ripples, resonance, perforation

[Stefan Hechenberger]

Let's talk about this in a more organized way and make sure we are all
on the same page. There are probably three different issues that get
convoluted often but may also be related to some degree:

1) ripples after sharp turns
2) ripples in the middle of a curve or line
3) perforation of a perfectly placed cut

1) and 2) are feedrate-related and most likely caused by some kind of
resonance. On our machine these ripples start getting very noticeable at
speeds above 4000mm/min. Others have reported to see significant ripples
at lower speeds.

3) is possibly related to idiosyncrasies of the Reci lasers or how we
pulse them. Generally I don't think of this as a big problem but for
some materials it may create rough edges. Perforation may be less when
using analog control etc ...

I think we need to start posting some images that we can compare. From
this we can then better differentiate between issues that are related to
design limitation and assembly/tweaking issues.

The following info would be highly useful:

- image of cut showing ripples
- feedrate, power
- version of LasaurApp and firmware
- new or old gantry (ref design changed about mid-2012)
- modifications from ref design


--
Stefan Hechenberger
studio: Nortd Labs - labs.nortd.com
resident: F.A.T. Lab - fffff.at
project: Lasersaur - labs.nortd.com/lasersaur

[Kevin]

Righty, so we've been running some testcards for documentation, we used the one from Till as it seemed a bit better for detailed testing.

Results here: http://www.flickr.com/photos/beyond_io/sets/72157632746756009/

For now, the results are "acceptable" below F1000, but I'm thinking they could be better still.

I'm also starting to wonder if the visual resonance at low speeds is still stepper related or if it's the laser pulse that gets really visible (http://www.flickr.com/photos/beyond_io/8468019448/in/set-72157632746756009). Then again, if it would be the laser pulse, it would probably persist over the entire cut so it's probably stepper/resonance related.

Next steps we're going to try: 

• Add some more robust signal filtering on the PCB

• Try to adjust the Gecko trim pots

• Try to adjust the acceleration variables in LasaurGrbl

Updates soon!

[Steve Baker]

Kevin wrote:

> Results here:
> http://www.flickr.com/photos/beyond_io/sets/72157632746756009/

Here is some analysis:

The laser pulse doesn't "know" which direction the head is moving in.
Since your results are clearly showing that the wobble is happening only
at some parts of the curved lines and some angles of the straight line,
it's hard to imagine how the laser can be the cause...it might cause
"perforations" - but it can't possibly cause "wobble".

A geometric "wobble" can't be to do with laser power - it's got to be
something mechanical...physically moving the head or vibrating a mirror
mount.

So what's causing the vibration?

Let's look *carefully* at the attached image...

I cropped and re-scaled the three photos so they are the exact same size.
Then I measured the number of ripples along that red line (which is the
same length in all three images).

It's pretty clear that the number of ripples per millimeter is directly
proportional to the speed. Low speeds have short wavelengths and high
speeds have long wavelengths.

BUT - bear in mind that the head is traversing those distances in
different amounts of time. In the 3000mm/min case, it traverses those 6
ripples in half the time it took the 1500mm/min case to make 12 ripples.
Twice the time - twice the number of ripples!

Which means that: THE FREQUENCY OF VIBRATION IS CONSTANT NO MATTER THE
SPEED OF THE MOTORS!!

It doesn't matter whether the head is moving fast or slow - you're getting
a constant vibration at the same hertz rate...NO MATTER HOW FAST THE
MOTORS ARE SPINNING.

That constant hertz rate means that this simply cannot be something that's
to do with a single motor doing something bad.

I can only put this down to some resonant frequency effect...that's an
inescapable conclusion.

Mechanical structures have a number of frequencies at which they'll
resonate - the source of the vibration that causes that resonance is going
to be hard to track down...it can be a very small source...but you can't
get away from it.

So, for example, when the head takes a sharp 90 degree turn, it'll start
the system vibrating at that fundamental rate...and that vibration will
die down as the system moves away. But at some angles, the vibration from
the two motors will produce "beat" frequencies at that same magic
frequency and the vibrations will build up and up the longer you move at
that constant speed and direction.

In my estimation (in my machine) this frequency is at around 16 Hertz.

Kevin: If you could measure the actual distance represented by the red
line in my diagram, you can calculate the approximate vibration frequency
in your machine...it looks kinda similar to mine.

So the question is what parts of the system might vibrate at 16 hertz -
and how do we prevent it?

-- Steve

[Erik Moon]

Is it my imagination, or do these photos make it look like only the
Y-axis is vibrating??

---------- -----=----- ----------
Erik Moon

> --
> You received this message because you are subscribed to the Google Groups "lasersaur" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to lasersaur+...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
> <ripples.png>

[Steve Baker]

I thought so too...but there are *some* obviously X-direction vibrations
in the vertical part of the motion in the close-up photos. However, my
machine doesn't produce those wobbles...I only see them happening on
certain angled diagonals.

I suspect the Y axis beam is vibrating - it's big enough and bendy enough
to have a 16Hz resonance.

My machine is too busy earning money (paying for itself!) to take time out
for testing right now - but what someone needs to do is to find out which
LONG diagonal motion (and at which speed) produces the worst vibrations.

Cut a bunch of foot-long lines radiating out at (say) 10 degree spacing
and figure out which line looks worst. Then make a test pattern made of
nothing but long parallel lines at that angle.

Now you should be able to run that pattern with the lid up and rest your
finger lightly on different parts of the machine to try to feel where the
vibration can be felt and where it can't.

A 10Hz to 20Hz vibration with a millimeter or so of amplitude should be
really easy to feel.

-- Steve

-- Steve

[Chris Uhlik]

When you do find the component that is vibrating, Sorbothane is pretty awesome stuff.  It's a sort of viscoelastic rubber with very high damping.  A bit of Sorbothane wedged between some parts or laminated with a piece of thin aluminum bar along the side of a beam can make a big difference in small vibrations.

Chris

[Gerald Anzalone]

How do you tell x from y in these images? Is the assumption x is parallel with top/bottom and y with left/right? If so, it looks to me as if the amplitude of the waves in y-axis is greater.

For what it's worth, I see the same "waviness" behavior with my Prusa printers and severity (amplitude of the waves) increases with print speed. It's also axis-specific, with typically higher amplitude in the y-axis. My conclusion was that it was a combination of effects (hysteresis in the belt, fundamental frequency of the frame, flexing of plastic components).The y-axis on the Prusa is the most massive moving component (something in common with lasersaur?). Applying what little I remember from Newtonian physics (F=MA, in particular), I choose to trim the mass of the y-axis as much as I could, which reduced the severity of ripples in my prints. Don't know how that's done with the big, stiff gantry on the lasersaur...

[Stefan Hechenberger]

Did some testing today ... and my ripples look very similar. The best
improvement came from reducing the acceleration setting in config.h a
bit. This takes care of the ripples in all the straight lines. They are
obviously caused by fast seek motion leading to the line.

http://www.flickr.com/photos/stfnix/8468625443/in/photostream/

The ripples in the curves are more enigmatic. The spiral is actually a
really good test to bring them out. They are strongest in circles of
50-80mm in diameter.

http://www.flickr.com/photos/stfnix/8469721812/in/photostream/

My theory is it's vibration caused by the acceleration pattern within
each line segment of the curve. At the right freq it finds a resonance
in the gantry. The strongest indication for this is that when I play
around with parameters in the motion planer (where the line segment
junction speeds are calculated) I still get the ripples but at different
areas in the spiral.

I also wonder if we can rule out low level stepper resonance. For that
they sell dampers:
http://www.linengineering.com/contents/stepmotors/Nema17_Damper.aspx

--
Stefan Hechenberger
studio: Nortd Labs - labs.nortd.com
resident: F.A.T. Lab - fffff.at
project: Lasersaur - labs.nortd.com/lasersaur

On 02/12/2013 12:35 PM, Kevin wrote:
> Righty, so we've been running some testcards for documentation, we used
> the one from Till as it seemed a bit better for detailed testing.
>
> Results here: http://www.flickr.com/photos/beyond_io/sets/72157632746756009/
>
> For now, the results are "acceptable" below F1000, but I'm thinking they
> could be better still.
> I'm also starting to wonder if the visual resonance at low speeds is
> still stepper related or if it's the laser pulse that gets really
> visible
> (http://www.flickr.com/photos/beyond_io/8468019448/in/set-72157632746756009).
> Then again, if it would be the laser pulse, it would probably persist
> over the entire cut so it's probably stepper/resonance related.
>
> Next steps we're going to try:

> � Add some more robust signal filtering on the PCB
> � Try to adjust the Gecko trim pots
> � Try to adjust the acceleration variables in LasaurGrbl

> studio: Nortd Labs - labs.nortd.com <http://labs.nortd.com>
> resident: F.A.T. Lab - fffff.at <http://fffff.at>
> project: Lasersaur - labs.nortd.com/lasersaur
> <http://labs.nortd.com/lasersaur>

[Steve Baker]

Gerald Anzalone wrote:
> How do you tell x from y in these images? Is the assumption x is parallel
> with top/bottom and y with left/right?

Yes - that's the way that test pattern is oriented.

> If so, it looks to me as if the
> amplitude of the waves in y-axis is greater.

Yes.

> For what it's worth, I see the same "waviness" behavior with my Prusa
> printers and severity (amplitude of the waves) increases with print speed.
> It's also axis-specific, with typically higher amplitude in the y-axis. My
> conclusion was that it was a combination of effects (hysteresis in the
> belt, fundamental frequency of the frame, flexing of plastic
> components).The y-axis on the Prusa is the most massive moving component
> (something in common with lasersaur?). Applying what little I remember
> from
> Newtonian physics (F=MA, in particular), I choose to trim the mass of the
> y-axis as much as I could, which reduced the severity of ripples in my
> prints. Don't know how that's done with the big, stiff gantry on the
> lasersaur...

The thing about resonance is that repeated application of a little nudge
builds up to a huge vibration. F=ma doesn't exactly apply because it's
all down to the timing of the little nudges.

I looked at the equation for calculating the resonant frequency of a beam
- and it wasn't much help. There are separate formulae for beams that are
clamped at the ends - or at one end - or not clamped. Ours are vertically
and X-direction clamped at both ends - but free to move in the Y
direction. For an absolutely solid, unclamped aluminium bar of the
dimensions of the lasersaur Y axis, you get a resonance at around 1Hz.
But vary any of the parameters and the answer you get is all over the
place...so I have no confidence in that idealized calculation.

This much is certain: The two motors each vibrate the machine at different
frequencies in the hundreds of hertz to kilohertz rates - depending on the
speed and slope of the line. On my machine (but not everyone's) I get no
waviness when cutting parallel to the axes - which meant that either there
is no vibration - or the vibration is parallel to the direction of cut.

With two vibration sources (such as when cutting diagonally) - you get
"beat" frequencies - an induced vibration with a frequency that is the
difference between the frequencies of the two individual vibrations. So
if one motor is vibrating at 1000Hz and the other at 1016Hz - then you get
a 16Hz vibration too.

A big chunk of metal can't move far in 1/1000th of a second - so a 1kHz
vibration doesn't matter - but a 1/16th second vibration could easily
produce the ripples we're seeing - and if that's the resonant frequency of
some major part of the machine - then it'll get amplified massively.

Recall the Mythbusters "Earthquake machine" episode where they used a tiny
machine to induce resonant vibrations in a bridge that could be felt 100
feet away. (It's on netflix - episode 60).

I think that's what's happening...but it's hard to prove.

> On Tue, Feb 12, 2013 at 11:16 AM, Chris Uhlik <chris...@gmail.com>
> wrote:
>
>> When you do find the component that is vibrating,

>> Sorbothane<http://www.sorbothane.com/>is pretty awesome stuff. It's a

[Steve Baker]

Worth a try...although if Stefan can understand what's going on, that's a
better approach.

-- Steve

Chris Uhlik wrote:
> When you do find the component that is vibrating,

> Sorbothane<http://www.sorbothane.com/>is pretty awesome stuff. It's a

[Kevin]

We've been running some simple motion tests, long angular lines along the entire table and noticed a ridiculously large resonance when we're starting at (0,0) and move to somewhere between (300,610) and ((360,610) at slow speed. The Y-cart starts shaking like crazy and is clearly visible on the X-cart.

There's a slight to no resonance when moving the X or Y individualy, it's really the combination of both steppers at a specific speed that make the structure go berserk.

We're thinking of adding dampers to both steppers, we'll try with some rubber between the attachment parts first or order the nanotec dampers which are pretty cheap.

On Monday, February 11, 2013 9:43:48 PM UTC+1, stefanix wrote:

[Steve Baker]

Wow! Thanks for that.

Something that builds up over such a long line really has to be resonance.

Defeating it is not fun.

Putting dampers on the stepper motors should reduce the amplitude of any
vibration transmitted from the X-drive motor to the Y cart - but I can't
see that this is the only (or even the main) source of problems because
the Y-drive motor isn't connected to anything that would be likely to
transmit vibration to the head - and it takes both motors working at the
same time to induce this resonance.

The Y-drive motors contribution to the resonance really has to be
transmitted via the drive belts...and they have an innate damping effect
due to the elasticity of the belt.

If the vibration is transmitted from the Y-drive motor to the resonating
parts via the belt - then it's likely that the X-drive motor's
contribution comes the same way...so these dampeners may not help.

Well, that's all speculation - and the dampers are cheap enough to make it
worth the experiment.

Bolting/gluing something to the Y-cart beam that absorbs vibrational
energy seems like the thing that's most likely to work. The amount of
energy put into the beam every second is quite small - which is why it
doesn't vibrate at other frequencies. It's only a problem at resonance
where a tiny amount of energy is transmitted, but it builds up over time.
Something that can leach that energy away only has to do so at a very
minimal rate to outpace the input vibrational energy and prevent it
building up via resonance.

Changing the dimensions/material/weight of the Y-beam will only change the
frequency at which it vibrates. That's unlikely to be much help unless we
can move the resonant frequency out of the range of frequencies that our
motors can produce...but if (as it seems) it's the beat frequency between
them that's causing the grief, there is no reasonable range of frequencies
between zero and about 1kHz that won't show up during normal operations.

I suspect that the solution may be the anti-vibration tape that someone
mentioned yesterday...applied along the length of the Y-cart
cross-beam...it would turn some of the vibrational energy into heat and
thereby prevent it from building up into a strong resonance.

I suppose that if we can characterize the frequency accurately enough, a
tuned mass damper would be another option.

It's all somewhat "black magic" though.

-- Steve

[Chris Uhlik]

On Wed, Feb 13, 2013 at 7:46 AM, Steve Baker <st...@sjbaker.org> wrote:

I suspect that the solution may be the anti-vibration tape that someone
mentioned yesterday...applied along the length of the Y-cart
cross-beam

I have found laminating an absorbing material to a stiff beam can sometimes remove enough energy to quench a resonance.  My airplane landing gear was a simple steel leaf spring which had a wobble under braking at about 30 mph.  After laminating it with a wood strip wrapped with fiberglass, the vibration went away.  I don't have a lasersaur yet, but the first thing I would try is to glue a strip of sorbothane+aluminum to the longer beams so that you have a sandwich

BEAMBEAMBEAM+sorbothane+ALUMSTRIP

BEAMBEAMBEAM+sorbothane+ALUMSTRIP

BEAMBEAMBEAM+sorbothane+ALUMSTRIP

BEAMBEAMBEAM+sorbothane+ALUMSTRIP

BEAMBEAMBEAM+sorbothane+ALUMSTRIP

BEAMBEAMBEAM+sorbothane+ALUMSTRIP

It will work even better if you can anchor ALUMSTRIP at one end of the beam.

As the beam flexes (left-right) in this illustration, it shears the sorbothane which is very resistant to high, small amplitude speed motion.

If you tap on the beam and hear any ringing, it is poorly damped at some frequencies.  After laminating a beam like this, when you tap it the beam will sound dead with no tendency to ring at any frequency.

Wood might work too, but sorbothane is a few orders of magnitude better.

To analyze this, you probably want a high frequency analog output accelerometer connected to an oscilloscope.  Tape it to the beams in various places and directions to look for resonances.  The "various places" means center (fundamental mode), quarter-length (second mode), etc.  The "various directions" means vertically and horizontally to look for vertical and lateral vibrations.

In a system like this with long drive belts, mounting motors on slabs of sorbothane sandwiched between a pair of mounting plates might be sufficient to suck energy out of belt vibration modes.  Damped idler pulleys might be more effective.

Chris

[Kevin]

If possible, I'm gonna try to test it this weekend. I'm thinking of getting a longer aluminum L profile and replacing the short one supporting the cable carrier on the y-drive.

I'll put whatever "rubber' I can find this weekend in between the L profile and the Y-drive, making it slightly more rigid and hopefully absorbing some of our ripples!

[Steve Baker]

That's been my thought too. It's really the only place to stiffen the Y
cart without interfering with the rollers on the X cart or reducing the
"ground clearance". I was thinking of bolting a heavier gauge length of
aluminium L channelling onto the full length of the cart and only
resorting to sorbothane tape between that L-channel and the cart if the
extra stiffening alone didn't help enough.

My hope is that the horizontal part of that "L" will strongly resist
Y-direction flexing - and therefore significantly stiffen the existing
beam without adding too much weight.

My concern is that doing that will just increase the resonant frequency -
simply moving the problem to vibration when cutting lines at some
different slope!

-- Steve

Kevin wrote:
> If possible, I'm gonna try to test it this weekend. I'm thinking of
> getting
> a longer aluminum L profile and replacing the short one supporting the
> cable carrier on the y-drive.
> I'll put whatever "rubber' I can find this weekend in between the L
> profile
> and the Y-drive, making it slightly more rigid and hopefully absorbing
> some
> of our ripples!
>
> On Wednesday, February 13, 2013 7:19:39 PM UTC+1, Chris Uhlik wrote:
>>
>>
>> On Wed, Feb 13, 2013 at 7:46 AM, Steve Baker

>> <st...@sjbaker.org<javascript:>

[Justin Krull]

The first mode of the actual section of 1426mm long aluminum (http://us.misumi-ec.com/item/10302244930/img/drw_02large.gif)  is at 120 Hz. This is based on FEA results.

[Steve Baker]

Justin Krull wrote:
> The first mode of the actual section of 1426mm long aluminum (
> http://us.misumi-ec.com/item/10302244930/img/drw_02large.gif) is at 120
> Hz. This is based on FEA results.

Hmmm - but it's more complicated than that because it's somewhat "clamped"
at both ends and it has that large chunk of X-cart hanging off of it - we
also have a "damper" of sorts in that the somewhat elastic X-drive belt is
pulled tight between the X-drive motor and the idler pulley. That's going
to absorb some of the energy of the vibration.

I think it's too hard to calculate accurately...even with Finite element
techniques.

-- Steve

[Justin Krull]

Even clamped at both ends, it only goes to 118 Hz (within margin of error). I really don't think that the beam is being driven to resonance to cause the variation we're seeing. I think the problem is in the steppers.

[Justin Krull]

If someone has the mass of the y-cart, I can calculate the required torque for the stepper and give an idea for whether it's underpowered for what we're asking of it. I'll work on it with an estimate, if I can't get the actual mass. The 1426mm beam mass is ~1.9kg, as a first estimate.

[Justin Krull]

Another interesting investigation might be the natural frequency of the cart-belt system:

http://www.ormec.com/LinkClick.aspx?fileticket=A_5dLGa-lCA%3D&tabid=145&mid=621

The allowable tension on the belt is: 127 N (reference below). The tension increases with changes in direction due to the inertia of the cart.

http://us.misumi-ec.com/us/ItemDetail/10300414430.html

[Justin Krull]

Here's some analysis of the steppers from a load and inertia perspective, estimating the y-cart mass at 3 kg. Everything looks okay there. It would be great to know the accelerations explicitly in m/s^2.

[Stefan Hechenberger]

I am curious to see if this helps. I doubt it's the source of the
problem but it might reduce the ripples quite a bit.

I have just ordered some steppers with half the step angle (0.9 instead
of 1.8 deg). We have initially decided against them because they are
less common but maybe they have a positive impact on the oscillation. I
also will be trying a damper from Nanotec.

--
Stefan Hechenberger
studio: Nortd Labs - labs.nortd.com
resident: F.A.T. Lab - fffff.at
project: Lasersaur - labs.nortd.com/lasersaur

[Kevin]

Right, after some pretty extensive testing today here are our conclusions:

- the L-extrusion (with/without foam,rubber) along the Y beam had no effect whatsoever.

- adjusting the X-drive trim pot slightly didn't change anything either

- adding rubber around the X-stepper didn't change a thing

- adjusting the acceleration variable in LasaurGrbl didn't change anything for us either

side-note: the rubber/foam we used probably isn't anywhere near as good as sorbothane, it's just what we could find during the weekend.

So what did: Loosening the Y-belts. Atm they are probably too loose but it did have a huge impact on the ripples. 

It seems like the Y belts were acting up as guitar strings (we had then really tight), which is what's generating the high frequency ripples.

That being said, the ripples are still there, they just have a way lower frequency than before but they definitely seem to be coming mainly from the Y axis.

pictures here:

http://www.flickr.com/photos/beyond_io/8481541313/in/set-72157632746756009/

Hope this helps!

[Kevin]

One additional question, how is the microstepping being handled in lasaurGrbl? Haven't dug through it intensively yet as I'm sure Stefan has the answer to that one.

[Steve Baker]

Wow!

That's great work.

So it's the Y belts that are resonating.

Stefan says that they should be tensioned until they vibrate at 60Hz when
you twang them...but we're getting ~15Hz wobbles.

So back to belt tensioning. The trouble with loosening the Y belts is
that you get hysteresis problems - and the risk of skipping a tooth which
can result in "Very Bad Things" because the #2 mirror gets misaligned.

Perhaps we should look again at using wider belts?

-- Steve

Kevin wrote:
> Right, after some pretty extensive testing today here are our conclusions:
> - the L-extrusion (with/without foam,rubber) along the Y beam had no
> effect
> whatsoever.
> - adjusting the X-drive trim pot slightly didn't change anything either
> - adding rubber around the X-stepper didn't change a thing
> - adjusting the acceleration variable in LasaurGrbl didn't change anything
> for us either
>
> side-note: the rubber/foam we used probably isn't anywhere near as good as
> sorbothane, it's just what we could find during the weekend.
>

> *So what did:* Loosening the Y-belts. Atm they are probably too loose but

>> > an email to lasersaur+...@googlegroups.com <javascript:>.

>> > For more options, visit https://groups.google.com/groups/opt_out.
>> >
>> >
>>
>
> --
> You received this message because you are subscribed to the Google Groups
> "lasersaur" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to lasersaur+...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
>

-- Steve

[Chris Uhlik]

So it looks like a drive belt resonance problem.  To take energy out of the drive belt resonance, you need some damping in the drive or idler pulley.

Here's a good picture of how that can be done.

http://www.litens.com/vibprod2.cfm

I don't know of an off-the-shelf damped timing pulley.  I wonder if this isn't something that could be 3D printed from ABS in a couple of concentric parts with a sorbothane bushing between them...

Chris

[Steve Baker]

The trouble with damping is that it may result in imprecise cutting.

The link you gave is for automotive belts...who cares if there is a little
lag in the driven system when you stomp on the gas pedal...but if all of
your cuts come out with rounded corners, it's not so much fun!

-- Steve

Chris Uhlik wrote:

>>> *So what did:* Loosening the Y-belts. Atm they are probably too loose

>>> but it did have a huge impact on the ripples.
>>> It seems like the Y belts were acting up as guitar strings (we had then
>>> really tight), which is what's generating the high frequency ripples.
>>>
>>> That being said, the ripples are still there, they just have a way
>>> lower
>>> frequency than before but they definitely seem to be coming mainly from
>>> the
>>> Y axis.
>>>
>>> pictures here:

>>> http://www.flickr.com/photos/**beyond_io/8481541313/in/set-**
>>> 72157632746756009/<http://www.flickr.com/photos/beyond_io/8481541313/in/set-72157632746756009/>

>>>> > BEAMBEAMBEAM+sorbothane+**ALUMSTRIP
>>>> > BEAMBEAMBEAM+sorbothane+**ALUMSTRIP
>>>> > BEAMBEAMBEAM+sorbothane+**ALUMSTRIP
>>>> > BEAMBEAMBEAM+sorbothane+**ALUMSTRIP
>>>> > BEAMBEAMBEAM+sorbothane+**ALUMSTRIP
>>>> > BEAMBEAMBEAM+sorbothane+**ALUMSTRIP

>>>> > an email to lasersaur+...@googlegroups.com**.
>>>> > For more options, visit
>>>> https://groups.google.com/**groups/opt_out<https://groups.google.com/groups/opt_out>.

>>>>
>>>> >
>>>> >
>>>>
>>> --
>> You received this message because you are subscribed to the Google
>> Groups
>> "lasersaur" group.
>> To unsubscribe from this group and stop receiving emails from it, send
>> an
>> email to lasersaur+...@googlegroups.com.
>> For more options, visit https://groups.google.com/groups/opt_out.
>>
>>
>>
>
> --
> You received this message because you are subscribed to the Google Groups
> "lasersaur" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to lasersaur+...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
>

-- Steve

[Chris Uhlik]

We'd have to do the calculations to be sure, but adding enough damping might not incur a large stiffness reduction.  If the current system is very high Q, the required damping might be very small.  I don't think this is the case as I suspect the belts are pretty well damped already in the tension direction.  Making the belts wider might increase their aerodynamic damping, but only if their thickness were also decreased so the mass didn't also increase.

Yes, there is a speed / vibration tradeoff.  If you want to cut very quickly, you can do so with a loss in precision by softening the system and adding damping.  Or you can stiffen the system to push resonant frequencies up.  That's why big, fast CNC machines are built with lots of cast iron.  If you need a very precise cut on a light-weight, inexpensive machine, you need to slow down.  But if you want a very fast cut on a lightweight, inexpensive machine, you can have that at by compromising precision by adding damping.  Making the tradeoff optimally requires a detailed understanding of what your target speed is, the dynamics of the machine, and the amount of money you are willing to pay.

Chris

[Stefan Hechenberger]

interesting ... that takes us back to trying out belt dampers. The way I
was imagining them is actually very simple and could be done quite
easily (especially with a 3D printer). I am thinking of plastic guards
mounted to the inside of the y-rail, in between the belt. They don't
touch the belt unless it start swinging in which case they constraints
the amount of swing.

It's also possible that the belt swing is not the cause but rather what
you can see about the resonance in which case the 0.9 deg stepper motors
or the Nanotec/LineE dampers may help. I'll be testing the latter two
things as soon as I get them.

--
Stefan Hechenberger
studio: Nortd Labs - labs.nortd.com
resident: F.A.T. Lab - fffff.at
project: Lasersaur - labs.nortd.com/lasersaur

On 02/17/2013 05:08 PM, Kevin wrote:
> Right, after some pretty extensive testing today here are our conclusions:
> - the L-extrusion (with/without foam,rubber) along the Y beam had no
> effect whatsoever.
> - adjusting the X-drive trim pot slightly didn't change anything either
> - adding rubber around the X-stepper didn't change a thing
> - adjusting the acceleration variable in LasaurGrbl didn't change
> anything for us either
>
> side-note: the rubber/foam we used probably isn't anywhere near as good
> as sorbothane, it's just what we could find during the weekend.
>

> *So what did:* Loosening the Y-belts. Atm they are probably too loose

> but it did have a huge impact on the ripples.
> It seems like the Y belts were acting up as guitar strings (we had then
> really tight), which is what's generating the high frequency ripples.
>
> That being said, the ripples are still there, they just have a way lower
> frequency than before but they definitely seem to be coming mainly from
> the Y axis.
>
> pictures here:
> http://www.flickr.com/photos/beyond_io/8481541313/in/set-72157632746756009/
>
> Hope this helps!
>
> On Sunday, February 17, 2013 12:16:12 PM UTC+1, stefanix wrote:
>
>
>
> I am curious to see if this helps. I doubt it's the source of the
> problem but it might reduce the ripples quite a bit.
>
> I have just ordered some steppers with half the step angle (0.9 instead
> of 1.8 deg). We have initially decided against them because they are
> less common but maybe they have a positive impact on the oscillation. I
> also will be trying a damper from Nanotec.
>
> --
> Stefan Hechenberger

> studio: Nortd Labs - labs.nortd.com <http://labs.nortd.com>
> resident: F.A.T. Lab - fffff.at <http://fffff.at>
> project: Lasersaur - labs.nortd.com/lasersaur
> <http://labs.nortd.com/lasersaur>
>

> > an email to lasersaur+...@googlegroups.com <javascript:>.

> > For more options, visit https://groups.google.com/groups/opt_out

> <https://groups.google.com/groups/opt_out>.

[Justin Krull]

I don't think it's the belts themselves. I think we need an inertia damper directly on the Y-axis stepper. Both would probably help, but I'm betting we'll see the largest improvement damping the stepper itself.

an email to lasersaur+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/groups/opt_out.

--
You received this message because you are subscribed to the Google Groups "lasersaur" group.

To unsubscribe from this group and stop receiving emails from it, send an email to lasersaur+unsubscribe@googlegroups.com.

[Justin Krull]

My hypothesis is based on the fact that loosening the belts helped: they're not resonating themselves, but they are dampening the resonance of the stepper motion that we see in the plots here:

http://www.linengineering.com/contents/stepmotors/pdf/Size23_Damper.pdf

http://www.linengineering.com/contents/stepmotors/pdf/Size17_Damper.pdf

[Justin Krull]

The trouble comes with fitting the Lin dampers onto our Lin motors. There isn't enough space on either end of the Y stepper shaft to add the standard damper Size 23 damper without modifying the coupling design.

[Ray Debs]

It looks like the dampener might possibly mount directly on the shaft outside of the couplers instead of directly next to the motor housing.  Any idea on how much these cost?

Ray

To unsubscribe from this group and stop receiving emails from it, send an email to lasersaur+...@googlegroups.com.

[Kevin]

As Justin is saying, I'm also guessing the actual source of the resonance is the Y-stepper, by loosening the belts the resonance just didn't get transmitted as badly as before.

Looking forward to your results with the dampers/0.9 steppers

Another thing I've been wondering, we tried etching acrylic this week but we get very visible dots/pulses in the engraving, any way to fix this? Does this have anything to do with PPI (pulses per inch) and is this implemented somehow atm.

[Justin Krull]

Pulses per inch is the stepper pulses per inch, not laser pulses per inch.

[T. Hergenhahn, Aeronauten]

Hi Justin,

i'd like to buy these ones

D6CL-8.0F

http://www.orientalmotor.de/Products/Stepper_motors/Controllers_accessories/Accessories/Dampers/?&dwn=artnr&arid=6271

they will fit on the 8mm driveshaft and also on top of the screw from nema17. I also don't think it has to do with the beams. 

i get the same ripples and have the old lasersaur hardware 11.08..

best till

To unsubscribe from this group and stop receiving emails from it, send an email to lasersaur+...@googlegroups.com.

[Kevin]

right, so what I'm seeing are steps? If I recall correctly the steppers should be able to take steps of 0.03-0.1mm but the ones we're seeing are a lot larger than that.

[T. Hergenhahn, Aeronauten]

hi kevin,

it is not about the steps but more the "stepper resonance". 

i can't translate these articles but maybe stefan can.

if this doesn't work maybe the stepper needs to use less of an angle,

for example 0.36deg.(the guy from orientalmotors told me this, he 

would like to come over and see the ripples)

http://www.goetz-automation.de/Schrittmotor/Resonanzen.htm

http://www.schrittmotor-blog.de/?p=96

best till

 

....homepage der AEROnautenwerkstatt www.aeronauten.org....

...........AEROnauten | Am Burghof 55 | 60437 Frankfurt............

................................tel 069 611 931 ....................................

[Justin Krull]

Here's the Nanotec page on stepper resonance, in case anyone hasn't seen it yet:

http://en.nanotec.com/faq_113_tips_handling_resonance.html

[Steve Baker]

How does this explain the observed fact that the resonance happens at all
speeds - and at specific drawing angles that change depending on the
speed? That's got to be an interaction between the X and Y drives.

-- Steve

>>>> http://www.flickr.com/photos/**beyond_io/8481541313/in/set-**
>>>> 72157632746756009/<http://www.flickr.com/photos/beyond_io/8481541313/in/set-72157632746756009/>

>>>>
>>>> Hope this helps!
>>>>
>>>> On Sunday, February 17, 2013 12:16:12 PM UTC+1, stefanix wrote:
>>>>
>>>>
>>>>
>>>> I am curious to see if this helps. I doubt it's the source of the
>>>> problem but it might reduce the ripples quite a bit.
>>>>
>>>> I have just ordered some steppers with half the step angle (0.9
>>>> instead
>>>> of 1.8 deg). We have initially decided against them because they
>>>> are
>>>> less common but maybe they have a positive impact on the
>>>> oscillation. I
>>>> also will be trying a damper from Nanotec.
>>>>
>>>> --
>>>> Stefan Hechenberger
>>>> studio: Nortd Labs - labs.nortd.com <http://labs.nortd.com>
>>>> resident: F.A.T. Lab - fffff.at <http://fffff.at>
>>>> project: Lasersaur - labs.nortd.com/lasersaur

>>>> <http://labs.nortd.com/**lasersaur

>>>> > BEAMBEAMBEAM+sorbothane+**ALUMSTRIP
>>>> > BEAMBEAMBEAM+sorbothane+**ALUMSTRIP
>>>> > BEAMBEAMBEAM+sorbothane+**ALUMSTRIP
>>>> > BEAMBEAMBEAM+sorbothane+**ALUMSTRIP
>>>> > BEAMBEAMBEAM+sorbothane+**ALUMSTRIP
>>>> > BEAMBEAMBEAM+sorbothane+**ALUMSTRIP
>>>> >

>>>> > For more options, visit https://groups.google.com/**
>>>> groups/opt_out <https://groups.google.com/groups/opt_out>
>>>> <https://groups.google.com/**groups/opt_out<https://groups.google.com/groups/opt_out>

>>>> >.
>>>>
>>>> >
>>>> >
>>>>
>>>> --
>>>> You received this message because you are subscribed to the Google
>>>> Groups "lasersaur" group.
>>>> To unsubscribe from this group and stop receiving emails from it, send
>>>> an email to

>>>> lasersaur+unsubscribe@**googlegroups.com<lasersaur%2Bunsu...@googlegroups.com>

>>>> .
>>>> For more options, visit

>>>> https://groups.google.com/**groups/opt_out<https://groups.google.com/groups/opt_out>

>>>> .
>>>>
>>>>
>>>>
>>> --
>>> You received this message because you are subscribed to the Google
>>> Groups
>>> "lasersaur" group.
>>> To unsubscribe from this group and stop receiving emails from it, send
>>> an
>>> email to

>>> lasersaur+unsubscribe@**googlegroups.com<lasersaur%2Bunsu...@googlegroups.com>

>>> .
>>> For more options, visit

>>> https://groups.google.com/**groups/opt_out<https://groups.google.com/groups/opt_out>

>>> .
>>>
>>>
>>>
>>
>
> --
> You received this message because you are subscribed to the Google Groups
> "lasersaur" group.
> To unsubscribe from this group and stop receiving emails from it, send an

> email to lasersaur+...@googlegroups.com.

> For more options, visit https://groups.google.com/groups/opt_out.
>
>
>

-- Steve

[Justin Krull]

The problem is specific (low) stepping frequencies on the Y-axis. The shallower the angle, the lower the stepping frequency. That's why you see the resonance at the more horizontal parts of these spirals. It's also symmetrical.

We have a damper on the way to install this week. I'll write back with results.

[Justin Krull]

Till has the right idea. Here's the page to order them in the US: http://catalog.orientalmotor.com/item/all-categories/ac-motor-accessories/d6cl-8-0f?plpver=11&categid=100&prodid=3001035&origin=keyword

"Hi Justin,

i'd like to buy these ones

D6CL-8.0F

http://www.orientalmotor.de/Products/Stepper_motors/Controllers_accessories/Accessories/Dampers/?&dwn=artnr&arid=6271

they will fit on the 8mm driveshaft and also on top of the screw from nema17. I also don't think it has to do with the beams. 

i get the same ripples and have the old lasersaur hardware 11.08..

best till"

[Kevin]

More questions!

As we've been trying some acrylic cuts/engraving lately, we saw quite a lot of "toothing" in the cuts and "dotting" in the engravings.

Is this a laser pulse related thing like mentioned here: 

http://www.buildlog.net/blog/2011/12/getting-more-power-and-cutting-accuracy-out-of-your-home-built-laser-system/

and can we fix it? We're using the analog laser PSU control, I wonder if people using the digital TH have the same issues.

Or is the laser beam constant and is it a stepper related thing, that we're actually seeing each step?

On Monday, February 11, 2013 9:43:48 PM UTC+1, stefanix wrote:

Let's talk about this in a more organized way and make sure we are all
on the same page. There are probably three different issues that get
convoluted often but may also be related to some degree:

1) ripples after sharp turns
2) ripples in the middle of a curve or line
3) perforation of a perfectly placed cut

1) and 2) are feedrate-related and most likely caused by some kind of
resonance. On our machine these ripples start getting very noticeable at
speeds above 4000mm/min. Others have reported to see significant ripples
at lower speeds.

3) is possibly related to idiosyncrasies of the Reci lasers or how we
pulse them. Generally I don't think of this as a big problem but for
some materials it may create rough edges. Perforation may be less when
using analog control etc ...

I think we need to start posting some images that we can compare. From
this we can then better differentiate between issues that are related to
design limitation and assembly/tweaking issues.

The following info would be highly useful:

- image of cut showing ripples
- feedrate, power
- version of LasaurApp and firmware
- new or old gantry (ref design changed about mid-2012)
- modifications from ref design

[Steve Baker]

None of the 'dotting/toothing' that I'm seeing can be related to the laser
pulsing...it's impossible.

The laser tube doesn't "know" how the head is moving - so the laser power
can't be related to that. A pulsing problem of this kind ought to be
visible in long, straight cuts that are parallel to the axis...and I've
*never* seen that.

So, what is it?

Well, there seems to be at least three separate effects here:

1) After sharp turns at high speeds or after repositions of an inch or
more in length, the system vibrates for the first centimeter or so of
motion afterwards. These seem to vary in amplitude depending on where you
are on the bed.

2) When cutting or etching at some fairly specific slope angles, you get
continuous vibrations with frequencies that are inversely proportional to
the head speed.

3) When the head is accelerating or decelerating, the laser is moving more
slowly than you specified. This deposits more energy into the cut
whenever the laser starts and stops or switches direction radically. This
is a particular problem when engraving rather than cutting because when
cutting, you don't care so much about excess energy in the cut.

I'm not aware of any issues that don't fall into one of those three buckets.

Fixing them is a bit tricky:

(1) may be fixed by the latest software release because Stefan has reduced
the acceleration and deceleration rates - hopefully this allows the
vibration to damp down more naturally.

(3) may also be fixed - I think Stefan said that he's now reducing the
laser energy during the acceleration and deceleration phases. I'd be
surprised if that worked perfectly the first time around - but it's
definitely a start.

The most problematic issue is (2) - and we're actively discussing a
solution for that. Motor dampers and belt tensioning seem like the most
likely solutions...but there are other possibilities such as using motors
with a smaller step size.

It's also possible that a fix for (2) would fix (1) also - which would be
good because decreasing acceleration and deceleration rates could make the
machine dramatically slower at making small objects with many lines in the
centimeter range.

So, I suggest upgrading your code to yesterdays' release.

I have yet to do that - so I can't say for sure whether it's a complete fix.

-- Steve

[Kevin]

Thanks again for the elaborate explanation Steve.

You're probably right, if it would be laser pulse related, there should also be a clearly visible difference between the dots at different speeds.

So if it's stepper related, as we're also seeing the "toothing" on straight cuts on both axes it might be due to speed. I'll do some speed tests.

I know "some" toothing is usually normal with a lasercutter, I'm not that experienced on the matter, but the cuts I have from the Epilog I used at ITP last summer showed some of it too. The main difference with what we're seeing is that the toothing is much coarser on ours, .3-.5mm dots/teeth vs .02-.05mm on the epilog (very nearly visible). Which makes me think it might also be a focus issue.. will have to test that in the morning. 

This brings us to another question! Focus. (Sorry for the slight off topic here) What's the best technique to do this, until now, we haven't used a specific technique to decently calibrate lens height and get the focus point just right for different materials. I've read about the "ramp test" today, looks like a decent way to get it just right. 

As we'll probably be switching materials quite a lot, it would be good to find a good & precise way to refocus. Atm I'm thinking of getting the focal point right on top of our honeycomb surface when the lens is fully screwed in, and adjust the lens by putting a small piece of material between the lens holder en adjustment ring (if that makes any sence). Basically just moving the lens up for the exact same height as the material that has to be cut.

[Ira Burton]

I have also watched the PWM to the power supply with my oscilliscope, and I do not see any variation or weird spikes, so that would also discount the possibility that it is coming from the tube itself.

[peter]

Yes, I have the same question. How to do the Laser System?

Is it a PPI mode?

Best Regards,
Peter
Founder and Business Director

-----Original Message-----
From: lasersaur+bncbd753on2t...@googlegroups.com [mailto:lasersaur+bncbd753on2t...@googlegroups.com] On Behalf Of Steve Baker
Sent: Wednesday, February 20, 2013 5:39 AM
To: lase...@googlegroups.com
Subject: Re: [lasersaur] Re: ripples, resonance, perforation

[Steve Baker]

(Sending again, to the group - and with the attachment!)

Kevin wrote:
> Thanks again for the elaborate explanation Steve.
>
> You're probably right, if it would be laser pulse related, there should
> also be a clearly visible difference between the dots at different
> speeds.

Yes - that too.

> So if it's stepper related, as we're also seeing the "toothing" on
> straight cuts on both axes it might be due to speed. I'll do some
> speed tests.

You might also want to play with the belt tension. We don't get any
problems at all on long straight cuts that are parallel to either axis.

> I know "some" toothing is usually normal with a lasercutter, I'm not
> that
> experienced on the matter, but the cuts I have from the Epilog I used at
> ITP last summer showed some of it too. The main difference with what
> we're
> seeing is that the toothing is much coarser on ours, .3-.5mm dots/teeth
> vs .02-.05mm on the epilog (very nearly visible).

I haven't used Epilog's much - but I had access to the ATX Hackerspace ULS
laser - and I didn't see any "toothing" at all on that machine. I don't
think it should happen on a perfectly set up machine.

> Which makes me think it might also be a focus issue.. will have to
> test that in the morning.

Again, I doubt it.

> This brings us to another question! Focus. (Sorry for the slight off
> topic
> here) What's the best technique to do this, until now, we haven't used a
> specific technique to decently calibrate lens height and get the focus
> point just right for different materials. I've read about the "ramp
> test"
> today, looks like a decent way to get it just right.

I've played with that...not my favorite technique.

Here is what we do:

1) In the top of the lens tube, screw the lower ring down as far as it
will go. Place the lens on top and loosely tighen the upper ring on top of
it. That places the bottom surface of the lens exactly 10mm below the top
of the tube.

2) Measure from the top of the tube to the bed of the laser cutter. The
lens has a 75mm focal length - but it's 10mm down from the top - so the
total should be 85mm. We're told to measure from about the center of the
thickness of the lens - so it's more like 86.5mm. However, for wood, you
want to be focussed about half the thickness of the material - we use 3mm
plywood - so we actually set the top of the lens tube to about 88mm above
the bed. If you're using acrylic, focus to the top of the material
instead.

3) To make this measurement easy, I designed (and laser cut!) a measuring
gizmo. I've attached the design to this email for you - you need to cut
it out of 3mm material. I put lots of them onto a sheet because we keep
losing them!

4) I make this measurement in all four corners of the bed and again in the
center and use an average. Heavy-handed banging of material onto the bed
has occasionally nudged it a millimeter or two downwards - and I re-adjust
it as a part of my weekly maintenance cycle.

5) Warped material is a continual pain - even seemingly flat material can
easily put the focus out by several millimeters. Also, as our honeycomb
sheet has aged, it's warped a little too. We use painter's tape to tape
the edges of the material down onto the bed frame every foot or so (or
closer if necessary) - thereby taking out the warp from both material and
frame.

6) When using somewhat warped plywood, we flip the sheet over so that the
edges curve upwards and the sheet is concave when viewed from above. Thus
taping the edges is fairly effective at removing the warp. If you place
the sheet so that it's convex when seen from above, taping the edges
doesn't work.

7) Where possible, we try to cut the gross shape of the parts out first so
that when there is slight warpage, the parts fall flat onto the bed when
we cut the fine details and do the etching in the second and third passes.
You can't always do that - and it doesn't always help...but it's better
than not doing that!

-- Steve

[douglas repetto]

On 2/20/13 12:16 AM, Steve Baker wrote:
> 3) To make this measurement easy, I designed (and laser cut!) a measuring
> gizmo. I've attached the design to this email for you - you need to cut
> it out of 3mm material. I put lots of them onto a sheet because we keep
> losing them!

I have a similar tool for my laser. It has a couple magnets glued to it
so that we can stick it to the side of the laser and it stays put!
Everyone who uses the laser gets in the habit of grabbing the bed/focus
adjust wrench with the left hand and the focusing tool with the right
hand, then putting them back in exactly the same places when done.

best,
douglas



--
............................................... http://artbots.org
.....douglas.....irving........................ http://dorkbot.org
.......................... http://music.columbia.edu/cmc/music-dsp
...........repetto............. http://music.columbia.edu/organism
............................... http://music.columbia.edu/~douglas

[Justin Krull]

Kevin,

Can you post preliminary pictures of the "toothing" to make sure we're all on the same page?

Justin

--
You received this message because you are subscribed to the Google Groups "lasersaur" group.

To unsubscribe from this group and stop receiving emails from it, send an email to lasersaur+unsubscribe@googlegroups.com.

[Kevin]

Hey Justin,

Here's one quick pic of the "toothing":

http://www.flickr.com/photos/beyond_io/8492416631/in/photostream/lightbox/

As you can see, toothing is an overly exaggerated word for this. This is a sample of a 45degree cut. It looks much worse than it actually is, if you look at it from the side, you can only see the toothing very very slightly with a pair of good eyes. You do feel it really well you scratch it with your nail.

On horizontals and verticals, it's also visible but it's much less. The biggest difference I'm seeing compared with epilog samples is the actual width of each "tooth" (which is why I was thinking it could be focus related).

I also mentioned PPI before as I'm thinking this toothing/dotting might be related to that, if you have a look at this:

http://www.buildlog.net/blog/wp-content/uploads/2011/12/IMG_4254.jpg

referenced from this blog post: http://www.buildlog.net/blog/2011/12/getting-more-power-and-cutting-accuracy-out-of-your-home-built-laser-system/

It seems like we're experiencing something like the 4th line from the top.

From what I understand they basically modulate the laser pulse frequency according to the distance travelled so they manage to get a nice and smooth path compared to the toothing/dotting.

We'll be doing some more testing after doing some decent focussing work.

To unsubscribe from this group and stop receiving emails from it, send an email to lasersaur+...@googlegroups.com.

[Steve Baker]

It's always nice to tell us the speed and power setting you were using -
and also photograph it next to a ruler so we get a more accurate feeling
of scale.

I don't see why focus would be involved...focus isn't a time-varying or
space-varying thing (certainly not at the 30Hz-ish/millimeter size that
your photo kinda suggests). If anything, a defocussed laser would make a
blurrier, softer edge - it would tend to blur out the toothing.

-- Steve

>>> ..............................**................. http://artbots.org
>>> .....douglas.....irving.......**................. http://dorkbot.org
>>> ..........................
>>> http://music.columbia.edu/cmc/**music-dsp<http://music.columbia.edu/cmc/music-dsp>
>>> ...........repetto............**.
>>> http://music.columbia.edu/**organism<http://music.columbia.edu/organism>
>>> ..............................**.
>>> http://music.columbia.edu/~**douglas<http://music.columbia.edu/~douglas>

>>>
>>>
>>> --
>>> You received this message because you are subscribed to the Google
>>> Groups
>>> "lasersaur" group.
>>> To unsubscribe from this group and stop receiving emails from it, send
>>> an

>>> email to lasersaur+...@**googlegroups.com <javascript:>.
>>> For more options, visit
>>> https://groups.google.com/**groups/opt_out<https://groups.google.com/groups/opt_out>
>>> .

>>>
>>>
>>>
>>
>
> --
> You received this message because you are subscribed to the Google Groups
> "lasersaur" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to lasersaur+...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
>

-- Steve

[Kevin]

I was thinking about focus as the teeth are so "large", focusing should make the "hit dotsize" smaller, hence make the teeth smaller. Then again that's just a guess.

I'll make some time during the weekend to do some more testing and take better pictures, this was just a quick shot of a sample I still had in my bag.

[Erik Moon]

Here's another example of the toothing / combing power problem.

http://dl.dropbox.com/u/43029292/20130220_Lasersaur1.JPG

The material is 6 mm acrylic (nominal thickness is closer to 5.5 mm). Cutting speed = 800 / Power = 30%

The intent was to "vector engrave" some line art. When I do this same engraving on a commercial 100w laser cutter (Hurricane Lasers Charley), the engraving depth is consistent across all the image. There are no significant variations in engraving depth. When "engraved" on the Lasersaur, we get this very distinct combed / toothed pattern.

-Erik Moon

[Justin Krull]

Thanks for the photograph, Kevin. I'm looking forward to your testing.

Just to clarify for everyone, the Reci CO2 lasers are not pulsed. They are continuously on, when they are on. We use pulse-width modulation (PWM) to feed the power supply an average control voltage that corresponds to an output power. The laser is not pulsed to achieve an average power, like the article Kevin posted.

Pulsing the laser is suggested by that article as an improvement on the existing continuous-on, when on, mode. Again, the laser is absolutely not pulsed directly by us with the current mode of operation.

[Justin Krull]

That "toothing" looks very regular in Kevin's picture, but less so in Erik's, as a comment. Maybe the laser is pulsed naturally by the power supply somewhat. A cutting speed test would quickly tell whether that is the case.

[Erik Moon]

Justin-

I suspect Kevin is seeing his "toothing" in a straight line.

Mine is a vector pattern (company logo) -- so there are lots of twists, turns and corners in the cut path. I personally think it is related to acceleration / deceleration causing uneven dwell-time -- but I don't really have any specific evidence to back this up other than the fact that sharp corners are always cut deeper than straight segments. Wide curves look more regular like Kevin's pictures. Engraving text is the worst because of all the little twisties and corners that are inherent to text...

-Erik

[Stefan Hechenberger]

Oh, nice way to look at this. I should do the same when playing with the
beam dynamics code.

To clarify ... in a standard Lasersaur the laser is pulsed for any
intensity not 100%. The reference design has the PWM line going in at
the TH pin which is directly coupled with the laser output. The exact
power that is generated depends on the IN pin and trim pot. The IN pin
is typically tied to 5V and this is trimmed to about 26mA (which is the
maximum the laser should be driven at).

The PWM frequency is 31.25kH (although we can can change this). Setting
the laser to 30% means the laser
switches on for 9.6us every 32us (in therory, not sure how well the reci
lasers translate this. At the output there may just be the average).
Depending on the head velocity this leads to different cutting depth.

The beam dynamics code we have currently in v13.02 adjusts the intensity
based on velocity (by default very conservatively, but it's in there so
people can start playing).

I am not sure the perforation/serration and varying cutting depth are
linked. I have pretty much the same serration on acrylic when running at
100% (there is no pulsing).

The PPI article was very interesting. Thanks for posting. When I have
time I will see how this exactly differs from the dynamics code we have
now and how this can be applied to the Lasersaur. This approach can do
two things: (1) give good compensation for varying speed (2) overdrive
laser. It may be possible to run the laser at 140W but restrict the PWM
dudty cycle to 50%. This means each individual pulse is stronger but in
average the laser is still under 100W. For some materials this leads to
better cuts and less HAZ.

--
Stefan Hechenberger
studio: Nortd Labs - labs.nortd.com
resident: F.A.T. Lab - fffff.at
project: Lasersaur - labs.nortd.com/lasersaur

[Justin Krull]

A pulse width of 9.6 us is likely invisible to the laser power supply, but maybe not. That's what I was trying to get at. The article is using pulse widths of ms.

--
You received this message because you are subscribed to the Google Groups "lasersaur" group.

To unsubscribe from this group and stop receiving emails from it, send an email to lasersaur+unsubscribe@googlegroups.com.

[Justin Krull]

We should probably reduce the PWM frequency to something more on the order 1 kHz to start playing with pixels per inch. 

(1000 mm/min)/(1000 PPI) = 1.524 milliseconds/pulse

The 32 kHz PWM frequency might also explain the percent output amps vs. percent input plot that we get. I'll post this later tonight.

[Justin Krull]

Here's the info from the Epilog manual. They can adjust from 10 to 5000 PPI. Looking at the LasaurGRBL code, it seems like some work to implement PPI, but definitely doable. We might consider calling the control_laser_intensity method with both a PWM value and another PPI-related value to set the PWM frequency prescaler.

[Kevin]

Any news on the dampers? Really curious about those testresults!

On Monday, February 11, 2013 9:43:48 PM UTC+1, stefanix wrote:

Let's talk about this in a more organized way and make sure we are all
on the same page. There are probably three different issues that get
convoluted often but may also be related to some degree:

1) ripples after sharp turns
2) ripples in the middle of a curve or line
3) perforation of a perfectly placed cut

1) and 2) are feedrate-related and most likely caused by some kind of
resonance. On our machine these ripples start getting very noticeable at
speeds above 4000mm/min. Others have reported to see significant ripples
at lower speeds.

3) is possibly related to idiosyncrasies of the Reci lasers or how we
pulse them. Generally I don't think of this as a big problem but for
some materials it may create rough edges. Perforation may be less when
using analog control etc ...

I think we need to start posting some images that we can compare. From
this we can then better differentiate between issues that are related to
design limitation and assembly/tweaking issues.

The following info would be highly useful:

- image of cut showing ripples
- feedrate, power
- version of LasaurApp and firmware
- new or old gantry (ref design changed about mid-2012)
- modifications from ref design

[Stefan Hechenberger]

yes, I have installed one of the oriental motors dampers ... no
discernible difference.

http://www.flickr.com/photos/stfnix/8524588046/in/photostream/
http://www.flickr.com/photos/stfnix/8523475031/in/photostream/

I have also gotten 0.9 deg (instead of 1.8) stepper motors which I
haven't tried out yet. I would expect they give us smoother motion or at
least move the resonance to a different frequency. It will probably be
another week until I have time to test them.

--
Stefan Hechenberger
studio: Nortd Labs - labs.nortd.com

work: Institut f�r Experimentelle Architektur, UIBK

resident: F.A.T. Lab - fffff.at
project: Lasersaur - labs.nortd.com/lasersaur

> studio: Nortd Labs - labs.nortd.com <http://labs.nortd.com>
> resident: F.A.T. Lab - fffff.at <http://fffff.at>
> project: Lasersaur - labs.nortd.com/lasersaur

> <http://labs.nortd.com/lasersaur>

>
> --
> You received this message because you are subscribed to the Google
> Groups "lasersaur" group.
> To unsubscribe from this group and stop receiving emails from it, send

> an email to lasersaur+...@googlegroups.com.

[Gabriel Helms]

Would putting the dampers, not on the steppers, but in the transmission rod help?

Sent from my iPhone

On Mar 3, 2013, at 5:08 AM, Stefan Hechenberger <ste...@nortd.com> wrote:

>
> yes, I have installed one of the oriental motors dampers ... no discernible difference.
>
> http://www.flickr.com/photos/stfnix/8524588046/in/photostream/
> http://www.flickr.com/photos/stfnix/8523475031/in/photostream/
>
> I have also gotten 0.9 deg (instead of 1.8) stepper motors which I haven't tried out yet. I would expect they give us smoother motion or at least move the resonance to a different frequency. It will probably be another week until I have time to test them.
>
> --
> Stefan Hechenberger
> studio: Nortd Labs - labs.nortd.com

> work: Institut für Experimentelle Architektur, UIBK

[Kevin]

Today we noticed something rather strange. 

We were cutting regular plywood (3mm) squares for testing purposes and saw that our verticals were nice and smooth but our horizontals showed vibration patterns.

Some of it is what we usually encounter, residual vibration from the y-movement, but there's also a constant vibration on the Y-beam when moving in the X-direction which seems related to the X-stepper. Would it be possible that because of the fact that the X-stepper is not aligned in the center of the Y-drive beam, that it's centrifugal force generates a resonance on the large Y-drive beam? It's just a random thought, might as well share it!

On Sunday, March 3, 2013 2:08:15 PM UTC+1, stefanix wrote:

yes, I have installed one of the oriental motors dampers ... no
discernible difference.

http://www.flickr.com/photos/stfnix/8524588046/in/photostream/
http://www.flickr.com/photos/stfnix/8523475031/in/photostream/

I have also gotten 0.9 deg (instead of 1.8) stepper motors which I
haven't tried out yet. I would expect they give us smoother motion or at
least move the resonance to a different frequency. It will probably be
another week until I have time to test them.

--
Stefan Hechenberger
studio: Nortd Labs - labs.nortd.com

work: Institut f�r Experimentelle Architektur, UIBK

[Stefan Hechenberger]

I don't think I can reproduce this. You may want to check if the roller
bearings clamp the rails tightly. Does the y-bar have play when you try
to rotate it along its long axis? How about the x-cart?

Do you get those ripples on the entire line, even when it's say 10cm
long? Is this specific to a certain freedrate?

--
Stefan Hechenberger
studio: Nortd Labs - labs.nortd.com

work: Institut f�r Experimentelle Architektur, UIBK

resident: F.A.T. Lab - fffff.at
project: Lasersaur - labs.nortd.com/lasersaur

On 03/03/2013 07:10 PM, Kevin wrote:
> Today we noticed something rather strange.
> We were cutting regular plywood (3mm) squares for testing purposes and
> saw that our verticals were nice and smooth but our horizontals showed
> vibration patterns.
> Some of it is what we usually encounter, residual vibration from the
> y-movement, but there's also a constant vibration on the Y-beam when
> moving in the X-direction which seems related to the X-stepper. Would it
> be possible that because of the fact that the X-stepper is not aligned
> in the center of the Y-drive beam, that it's centrifugal force generates
> a resonance on the large Y-drive beam? It's just a random thought, might
> as well share it!
>
> On Sunday, March 3, 2013 2:08:15 PM UTC+1, stefanix wrote:
>
>
> yes, I have installed one of the oriental motors dampers ... no
> discernible difference.
>
> http://www.flickr.com/photos/stfnix/8524588046/in/photostream/

> <http://www.flickr.com/photos/stfnix/8524588046/in/photostream/>
> http://www.flickr.com/photos/stfnix/8523475031/in/photostream/

> <http://www.flickr.com/photos/stfnix/8523475031/in/photostream/>
>
> I have also gotten 0.9 deg (instead of 1.8) stepper motors which I
> haven't tried out yet. I would expect they give us smoother motion
> or at
> least move the resonance to a different frequency. It will probably be
> another week until I have time to test them.
>
> --
> Stefan Hechenberger

> studio: Nortd Labs - labs.nortd.com <http://labs.nortd.com>
> work: Institut f�r Experimentelle Architektur, UIBK

> > <http://labs.nortd.com/lasersaur

> <http://labs.nortd.com/lasersaur>>
> >
> > --
> > You received this message because you are subscribed to the Google
> > Groups "lasersaur" group.
> > To unsubscribe from this group and stop receiving emails from it,
> send

> > an email to lasersaur+...@googlegroups.com <javascript:>.

> > For more options, visit https://groups.google.com/groups/opt_out

> <https://groups.google.com/groups/opt_out>.

[Justin Krull]

We can also confirm the damper did next to nothing. The machine sounds quieter with it, but the ripples are completely unchanged.

I'm looking forward to the testing with a different stepper. We're going to look for a drop-in replacement.

On Sun, Mar 3, 2013 at 11:17 AM, Stefan Hechenberger <ste...@nortd.com> wrote:

I don't think I can reproduce this. You may want to check if the roller bearings clamp the rails tightly. Does the y-bar have play when you try to rotate it along its long axis? How about the x-cart?

Do you get those ripples on the entire line, even when it's say 10cm long? Is this specific to a certain freedrate?

--
Stefan Hechenberger
studio: Nortd Labs - labs.nortd.com

work: Institut für Experimentelle Architektur, UIBK

    work: Institut f�r Experimentelle Architektur, UIBK

an email to lasersaur+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/groups/opt_out.

--
You received this message because you are subscribed to the Google Groups "lasersaur" group.

To unsubscribe from this group and stop receiving emails from it, send an email to lasersaur+unsubscribe@googlegroups.com.

[Justin Krull]

Does anyone know what the "gold standard" in steppers is? I know the stepper drivers we're using are pretty well regarded, but it doesn't seem like the Lin Engineering or Nanotec steppers are what people regard as high quality.

[Chris Uhlik]

For a linear system to resonate, there are two issues to consider:

1. a periodic forcing function, i.e. some energy source at a given frequency

2. a low damping mode where energy is exchanged between two states, e.g. spring energy and kinetic energy

Choosing a "better" stepper motor or using a smoother microstep strategy attempts to address the problem by path 1, reducing the energy put into the vibratory mode.

Putting a damper on the stepper also mostly is about reducing the energy going into the system.

But, if the Q is high enough, energy reduction won't solve the problem.  Q is high when damping is low.  Think of Q as the ratio between energy stored in the spring to energy removed by the damping over one cycle of vibration.  Putting a damper on the stepper doesn't really take energy out of a vibration cycle --- the damper is on the wrong side of the drive belt to do that.

I suspect this problem will only be solved by damping the structure.  We need to understand what is moving and then add something that dissipates energy of that unwanted motion.

Chris

P.S. it gets more complicated for nonlinear systems.  Nonlinear systems can transfer energy from one frequency to another.  Our drive system is non-linear, but I don't think the structure is significantly nonlinear.

To unsubscribe from this group and stop receiving emails from it, send an email to lasersaur+...@googlegroups.com.

[Erik Moon]

The Hurricane Lasers commercial machine I have access to uses a very different drive system. It has a bed that is approximately 24" x 40".

Both the x and y axes are driven by significantly larger steppers with a reduction gear and significantly wider belts.

There is a small pulley on the stepper (approx 15-20 mm diameter) with a short (~150 mm) / wide (~20 mm) closed loop belt that connects to a larger sized pulley on a two-sided reduction gear (looks like approximately 5:1 -- both gears are ~ 20 mm wide). The smaller side of the reduction gear is about the same size as the first pulley ~15-20 mm diameter. It drives the belt connected to the carriage - also about 20 mm wide. Looks like both these belts use a GT2 or similar tooth profile.

I'm thinking this system would allow the reduction system to reduce vibration input to the system due to smoother stepping (e.g. reduction requires far more rotation of the stepper for even a very short distance) and because the wider belts would vibrate at a lower frequency. The short belt would also help isolate vibration of the carriage from vibration of the motor shaft.

If you guys would find it useful, I can try to post some photos of that machine's guts.

-Erik

[Steve Baker]

I agree entirely with Chris.

We need to have the ability to send high frequencies through the drive
system in order that the machine can cut sharp corners without rounding
them off. So filtering the input in an effort to remove energy at the
resonant frequencies is not such a great thing.

Worse still, I'm also beginning to suspect that the drive belts are what's
vibrating - and because the length of belt between drive sprocket and the
X or Y cart depends on where you are on the table...I'm beginning to think
that the resonant frequency might be different depending on where you are
in X/Y space.

I have anecdotal evidence for that from some of our work where we're
cutting and etching 200 identical parts out of one 2'x4' sheet.

With SVG optimization turned off - so that the RELATIVE motion of the head
is identical from one part to the next - we find severe "wobble" in the
etch path on some parts and no wobble whatever on others. With all other
things being equal, that says that the amount of vibration is X/Y
coordinate-specific. Sadly I haven't yet had time to look to see where
those parts came from out of the sheet...but I bet it would be instructive
to try some of our test patterns methodically over the entire bed of the
machine.

But if the drive system is adding energy at some specific frequency - and
the drive belt goes into resonance only when it's at the right length so
its' natural frequency matches that - then the nature of these vibrations
could very well be different depending on where in X/Y space you run your
tests!

I think the right thing to do here is to use a strobe lamp or something to
try to see the resonance happening in the structure - and then figure out
how to bleed away that energy to stop it from building up.

-- Steve

>>>>> http://www.flickr.com/photos/**stfnix/8524588046/in/**photostream/<http://www.flickr.com/photos/stfnix/8524588046/in/photostream/>
>>>>> <http://www.flickr.com/photos/**stfnix/8524588046/in/**photostream/<http://www.flickr.com/photos/stfnix/8524588046/in/photostream/>
>>>>> >
>>>>> http://www.flickr.com/photos/**stfnix/8523475031/in/**photostream/<http://www.flickr.com/photos/stfnix/8523475031/in/photostream/>
>>>>>
>>>>> <http://www.flickr.com/photos/**stfnix/8523475031/in/**photostream/<http://www.flickr.com/photos/stfnix/8523475031/in/photostream/>

>>>>> >
>>>>>
>>>>> I have also gotten 0.9 deg (instead of 1.8) stepper motors which
>>>>> I
>>>>> haven't tried out yet. I would expect they give us smoother
>>>>> motion
>>>>> or at
>>>>> least move the resonance to a different frequency. It will
>>>>> probably
>>>>> be
>>>>> another week until I have time to test them.
>>>>>
>>>>> --
>>>>> Stefan Hechenberger
>>>>> studio: Nortd Labs - labs.nortd.com <http://labs.nortd.com>
>>>>>
>>>>> work: Institut f�r Experimentelle Architektur, UIBK
>>>>> resident: F.A.T. Lab - fffff.at <http://fffff.at>
>>>>> project: Lasersaur - labs.nortd.com/lasersaur

>>>>> <http://labs.nortd.com/**lasersaur<http://labs.nortd.com/lasersaur>

>>>>> <http://labs.nortd.com/**lasersaur<http://labs.nortd.com/lasersaur>
>>>>> >
>>>>> > <http://labs.nortd.com/**lasersaur<http://labs.nortd.com/lasersaur>
>>>>>
>>>>> <http://labs.nortd.com/**lasersaur<http://labs.nortd.com/lasersaur>

>>>>> >>
>>>>> >
>>>>> > --
>>>>> > You received this message because you are subscribed to the
>>>>> Google
>>>>> > Groups "lasersaur" group.
>>>>> > To unsubscribe from this group and stop receiving emails from
>>>>> it,
>>>>> send
>>>>> > an email to lasersaur+...@googlegroups.com <javascript:>.
>>>>>

>>>>> > For more options, visit https://groups.google.com/**
>>>>> groups/opt_out <https://groups.google.com/groups/opt_out>
>>>>> <https://groups.google.com/**groups/opt_out<https://groups.google.com/groups/opt_out>

>>>>> >.
>>>>>
>>>>> >
>>>>> >
>>>>>
>>>>> --
>>>>> You received this message because you are subscribed to the Google
>>>>> Groups "lasersaur" group.
>>>>> To unsubscribe from this group and stop receiving emails from it,
>>>>> send
>>>>> an email to

>>>>> lasersaur+unsubscribe@**googlegroups.com<lasersaur%2Bunsu...@googlegroups.com>

>>>>> .
>>>>> For more options, visit

>>>>> https://groups.google.com/**groups/opt_out<https://groups.google.com/groups/opt_out>

>>>>> .
>>>>>
>>>>>
>>>>>
>>>> --
>>>> You received this message because you are subscribed to the Google
>>>> Groups "lasersaur" group.
>>>> To unsubscribe from this group and stop receiving emails from it, send
>>>> an email to

>>>> lasersaur+unsubscribe@**googlegroups.com<lasersaur%2Bunsu...@googlegroups.com>

>>>> .
>>>> For more options, visit

>>>> https://groups.google.com/**groups/opt_out<https://groups.google.com/groups/opt_out>

>>>> .
>>>>
>>>>
>>>>
>>>
>> --
>> You received this message because you are subscribed to the Google
>> Groups
>> "lasersaur" group.
>> To unsubscribe from this group and stop receiving emails from it, send
>> an

>> email to lasersaur+...@googlegroups.com.
>> For more options, visit https://groups.google.com/groups/opt_out.
>>
>>
>>
>
> --
> You received this message because you are subscribed to the Google Groups
> "lasersaur" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to lasersaur+...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
>

-- Steve

[Steve Baker]

I'm definitely in favor of trying a wider belt because that would also
reduce the problem of tooth-skipping.

-- Steve

> --
> You received this message because you are subscribed to the Google Groups
> "lasersaur" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to lasersaur+...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
>

-- Steve

[Justin Krull]

It's not the structure or the belts. It's the stepper settling about its set point at lower frequencies. The problem will only be fixed by reducing the inertia of the cart significantly or changing out the stepper.

I am looking to Stefan here with his new Nanotec 0.9 degree stepper test.

Can anyone suggest a manufacturer of high quality steppers? We are looking for one to order a drop-in replacement from to also perform some tests. I'll do some research tonight if no one has suggestions.

[Justin Krull]

Put your hands on various parts of the structure while running the machine on a job that has a lot of varying step frequencies. You'll see that there just isn't enough energy going into the system to resonate the structure to produce the ripples we're seeing.

I am certain that testing a different stepper is the right first path.

[Chris Uhlik]

Justin,

Perhaps you are right about there being no significant structural vibration.  I suspect Steve is right about the problem being drive belt banjo mode vibrations.  Thin, high-tension drive belts more closely approximate banjo strings than thicker, wider belts.  More belt mass lowers frequency, less tension lowers frequency, and more belt material means more visco-elastic damping taking energy out of the belt.  

Changing stepper motors is an easier design change, so go ahead and try it.  But if (when?) that doesn't stop the drive belts from vibrating because it didn't lower the Q of the drive belt system, then I suggest trying wider, thicker belts.

Chris

[T. Hergenhahn, Aeronauten]

Hi Justin,

the guy from oriental suggest me the controller stepper packages with 0,36 deg for round about 250,-Euro
maybe less. I ask for dampers but he just have the one stefan already bulit in.


best till

[Dion]

Any updates on this topic yet...?

It's been very quiet on here of late, is anyone still working on the Lasersaur?

[Justin Krull]

Stefan,

Have you been able to get the 0.9 deg Nanotec on your machine? It'd also be great to know whether it's a drop-in replacement on the NA BOM.

Justin

On Thu, Mar 14, 2013 at 5:21 AM, Dion <dion....@gmail.com> wrote:

Any updates on this topic yet...?

It's been very quiet on here of late, is anyone still working on the Lasersaur?

--

[Stefan Hechenberger]

not yet ... next week I will probably have a chance to try it out.

--
Stefan Hechenberger
studio: Nortd Labs - labs.nortd.com

work: Institut f�r Experimentelle Architektur, UIBK

resident: F.A.T. Lab - fffff.at

project: Lasersaur - labs.nortd.com/lasersaur

On 03/14/2013 05:49 PM, Justin Krull wrote:
> Stefan,
>
> Have you been able to get the 0.9 deg Nanotec on your machine? It'd also
> be great to know whether it's a drop-in replacement on the NA BOM.
>
> Justin
>
>
> On Thu, Mar 14, 2013 at 5:21 AM, Dion <dion....@gmail.com
> <mailto:dion....@gmail.com>> wrote:
>
> Any updates on this topic yet...?
>
> It's been very quiet on here of late, is anyone still working on the
> Lasersaur?
>
> --
> You received this message because you are subscribed to the Google
> Groups "lasersaur" group.
> To unsubscribe from this group and stop receiving emails from it,
> send an email to lasersaur+...@googlegroups.com

> <mailto:lasersaur%2Bunsu...@googlegroups.com>.

[Kevin]

We ran a few more tests regarding the beam/visible pulsation.

Results:

- latest software/PWM on TH: etching long lines at high speeds makes the pulses visible (which is logical):

http://www.flickr.com/photos/beyond_io/8577373798/in/photostream
- latest software reverted to 31Khz but still with power modulation depending on speed/PWM on IN: beam is always on and power is modulating which is better for clean cuts (definitely in acrylic). One of the obvious problems we're having again is the power setting limitation of S30.

Controlling through the IN seems better in most use cases, no more need for PPI as the beam is constantly on with a variable power. Atm there are two obvious issues though: the power setting limitation of S30, etching on S30 is much too high for thin/soft materials, and safety, controlling through IN bypasses some of the current failsafes which leads to laser firing at unexpected times.

Another thing we encountered today, there's a huge difference on the beam pulse depending on the movement direction:

- Y-axis: http://www.flickr.com/photos/beyond_io/8577373730/in/photostream/
- X-axis: http://www.flickr.com/photos/beyond_io/8577373766/in/photostream/

Both cut at F1500 S255. We also tested this with a longer line going in just 1 direction to make sure the results weren't affected by resonance caused by previous motion and had the exact same results. A lot of "toothing" in X-direction, a pretty stable beam in Y. Any idea what could be causing this? I see 2 possible reasons: there's something off with the X-stepper, or activating the X-stepper is influencing the PWM of the laser because of shielding issues.

Anyhow, food for thought!

On Monday, February 11, 2013 9:43:48 PM UTC+1, stefanix wrote:

Let's talk about this in a more organized way and make sure we are all
on the same page. There are probably three different issues that get
convoluted often but may also be related to some degree:

1) ripples after sharp turns
2) ripples in the middle of a curve or line
3) perforation of a perfectly placed cut

1) and 2) are feedrate-related and most likely caused by some kind of
resonance. On our machine these ripples start getting very noticeable at
speeds above 4000mm/min. Others have reported to see significant ripples
at lower speeds.

3) is possibly related to idiosyncrasies of the Reci lasers or how we
pulse them. Generally I don't think of this as a big problem but for
some materials it may create rough edges. Perforation may be less when
using analog control etc ...

I think we need to start posting some images that we can compare. From
this we can then better differentiate between issues that are related to
design limitation and assembly/tweaking issues.

The following info would be highly useful:

- image of cut showing ripples
- feedrate, power
- version of LasaurApp and firmware
- new or old gantry (ref design changed about mid-2012)
- modifications from ref design

--
Stefan Hechenberger
studio: Nortd Labs - labs.nortd.com

[Steve Baker]

Kevin wrote:

> Another thing we encountered today, there's a huge difference on the beam
> pulse depending on the movement direction:

...I doubt this is a PWM issue.

If you watch the X-drive belt, it visibly vibrates like a plucked string
when the machine is in motion. That must cause a cyclic change in the
belt tension - which in turn must exert an X-direction force on the head -
resulting in a sinusoidal acceleration of the head along the X axis.

When the vibration momentarily pulls the head in the same direction as the
cut, the head briefly moves faster than it should - so there is less
energy deposited at each point along the cut than there should be. When
the vibration direction reverses a fraction of a second later, it pulls
the head back the other way, the velocity along the cut line is reduced
(maybe briefly halted or even reversed) and so there is more energy
deposited at each point.

That effect would mimic a PWM issue by varying the amount of total energy
deposited at each point along the line as a function of time rather than
distance moved.

It would be surprising if this *didn't* happen - at least to some degree.

The only question is whether the magnitude of the force is enough to
explain the effects you're seeing.

An interesting test would be to dramatically change the belt tension and
thereby (presumably) alter the frequency of the vibration, then see if the
frequency of these ripples changes.

If your theory about power supply interference is correct, then the
distance between the ripples shouldn't change.

-- Steve

> --
> You received this message because you are subscribed to the Google Groups
> "lasersaur" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to lasersaur+...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
>

-- Steve

[Kevin]

Good catch Steve :)

Lowering the belt tension changed the result but finding a sweet spot seems like a harder thing to do.

Anyhow, the X belt/stepper seems to be te culprit of most of the resonance issues, wonder how we're going to fix that.

[Steve Baker]

Awesome! My half-assed knowledge of physics produced results!!

I didn't expect that changing the belt tension would fix the problem...but
it does help to eliminate electrical/laser issues from the list of
possible causes.

I think there are resonance and vibration issues *everywhere* here - and
I'm not optimistic about fixing most of them.

Heavier/wider drive belts sound like they might be a good thing -
something to reduce the available energy for vibration...they might also
allow us to lower the belt tension without so much risk of skipping a
tooth...lower tension moves the resonant frequency someplace else - maybe
far enough away from the stepper motor frequencies to solve the problem.

Some kind of automatic belt tensioner would allow us to control the
problem better - right now, getting the tension "right" is a bit of a
crap-shoot.

More drastically, I suppose one could consider switching from belt drives
to lead-screws.

I doubt there is a quick fix.

Ridding a complex machine of resonance problems seems to be a bit of a
black art - there don't seem to be well-established principles here.

-- Steve

>> > email to lasersaur+...@googlegroups.com <javascript:>.

[Stefan Hechenberger]

Interesting, thanks for posting.

I have yet to try out the 0.9 deg steppers but another thing high on my
list is belt rails/dampers. It's quite possible the extensive swing of
the belts (which is visible) functions as an amplifier for other sources
of vibration. It would actually be a really easy fix. Anybody has a 3d
printer who can prototype these quickly?

> difference in cutting quality

It's probably a reasonable conjecture that it may have to do with the
belt vibration. It's also possible it's interrelated with beam
polarization. A linearly polarized laser beam has varying performance
depending on the cutting direction:

http://www.lasermech.com/polarizer.html
http://www.synrad.com/duolase/duolase_main.htm
http://www.laser.ru/personal/niziev/papers/cutting.pdf

--
Stefan Hechenberger
studio: Nortd Labs - labs.nortd.com

work: Institut f�r Experimentelle Architektur, UIBK

resident: F.A.T. Lab - fffff.at
project: Lasersaur - labs.nortd.com/lasersaur

> studio: Nortd Labs - labs.nortd.com <http://labs.nortd.com>

> resident: F.A.T. Lab - fffff.at <http://fffff.at>
> project: Lasersaur - labs.nortd.com/lasersaur

> <http://labs.nortd.com/lasersaur>

[Steve Baker]

Hmmm - I hadn't thought of beam polarization...and I've always seen
identical power from both X-direction and Y-direction cutting on the
lasersaur.

However, on the 60 watt ULS laser cutter that I used at ATXHackerspace, I
often noticed different cut depths when moving in the X and Y directions.
I'd even make two identical parts, one rotated 90 degrees to the other to
check this.

I need to read more about this phenomenon.

-- Steve

Stefan Hechenberger wrote:
>
> Interesting, thanks for posting.
>
> I have yet to try out the 0.9 deg steppers but another thing high on my
> list is belt rails/dampers. It's quite possible the extensive swing of
> the belts (which is visible) functions as an amplifier for other sources
> of vibration. It would actually be a really easy fix. Anybody has a 3d
> printer who can prototype these quickly?
>
> > difference in cutting quality
>
> It's probably a reasonable conjecture that it may have to do with the
> belt vibration. It's also possible it's interrelated with beam
> polarization. A linearly polarized laser beam has varying performance
> depending on the cutting direction:
>
> http://www.lasermech.com/polarizer.html
> http://www.synrad.com/duolase/duolase_main.htm
> http://www.laser.ru/personal/niziev/papers/cutting.pdf
>
> --
> Stefan Hechenberger
> studio: Nortd Labs - labs.nortd.com

> work: Institut für Experimentelle Architektur, UIBK

-- Steve

[Kevin]

While cutting a streetmap yesterday, we noticed obvious ripples at certain angles (like in the spirals we tested a while ago).

It turned out to be exactly at around 18 deg. which happens to be exactly 10 times the step angle. So we did some more testing:

http://www.flickr.com/photos/beyond_io/8589120988/in/set-72157632746756009/

http://www.flickr.com/photos/beyond_io/8589120774/in/set-72157632746756009/

http://www.flickr.com/photos/beyond_io/8589120610/in/set-72157632746756009/

It's only happening for the horizontal diamonds, so it seems to be related to the Y-axis (seems like I blame a different axis every other week).

I also tested 90deg angles at high speed, and I only get overshoot from the Y-axis: cutting the vertical at high speed, followed by the horizontal, gives waves at the start of the horizontal.

This doesn't happen on the X-axis, no matter what the speed is (I tested up to F8000). 

Could the overshoot be related to Y-stepper torque?

I have no clue what the 18deg angle means, all I can say is that damping the X and Y cart by hand didn't change the results, so it doesn't seem to be a resonance issue.

[Dion]

I'm not as far as you guys are with your builds, but I'm very interested in this thread, because I'll have to implement what ever changes are needed down the line. So I'm by no means claiming to know what I'm talking about here... :-)

However, there's been a lot of talk about the stepper motors being the root cause of the problem, but has anyone given a thought to it possibly coming from somewhere else e.g. from the tube itself?

Feel free to flame me or slap the back of my head, but I just think that how ever the vibration is being caused, it is being transmitted into the frame and also into the tube...and that can't be good.

Perhaps the tube must decoupled from the entire frame to ensure that it doesn't aid in propagation of the vibration?

[Steve Baker]

The vibration issue is horribly complicated because the "diagnosis tree"
is both broad and deep.

There are two core issues there:

1) What is the source of the vibration?
2) How is that vibration affecting the path of the laser beam as it hits
the target material?

It seems to me that if the laser tube were the *CAUSE* of the vibration -
then it would also be at least one of the *EFFECTS* of it. So if we can
show that the tube itself isn't vibrating, then it's hard to see how it
could be the cause of vibration in something else...right?

Let's just break down question (2) and look at how vibration changes the
path of the laser beam across the bed...

2.1) By vibrating the laser tube itself.
2.2) By vibrating one or more of the mirrors.
2.3) By vibrating the Y cart.
2.4) By vibrating the X cart.
2.5) By vibrating the bed (and hence the material itself).

This ends up being a complicated matter to sort out...but if the laser
tube were somehow the cause of the vibration then (2.1) would be true. If
so, then there are two possibilities:

2.1.1) The laser tube is vibrating "rotationally".
2.1.2) The laser tube is vibrating "translationally".

If (2.1.1) is true then we're supposing that the laser tube is being
vibrated more at one end than the other - or both that both ends are
vibrating - but out of sync with each other. Under those circumstances,
the tube would be rotating/twisting and the laser output would be being
aimed in a variety of different directions over time. But if that were
true then the positional error of the beam as it hits the target would be
much larger when at the bottom-left corner of the bed than at the
top-right corner - because the displacement of the beam increases the
further you get from the point of rotation. We don't see that symptom,
the wobble seems to be fairly constant in magnitude over the entire bed.
So that cannot be happening and (2.1.1) must be FALSE.

If (2.1.2) is true then we're supposing that the laser tube is vibrating
but only "translationally". It's moving - but the beam is staying
perfectly parallel with it's designed path. That would explain the
symptoms - but because there are two brackets holding the laser tube,
spaced several feet apart - then for a mechanical vibration to cause
purely translational motion, it would have to move both brackets by
precisely the same amount at precisely the same time! This seems
spectacularly unlikely. Even if both ends were moving...what are the odds
that there would be no phase delay and no attenuation of the vibration
from one end of the tube to the other and thus induce a rotational change
(ie 2.1.1). This seems impossible, so (2.1.2) is almost certainly FALSE.

Hence (2.1) is FALSE - and we may discount tube vibration as the effect -
it's got to be the mirrors, the X or Y cart or the material itself that's
vibrating.

OK - so logic and observation rules out (2.1.1) and (2.1.2)...so I really
don't think the tube is the culprit. Other reasons why it seems unlikely
are:

* The tube is really heavy - and it's full of water!
* The sloshing of the water inside the tube would tend to absorb energy
and dampen out any resonant vibration.
* The tube is bolted to the back wall - which is one of the stiffest parts
of the structure.
* It has some vibrational damping due to the soft rubber between the
mounting brackets and the tube itself.
* The laser tube doesn't "know" which way the head is moving or at what
speed. However, it's clear that this "wobble" only happens at certain cut
speeds and angles...so it can't be the *CAUSE* of the problem.

Figuring out what the *EFFECT* of the vibration is may not help us much -
what we need to fix is the *CAUSE*.

-- Steve

[Dion]

What about trying to figure the amount of energy required to cause this vibration?

I mean, the entire frame weighs around 100kg and it just seems very unlikely to me that a small stepper motor can vibrate a 100kg stationary object.

Perhaps it's a resonant frequency that's hitting a "sweet spot" and making the whole frame resonate?

However, the results seen by dampening the x-axis and y-axis tend to suggest that this is not true, so then it most probably has to lie with a much lighter object, like a mirror or something in the tube or is it maybe the work surface that's vibrating?

What if the vibration is in the frame? How does one go about decoupling the important parts from the frame? This is what I'm most interested in, because if the problem is caused by a fundamental design issue, then I'll have to consider if I continue and then I'll have a very expensive aluminium frame in my workshop...

[Kevin]

It's testing bonanza today!

To make sure our results were somewhat conclusive I made another pattern, to find out our previous results weren't "conclusive" at all:

Every angle is a multiple of 1.8deg.

- F600: http://www.flickr.com/photos/beyond_io/8589762314/in/photostream/

Still ripples, different wavelength, different spots.

- F1200: http://www.flickr.com/photos/beyond_io/8588662133/in/photostream/

Same ripples as with previous tests obviously, extra ripple visible at 1.8deg.

- F2400: http://www.flickr.com/photos/beyond_io/8588661899/in/photostream/

Ripples at different angles again + after tightening our Y belts, the ripples got amplified,wavelength changed.

We tightened the Y belts to see if the overshoot would disappear on the 90deg. angles, which it did, so the overshooting I mentioned earlier has nothing to do with stepper torque, but with belt tension.

To sum up, back to square one, it still does seem to be related to the Y stepper though, the Y belt tension just amplifies the source of the problem, then again, this starts to feel more like a guessing game.

[Justin Krull]

What is the angle with the large "amplified" ripples in the right picture (better yet, what are the angle increments between lines)?

--

[Jonathan Buford]

Has anyone tried a dampening idle wheel anywhere it would fit. Basically something that rests against the belt and would dampen out oscillation?

--

[Justin Krull]

Guys, it isn't the belts. It's the stepper overshooting and undershooting position at specific stepping frequencies. The tests by Kevin prove the hypothesis that was presented earlier on this..

[Kevin Verelst]

Every angle is a multiple of 1.8degrees. So the "amp'd" one is at 11.8degrees (actually 191.8deg).

-- 

Kevin Verelst

Sent with Sparrow

[Steve Baker]

The thing that keeps coming back to haunt me about this is that there is
this sensitive dependence on the angle of the line.

I can think of two reasons why this might be:

1) Perhaps the beat frequencies (the sum and difference of the frequencies
of the two motors) generates a new frequency of vibration that hits the
resonant frequency of some belt or other part of the structure.

2) When you do your testing at (say) 1200mm/min, then draw lines at all
sorts of angles, the stepper motors are not both rotating at 1200mm/min
rotation rates. At 45 degrees, for example, both X and Y motors are
moving at only 850mm/min. At every angle, you're moving each motor at
some different speed.

So your tests aren't testing what you think...suppose you got resonance on
a 45 degree line at 1200mm/min. That could be because the system
resonates when either motor runs at 850mm/min. When you run the same
pattern at higher speed, then the slope of the line at which the offending
motor is running at 850mm/min will be different.

But it's hard to tell at what X and Y motor speed the resonance is bad
with the tests that you're doing.

So, if you have a moment, try drawing a bunch of lines at a range of
angles - but this time, calculate the speed such that the Y motor always
moves at the same speed and the slope of the line is determined solely by
the speed of the X motor. You'll also need to vary the laser power
proportionate to the actual speed to avoid making big scorch marks at some
angles and not making any mark at all at other angles - so you'll probably
need to write some software to generate the G-code).

Then, just for grins, draw the exact same set of lines but with the X
motor constant in speed and the Y motor varying.

Doing that will truly isolate what one motor is doing from the other...and
if one of them is the specific culprit at specific speeds - then that will
stand out clear as day.

-- Steve

> --
> You received this message because you are subscribed to the Google Groups
> "lasersaur" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to lasersaur+...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
>

-- Steve

[Justin Krull]

I'm not sure whether the problem changes with microsteps or full steps, so here's analysis that includes both. My current thinking is that it has to do more with the full steps, so the 33 Hz step frequency on the Y-axis appears to be the problem. If you change the cutting speed, the angle you see those resonant vibrations will shift because the Y-axis mm/min will change, which will confirm the 33 Hz issue.

Does anyone besides Stefan have another drop-in replacement stepper they can try? Kevin, can you post the cut files, so we can reproduce your cuts?

[Kevin Verelst]

Here's the angle tester!

Steve, I'll do a manual test, disabling the X-stepper altogether see if the ripples are still visible.

[Justin Krull]

Here's the analysis for the most prominent lines on the other two speeds. Looking at the common frequencies between the speeds, it looks like 30-40 Hz is the issue, regardless of full step frequency/micro-stepping frequency for the current micro-stepping frequency of 2000 microsteps per revolution.

[Kevin]

So we disabled the X-stepper, and manually moved the X-cart running the Y at the resonant frequency we found earlier, we also ran the same test on another "angle" which wasn't showing the ripples earlier and the difference is clearly visible. So I guess we can conclude there's something off in the Y-stepper.

test lines: http://www.flickr.com/photos/beyond_io/8589100499/in/set-72157632746756009/

scientific test setup: http://www.flickr.com/photos/beyond_io/8589101391/in/set-72157632746756009/

Now that we "discovered" this problem, how do we go about fixing it?