Ridges on cut edges using Shenhui SH-350 (50W Chinese laser)

[Storen]

If it is PSU/modulation wouldn't the effect be the same on X and Y axis? On my machine the ridging is more pronounced at the diagonal areas of the round shapes (see the attached sample). And it repeats itself at the exact same location every time. Shouldn't the location be more random and the effect more uniform if it is related to the PSU?

[RogerClark]

If it was laser modulation I'd expect the same amount of ridging in X and Y.

Actually, I cut some acetate sheets today and strangely I'm getting much better cuts at 45mm / sec than I'm getting at 30mm / sec.

I wonder if what we are seeing is the effect of vibrations in the final optics / lens assembly, i.e it will have an oscillation frequency at which vibrations have most effect, (this may be different in X and Y - I'm not sure).

Do the higher cost machines have better tracks of better bearings etc or perhaps they have damping in connection between the X carriage and the final optics assembly (i.e laser head) ??

I'm trying to think of a low tech way to measure these vibrations. I think the frequency of the vibrations is very low e.g. perhaps 10Hz.

I would suggest that perhaps bolt a loud speaker to the laser head and try to record any sound, using the speaker, as it would be efficient at recording at 10Hz, but you'd probably need to modify the speaker, but attaching a weight to the cone.

I have some motion sensor module which may be able to measure this sort of vibration, but they are basically the same sensor that most mobile phones have in them

MPU-9250 | InvenSense

So you may find there is an App that you could run on your mobile phone (perhaps this one)

https://play.google.com/store/apps/d...kinetics&hl=en

which you could use to monitor the vibration.

However it sort of depends on how fast the App samples the data (whether its designed to look for vibration or just for slower varying inputs). I know the sensors themselves are very fast and quite sensitive.

If the problem is vibration transfer into the laser head, there are 3 solutions.

1. Stop vibrations being generated in the first place (I think this is impractical)
2. Increase the mass of the laser head, so that input vibrations have less effect (However the downside of this, is that the machine will struggle to move the head quickly)
3. Fit damping between the laser head and the X carriage

But I'm not a mechanical engineer (I'm an electronics engineer / software developer) so I am not an expert on vibration damping

[RogerClark]

Another thought...

I don't bolt my workpieces down and they are one a flexible mesh (a heating duct grill)

The vibration may not be in the laser head, but may be in the workpiece

I noticed a scientific article on this, but it was behind a paywall.

Perhaps making the bed heavier would be worth a try. I suspect it would not be too hard to do. And also try making sure the work piece is fixed to the heavy items on the workbed

[Labmaster]

If this are vibrations from the machine itself they could even be induced on the first or second mirror stage.

I hope i can get my machine up and running again soon, so that i could do some test myself.

Will be quite interesting, as i have already modified the laserhead mount and made some other mechnical parts stiffer than they where originaly.
I also modified the table mounts.
Here you could see what i already have done:
http://www.cnczone.com/forums/genera...tor-setup.html

...But I'm not a mechanical engineer (I'm an electronics engineer / software developer) so I am not an expert on vibration damping

Me too, at my daily work i mainly have to deal with STM32 processors and Invensense sensors :-)

best regards,
Walter

[Storen]

@Roger

//1. Stop vibrations being generated in the first place (I think this is impractical)

I think that is why the high-end brands like Trotec use expensive analog servo motors. They don't vibrate as steppers do, and you can't even hear them while they work. Last week I had a chance to test their "Speedy" machine - quiet, sturdy, no vibrations, perfectly smooth cuts at very high speeds. Quite amazing.

//2. Increase the mass of the laser head, so that input vibrations have less effect (However the downside of this, is that the machine will struggle to move the head quickly)

Actually extra weight can make the situation worse because of the inertia - the machine will shake every time the head changes its direction. All precise machines I've seen tend to have minimal head-weight. Some as Argus and Epilog have their lens and mirror exposed and held only by a very light aluminum bracket. Trotter's head I saw was also aluminum and super light.

Another thing that helps precision is using ball-screw linear actuators instead of geared-belts. No vibrations, perfectly smooth motion, very affordable. They will reduce the overall speed and require higher torque than belts. As far I know the max. speed you can have with normal stepper motor and ball-screws is around 200-300mm/s. That is perfectly fine for cutting but it will slow down the engraving. Thats why typically they are used in specialized machines for cutting.

[RogerClark]

@Storen

Ok about analog motors etc.

I would have thought lead screws will make the machine very slow. I'm not an expert at laser CNC but I also have a 3D printer and have looked at various designs, and lead screws are only used on the Z axis of most 3D printers as they are too slow.

I see what you mean about weight of the head assembly, but I think what you are saying is that some machines effectively isolate and damp the laser head, but having it connected though a light weight bracket.


However I think the problem is that we really don't know where the vibrations are coming from. The stepper motors will by their nature cause vibration, But this may not be the cause of the problem. It could be oscillation in the drive belt.

I will see how hard it is for me to build a vibration monitoring device, that I can attach to various parts of my machine to determine exactly what is really vibrating.

[Storen]

@Roger

//I will see how hard it is for me to build a vibration monitoring device, that I can attach to various parts of my machine to determine exactly what is really vibrating.

That would be awesome! With that we should be able to make some progress

[RogerClark]

OK.

I'll need to work out the best way to do this, as the data capture needs to be about 100 times per second to get any meaning full data. So I'll probably need to build something that has a SD card to record the data onto, rather than trying to send data in real time to the PC.

Actually, I'll also look at at a low tech solution, i.e I some USB audio dongles, which I think can be modified so that they can accept very low frequency inputs e.g. 10Hz or lower, and perhaps I could just record into my PC and view in Audacity etc.
However I'm not entirely sure what I could use for a sensor in this case.

I know that a speaker with a weight attached to the centre of the cone can be used to detect motion, as the speaker will create electrical voltage (and current) if you move it, i.e the reverse of when you send voltage / current to it from a amplifier

So perhaps if I attach the cone of a small speaker to the bed or the laser head, and let the outer part of the speaker "float", it may work to pick up vibrations.

But obviously a nice 3 axis accelerometer like the MPU9250 recording its data onto a SD card would probably be the best device possible.


Edit.

It looks like I can just send the data straight to the PC.

I've just run a test sending sequential numbers, and I'm managing to send 250,000 characters per second to a virtual serial device on the PC (albeit a fast PC so I'll need to test my laptop that I'd be running near the laser)

But 100 samples per second, I only need to send the 16 bit values from the 3 accelerometers to the PC, with commas separating their values.

so assume worst case 5 chars per number e.g. "12345,12345,12345" thats only 15 chars per sample, so 100 samples per second would only need 1500 chars per sec and I can do 250,000 chars per sec

OK. I need to hook up the accelerometer module to my mircocontroller board (72Mhz STM32 based board ) with built in USB

[RogerClark]

Hi @Storen

Well, I don't know how useful this will be, but...


I wired up a MPU9150 motion sensor board to a micro-controller which is connected to the PC via USB and shows up a Serial connection

The sensor is this one from eBay

9DOF MPU 9150 3 Axis Gyroscope Accelerometer Magnetic Field Replace MPU 6050 | eBay

Here is the data sheet for the sensor

http://dlnmh9ip6v2uc.cloudfront.net/...-MPU-9150A.pdf

When the sensor is just resting on my desk I get a lot of noise.

e.,g. here is some accelerometer data samples

-24 416 16368
-76 440 16296
-84 512 16436
-76 392 16216
-72 404 16260
-12 392 16172
-40 408 16264
-60 400 16272
-88 384 16380
-36 424 16192
-72 412 16264
-84 452 16208
-48 364 16344
-28 340 16304
-72 360 16208
24 388 16228
-12 416 16308
-84 384 16348
-120 388 16296
-20 368 16220
-60 412 16240
-116 380 16400
-60 368 16424
0 352 16340
-48 368 16324
-68 364 16240
-16 420 16172
28 380 16208
-16 384 16252
4 440 16336
-28 484 16388
-28 388 16280
-36 440 16412
-24 440 16264
-24 340 16112
-20 368 16240
-32 464 16392
-120 384 16432
-76 348 16288
68 352 16164
-80 424 16412
-32 372 16264
-104 376 16252
-76 396 16328
-40 324 16252
-40 284 16300
8 340 16308
-48 448 16380
-104 456 16192
-40 380 16160
-44 380 16144
-60 380 16328
20 432 16456
-132 364 16324
-56 420 16296
-16 352 16312
-56 428 16308
-64 412 16344
-88 296 16240
-44 308 16208
-84 344 16252
-80 352 16304
-28 384 16332
-32 384 16404
-36 376 16300
20 460 16192
40 412 16324
8 376 16364
4 368 16256
52 344 16092
-48 404 16212
-20 392 16372
-32 480 16308
24 384 16316
-24 360 16240
-124 432 16312
0 472 16416
4 444 16412
40 324 16400
-80 344 16300
-72 368 16196
12 380 16244
4 412 16288
-16 384 16392
-80 392 16340
-24 432 16392
-96 376 16380
-124 380 16340
-36 368 16388
-8 416 16300
4 452 16168
-32 336 16236
48 368 16324
-52 312 16428
-88 384 16308
-48 436 16264
-80 384 16288
-24 456 16280
44 428 16392
-20 464 16364
-16 424 16312
12 428 16288
44 444 16128
-56 404 16240
-44 388 16120
-28 408 16280
-32 344 16248
-24 340 16328
4 428 16404
-44 428 16372
-4 396 16364
-12 340 16408
-128 484 16332
-68 468 16228
-40 428 16240
-16 352 16364
-28 432 16184
-104 460 16088
-120 376 16208
-44 384 16296
-52 420 16392
-80 360 16220
-48 364 16248
-120 372 16292
-76 380 16356
-8 488 16336
24 468 16248
-40 468 16360
-12 408 16352
32 332 16280
28 332 16328
12 384 16340
-48 396 16296
-4 396 16232
36 380 16244
8 396 16392
-100 476 16484
44 344 16328
-40 288 16292
-56 304 16216
-44 444 16104
-60 336 16252
64 380 16388
72 380 16400
-52 352 16252
-48 340 16320
-52 424 16340
-60 412 16272
32 432 16280
-88 356 16352
-56 364 16364
-48 504 16312
28 364 16292
32 360 16380
-48 376 16424
-148 348 16328
-80 448 16116
12 376 16024
104 388 16188
24 492 16296
-76 388 16284
0 464 16316
60 388 16388
-24 360 16272
-120 424 16372
-24 504 16368
-44 360 16240
-8 244 16364
-40 388 16336
-36 432 16268
-100 412 16176
-72 372 16252
84 392 16160
-72 336 16260
-28 376 16272
-64 304 16240
0 432 16372
28 456 16308
-20 412 16332
-52 396 16320
-48 436 16284
-4 420 16240
52 528 16300
-24 456 16308
-148 456 16376
-96 444 16344
0 352 16272
76 424 16180
36 448 16188
-116 388 16224
-76 388 16272
0 404 16256
0 368 16288
-40 440 16240
-32 436 16292
8 396 16348
76 380 16324
32 388 16148
-20 376 16320
-100 368 16284
-28 308 16188
-56 364 16120
-128 304 16212
-112 348 16264
-116 368 16400
-60 244 16324
-52 356 16192
-32 324 16200
-48 400 16320
-28 364 16452
-44 340 16444
-76 396 16224
-28 332 16248
-24 392 16356
16 428 16352
-24 376 16396
-104 332 16356
-24 376 16332
16 412 16276
-48 384 16204
-52 448 16212
-68 380 16252
-128 252 16268
-120 360 16300
-88 336 16304
-92 336 16228
-36 424 16276
68 448 16312
-160 432 16300
8 352 16348
28 408 16220
40 396 16220
-20 404 16316
-60 348 16328
-12 416 16272
16 408 16252
-20 384 16276
76 372 16228
100 436 16184
68 344 16348
-16 352 16448
44 320 16400
16 408 16308
24 380 16252
28 344 16248
-16 404 16340
-16 412 16276
20 428 16364
48 412 16308
-48 420 16348
-28 368 16320
-4 376 16360
-28 316 16324
-40 380 16316
28 428 16300
-32 432 16168
20 416 16228
-84 352 16180
-4 412 16160
16 356 16320
-28 340 16168
-8 364 16228
-20 312 16388
-36 392 16412
-20 388 16340
-16 416 16420
-84 368 16188
-4 424 16228
-84 376 16316
-40 436 16324
-64 364 16280
0 284 16276
16 344 16304
-12 352 16292
4 392 16308
-108 452 16328
-40 432 16396
-20 404 16460
24 376 16336

(you may be able to cut and paste into excel)

But I am getting over 800 samples a second, which should be enough to detect low frequency vibrations on the machine

But its getting too late to do any more today. I will try to attach (bolt) it to the laser cutter tomorrow and take some readings

[Storen]

Hi Roger,

Thank you for doing this! It is too advanced for me, so I'll wait for your conclusion.

BTW, I contacted Shenhui about this issue. What they were able to suggested is:

- Check if the graphic files have straight and smooth curves

- Check if the belts are too tight or loose

- Check if the lens I securely attached to the head

Sadly, none of that is the problem.

[RogerClark]

Hi Storen

I think Shenhui sells these machines as laser engravers, so the quality of the operation is probably what they would expect.

I wonder if all of these machines have the same issues with, what appears to be vibration.

I'll see if I can find time today to bolt the sensor to the machine and take some readings

[Labmaster]

Today i diassembled the whole y and x axis mechanics to find out how smooth the linear rails are moving, what should i say, its horrible.
One problem are the linear feeds themself, even when moving a single slider on the rail it does not run smooth, it is more like as if you turn a defekt ball bearing. I spend hours to find out how to get it to run smoother but no success.
What makes this even worse is, when the two y axis rails are linked together with the x axis gantry.
There is a huge geometry error on the mounted rails.
It is quite impressive that the small NEMA17 stepper is able to move the y axis that precise.

On friday i will try to fix at least the geometry error but even think about buying new higher end linear rails.
We will see.

best regards,
Walter

[Storen]

@ Roger

//I think Shenhui sells these machines as laser engravers

Hmm... you may be right, but they should not

Actually I would rather suggest these machines for cutting. They are doing pretty good job for cutting opaque materials at low speeds, At 20mm/s and below the ridging is minimal and very difficult to notice. They give you the same quality as the more expensive machines.

It is more problematic when you cut transparent acrylic where the reflections and refractions make it more obvious. And it is terrible for cutting paper because to have clean-light edges you have to cut at high speeds and you know what happens then...

The uniformity of the engraving is the worse I've seen. Especially in the depth of the engraving. I could recommend them only for engraving wood and only to less demanding clients.

[RogerClark]

Hi Walter

What error do you have in the geometry ?

I have a "Bevel Box", which I used to check the head to bed alignment and it was fine.

I also cut a 10cm square from acrylic and measured the angles of the sides using the bevel box and they were about 0.1 deg out, - which is better than I could align them to

Re: Knackered bearing

Do you think if you bought a new carriage it would be any better ?

I'm not sure what the intended use of the rails really is.

I know that you can buy rails that are intended for yachts and also for large sliding window panels, but I think in both cases the rails would not be mild steel (which is what I presume our rails are made of)

I have seen videos of really big sliding roof panels, moving on carrages and rails very similar, which seem to glide effortlessly

I wonder how much it would cost to either replace the carriage and / or rail (probably both)

[Storen]

@ Walter

My carriages make some very scary sounds at certain locations. I wonder where are they getting these parts from... I think most Chines manufacturers (including Argus and HSG) are using the same Taiwanese Hiwin rails. Maybe Argus are getting only the selected ones, that could pass their tolerance tests. And low-cost brands as Shenhui are buying the ones with largest deviation. Or they screw them during assembly. That is quite possible considering the condition of the housing. I don't know.

//There is a huge geometry error on the mounted rails.

What exactly is the error? I noticed my X and Y are not perfectly perpendicular and the Y rails are not perfectly parallel. All that resulting to additional friction.

On the positive side - 70% of the samples I got from different Chinese vendors were no different. In fact some were worse. At least, you know, we didn't spent much

[RogerClark]

I've not dissassembled either axis, but to me the X Axis seems to run quite smoothy

The Y axis seems stiff, in fact I had to slow the "idle" speed and acceleration to prevent the motor slipping

I really recommend getting one of these

±180° Accuracy Digital Bevel Box Inclinometer with 3 Disk Strong Magnets | eBay

It won't directly measure the X to Y axis alignment, but it will do, if you print a square (or rectangle) and use it to measure the relative angles of the sides

Note the Bevel Box does not measure absolute level, its not like a spirit level, its purely a measure of relative angles of inclination.

[Labmaster]

Maybe geometry error was not the correct word.
It is hard for me to explain.
The surface of the black steel frame where the rails are mounted on have no flat top angles.
Both Y rails are not flat to each other, the left one is tilted to the left, the right one is tilted to the right.
That means, if i would mount the X axis only on the left slider the X axis reaches the other side about 20mm above the slider.
On friday i can make photos of it.

As a quickfix i put some thin washers on top of both outside top holes of the sliders, but this fixed the issue only slightly, there is still some force on the whole system when tightening up the screws.

Regarding another set of linear systems, there is a big range of such linear systems on the marked, specialized for industrial use on machines, they are available in different qualities and prices.
The linear system on this machine are quite cheap (no hiwin...)
The balls ( i disassembled the sliders today ) are rollin very rought in a not so precise plastic cage.
The slower you move them on the rails the more hickups you feel. I tried different types of greas, oils... today and every time cleaned them in between, and found that the thinest oil gave the best result but still far away from that what i would expect.

As already wrote,
on friday i will move on, in the meantime i only have my mobilephone here for typing ( as now )

Btw: i also have bevelboxes here, we mainly use them for setting up rc helis :-)
https://www.freakware.de/p/bevel-box...005-a34385.htm

Best regards,
Walter

[RogerClark]

Hi Walter

OK.

I think you are saying that your Y axis rails are not level in the X axis.

As they are not very wide, I think its difficult to measure this.

I just took some measurements on my machine, but as far as I can tell everything is in alignment.

The only thing I can not measure accurately is the distance that the 2 Y axis rails are apart, e.g. if they are further apart at the back (or the front)

I will need to use a tape measure to check the distance of the diagonal lengths, as this is the best method to see if something is not rectangular, as one diagonal will be longer than the other

I'm not sure how much use this will be , but I will attach my accelermoeter to the machine this evening and collect some data

[Labmaster]

Yes, correct, this is what i wanted to explain.
On my machine the level error can not be maesured by the bevel box (put on top of the rail) but over the half a meter of the xrail lenght the error is noticable quite well when you trie t screw it down on the other side.

When measuring vibrations will you do some kind of FFT on the data ?
I think without this the data are quite hard to understand.

A FFT for each axis would be very usefull but then even the normalization, peak measurement and a couple of other small things are very important.
Don't forget to configure the internal Filter of the IMU and set a usefull fullscale range ( i guess the lowest G settings migth here be the best )

best regards,
Walter

[RogerClark]

Hi Walter

OK. Perhaps my machine also has this problem. Its hard to see, as the rails are narrow in X.

I will post data. At the moment I can get about 850 samples per second, of +/- 2g (this is maximum sensitivity on this sensor),

I am not sure what the max number of samples per second is for this device. I think if I send more data per line I get a higher data rate e.g. rather than
x1,y1,z1
x2,y2,z2
etc

if I send
x1,y1,z1,x2,y2,z2

(but I have a maximum line length for the data which I think is probably 256 digits (including commas ",")

Anyway, I am about try now, but I need to find a good way to attach the module to the machine.

Edit

Its taken me a while to make a strong bracket for the sensor from some 3mm 90 deg angle aluminum extrusion.

So I'm not going have time to take any readings this evening (here in Melbourne Australia)

Also need to work out how I can get the USB cable from the cutting head to my laptop without it getting in the way.
Fortunately I have a 5m USB extension cable I can use.. So it should be OK and I wont need to put the laptop inside the laser cutter ;-)