[Nails]

masochist.....I know....I hoped to be done by the end of last year.

Eclipze - Is that small timing pulley a MXL trapezoid tooth profile pulley? If so, you might want to check into the GT2 profile, 2mm pitch. It's a "modified curvalinear" tooth profile which provides significantly greater load carrying capacity, improved positional registration (I think only slightly), less vibration and noise. I just learned about these last week, and changed the pulleys in my design.

You can get them at sdp-si.com

[atlab]

... and 3 usb microscope cameras... Programming seems trival ...

You tried to connect 3 usb cams simultaneously? This is not trivial...

[Eclipze]

Nails... I have NFI what shape the tooth pulley's use. I already have the pulleys and belts, and have the pulleys press fit on the steppers. So won't consider alternatives at the moment, but will keep your comments in mind if they aren't meeting the requirements.

RomanLini... the carriages and rail for the X-axis are actually smaller! The Y-axis the size is similar. The major difference however is the weight. The previous y-axis rail was under 200grams. The new one (which I still have to trim to length) is probably going to be over 3 times this amount. Time will tell just how much sag there will be, however with an overhang of only 220mm I'm hoping it won't be significant. Never really been that concerned about height accuracy... at least know where near as the requirement for positional x/y/c accuracy.

The Z-axis carriage brackets are done, but still need to have the vacuum part to go. However the tiny stepper had a hole done the shaft for the airflow Also made another y-axis main carriage to accommodate some minor changes.

[Nails]

atlab - No, I have not. Can you elaborate on what the problem will be?

[nisma]

I don´t see any problem. There is a hardcoded limit of 10 cameras that can be adjusted. Why with only tree webcams there should be a problem ?
It´s possible to make a dataset of different stripes images with a calibrated camera. If yes, i make available a applicaktion (~3Mbyte) and Webspace for
the images including the calibration sheet (using the usual opencv chessboard for that, so nothing special).

[Eclipze]

Y-axis motor mount is now done. I'm also using a more powerful stepper than the previous model. The weight should help with the balance of the axis too.

Have the aluminum profile for the base. Going to makes the plates that will hold the X-axis to the base today.

[neilw20]

With all that overhang, I hope the shaft does not fatigue and break off as they do.
Looks nice, but I think you should consider supporting the outer end of the shaft.

[Eclipze]

With all that overhang, I hope the shaft does not fatigue and break off as they do.
Looks nice, but I think you should consider supporting the outer end of the shaft.

I'm not expecting the shaft to fatigue/break. The overhang is only 16mm. If I changed, I'd have to get new motors, new pulleys and re-design both x and y axis brackets... which isn't going to happen unless they do break.

[RomanLini]

I think the unsupported stepper shaft will be fine. The overhang is not that much, and you only have light load and ideally you should be using a very well controlled and gentle acceleration.

The stepper (I'm assuming size 17?) wil be rated for about 2 kg constant radial force (from memory) and you will probably only have a few hundred grams total belt tension so it looks to all be well within specs. That's assuming you run a sprung belt tensioner, which is the only way to do it for this application.

If I was to pick a fault I'd say the motor is maybe a little underpowered for that weight of your machine head and the gearing. I've just received my Y axis motor, it is a larger (longer) size 17 with 45 oz-in torque. Yours looks like maybe 25 or 30 oz-in? With about 5mm pull radius and your total head weight must be about 1 to 1.2 kg? I don't think you are going to have much margin to play with there.

[Eclipze]

The stepper has a maximum radial force of 28N, 20mm from the flange. So I'm well within specs. It's a NEMA 17. The previous stepper was shorter, and I'd posted a video of it ripping back and forth very fast. This one I'm current using has a longer body and is 50 oz/in.... I'm thinking it's a little on the overkill side, but should give plenty of head room. The x-axis is NEMA 23, 130oz/in.

As far as weight goes, I don't know the weight of the carriages, but the rest of the head is around the 370g mark at the moment. Haven't got the camera on there yet, which will be another 50g. But the head weight won't be anywhere near 1kg.

The y-axis rail is much heavier than the previous one I was using. Once I trim it to length, I expect it will be around ~450g (old one was 150g). The motor is 320g. I'm expect the x-axis will be moving around 1.5kg

[Eclipze]

To hold the rail to the aluminum profile, I made some snug fit plates to run along within the profile section. I hex pocketed the holes and press fit 3mm lock nuts in place. With 3 of them in place, it holds the rail quite firmly in place on the aluminum profile.

[RomanLini]

...
As far as weight goes, I don't know the weight of the carriages, but the rest of the head is around the 370g mark at the moment. Haven't got the camera on there yet, which will be another 50g. But the head weight won't be anywhere near 1kg.
...

Sorry for the miscalc! My head with 3 RC servos and all lightweight hollowed out plastic construction runs about 400g.

I remembered yours having a stepper motor and metal construction and a lot more features like solder paste dispenser, etc so I assumed the weight of abotu 1kg. I expected your stepper motor alone to be 300g, or is it a tiny size?

[Eclipze]

No probs ;-) I don't have the solder paste dispenser on there... no immediate plans for it. It wasn't good enough, so put that on the back burners. The head rotating stepper is rather small. 20x20x33mm. Weighs 60grams. But it's only a feather weight 2.5 oz/in. The big change I had to make was to remove the entire assembly with the rotary pneumatic fitting and gear drive, simply because the motor couldn't turn it. I had bearings supporting the rotating section, but the friction of the gears and the rotary fitting was too much. So this design went for a hole through the shaft.

Haven't fully re-designed this yet. I wanted to make an alternative tool changing tip design. The previous one used acrylic, but the mating surfaces binded too easily.

[atlab]

I don´t see any problem. There is a hardcoded limit of 10 cameras that can be adjusted. Why with only tree webcams there should be a problem ?

I've wrote test prog for Windows XP, see attachment.
The troubles are:
- USB bandwidth;
- unknown problem with very long switch time among cams.

You can connect up to 4 webcams 640*480 to you PC.
Press on small picture to enlarge video.

P.S. VirusTotal tested
http://www.virustotal.com/file-scan/...917-1298086072

[alphatroniqu]

HI

that fist beta of software for drive my machine

http://www.alphatronique.com/Beta_PMx60.zip

the online tool was for got user feed back and sujestion

for quit demo just open "savon13.pnp.txt"(cad ASCII file) or *.pro (zevatech binary)
then use pull down menu for set feeder type in bank
then click on one of the feeder bank box for setup feeder
feeder pick point preset was on .ini whit part definition and machine size in mm

that version start to control machine , next step was to made it assemble a pcb ;-)


Best regard

[Eclipze]

The use of a AV card for multiple cameras may be the best option. I believe even the Madell basic machine uses a PC card for the camera.

alphatroniqu, I had a look at the Zevatech tool. Neat So do you plan to use a serial port for driving the data?


I've cut up the aluminum profiles and assembled the base I CNC machined all the ends, so not only are the lengths very accurate, but also square and flat. X-axis is mounted. Took a lot longer than I'd thought!

[Nails]

>>I've wrote test prog for Windows XP, see attachment.
>>The troubles are:
>>- USB bandwidth;
>>- unknown problem with very long switch time among cams.

atlab - I downloaded and ran your program on my laptop which has a usb webcam. Right now I only have one usb microscope, so I got two images.

They both were updating at the same time in the small windows. I did notice about a second delay enlarging the usb microscope, but not the webcam.

I suspect the delay might be related to closing the camera, and re-opening it with the larger resolution. There is possibly some request to reset the USB camera device upon (re)opening, causing the variable delay - related to the specific hardware. This wouldn't be the case in normal operation. All cameras would be set to deliver the highest resolution and opened once at the start of the program.

I understand the bandwidth problem would be an issue if all 3/4 cameras were on the same root usb hub, as they would all have to share the bandwidth of one controller. On separate root hubs, or by using additional usb add-on card(s), each one would be isolated with their own usb controller hardware, so I wouldn't expect it to be an issue.

Eclipze - Nice work...It is great encouragement to see someone with the energy to see this through.

A method I like to use when using extruded aluminum for a base, like you did, is to mill a shallow slot for the ends of the cross-members (as well as the sides/ends) to fit into. This eliminates the need to measure between the cross-members/ends and keeps the top sides parallel, by not allowing any movement nor twisting action that might happen as you secure them together with the L brackets.

Extruded aluminum is great for the adjustability/fexibility it allows....until you don't want it!

[Eclipze]

A method I like to use when using extruded aluminum for a base, like you did, is to mill a shallow slot for the ends of the cross-members (as well as the sides/ends) to fit into. This eliminates the need to measure between the cross-members/ends and keeps the top sides parallel, by not allowing any movement nor twisting action that might happen as you secure them together with the L brackets.

That's a top idea! I had to drill the holes for accessing the bolt heads, and could have accommodated machine pockets at the same time on that face of the profile. I'II probably use this approach in the future, as it would certainly make it neater to assemble.

I've actually left the design a little open to extending the y-axis another 40mm... given those two short length were replaced. I almost went the extra distance, but the belt had already been trimmed (from last design) and didn't want to stretch the confidence with the whole unsupported rail approach hahaa... Aiming for a win here, not egg on the face :-)

[atlab]

...I suspect the delay might be related to closing the camera, and re-opening it with the larger resolution.

The problem with my webcams is that delay increase at several switchings. I don't know why.

...I understand the bandwidth problem would be an issue if all 3/4 cameras were on the same root usb hub, as they would all have to share the bandwidth of one controller...

I thought about using:
- ethernet webcams;
- mini-ITX PC for every webcam.
I think, these are the decisions comparable at the price.

[alphatroniqu]

Hi Eclipze

ya i use modbus over serial port for drive it (Lan Telnet was planed for future)
since controller pcb inside machine handle low level task
only hight level command to pass over modus

i need 2 command per placed part
first contain feeder position coordinate ,Speed ,Feeder type ,Bit to use
last contain part drop coordinate ,speed ,rotation

velocity ,motor driver etc etc was handle by controller ;-)
as a hardware guy i try to keep it as hardware as possible

Best regard