gavztheouch wrote:Here are a few pics of the laser setup with the old y-axis HTD 3mm belts. The run of belting was chopped down to about 300mm as a test to see if it would reduce the backlash, which it did suggesting it is possibly belt stretch

Hi Gav

Worrying, that's a similar type of belt I'm using (from motionco - HTD 9mm wide 3mm pitch , aluminium pulleys). After replacing my slotted shaft couplers for the same ones you are using (marchantdice), my backlash virtually disappeared. I say disappeared in the loosest sense, there is obviously still some, but it's barely visible (it's measurable of course- approx 0.1/0.2mm 'ish). To be honest, visible is more important to me than measurable these days as I no longer build parts for the hubble

I'll follow this new thread with interest, are you considering trying HTD in a different type of material? If so, then I'd be interested in any improved results (so I can change mine if it works, or let you spend the money if it doesn't )

Ballscrews should prove interesting, high helix screws are certainly not cheap, and in the size you need, they'll need to be rigid - you won't get away with little teflon coated jobs at 6mm diameter and 20mm pitch I'm guessing min 15mm in X?

Good luck with the mk2, it's looking good. I do fancy another build, slightly smaller perhaps (700-900mm X) and with decent quality slides like yours. I have a huge THK linear slide sitting in the garage, but it's just too hefty for a laser - more suited to that heavy duty cnc router.

Cheers

Neil

Had a quick look at the belt specs - not that much info really - mine aren't overprinted with a manufacturer.
As far as I can see on their site, tension should be 100N. Without getting involved, let's call it 10Kg or 22lb if you are metrically challenged. That should be easy to pull by hand without resorting to a screwdriver lever
Similar specs at gates show +\- 0.03mm at 1200mm belt length variation (I'm assuming manufacturing variation).
As for stretch, I don't know.
Perhaps someone with higher mechanical knowledge can suggest what a ballpark value might be for HTD 3M style (not the company) 9mm wide, 3mm pitch. I believe the core is glass with neoprene casing - maybe kevlar is better (although I can't think why kevlar would stretch less than glass).
As an aside, gates have a wee tension gadget where you pluck the belt and it listens for th tone - you put in some values and it tells you what tension is. I have an app (free I think) on my iphone that I use to tune my guitars, maybe that would work - given a similar overall length ?
Need to ask the guys at motionco methinks.

BTW, locking your two Y leadscrews together when both are being driven is not a great idea. However, driving one and belt to the other is ok - but you're then back at square one Its really only a problem when one fails (note the word 'when' .
I know of lots of systems that use two motors 'though - shopbot springs to mind.

Cheers

Neil

Those belts look about 30mm?

Like most things in life, I didn't give belt width much thought for mine - just picking something that was available rather than researching what would be best (cost is a huge factor of course )
I suppose (like resistors) a large x sectional area will give less 'resistance' - in our case stretch.

That's what you should do then - buy wider belts and pulleys, fit, test and report back to us here

I can't imagine epilog fitting wider belts for no good reason (unless they buy stock on a one size fits all scenario where the price cut is high )) I'm assuming that they have found this a problem too.

Cheers

Neil

Sorry to hear you've lost faith in timing belts.

I use a friend's VersaLaser. With a 12" X 24" working area, it's somewhat smaller than your 1200 mm X 900 mm laser, but the VersaLaser does not seem to have any problems with backlash due to belt stretch, and it uses a tiny belt. I could take another look, but it looks like a much longer version of the .25" MXL timing belt used in the ORD Bot. It's a very light duty belt, but the carriage of the laser has very little mass and it glides easily with little friction. Maybe VersaLaser is doing some sort of backlash compensation? It cuts very round circles, though.

When I've finished my current projects, I'd like to build a 60W laser with a 12" X 24" or slightly larger working area, so I'm reading the build logs trying to get an idea of what works and what doesn't. Thanks for documenting your build.

Thanks for that hobby CNC ball screw turning tutorial. Bookmarked! That was pretty much what I had envisioned, but the tips and tricks were excellent. I've been procrastinating the ball screw update on my CNC lathe project.

Too many projects, not enough time! Too much time on BuildLog.net?

they are usually hardened, so you need to grind off the top layer of metal. once you are through that, they cut very easily. it's also a great excuse for buying more tooling for your lathe

we used a lathe mounted grinder, but you can even use something like a dremel grinder as long as you set it up right and take your time.

a quick search led to this: http://alisam.com/page/1gu8y/Metalworki ... lders.html

something like that would work, and I would imagine you could fab your own rather than buying one.

That begs the question... how does someone bend a 15mm diameter ball screw? Did they use it as a pry bar to lift concrete beams to free victims trapped in earthquake wreckage before shipping it to you? I know kids with that special power. Give them an anvil for Christmas and they'll tear it up Christmas morning.

Bummer about your ball screw. Hopefully there is a replacement coming soon.

Good luck straightening one. It is an art to straighten a ball screw to a point where it does not wobble when being used.

To machine them you just need a carbide insert. You can cut past the hard surface without too much of a problem. I had the do the 1-1/2" diameter one on the Z axis of my mill. Hardened steel is not too big of a deal as long as you have a solid machine. Also cermet inserts work even better.

Bent ballscrew ? RETURN TO SENDER! You don't pay that amount of $$$ for a bent screw

Bit of an aside: what spindle-speed, depth-per-pass, ipm and drill do you use for routing out Al ? I'm going to have to do some soon, and I've never routed Al. Are you using any coolant or directed air blown / sucked to help out ?

Cheers,
Simon.

Thanks Much appreciated

Simon

Using a laser to monitor a laser. Cool. That reminds me of a joke. Paraphrased: My laser has it's own laser, and even its laser is cooler than your laser.

I'd be tempted to replace that Arduino with an op amp comparator to set the trip level voltage from the photocell. Of course, I don't have an Arduino in my parts bin, but I have lots of inexpensive op amps. Use what you have. And as Steve Ciarcia would say on the issue of what to implement in hardware and what to implement in software... "I program in SOLDER."

My first inclination would generally be to avoid designs where two motors could be out of sync and rack the gantry at a harmful angle, but given the advantages of the dual motor design, this looks like a very appropriate solution to me. The Hadron printer I'm building uses two Z stepper motors driving a threaded rod on each end. Fortunately, I don't think those motors are powerful enough to damage the stouter Hadron mechanics because the Hadron is small and the motors don't have the leverage that you have on a long laser gantry.

gavztheouch wrote:Because it's aluminium I need to baby sit it, which takes all the fun out of it

Yeah, but those parts look great!


When I have a big laser job, I web surf (often BuildLog.net) while babysitting the laser. Yet another advantage to having internet access in the shop and Firefox on every PC based CNC machine tool.

I'm going to go work on my Hadron Ord Bot. I've been away from it for a week or so, and I'm eager to wrap it up.

If you'd like to repurpose that Arduino (as the controller for a 3D printer!), here's an op amp based laser trip wire detector circuit that definitely SHOULD work. I didn't actually build it, but it's a simple circuit. The values aren't critical. They're just voltage dividers. You could use a 10K potentiometer, or a 47K, or 1K. Once you determine the correct ratio, you could substitute fixed resistors. If you need more current than the op amp can supply, you can add a transistor to the output to amplify the current. Swap the + and - inputs to the op amp and it's a laser detector instead of a laser absent detector.

You have a light load and a relatively heavy ball screw, but the screws seem fairly long. Are you seeing screw windup with fast acceleration? Ball screws have a large pitch, so it wouldn't take much torsional deflection to result in some axial motion that would appear similar to dynamic backlash. The other possibility I can conceive would be flexing in relatively long and unsupported ball screws. Either of these problems should be dynamic problems, causing following error when in motion, but not a loss of accuracy when the axis has reached a destination and come to a stop, and any twisting or bending of the ball screw has been eliminated. Are you seeing following error based on observations of cut quality, or are you measuring positioning errors where the commanded motion doesn't match the actual motion after the axis has stopped moving?

One trick for eliminating screw windup is to put an encoder on each end of the ball screw to measure the windup (difference between the two encoders) and compute the axis position based on the screw windup evenly distributed across the length of the screw. The farther the ball nut is from the motor, the larger the windup error. But I doubt your controls are up to that sort of ball screw windup compensation.

Of course, in an open loop stepper motion system, losing steps is always a possibility. More or less microstepping could help, or slowing the motion (probably not what you want to do), or more motor current, or bigger stepper motors. Occasionally, faster motion will prevent lost steps if the steady state speed results in harmonics the stepper motor doesn't like, and the ramp up and ramp down isn't in the harmonic zone long enough for the amplitude to build to a level that causes missing steps.

Hi Gav

Just noticed your new setup - looking really good ! Loved the anti- whip support for the ballscrew end machining - looks like my old music stand
50um, I wish that was what I had

Regarding the snake data to show up backlash, I swapped to 'waves' of half arcs - the backlash showed up instantly - it's amazing how accurate your eyes are.

Anyway, regarding the dsp and 'backlash' (although this isnt realyl backlash). On my system, I had to set the Y axis steps/um different from the X (X was as calculated using the pitch circle diameter of the pulleys - bang on) - to date, I haven't found out why (I did actually swap drive pulleys and drives to see if that was a cause). Both axes now use exactly the same steppers and belts - the belts are even from the same reel.
I did the settings over 500mm on both axes - checked the diagonals were equal too. I also measured multiple square test pieces. All dead nuts on, but the axis steps are different. I began to think like you ,that there was some issue in the pulses from the dsp. I no longer have access to any decent test equipment - a pulse counter would be a good thing to hang on the step outputs during a 45deg diagonal move with the same steps/um on each axis. I also tried a change in step polarity

I don't remember having this issue using the lasersaur board, but I had a lot of backlash at that time caused by those split couplers which may have masked it I no longer have the board, I mailed it to a friend for his laser.
Hopefully, you'll get to the bottom of it using mach - look forwrd to yourr results.
Cheers
Neil

So, are you saying that what you were observing and reporting as backlash may have been uncoordinated motion by the DSP controller?

Stuff like that is why I was recommending a static measurement of backlash rather than inferring backlash from print quality or some other indirect method. True backlash should be relatively easy to measure. If something can be measured directly, that is by far my preference. No point introducing complexity and uncertainty about what you're measuring or observing. That can lead to replacing belts with ball screws when the belts were actually OK.

Of course, hind sight is always 20/20.

My current CNC retrofit project uses LinuxCNC and I've been bumbling along on the edge of effectiveness for too long. I'm finally starting to see some progress. If I knew what I was doing when I started I could have saved at least 90% of the effort so far. That knowledge comes at a high price. Welcome to the bleeding edge of project build craziness. On the plus side, my next two (or more) LinuxCNC projects should be much easier, but you probably aren't planning on building another laser engraver.

Maybe I can learn from you and others on BuildLog.net, so my future laser build goes a bit more smoothly. It'll probably be a LinuxCNC project!

Liberty4Ever wrote:So, are you saying that what you were observing and reporting as backlash may have been uncoordinated motion by the DSP controller?
Stuff like that is why I was recommending a static measurement of backlash rather than inferring backlash from print quality or some other indirect method. True backlash should be relatively easy to measure. If something can be measured directly, that is by far my preference. No point introducing complexity and uncertainty about what you're measuring or observing. That can lead to replacing belts with ball screws when the belts were actually OK.

This is a slim possibility, but I have no real way of proving it - unless I can get my hands on a pulse counter - I've attached a screendump of my system settings. I'm not entirely sure it's actually the case. I have also cut a 500mm L shape where the line is cut contiguously - effectively nulling any backlash as that would have been taken up in it's initial positioning (I purposely set the directions then nailed the datum) and there have been no reverse directional changes made - the lines were orthogonal and of equal length.
Doing a quick calculation, over 500mm, the step count difference between the two axes is 22328 steps. At the X step size, that equates to just over 1.9mm - which seems quite a lot over 500mm. I did some further testing last night, and my machine cuts dead square and size accurate to within 0.05mm 'ish - over the 150mm my vernier works. Over 150 to 500, it's difficult to gauge accurately - flat ended steel rules aren't really that accurate (dimensionally), but are fine for consistency - my sizes are consistently the same in X & Y. Different materials give differing results of course as expected. I do see an amount of backlash on wee circles (3mm or less) - it's visible, hardly measureable accurately - but I now cheat and set those circles to be cut at a much slower rate which improves matters.
I use an offset of 0.1mm which is fine for any work that I need an accurate fit on (mainly balsa/liteply/acrylic) - this is with a 63.5mm lens.
To be honest, it can all get a bit anal can't it At the end of the day, I'm only cutting bits of wood and what I get from the system is perfectly OK. My real feeling is that this is all mechanical imperfection, but I'd like to just count the pulses to give me a nice psychological warm feeling anyway

I apologise if this is hijacking Gav's thread, I'm not intending to - I just don't want to see his ballscrews go to waste LOL It may all be a factor.

I'm not sure I understand what a 'static backlash' measurement is 'though ? By inference, 'backlash' has to include movement?

Cheers

Neil
Attachements...

System settings

iGull wrote:I'm not sure I understand what a 'static backlash' measurement is 'though ? By inference, 'backlash' has to include movement?

http://www.cncexpo.com/MeasuringBacklash.aspx shows the procedure. In summary, move into a dial indicator, zero the indicator, then move off. The difference between what the computer thinks it moved and the reading on the dial is the backlash.

This is obviously easier to do on a cnc machine that has jog or mdi support.

Have you tried your test while setting Step Edge to "Falling"?

Greolt

I made up the term "static backlash". There is probably a real static backlash, and my made up version is probably different from the real version.

It's true that backlash inherently requires motion, so in that sense, I don't see how there could be static (no motion) backlash.

What I meant by "static backlash" is real, mechanical backlash, the measurement of which was described in the post by jv4779 a couple of posts earlier in this thread... as opposed to what was being described as backlash but seemed to be inferred from laser cutting errors. Some forms of control errors or electrical problems could mimic backlash, and if they were assumed to be backlash when they weren't, a lot of time and money could be wasted trying to fix nonexistent backlash.

I plan on using LinuxCNC when I eventually build a laser, so I'll be able to enter MDI commands or manually jog the axes, so measuring backlash (and entering backlash compensation) should be very easy. Of course, engraving with LinuxCNC requires a multi-step image-to-Gcode conversion. I do a lot of engraving, but it tends to be the same engraving over and over again, as opposed to the custom work that a sign shop, trophy shop, or electrical machinery label maker would do where every job is different. The tedious but infrequent conversion for engraving won't bother me too much and I can just run the same G code each time I make new parts.

So all the ballscrew-effort wasn't neccesary at all????

Kees