tests showed otherwise.

It seems I am dramatically overpowering everything (laser power) but
it only shows at higher speeds where I tend to increase the power in
proportion to the speed but on some materials like acrylic and slate
it does the splattering. I dropped the power from estimated 55W to 15W
in many stages, finding that 10W was too little and now the splatter
has gone. At 600 and 800 mm/s I now just have a little feathering of
edges of engraved text but in proportion to what you would expect now and like Rich. The 800mm/s may or may not be sellable depending on the price, file, quantity, material etc. I hope Rich's upgrades make everything sellable at 1000mm/s.

All afternoon chasing other things because I didn't realise you could get away with overpowering things at 400mm/s but not 800mm/s. Any ideas why?

I keep learning this lesson repeatedly, I hope it sticks now.

Rodne Gold
02-27-2012, 12:51 AM
We have timed thruput and were often quite shocked to find SLOWER speeds are faster - especially with small stuff. The problem with running large jobs at the machines limits or beyond is that the chances of ruining a large piece is also increased - miss a step and its all gonners. Chinese are majorly competitive , they wont "give away" a distinct speed advantage over others if it was really reliable.
My thinking on these machines is that you can afford 2 or 3 of em at 1/2 the price or less than alternatives , would rather have multiple machines reliably plodding along.
This however takes nothing away from your research - you guys are at the face and your optimization will help all others.

Dan Hintz
02-27-2012, 8:07 AM
Looking at this close up the unidirectional at high speed is sharp on one side of the vertical edges and splattered/dragged on the other but even at 800mm/s one edge is sharp suggesting good timing from the laser/power supply. On the bidirectional it splatters/drags in alternate directions on each scan line.
For the unidirectional, the crisp lines show good initial timing with the tube... the "splatter" where the tube should be off shows it's not so easy to get the control you want. The bidirectional "splatter" is showing the same "off" issue, just in both directions and muddles the results.

john banks
02-27-2012, 12:36 PM
225415

Here is the latest. Unidirectional. 800mm/s. 30% power.

So why is it so sensitive to power at 800mm/s? At 400mm/s you have a much wider acceptable range of laser power for good results.

matthew knott
02-27-2012, 2:10 PM
The tube ramp up and decay times are going to be longer at the higher powers, also they will be prone to being more variable as the plasma decays. If you kill the power to one of these tubes the laser does not just instanly switch off, same with powering up. You need a pyroelectric detector and oscilliscope to look at the beam and the laser gate signal (i have one somewhere) this would show you the relationship between when you tell the laser to turn on/off and what the laser is actually doing. You getting into the limitations of the tube/power supply. diffusion cooled rf lasers (as you find in "western" lasers) are going to be much better at following the input signal IMHO but they cost 20 times the money. Horses for courses as they say. As Rodney says the price is so cheap that being a bit slower is not a deal breaker, 3 lasers will be quicker than one "western" laser and still cost less.

Dan Hintz
02-27-2012, 3:01 PM
The tube ramp up and decay times are going to be longer at the higher powers, also they will be prone to being more variable as the plasma decays. If you kill the power to one of these tubes the laser does not just instanly switch off, same with powering up. You need a pyroelectric detector and oscilliscope to look at the beam and the laser gate signal (i have one somewhere) this would show you the relationship between when you tell the laser to turn on/off and what the laser is actually doing. You getting into the limitations of the tube/power supply. diffusion cooled rf lasers (as you find in "western" lasers) are going to be much better at following the input signal IMHO but they cost 20 times the money. Horses for courses as they say. As Rodney says the price is so cheap that being a bit slower is not a deal breaker, 3 lasers will be quicker than one "western" laser and still cost less.
I'm going to let you answer for me from now on, Matt ;) Excellent description of the problem...

john banks
02-27-2012, 4:04 PM
Fantastic info and ties in exactly with what I'm seeing. Thanks!

The splattering on turn off seems to occur over 40% power. May be a feature of my tube since Rich's 80W doesn't do it at 80% power.

Should I try my other tube and/or power supply? Otherwise they will just sit for however long this one takes to die, and it would be nice to find out how much variation there is between examples of the same product.

matthew knott
02-27-2012, 5:19 PM
I would think the power supply is more likely to be the weak area, fast "western" laser has a rise time of about 150us, the chinese machines seem to be in the 1-2ms range (for an alleged fast supply). Maybe the reci supply and tubes are better (is that what you have?). Dont give up just yet, its really intersting to see what you can get out of one of these machines. Race tuned chinese lasers, love it :)

john banks
02-27-2012, 5:39 PM
Last year GTRs, this year Chinese lasers. Next year, Yugo Turbo? ;)

Yes it is RECI tube and PSU. Not sure if the different tubes have different PSUs or whether they just turn a few pots. One of the pots is revealed externally. Documentation claims "<<1ms" response time but I don't know if that is to fire up the laser from "off" or to pulse it rapidly once it is already "on". Suggest not the latter which is much faster since I can see the dither pattern rendered at 6300 pixels per second from a test I mentioned earlier.

matthew knott
02-27-2012, 6:00 PM
1ms at 1000 mm/s is 1mm, (gcse maths) is your error about that, bet there are good and bad tubes and good and bad power supplies, (as with all things) but @ 1ms rise time 1000mm/sec is really pushing your luck. Nothing wrong with that mind

Rich Harman
02-27-2012, 6:45 PM
If you look at the picture I posted near the top of page two (and below) you can see three places where the laser fired a dot after after the character at 800mm/sec 80% power. The line of text was 13 characters long, the only place that it fired late like that was on the "8" which was the first character. The laser fires left to right.

The power supply that came with the machine appears poorly constructed, the sheet metal bends for the case are not well done. It does not look like the Reci power supplies pictured on the Reci website. It doesn't mean the internals of the power supply are sub standard but it does make me wonder.

http://www.sawmillcreek.org/attachment.php?attachmentid=225219&d=1330248443

john banks
02-27-2012, 6:53 PM
It is much finer than that. 1ms at 1000mm/s would mean you could only have 25 DPI? I can see the holes in 200 DPI with alternate dots on/off rastered at 800mm/s. The fuzziness of font edges is up to about 0.1mm at the same speed I think, need to do some measurements on that "800" image above which has 12mm high characters.

john banks
02-27-2012, 6:54 PM
Rich can you try less power at 800mm/s and see if your splatter disappears? I think my limit is about 40%.

Rich Harman
02-28-2012, 12:31 AM
I tried it again, 800mm/sec and 40,60,80 and 98% power.

40% is terrible, the laser doesn't develop full power for the first half a mm or so.
60% is better but still shows ragged leading edges. The trailing edge is fine. There are several splatters after the first two characters.
80% is good, a few splatters after the first character.
98% is best. Zero splatters.

I'll try to get some pictures...

I have a bellows attachment for some super macro shots - do you want?

john banks
02-28-2012, 2:53 AM
Not sure what the bellows involved, whatever you think shows it, the previous shots were great.

Interesting pattern though!

Rich Harman
02-28-2012, 6:48 AM
Bellows are used between the camera and a lens in order to separate the two. It allows for greatly magnified pictures.

matthew knott
02-28-2012, 7:47 AM
I thinks its the initial laser on, not the pulse to pulse stabilty thats causing you the problem, try marking vertical lines starting really thin, then for each stripe make it a little wider, also try marking thin verticle stripes with very close spacing and some with much larger spacing, this will show you alot about how the laser performs at the high speeds.

john banks
02-28-2012, 12:16 PM
The splatter I'm getting is as the laser turns off when I use too much power. It sounds like Rich is reporting it as the laser turns on. I will retest to be certain I have the direction correct but my "on" stroke when unidrection is from right to left when I checked before.

Stepper motor driver which is now on the y-axis went rough again last night (I never turned up the power on this one), but the replacement has arrived from China today, so will fit and test more.

john banks
03-02-2012, 7:51 AM
Still need to test more to confirm the direction and the power at which I get the splatter.

I did read in my searching that RDCAM supports a max engraving speed of 90 metres per minute (1500mm/s)

john banks
03-05-2012, 7:54 AM
A few more thoughts on this to see if the splatter at higher power (which marks the material at high speed better) can be mitigated which I intend to try.

1. Try minimum power at a lower value than maximum power (although the manual suggests this won't make a difference).
2. Try the "special" engraving mode that has a default 98% max power (I've lowered it to 90% already) that is designed for use to flatten areas. The manual is vague.
3. Try the stamp engraving mode to ramp up and down the power on edges.
4. Try a tickle pulse embedded as gray values of say up to 5% in the image and use "output direct" which varies the PWM signal for power which is how the 3d engraving was done. Try to deduce whether TTL is being bit-banged in this mode or whether it stays on during the entire scan line. This may or may not improve stability and responsiveness of the tube.

It could be that a 100W Chinese tube is just too blunt an instrument to engrave with high power and speeds over 400mm/s, but most of our intended work is cutting but the early requests we're getting are engraving! One of the above might work, and then I might try my other tube and power supply to see if they are different.

I also want to investigate the "dot" setting more when cutting. My first attempt didn't cut as well as "cut" but I need to play with the settings more.

Dan Hintz
03-05-2012, 9:43 AM
It could be that a 100W Chinese tube is just too blunt an instrument to engrave with high power and speeds
The bigger the tube, the more "oomph" you need to get it to lase, tickle or no tickle. This means a tighter-spec power supply to handle that surge of power, a larger surge of power to create a long tube of plasma, etc. It's diminishing returns the bigger you size the tube, particularly if you can't control (and tighten) the specs on everything in the control line (e.g., power supply).

john banks
03-05-2012, 6:20 PM
1. Try minimum power at a lower value than maximum power (although the manual suggests this won't make a difference).
2. Try the "special" engraving mode that has a default 98% max power (I've lowered it to 90% already) that is designed for use to flatten areas. The manual is vague.
3. Try the stamp engraving mode to ramp up and down the power on edges.
4. Try a tickle pulse embedded as gray values of say up to 5% in the image and use "output direct" which varies the PWM signal for power which is how the 3d engraving was done. Try to deduce whether TTL is being bit-banged in this mode or whether it stays on during the entire scan line. This may or may not improve stability and responsiveness of the tube.


1&2 make no difference.

3 helps slightly but if you go for a too aggressive ramp the engrave is not as deep as you'd expect from the peak power level it is supposed to hit. Too wide a ramp and the edges aren't crisp which is kind of what it is desined for - stamps.

4 is not fully tested so far, but at 4% the tube is making a mark on acrylic, at 3% it is not. With a 1 bit image rastered in this output direct/3d mode, the speed needs to be REDUCED to get sharp edges as it seems to only vary the power of the tube and not bitbang the TTL line, and the response is slower. This is a shame, I hoped it would turn off the TTL when the intended power was minimum or 0, but it leaves it at minimum power for the entire scan line and TTL is on. If I engrave text in this mode and go too fast then you get a pseudo backlash effect if bidirectional and fonts are widened considerably as the tube ramps up and down too slowly. The maximum useful speed in this mode is 200mm/s.