[scsi]

Looks very interesting. Please try to fit that rotating head and I'll hook you up with the software. I do handle automatic feeders through TTL I/O's or serial port commands.

[RomanLini]

That's a nice big machine there Flippersplace! Congrats!

Do you really need the 0.9 degree steppers? Your pulleys look like about 22m diameter so that's a lead of 69 mm, with typical 16th ustep drives you should get a resolution of 69/3200 which is 0.02mm per ustep. I would have thought a 50th of a mm would be enough resolution?

I was interested in the Wikiplacer too but the web page just said there was nothing done yet. Or is there some trick to get in?

[stewi]

We would love to have you on the Wikiplacer project. Get a password and sign in or write a PM to me. At the moment, we are discussing how much Wiki we want to be, before we post any meaningful content.
However, it looks like we agree on an entry level machine, single cylinder engine rather than a Porsche.

[flippersplace]

..Do you really need the 0.9 degree steppers? Your pulleys look like about 22m diameter so that's a lead of 69 mm, with typical 16th ustep drives you should get a resolution of 69/3200 which is 0.02mm per ustep. I would have thought a 50th of a mm would be enough resolution?

Hi Roman, when planning this machine with a T5 5mm pitch belt, I decided on 16 tooth pulleys which with 1.8 deg stepper = 2.5 steps per mm, which if I then microstepped at 1/16 came to a nice round 40 steps per mm which equals the mil units I'm used to. (Actually I'd be happy with a reliable 20 steps per mm)

In my initial testing once the gantry was in motion, with microstepping at 1/16 (experimenting) I wasn't getting consistant movement when single stepping. It was like every 8th step or so was showing no movement then it would catch up on the next. So because I needed a bigger stepper for the X anyway I figured I'd go for the 0.9 degree ones. and see if a higher physical resolution hence less reliance on microstepping would help.

Not very scientific but I am curious to see what the 0.9 deg steppers go like. I had no idea what the belts would be like / will perform like set up this way so its all a bit of a fun experiment..

[flippersplace]

And now SCSI has convinced me that rotating heads are the way to go so I've this morning drilled 2 different stepper shafts to cobble together a revised head with rotation & pneumatic Z movement.

Just gonna struggle with keeping the gantry weight down with the addition of a stepper & larger linear slides / pneumatic cylinder.

[scsi]

We want pictures!

[RomanLini]

Stewi, thanks for the offer with the Wiki project. I may join in with you there and offer a few suggestions, whether or not the suggestions will help much is another matter.

Flippersplace- Yeah 80mm lead is getting a bit high to hold ustepping tolerance, the 0.9 degree motors will help a bit with that although they usually have speed penalties. Have you considered adding a 2:1 or 2.5:1 pulley setup to your existing motors? It might not cost much more, and will have payoffs in smoothness (better decoupling of ustep resonance from the moving load) and the pulleys will add a bit of flywheel damping too.

[flippersplace]

Hi Roman, yes I have step down pulleys in the back of mind but only if I end up needing to.

I've never played with belts, pulleys , steppers etc so this is an empirical exploration process to satisfy my curiosity..

I also got some 10 tooth pulley stock when I got the 16, but thought I'd start with the 16's so the belt isn't wrapped so tight..

Initially I imagined a pretty light gantry might let me get away with the microstepping, but its all a bit heavier now...

[flippersplace]

[ame=http://www.youtube.com/watch?v=ODySJj3vx10]2012 March MK2 DIY Homemade PCB SMT Pick Place machine - YouTube[/ame]

[Eclipze]

GREAT TO SEE!!!

Awesome work! What are you using for software at the moment?

[flippersplace]

SCSI's software. Its pretty good & is quite mature. Very hardware agnostic as well.

Is quite happy to work with my el-cheapo USB microscopes for up & down vision as well as my assorted crappy stepper drivers.

It's been a mission but this machine is now repeatedly & accurately placing 0805 parts. Any limitations are not due to his software but rather my machine as the hardware list includes a chopping board, nail plates & wood & aluminium extrusions from the local hardware store (Bunnings)...

I've still some tweaking of the driver settings so I can ramp speed up a little but no sense pushing it. (Next upgrade is leadshine psu & drivers)

Still no nozzle changer so not placing larger parts (yet)

[phomann]

Hi,

I missed this, but there is a misconception about micro stepping.

Microstepping will not improve accuracy as you have seen. You should do all your calculation on the distance a full step moves an axis. At a pinch you could use the 1/2 step distance.

The reason is that with microstepping, the driver is attempting to hold the rotor between the full step detent position, by adjusting the current in the two windings, using the sine and cosine of the required angle.

A number of things work against this method from being accurate.

• The first is the the two windings in the stepper are not absolutely identical, therefore, even with the winding currents being correct, the resulting rotor will be off a bit and not linear across the range of micro steps.
• The detent force of the motor is trying to pull the rotor towards it. This is stronger the closer the rotor is to the full step position.
• The forces of the axis pushing against the motor pushes the rotor from it's theoretical position.

Some of what you have seen is explained by the above. There is no movement on the 1st microstep as the force is not enough to overcome the detent force and the axis resistance. As the microsteps increase the rotor springs into position.

Microstepping was introduced to provide a smoother movement motion at low speeds, rather than cogging from one fullstep position to another.

If you need greater accuracy go to a 0.9 degree stepper than than 1.8 degree stepper, or gear down the drive.

BTW, great video.

Cheers,

Peter

Hi Roman, when planning this machine with a T5 5mm pitch belt, I decided on 16 tooth pulleys which with 1.8 deg stepper = 2.5 steps per mm, which if I then microstepped at 1/16 came to a nice round 40 steps per mm which equals the mil units I'm used to. (Actually I'd be happy with a reliable 20 steps per mm)

In my initial testing once the gantry was in motion, with microstepping at 1/16 (experimenting) I wasn't getting consistant movement when single stepping. It was like every 8th step or so was showing no movement then it would catch up on the next. So because I needed a bigger stepper for the X anyway I figured I'd go for the 0.9 degree ones. and see if a higher physical resolution hence less reliance on microstepping would help.

Not very scientific but I am curious to see what the 0.9 deg steppers go like. I had no idea what the belts would be like / will perform like set up this way so its all a bit of a fun experiment..

[flippersplace]

Thanks Peter, yeah that post you mention is soooo last year.

Yes the machine is still direct drive but is now 0.9 deg steppers & also the belt is now gates GT2 (2mm pitch) with only 20 teeth (quite small pulleys so I had to turn down the 0.9mm stepper motor shafts to be able to use them..) all in the interests of upping the physcial accuracy before any potential help by microstepping so in theory my full step accuracy is now 0.1mm

That has helped a lot. It also now has magnetic strip / hall sensor positioning encoders which has really helped make up for a lot of other issues (belt slip due to inertia, shafts/pulley holes not true, phase of the moon, timber warp & MDF swelling.. (I'm kidding about the last one, maybe...)

I've come to understand that a dimension of 1mm is like a meter where a machine like this is concerned with regard to placing parts

I have left enough physical room to gear down the drive but I think I'll probably leave that for a Mk3 machine if that should happen on day. But we shall see.

[flippersplace]

From this to this..

[Steven C.]

O.K., just had to show off what 5 minutes of tuning my KFlop resulted in. This is the guts from a mid 1980s machine using Vexta 5 phase 1/2 step motors/drive and a KFlop pulse generator. I don't remember how fast it ran before I decommissioned it (it ran off a '386 computer) but no where near this fast!

I was just showing it off to my hard-to-impress Daughter and accidentally clicked "Step" twice so it stalled and then slammed into the side. Tomorrow morning I wire up the limit switches before I break this thing!

And people made fun of how slow my Taig pick and place machine on YouTube was :-) (Insert evil mad scientist laugh here).

- Steven Ciciora

[Steven C.]

I forgot to mention, that's moving 21.5" and back.
- Steven Ciciora

[flippersplace]

Steve, no link...

[Steven C.]

Sorry about that... I clicked on the little paper clip and it took a little while for bits to upload, I'm not use to CNCZone so I'm not sure what happened. Here it is on YouTube:

Pick and Place machine First Light - YouTube

- Steve

[flippersplace]

Wow fast..

[Eclipze]

I couldn't tell if the wobble was the bench, the camera or my screen playing back the video hahaa...