fall fashions for the LitePlacer: outerwear edition

[WayOutWest]

So my pick and place was complaining to me that all the other kids at school had cool clothes... so I got him a hat and a backpack.

That cabling you see running through the core of the Z-axis makerslide is Chainflex (thanks @mawa!). I'll probably swap out the Cat5-STP for Chainflex (but keep using RJ45-STP connectors, which I love). The pressure sensor is the one @thereza recommended, and it works wonderfully. The backplate has a FET to allow TinyG control of the downcam LEDs, since mine are blindingly bright and generate a lot of heat. The USB hub is this one with the plastic casing removed; luckily the PCB inside there has mounting holes. It holds up wonderfully even under full-framerate 1280x760@30fps from three cameras at the same time.

More details soon. Lots of bugs on this board (some you can see), so I will be sending out rev 1.1 soon. It will have PCB-mounted limit switches on the backplate for the X-axis to further reduce wiring clutter. I will post GERBERs once I'm happy with the results, and since my board fab has a minimum run of 10 pieces there will be free extras for anybody who wants to pay for postage.

Also, my !@()*@*%^@ solenoid died, and I can't find anything even remotely close locally. Ugh. Four days of downtime (due to weekend) even with overnight shipping.

hat.jpeg (111.94 KiB) Viewed 7262 times

backpack.jpeg (413.02 KiB) Viewed 7262 times

[mrandt]

Hey Adam,

looks nice - I like the idea

Will you share the source files (what eCAD are you using btw?) or Gerbers?

Have you considered adding a microcontroller and USB-RS232 converter to the hat PCB?

That might be a nice, clean way to support and interface:
- vacuum measurement
- variable high-speed PWM (needs to be in kHz range to avoid shutter effects) for down lights
- some spare input / outputs, e.g. for head to automatic feeder communication (IR / laser diode)

To make wiring even easier, you might want to use wire clamps rather than screw terminals on the boards.

Wago (German brand) makes about the best I could find - they have many different models, both for THT and SMT.

Examples:
https://eshop.wago.com/JPBC/singleview/ ... FS=Catalog
https://eshop.wago.com/JPBC/singleview/ ... FS=Catalog

Looking forward to Rev 1.1

Regards
Malte

[WayOutWest]

Will you share the source files (what eCAD are you using btw?) or Gerbers?

Yes, I'll certainly post GERBERs once I have something that's reasonably debugged. This version required me drilling several post-manufacturing holes in the PCB

I use a pretty obscure CAD flow originally intended for chip design, so I doubt the original files would be very useful. I can post unpoured GERBERS so it should be only 10-15 minutes work to "re-trace" the traces into any other CAD tool, it's a pretty simple board.

Have you considered adding a microcontroller and USB-RS232 converter to the hat PCB?

Yes, for the ultimate along those lines, I plan to put a tiny PCB and AVR on the back of every stepper.

- vacuum measurement

You bet -- no more capacitor bodge.

- variable high-speed PWM (needs to be in kHz range to avoid shutter effects) for down lights

Is the dragchain wire inductance too high to drive these frequencies from the TInyG? Coming from a VLSI background I have very little intuition for inductance we are spoiled and get to pretty much ignore it inside the padframe... it's a DC world in there.

To make wiring even easier, you might want to use wire clamps rather than screw terminals on the boards.
Wago (German brand) makes about the best I could find - they have many different models, both for THT and SMT.

Oh cool! I had seen the "free-hanging" version of those but did not know they had through-hole mounted blocks! I will definitely use those. But:

Examples:
https://eshop.wago.com/JPBC/singleview/ ... FS=Catalog
https://eshop.wago.com/JPBC/singleview/ ... FS=Catalog

Links seem to only work for your account (maybe depend on cookies in your browser)?

[mrandt]

Yes, I'll certainly post GERBERs once I have something that's reasonably debugged. This version required me drilling several post-manufacturing holes in the PCB

Thanks, looking forward to that!

Yes, for the ultimate along those lines, I plan to put a tiny PCB and AVR on the back of every stepper.

Understand you're idea. I personally would like to keep it simpler:

I am happy with TinyG drinving the motors and interfacing the limit switches, but it lacks input / output for anything else. I don't want to mess with the firmware or solder extra pins.

So my plan is to use a second DIY controller board with USB interface and implement simple protocol to control solenoid valves, vacuum sensors, LED lighting and hence forth at some point.

Should be easy to integrate that into LitePlacer or OpenPNP software.

- variable high-speed PWM (needs to be in kHz range to avoid shutter effects) for down lights

It is not about the inductance of the wires, I am not an expert either but guess it should not be a problem, even if we went up to a few kHz. It is more about generating a proper "high frequency" PWM for the lights. If PWM frequency is too low, you get "rolling shutter" effects on most CMOS cams which makes computer vision unusable. AFAIK the TinyG PWM frequency is too low - but that is an assumption based on the fact that AVR builtin PWM frequency is in fact too low for our purposes. Would need to hook up my scope to PWM output and measure. I am currently using external LED driver to prevent these kind of issues in the first place.

Oh cool! I had seen the "free-hanging" version of those but did not know they had through-hole mounted blocks! I will definitely use those.

I think you're refering to the ones used in electrical installation (household and industrial AC wiring)? These are absolutely great, save you a bunch of time and are much more secure than any "screwable" clamps. They even come in special versions for braided wires.

Sorry for the links not working, checkout Wago's international website here:
http://global.wago.com/en/index-en.jsp

For the parts I was referring to - searched and found them at Farnell / Newark. Seems Digikey and Mouser don't have them.

SMT:
http://www.newark.com/MarketingProductL ... 79,79T9388

THT:
http://www.newark.com/webapp/wcs/stores ... reId=10194

For small wires, the 233 series with white levers to release wires works great.

By the way: I am not affiliated with Wago, just a happy customer

[WayOutWest]

- variable high-speed PWM (needs to be in kHz range to avoid shutter effects) for down lights

It is more about generating a proper "high frequency" PWM for the lights. If PWM frequency is too low, you get "rolling shutter" effects on most CMOS cams which makes computer vision unusable. AFAIK the TinyG PWM frequency is too low - but that is an assumption based on the fact that AVR builtin PWM frequency is in fact too low for our purposes.

Hrm, I thought the AVR hardware PWM could toggle a pad at up to 50% of the clock frequency with no CPU overhead. Am I mistaken? In that case the 32mhz TinyG should be able to produce output PWMs with 62.5ns granularity.

http://www.newark.com/MarketingProductL ... 79,79T9388
http://www.newark.com/webapp/wcs/stores ... reId=10194

Awesome. I also made the breakout boards below, but now I'll re-run them using the Wago blocks instead.

By the way: I am not affiliated with Wago, just a happy customer

Same here, with respect to the various companies/brands I recommend (there have been a couple so far). I go out of my way to do this only because I myself rely on obscure forum postings to find good suppliers/products/etc.

DSCF1153.JPG (324.07 KiB) Viewed 7226 times

DSCF1152.JPG (333.23 KiB) Viewed 7226 times

[mrandt]

Hrm, I thought the AVR hardware PWM could toggle a pad at up to 50% of the clock frequency with no CPU overhead. Am I mistaken? In that case the 32mhz TinyG should be able to produce output PWMs with 62.5ns granularity.

That may be true for the AVR hardware, but if you use Arduino abstraction (as TinyG does) the frequency drops to about 490 Hz according to the API reference:
https://www.arduino.cc/en/Reference/AnalogWrite

Problem is that CMOS camera sensors are read "line by line", which creates the "rolling shutter" effect if PWM is too slow.

Find a very good explanation in this PDF: http://www.mikewoodconsulting.com/artic ... utters.pdf

Rule of thumb: Use PWM frequency of 1-2kHz or faster to avoid problems with CMOS cams - many LED drivers buck controller ICs can do that, and of course most microcontrollers if you use lower level programming.

[JuKu]

> but if you use Arduino abstraction (as TinyG does)

I don't think this is correct?

[mrandt]

> but if you use Arduino abstraction (as TinyG does)

I don't think this is correct?

Not sure - I just assumed that they coded the firmware in Arduino to have less redundancy with grblShield and the like?

But even if they don't, the PWM frequency will depend on their configuration. As it is intended for spindle control, it is probably not the highest frequency possible...

Seems there is only one way to find out PWM frequency for sure: If I find some time, I will hook up my scope to the PWM output and obtain a frequency reading

If anyone feels like doing that before I get to it, please let us know the results.

Regardless of the result, I think we could use additional input / output (both digital and analog) to control features on the machine that are beyond moving the motors.

Vacuum sensor is the most prominent, but LED lighting PWM and special use cases (e.g. solenoid pin to advance tape / drag feeder or signalling an automatic feeder) might be helpful as well.

In case we had a nice "hat" with USB connection for that, it would further reduce wiring effort and could be integrated into the software nicely.

[mawa]

Have I missed something or why should I dim the down light instead of just turning it on when it is needed (optical homing, fiducial measurement, tape measurement, tape hole progression)?

So when your LEDs get too hot why not just turn 'em off?

[mrandt]

Have I missed something or why should I dim the down light instead of just turning it on when it is needed (optical homing, fiducial measurement, tape measurement, tape hole progression)?

So when your LEDs get too hot why not just turn 'em off?

Depending on your setup, they might get too bright - then you have to dim them