Introducing MotionBoard!

[Paul Jones]

Hi All,

 

I’ve finally finished it – and given it another new name. MotionBoard.

This is a motion processor module with an Arduino footprint that runs Smoothie. The pinout is compatible with both Firepick EMC02 and Ramps 1.4 boards.

Douglas Pearless has the original one and has his Firepick running on it.

 

Currently it has an LPC1769 Arm Cortex M3 processor. NXP is releasing a Cortex M4 later in the year (sometime between July and Nov I think??) that has a very similar footprint to the M3 chip, so I’ve designed the board so it works with both.

 

Design file are available at: https://github.com/PeeJay/MotionBoard

 

Price is $55 + $25 postage (USD)

If you want more than one contact me for a postage quote.

Payments can be made via PayPal to pa...@pauljones.id.au

 

ETA is boards being sent last week of June/First week of July.

 

Special thanks go to Douglas Pearless for implementing the RotaryDelta arm solution in Smoothie, and of course our sleepless leader Neil Jansen for creating such an awesome open source project!

 

Cheers,

Paul.

 

 

 

[Neil Jansen]

On Thu, Jun 11, 2015 at 9:37 PM, Paul Jones <pa...@pauljones.id.au> wrote:

I’ve finally finished it – and given it another new name. MotionBoard.

This is a motion processor module with an Arduino footprint that runs Smoothie. The pinout is compatible with both Firepick EMC02 and Ramps 1.4 boards.

Douglas Pearless has the original one and has his Firepick running on it.

 

Currently it has an LPC1769 Arm Cortex M3 processor. NXP is releasing a Cortex M4 later in the year (sometime between July and Nov I think??) that has a very similar footprint to the M3 chip, so I’ve designed the board so it works with both.

 

Design file are available at: https://github.com/PeeJay/MotionBoard

 

Hi Paul, 

I sent you a private msg about buying a few of these.

Just wanted to say publicly THANKS so much for getting this done and debugged!!  This is going to be a great path forward for FPD, and I like the name as well.  Really excited to get my hands on these.

Also, just curious, are these kits or fully assembled?  If assembled, do you have access to real PnP equipment?  It's another chicken and egg problem as far as FPD is concerned.  (I am confident though, we're closer than ever to having reliable PnP.. it's been painful getting there but this is one of those last steps to the inevitable).

Cheers,

--

Neil Jansen, Co-Founder

Tin Whiskers Technology, LLC

http://tinwhiskers.io

[Peter Shabino]

So if this works as advertised will this be the solution of choice for the FPD?

Also is there a BOM / gerber data already available? If not will extract the data from the eagle files. 

Thinking of doing a build here to save on shipping across the pond. 

Thanks,
Peter

My projects:
http://www.wire2wire.org/

---

Date: Thu, 11 Jun 2015 21:49:17 -0400
Subject: Re: [FirePick] Introducing MotionBoard!
From: ne...@tinwhiskers.io
To: fire...@googlegroups.com

--
You received this message because you are subscribed to the Google Groups "FirePick" group.
To unsubscribe from this group and stop receiving emails from it, send an email to firepick+u...@googlegroups.com.
To post to this group, send email to fire...@googlegroups.com.
Visit this group at http://groups.google.com/group/firepick.
For more options, visit https://groups.google.com/d/optout.

[Neil Jansen]

On Thu, Jun 11, 2015 at 10:24 PM, Peter Shabino <wi...@hotmail.com> wrote:

So if this works as advertised will this be the solution of choice for the FPD?

I'm envisioning that Smoothie slowly becomes the controller of choice, until something better comes along (Charles's MachineKit, for example).  There was no way for me to include the Smoothie / MotionBoard in the beta kits, because it didn't exist at the time (although I could play the hindsight game and ask why I didn't think of it, lol).  It's better than the Arduino Mega / Marlin we're currently using, and should hold us over for a while. 

Support for the Arduino Mega, however, won't officially go away, at least not yet.. That will happen eventually, but not within weeks of people having their beta kits in-hand.  This will be a gradual thing.. . To start that gradual process off, I want to try Smoothie myself and see what all the fuss is about :)  Those videos from Douglas look like it lives up to the name.

 

Thinking of doing a build here to save on shipping across the pond. 

Depending on how much your time is worth, it might be better just to buy it from Paul.  I have never, ever came out ahead on any board I built myself, if I bill my time at what it's worth... Let alone with all the MOQ's on Digikey parts and custom PCB's.

 

[Paul Jones]

It will be fully assembled, except for the pin headers (at least for now). My pick and place equipment currently consists of two hands J It doesn’t take long to assemble with a bit of practice. Hopefully my firepick will be up and running soon, but this first batch I’ll assemble by hand.

 

I’m thinking about doing a kickstarter when good quantities of the M4 chips are available, they will most likely be done by an assembler.

 

Paul.

 

From: fire...@googlegroups.com [mailto:fire...@googlegroups.com] On Behalf Of Neil Jansen
Sent: Friday, 12 June 2015 11:49 AM
To: fire...@googlegroups.com
Subject: Re: [FirePick] Introducing MotionBoard!

 

On Thu, Jun 11, 2015 at 9:37 PM, Paul Jones <pa...@pauljones.id.au> wrote:

--

[Peter Shabino]

What is the max step rate of the smoothie when running 3 axis? Starting to think of replacing my PICnc plan with this for my laser cutter as well. Too bad they don't have raster support working yet.

What is the user interface like for smoothie? Don't know much about them. 

Thanks,
Peter

My projects:
http://www.wire2wire.org/

---

From: pa...@pauljones.id.au
To: fire...@googlegroups.com
Subject: RE: [FirePick] Introducing MotionBoard!
Date: Fri, 12 Jun 2015 02:47:39 +0000

[Paul Jones]

I have a BOM on octopart for the Rev A board, but it needs updating. You’re best to just extract it from the Eagle files, as they are always the reference for everything else.

I should also point out that there is not a lot of fat in the price so you may come out behind in the end. A more sensible price would be $95 ish at low quantities (under 50). Also, 4 layer one-off boards are expensive! If I end up doing a kick starter and sell a few hundred that dramatically lowers the price, and I don’t want any early adopters (read Guinea pigs) to feel ripped off!

I can take $5 off the postage if you are happy with no tracking number, but it doesn’t seem worth it.

 

Cheers,

Paul

 

From: fire...@googlegroups.com [mailto:fire...@googlegroups.com] On Behalf Of Peter Shabino
Sent: Friday, 12 June 2015 12:25 PM
To: fire...@googlegroups.com
Subject: RE: [FirePick] Introducing MotionBoard!

 

So if this works as advertised will this be the solution of choice for the FPD?

[Peter Shabino]

Nope not at all just happen to have a large stock of parts and more than likely have everything in house other than the processor and the PCB... So not quite the normal condition. Just looking at the layout and size $55 is very reasonable maybe even under priced. 

Yea would be about a wash PCBs would be about $28 each and the processor is $12  so at $40 just with that.  Only upside is if I tossed it on one of my rush panels it would get here sooner :). 

So this is for motion control only correct? Sill will need some sort of PC on the outside to do the vision processing and send the move commands?

Thanks,
Peter

My projects:
http://www.wire2wire.org/

---

From: pa...@pauljones.id.au

To: fire...@googlegroups.com
Subject: RE: [FirePick] Introducing MotionBoard!

Date: Fri, 12 Jun 2015 02:57:53 +0000

[Paul Jones]

Ahh you sound like me then! In that case go for your life J

Yes, you need a pc to run OpenPnP.

 

Paul.

[Roy Nielsen]

Looking forward to getting one!

-Roy

--

[Neil Jansen]

On Thu, Jun 11, 2015 at 10:57 PM, Paul Jones <pa...@pauljones.id.au> wrote:

I have a BOM on octopart for the Rev A board, but it needs updating. You’re best to just extract it from the Eagle files, as they are always the reference for everything else.

I should also point out that there is not a lot of fat in the price so you may come out behind in the end. A more sensible price would be $95 ish at low quantities (under 50). Also, 4 layer one-off boards are expensive! If I end up doing a kick starter and sell a few hundred that dramatically lowers the price, and I don’t want any early adopters (read Guinea pigs) to feel ripped off!

I can take $5 off the postage if you are happy with no tracking number, but it doesn’t seem worth it.

Welcome to the economics of open-source beta kits!  ;)  It's certainly not a way to get rich quick... ask me how I know :)

[Michael Anton]

On Thursday, June 11, 2015 at 8:58:02 PM UTC-6, Paul Jones wrote:

Also, 4 layer one-off boards are expensive! If I end up doing a kick starter and sell a few hundred that dramatically lowers the price, and I don’t want any early adopters (read Guinea pigs) to feel ripped off!

You can get 4 layer boards fairly inexpensively from smart-prototyping.com.  I've had pretty good luck so far ordering from them, and they are very good quality.  They have decent solder paste stencils as well, and again, very cheap.

Mike

 

 

Cheers,

Paul

 

[Michael Anton]

For reference, 10 pieces of a 4 layer Mega 2560 sized board (actually a 15x10cm process size) from smart-prototyping works out to $9.20 each plus shipping ($10.80 each with ENIG), which is usually about $30.  Delivery is usually within two weeks of ordering.  There are a few gotchas I have discovered in ordering, which I'm happy to share if there is interest.

Mike

[Peter Shabino]

What is the projected max step clock rate (kHz/MHz) when running 4 axis? Back to thinking about my laser cutter plan again... 

Thanks,

Peter

My projects:
http://www.wire2wire.org/

---

From: pa...@pauljones.id.au
To: fire...@googlegroups.com
Subject: RE: [FirePick] Introducing MotionBoard!

Date: Fri, 12 Jun 2015 03:14:16 +0000

[ric...@hornbaker.com]

I'm still an outsider on this project, working to roll my own using FPD as the starting point. I'll admit to some confusion still.
My prior inquiry kinda jumped the tracks, but I'm trying to understand what functions are being run on which hardware, and why.
There is the driver hardware, a la RAMPS. And apparently Neil's EMC boards.
There seems to be the MCU running the driver, a la 2560. Or apparently Smoothie, which I gather has the driver hardware integrated, and MotionBoard (which does or doesn't integrate the drivers?) but has an Arduino header.
MachineKit keeps getting mentioned as a goal state, but it's software not hardware.
And this is before Pi comes into the picture for the camera, LCD, and...?
Then there is the PC for the vision processing? And OpenPnP?

Surely I'm missing plenty here. But what I'm looking for is a hardware version of Neil's architecture / function diagram - which part is doing what?

Am I the only one lost in all the excitement?

[Douglas Pearless]

Hi Richard,

There has been a lot of evolution going on.

I would recommend you consider the beta kit pack and later the motionboard which replaces the 2650 board and is based on Smoothie; as that will get you a lot of the tricky parts and you will be very close to everyone else which will make your like a lot easier to debug any issues.

Just my opinion though 😊

Sent from my iPhone

[Ristola]

Paul,

Looking at your boards, they look great !  

I wish i could route like that.

Looking at the SD card socket, and comparing the footprint to the different variation of SD sockets I have on hand, Please Verify this socket isn't on BACKWARDS !

I hope I'm wrong !

Loren

On Thursday, June 11, 2015 at 9:37:51 PM UTC-4, Paul Jones wrote:

Hi All,

 

I’ve finally finished it – and given it another new name. MotionBoard.

This is a motion processor module with an Arduino footprint that runs Smoothie. The pinout is compatible with both Firepick EMC02 and Ramps 1.4 boards.

Douglas Pearless has the original one and has his Firepick running on it.

 

Currently it has an LPC1769 Arm Cortex M3 processor. NXP is releasing a Cortex M4 later in the year (sometime between July and Nov I think??) that has a very similar footprint to the M3 chip, so I’ve designed the board so it works with both.

 

Design file are available at: https://github.com/PeeJay/MotionBoard

 

Price is $55 + $25 postage (USD)

If you want more than one contact me for a postage quote.

Payments can be made via PayPal to p...@pauljones.id.au

[Ristola]

PS - Is it Mirrored Also ?

I  only ask because SDcard sockets has a gotten me more then once ! hehe

Loren

On Thursday, June 11, 2015 at 9:37:51 PM UTC-4, Paul Jones wrote:

Hi All,

 

I’ve finally finished it – and given it another new name. MotionBoard.

This is a motion processor module with an Arduino footprint that runs Smoothie. The pinout is compatible with both Firepick EMC02 and Ramps 1.4 boards.

Douglas Pearless has the original one and has his Firepick running on it.

 

Currently it has an LPC1769 Arm Cortex M3 processor. NXP is releasing a Cortex M4 later in the year (sometime between July and Nov I think??) that has a very similar footprint to the M3 chip, so I’ve designed the board so it works with both.

 

Design file are available at: https://github.com/PeeJay/MotionBoard

 

Price is $55 + $25 postage (USD)

If you want more than one contact me for a postage quote.

Payments can be made via PayPal to p...@pauljones.id.au

[Paul Jones]

Yes I have double checked that, I got it backwards the first time!!

 

My other common mistake is using the wrong pin pitch for IC footprints – thankfully I avoided that one this time!

 

Paul.

 

--

[Paul Jones]

That’s where I’ve got my last few board orders from. They are surprisingly cheap, and the stencils are excellent.

The only issue I have with them is they leave a wide clearance between the soldermask and the pads, so most ICs smaller than SO8 end up with no soldermask between pins. This hasn’t caused me any problems yet, but I have one board with an 80 pin Molex connector and 7 mils pin spacing, that was a bit messy if the paste is not perfectly aligned.

Same issue with the silkscreen, put it too close to a pad and it disappears.

 

Paul.

 

From: fire...@googlegroups.com [mailto:fire...@googlegroups.com] On Behalf Of Michael Anton
Sent: Friday, 12 June 2015 2:51 PM
To: fire...@googlegroups.com
Subject: Re: [FirePick] Introducing MotionBoard!

 

For reference, 10 pieces of a 4 layer Mega 2560 sized board (actually a 15x10cm process size) from smart-prototyping works out to $9.20 each plus shipping ($10.80 each with ENIG), which is usually about $30.  Delivery is usually within two weeks of ordering.  There are a few gotchas I have discovered in ordering, which I'm happy to share if there is interest.

--

[Michael Anton]

If they are modifying your soldermask layers, that is not good.  You can probably have them run it as you have specified, if you include a note to that effect in the comments when you order (they do not read readme files, as I found out the hard way).

I just ran a set of boards that have a 25mil pitch part on them, and there is solder mask between the pads.  I didn't specify anything special when I sent the job to them, and they came out fine.  I would check your pad stack rules, to make sure you don't have the solder mask backoff set too high.  I would imagine that they use a photoimageable mask, so they should be able to handle quite fine masks (though there will still be some practical limit obviously).  I used a backoff of 5mil on the boards I just got back.

Really, a connector with a 7mil pin pitch?  That sounds really odd, and extremely difficult to route, to say nothing of soldering it reliably.  Are you sure that isn't a 7mil pad gap?

Eliminating silkscreen that is too close to pads is pretty standard.  Silkscreen is usually aligned by eye, and can have some pretty wide errors, so keeping the silkscreen well away from pads is a good plan.  I usually use 10mil backoff from pads, and I've never had any eliminated.

Mike

[Paul Jones]

Whoops, 7 mils spacing between pads, 19 mills pitch. Darn small anyway…

Now that I think about it I used a different cam processor than what I usually do. I didn’t think to check that! Thanks.

[Alan]

Do you reckon smoothie could run 3 servo motor's / Brushless DC ? I mean horse power wise + code configuration. I'm keen to write an open source BLDC FOC mbed module to get it working

[Paul Jones]

The processor could do it, but Smoothie would need a bit of work. You would have to implement a position control loop and then somehow integrate that into the motion planner. Best talk to the Smoothie devs about how to do that.

It would be interesting to try running a BLDC motor using stepper driver ICs. You would have to turn off microstepping, then you could use 1.5 of them as a 3-phase H bridge. It would be awesome if you could pull it off!

 

Paul.

 

From: fire...@googlegroups.com [mailto:fire...@googlegroups.com] On Behalf Of Alan
Sent: Friday, 12 June 2015 8:55 PM
To: fire...@googlegroups.com
Subject: Re: [FirePick] Introducing MotionBoard!

 

Do you reckon smoothie could run 3 servo motor's / Brushless DC ? I mean horse power wise + code configuration. I'm keen to write an open source BLDC FOC mbed module to get it working

 

--

[Bruce Bock]

Looks like the boards are $81.80 for 3 from OSH Park. 

--

Life at the speed of Moore.

[Dustin Kasel]

I would recommend considering www.pcbcart.com. I use them for all of my prototype PCBs, never had an issue with quality or lead-time.

Dustin

[The Guru]

I've been out of the loop the last couple days - so this replaces the mega? Plugs into the EMC02?

[felix....@gmail.com]

Yes Tom

[Neil Jansen]

On Sat, Jun 13, 2015 at 10:43 AM, The Guru <cyberr...@gmail.com> wrote:

I've been out of the loop the last couple days - so this replaces the mega? Plugs into the EMC02?

This is not an "official" upgrade for the FPD beta kits, at this time.  But we're welcoming everyone that wants to stay on the bleeding edge to go ahead to purchase a MotionBoard from Paul.  The Arduino Mega / Marlin firmware will still be supported to some extent for a while, but long term, it will fade away into the sunset, after the laggards have finally upgraded.

[David Granz]

Looks great Paul! Paypal order sent :) -DaveG

[Alan]

Just sent another paypal order your way, feeling excited! 

This might of been covered but do I need a special programmer to program these?

On Friday, 12 June 2015 13:37:51 UTC+12, Paul Jones wrote:

Hi All,

 

I’ve finally finished it – and given it another new name. MotionBoard.

This is a motion processor module with an Arduino footprint that runs Smoothie. The pinout is compatible with both Firepick EMC02 and Ramps 1.4 boards.

Douglas Pearless has the original one and has his Firepick running on it.

 

Currently it has an LPC1769 Arm Cortex M3 processor. NXP is releasing a Cortex M4 later in the year (sometime between July and Nov I think??) that has a very similar footprint to the M3 chip, so I’ve designed the board so it works with both.

 

Design file are available at: https://github.com/PeeJay/MotionBoard

 

Price is $55 + $25 postage (USD)

If you want more than one contact me for a postage quote.

Payments can be made via PayPal to pa...@pauljones.id.au

[Paul Jones]

Thanks!

 

No programmer needed – You can use an FTDI type serial cable to program them via ISP if you wish, but I’ll supply them pre-programmed as I have to test them anyway. After the initial programming the SD card can be used as a mass storage device, so then you just drag and drop the firmware onto it and reboot.

 

 

Paul.

 

From: fire...@googlegroups.com [mailto:fire...@googlegroups.com] On Behalf Of Alan
Sent: Sunday, 14 June 2015 12:57 PM
To: fire...@googlegroups.com
Subject: [FirePick] Re: Introducing MotionBoard!

 

Just sent another paypal order your way, feeling excited! 

--

[Ristola]

I also sent payment via paypal.  

This will be a sweet improvement over Marlin

Loren

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

[Dustin Kasel]

I have also sent my payment, thanks for coordinating this Paul!

Dustin

[Peter Betz]

Looks great!

I too am impressed with the routing. I am about to send you a payment as well for one, it sure did look smooth on the video. (I put an Ethernet SmoothStepper on my CNC mill. That made a huge difference in the smoothness of the machine). Pretty sure smoothie is aimed at CNC as well if I recall, haven't checked out their site.

I haven't had a chance to examine the software architecture, am I correct in assuming the PNP programming side (That Neil is about to start focusing on again) is independent of the motion controller? Open PnP will essentially issue commands to the motion controller (G code? like cnc??) so it doesn't care if the receiver is the Mega or the Smoothy board? I just want to make sure I keep the machine in line with the way the focus is going to be currently applied.

Just finished a new panel, I added fiducials in the corners of each individual board so hopefully I will be testing this all out on the firepick at some point!

 

 

 

Peter.

[Neil Jansen]

On Sun, Jun 14, 2015 at 2:22 AM, Peter Betz <betzt...@gmail.com> wrote:

am I correct in assuming the PNP programming side (That Neil is about to start focusing on again) is independent of the motion controller? Open PnP will essentially issue commands to the motion controller (G code? like cnc??) so it doesn't care if the receiver is the Mega or the Smoothy board? I just want to make sure I keep the machine in line with the way the focus is going to be currently applied.

OpenPnP is agnostic to motion controller type, all of this gets abstracted via motion controller interfaces.  So at the end of the day, it doesn't care if there is a Smoothie board or an Arduino or whatever at the end of the line, however it will expect the motion controller to do its job at the end of the day.  Currently, Marlin isn't doing its job (velocity calculations are incorrect, and the pulses aren't smooth), we think we can whip it into shape, but that will take a bit of work, and will try to keep supporting this for at least a few more months, but overall Smoothie looks much more promising.  The time investment into Smoothie will be a much better bang for the buck.  

We actually just sent out a survey as we're trying to gauge how our beta test community feels about upgrading or staying with arduino.  I'm predicting around 50% will want to upgrade and the rest won't.. which makes our job tough :)  lol.  

Being that the Smoothie MotionBoard is only $50 or so plus S&H, I'm really hoping that everyone is able to upgrade, but I understand either way, and will try to support everyone as best as I can.  Everyone following the day-to-day conversations in the google group has probably seen this coming for months and months now, but I also understand the new folks that might not have been up on all these proposed changes.

[Peter Betz]

Thanks for clearing that up. I see benefit in proper motion controllers so I am comfortable upgrading now (It also makes it easier seeing the machine function with it, it looked like in the video it was dry running a 3D print, so that boosts confidence!). It will be interesting to see how the others answer!

I completed the survey. The stars part was confusing a little bit. The first question using stars had 1 star being the most desireable (easy to build). The next ones had no explanation on which way the scale went. So I assumed more stars the better. It is either I way over thought it, or your results may be opposite depending on the person.

Keep up the good work Neil and Felix.

Peter.

[Assaf Inbal]

Just sent my payment as well. Thanks for all of your hard work!

On Friday, June 12, 2015 at 4:37:51 AM UTC+3, Paul Jones wrote:

Hi All,

 

I’ve finally finished it – and given it another new name. MotionBoard.

This is a motion processor module with an Arduino footprint that runs Smoothie. The pinout is compatible with both Firepick EMC02 and Ramps 1.4 boards.

Douglas Pearless has the original one and has his Firepick running on it.

 

Currently it has an LPC1769 Arm Cortex M3 processor. NXP is releasing a Cortex M4 later in the year (sometime between July and Nov I think??) that has a very similar footprint to the M3 chip, so I’ve designed the board so it works with both.

 

Design file are available at: https://github.com/PeeJay/MotionBoard

 

Price is $55 + $25 postage (USD)

If you want more than one contact me for a postage quote.

Payments can be made via PayPal to p...@pauljones.id.au

[Karl]

Me two and thank you.

[Peter Shabino]

Sent mine in last night as well. Missing all the parts of any cost so may as well get in the group buy. :)

Later,
Peter

My projects:
http://www.wire2wire.org/

---

Date: Sun, 14 Jun 2015 06:29:22 -0700
From: ka...@firepick.org
To: fire...@googlegroups.com; shmu...@gmail.com

Subject: [FirePick] Re: Introducing MotionBoard!

Me two and thank you.

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

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

[Peter Shabino]

Did you use a mini or a micro usb connector? Need to pick up a hub for my firepick delta so want to get the cable at the same time. 

Thanks,

[Reef Morse]

Paul, count me in.  I'll buy one.  Am working on getting PayPal to send you money.  The board is stuffed??

Reef

[SeanB]

I'm in too, thanks Paul!

[Paul Jones]

Yes-ish, all the SMD will be populated, but not the headers. Some people may want them on a different side of the board, although FirePick needs them on the top side.

 

Paul

 

From: fire...@googlegroups.com [mailto:fire...@googlegroups.com] On Behalf Of Reef Morse
Sent: Monday, 15 June 2015 2:04 AM
To: fire...@googlegroups.com
Subject: [FirePick] Re: Introducing MotionBoard!

 

Paul, count me in.  I'll buy one.  Am working on getting PayPal to send you money.  The board is stuffed??

--

[Paul Jones]

Micro.

 

From: fire...@googlegroups.com [mailto:fire...@googlegroups.com] On Behalf Of Peter Shabino

Sent: Monday, 15 June 2015 12:48 AM
To: fire...@googlegroups.com

--

[Douglas Pearless]

Correct me if I am wrong Paul, people will need tall headers to enable the Ethernet port sufficient clearance from the EMC02 board.

I am going to try the Raspberry Pi 20 way tall header, plus tall single was headers for the rest.

And on the topic of Ethernet, I will soon be starting work on TCP/IP by enabling this on Smoothie for FirePick…

Cheers

Douglas 

[Paul Jones]

I haven’t quite got to the headers yet, but I’ve tested the Arduino double stacking headers and they are plenty long enough if they are soldered offset from the board a little. Hopefully I’ll obtain something less fiddley like the Rpi ones.

 

Paul

 

From: fire...@googlegroups.com [mailto:fire...@googlegroups.com] On Behalf Of Douglas Pearless
Sent: Monday, 15 June 2015 10:19 AM
To: fire...@googlegroups.com
Subject: Re: [FirePick] Introducing MotionBoard!

 

Correct me if I am wrong Paul, people will need tall headers to enable the Ethernet port sufficient clearance from the EMC02 board.

[Peter Betz]

Hi Paul,

 Hopefully you will be able to supply the headers loose for the boards??

Peter.

On Sunday, June 14, 2015 at 7:12:58 PM UTC-7, Paul Jones wrote:

I haven’t quite got to the headers yet, but I’ve tested the Arduino double stacking headers and they are plenty long enough if they are soldered offset from the board a little. Hopefully I’ll obtain something less fiddley like the Rpi ones.

 

Paul

 

/d/optout.

[Paul Jones]

Yes, sorry I didn’t make that clear. Everything is included except the micro sd card and micro usb cable. All parts will be assembled, programmed and tested. Pin headers will be supplied loose, so those who are using them for things other than FirePick can do whatever suits their application.

 

If anyone wants me to leave the Ethernet connector off the board just let me know. It makes the board sit about 8mm or so (0.3”) above where an Arduino would be, hence the long headers.

 

Paul.

 

From: fire...@googlegroups.com [mailto:fire...@googlegroups.com] On Behalf Of Peter Betz
Sent: Monday, 15 June 2015 12:21 PM
To: fire...@googlegroups.com
Subject: Re: [FirePick] Introducing MotionBoard!

 

Hi Paul,

--

[Douglas Pearless]

Hi Paul,

That’s fine with me.

Cheers

Douglas

[betzt...@gmail.com]

Yes, for sure, that’s great!

 

I may end up buying one for my MendelMax down the road....

 

Peter.

 

From: Douglas Pearless

Sent: Sunday, June 14, 2015 8:13 PM

To: fire...@googlegroups.com

Subject: Re: [FirePick] Introducing MotionBoard!

 

Hi Paul,

 

[NothingButGunpowder]

Thanks for all the hard work! I just sent my payment as well... Can't wait to try this out:)
Cheers,
Adam

[velias]

On Thursday, June 11, 2015 at 10:11:41 PM UTC-7, ric...@hornbaker.com wrote:

I'm still an outsider on this project, working to roll my own using FPD as the starting point.  I'll admit to some confusion still.
My prior inquiry kinda jumped the tracks, but I'm trying to understand what functions are being run on which hardware, and why.
There is the driver hardware, a la RAMPS.  And apparently Neil's EMC boards.
There seems to be the MCU running the driver, a la 2560.  Or apparently Smoothie, which I gather has the driver hardware integrated, and MotionBoard (which does or doesn't integrate the drivers?) but has an Arduino header.
MachineKit keeps getting mentioned as a goal state, but it's software not hardware.
And this is before Pi comes into the picture for the camera, LCD, and...?
Then there is the PC for the vision processing?  And OpenPnP?

Surely I'm missing plenty here.  But what I'm looking for is a hardware version of Neil's architecture / function diagram - which part is doing what?

Am I the only one lost in all the excitement?

Your not the only one!

I'm so confused by the mix of boards that people are using, I'm sitting this all out until it gets stable.

Sometimes in a project people have to put their foot down and say this is it, the hardware we are using and thats that, or things will never settle down.

Seems like there are too many cooks and the project needs a dictator.

 

[Joshua Pritt]

I think the reason we're still discussing the main processing board(s) is because it's still experimental to find the best solution for the least $ and still keeping it simple for the DIY crowd all while making sure the FPD is an awesome swiss army knife of a machine to do the PnP, 3D printing, solder paste extrusion, etc.  Am I right, Neil?

--

[Neil Jansen]

On Monday, June 15, 2015, velias <vice...@gmail.com> wrote:

Your not the only one!

I'm so confused by the mix of boards that people are using, I'm sitting this all out until it gets stable.

Sometimes in a project people have to put their foot down and say this is it, the hardware we are using and thats that, or things will never settle down.

Seems like there are too many cooks and the project needs a dictator.

Hi,

I'm the self-appointed benevolent dictator of FPD.  I can help with the confusion, I've said all this before but here goes: currently the officially supported board is an Arduino Mega 2560 on an EMC02 motion controller board.  This will stay an official board for a while. However, we've identified some weaknesses in the Arduino Mega platform so we're simultaneously working on replacing the 8-bit AVR based Arduino with an ARM Cortex M3 board.  It plugs into the EMC02 and is otherwise a very benign change.  The arduinos only cost $20-45 and have general purpose use even after they've been removed, so it's not like this is a big deal for most people.  We have eager early adopters that will try the Smoothie, and folks that want to keep the Arduino.  I see no problem with this approach.

I'm also hesitant to make a roadmap because it changes so often.

Hope this helps.  I can try my best to answer any otrt questions that you might have, just let me know.

[ric...@hornbaker.com]

Hi, Neil.

I dug back to the first ~30 posts in this list where you shared a fairly lengthy explanation of what's what, and that helped a bit.  I realize much has changed since then, but conceptually I think the bits still line up.  A lot of this is probably old hat for folks who've got prior experience in the 3D printing space.  Indulge me in a recap here, and let me know what's off the mark...

* Driver board - at the lowest level, a board like EMC01/02 or RAMPS is hosting the electrical drivers that take pulse input and translate to motor behavior.  Modules like 8825 provide that physical linkage; the board aggregates signals to a header, adds connectors for limit switches, etc.  Basically, this board is the interface to the electromechanical part of the robot.

* Real-time processor - links to the driver board and handles the timing of pulses, etc.  Examples being an Arduino 2560.

* Raspberry Pi2 - providing some amount of the application-level behavior, user interface, camera inputs, etc.

* Host PC - providing even more of the application-level functionality.  Maybe image processing is begin done here?  OpenPnP seems to be running here.

I gather the Marlin code typically runs on the real-time processor, but is certainly challenged by the speed, bit depth, FPU needs, etc. for a delta application

Paul's MotionBoard seems to displace the Arduino 2560 as the real-time processor, and is running Smoothie code instead of Marlin.  Apparently he's accomplished this at a fraction of the price of Smoothie's hardware - I gather because the physical interfaces and drivers are still on EMC01/02.

MachineKit seems to be highly regarded as an end goal, but it's software not hardware.  Would it run on MotionBoard, or...?

The architectural questions that come to my mind are things like...

• Realtime doesn't require a lot of horsepower, unless it's being forced to do G-code conversions and delta kinematics.  Why not offload that to a beefier CPU and host a simple/cheap MCU on the driver board for the realtime steps?
• Why are OpenPnP and machine vision functions being done on PC instead of the RPi?
• With the PC present, is the Pi really being used for much?  Could the RPi do more, and eliminate the PC?
• Could the RPi be the "beefier CPU" feeding a simple driver MCU for the RT timing - even simpler than the 2560?
  

Not looking to throw rocks or Monday-morning-quarterback, just understand the logic that led us to where things are today, and whether I'd be wasting time exploring tangents.  I realize it's a moving target, and you've got to put a stake in the ground at some point to make progress.

At this point, I'd just be curious if my grasp above is close to correct.  If anyone has time to expound on the architectural questions, that'd be interesting, but I appreciate this might not be a convenient time for academic discussions.

As someone else has suggested, I've got Arduino 2560, RAMPS 1.4, and 8825 modules in-hand just to get off the ground.  Once I get motors turning and can appreciate the pro's and con's, I'll explore alternatives like EMC and MotionBoard.  At least, I think that's the plan. ;-)

Cheers,

Richard

[Neil Jansen]

Hi Richard,

My comments are below.

On Mon, Jun 15, 2015 at 9:26 PM, <ric...@hornbaker.com> wrote:

Hi, Neil.

I dug back to the first ~30 posts in this list where you shared a fairly lengthy explanation of what's what, and that helped a bit.  I realize much has changed since then, but conceptually I think the bits still line up.  A lot of this is probably old hat for folks who've got prior experience in the 3D printing space.  Indulge me in a recap here, and let me know what's off the mark...

* Driver board - at the lowest level, a board like EMC01/02 or RAMPS is hosting the electrical drivers that take pulse input and translate to motor behavior.  Modules like 8825 provide that physical linkage; the board aggregates signals to a header, adds connectors for limit switches, etc.  Basically, this board is the interface to the electromechanical part of the robot.

Correct.  In the case of both the RAMPS and EMC02, both were mere breakout boards or "glue" that held all the other stuff together.  By itself it's useless... only when you add the brain and stepper driver modules and hook it up to a PC, does it become useful.

 

* Real-time processor - links to the driver board and handles the timing of pulses, etc.  Examples being an Arduino 2560.

Correct.  Smoothie, Arduino Due, etc., are a few others that fit in this group.

 

* Raspberry Pi2 - providing some amount of the application-level behavior, user interface, camera inputs, etc.

Correct.  This I would call an "application processor", they are not real time (with the one exception being the BeagleBone Black).  There are a few sneaky ways to do real-time on a conventional application processor, for example Mach3 CNC for PC, which pulses the stepper driver from windows (YUCK but it's been done for about a decade now on expensive CNC machines around the world).  Raspberry Pi also sort of fits in here as they're using DMA, but again, I try to avoid doing this whenever possible.

 

* Host PC - providing even more of the application-level functionality.  Maybe image processing is begin done here?  OpenPnP seems to be running here.

Yes.  Eventually we want to go with a single-board computer like a RasPi or BBB.. But for prototyping, this makes no sense, we can get more done on a normal personal computer or laptop with two 24" screens and a nice keyboard, and Eclipse / IntellJ IDE, all in one spot, rather than worrying about cross-compiling or memory shortages or whatever.  I want to get it working on a normal PC without any constraints, and then try to make it work on a single-board computer.

 

I gather the Marlin code typically runs on the real-time processor, but is certainly challenged by the speed, bit depth, FPU needs, etc. for a delta application

Yes.  Roughly:

• Marlin was designed to run on an Arduino Mega 2560
• Smoothie was designed to run on an NXP LPC 1768/9 ARM Cortex M3. 
• MachineKit was designed to run on a BeagleBone Black's PRU (and was based on LinuxCNC / EMC before that)

 

Paul's MotionBoard seems to displace the Arduino 2560 as the real-time processor, and is running Smoothie code instead of Marlin.  Apparently he's accomplished this at a fraction of the price of Smoothie's hardware - I gather because the physical interfaces and drivers are still on EMC01/02.

Yes, indeed :)  

 

MachineKit seems to be highly regarded as an end goal, but it's software not hardware.  Would it run on MotionBoard, or...?

Charles might be able to give a better explanation.  MachineKit is an attempt to get one board (BeagleBone Black) to run the application processor and the real-time motion control.  This is accomplished because the BBB has PRUs (programmable realtime units), little bits of silicon that allow real-time stuff to be done while the application process is off doing non-real time stuff like file and network IO, and user-level stuff.  The code for these PRUs aren't easy to write (they're assembly-based IIRC and rather terse), unlike an Arduino or the Smoothie dev environment.

 

The architectural questions that come to my mind are things like...

• Realtime doesn't require a lot of horsepower, unless it's being forced to do G-code conversions and delta kinematics.  Why not offload that to a beefier CPU and host a simple/cheap MCU on the driver board for the realtime steps?

We're working on it :)  That's one of the things on my list, and others are looking at this too.  I plan to have a rough working version of this after I get the documentation done. I'm behind obviously, as kitting all those beta machines and doing documentation is taking up more of my time than I had estimated.  But my goal is to offload the kinematics and have a workable Arduino based solution that will be "good enough".  Karl's working on this too but his work is more ground-up and might take months or longer. I'm a practical guy so I'm going in with a machete and a bandana.

• Why are OpenPnP and machine vision functions being done on PC instead of the RPi?

Because we're in beta testing right now, and doing Java development on a real PC is the smarter choice.  To me this is a no-brainer but I could elaborate on it if needed.  Some folks might have to actually try to do this before they would actually understand how much time and frustration this approach saves.

 

• With the PC present, is the Pi really being used for much?  Could the RPi do more, and eliminate the PC?

No.  the single-board computer isn't really needed until we've gotten OpenPnP to a usable state to actually start trying to "shrink" it down to fit on the Pi.  We'll probably do all this through package management to make it easy... like debian synaptic package manager, etc... that's not good to do yet but will be when we get openpnp usable.

 

• Could the RPi be the "beefier CPU" feeding a simple driver MCU for the RT timing - even simpler than the 2560?
  

Possibly but my approach was to use whatever processor ecosystem had the most usable and ready motion control code available.  That's why we started with Arduino.  It wasn't the "best" or the "fastest".  But the code was ready and well tested.  It is prohibitive to start thinking about other simple motion control platforms if there's not code currently available for them.  That's too much development work and we'd have better luck using what's already available to us. For example, Douglas and Paul spent literally weeks / months just modifying Smoothieware and Smoothieboard to the point where it would work with FPD.  Now imagine trying to do all that from scratch on a different board.  Most motion control platforms have man-years of work invested into them already, lots of trial and error and bug fixes and whatnot.

 

Not looking to throw rocks or Monday-morning-quarterback, just understand the logic that led us to where things are today, and whether I'd be wasting time exploring tangents.  I realize it's a moving target, and you've got to put a stake in the ground at some point to make progress.

Those are all fantastic questions.  I'm quite isolated right now having already understood all this, I've no idea what the newcomers know and don't know.  I don't want to spend a second of my time explaining what is already known to folks, but I won't know what to write about until I know that people don't know something .. if that makes sense.  I think what you wrote was actually a pretty good explanation, so I may borrow that with your permission to post somewhere officially on the delta.firepick.org site.

 

 

At this point, I'd just be curious if my grasp above is close to correct.  If anyone has time to expound on the architectural questions, that'd be interesting, but I appreciate this might not be a convenient time for academic discussions.

You pretty much nailed it.

 

As someone else has suggested, I've got Arduino 2560, RAMPS 1.4, and 8825 modules in-hand just to get off the ground.  Once I get motors turning and can appreciate the pro's and con's, I'll explore alternatives like EMC and MotionBoard.  At least, I think that's the plan. ;-)

The only other thing I'll mention here as it's sort of relevant, Marlin never did work well with the DRV8825's because it outputs non-standard timing pulses (lots of cheating and cleverness there, which is a story for another day).  That's one of the reason why the beta kits have the A4988's, is they're more tolerant to Marlin's butchery of the stepper pulses.

 

[Peter Shabino]

By my understanding the RPi2 is out of the picture right now.

Current default setup sounds like

cameras -> PC -> 2560 -> EMC01/02 -> motors

Now all the other chatter and upgrades:

2560 -> Soothie  (well in progress beta boards due at the end of the month) 

+ smoother path movements and on card Cartesian to rotary conversion. 

+ code working demos videos available 

+ code pull accepted by the base smoothie project

+ custom board survive plug and play

+ USB / Ethernet / SD card support

- more cost not included in original kits.  

PC + 2560 -> machinkit running on a BBB or RPi2(with additional hardware)

+ self contained package and full linuxCNC  support

+ no external PC needed with all the trimmings (Ethernet, SD, KB, mouse, video)

- more cost not included in original kits.  

- new custom cards needed to be designed 

- no new news on this in quite a few weeks. Not sure if anyone is still working on it yet. 

If I have it wrong someone will chime in :)

Later,
Peter

My projects:
http://www.wire2wire.org/

---

Date: Mon, 15 Jun 2015 18:26:47 -0700
From: ric...@hornbaker.com

To: fire...@googlegroups.com
Subject: Re: [FirePick] Introducing MotionBoard!

[Neil Jansen]

On Mon, Jun 15, 2015 at 9:57 PM, Peter Shabino <wi...@hotmail.com> wrote:

By my understanding the RPi2 is out of the picture right now.

Current default setup sounds like

cameras -> PC -> 2560 -> EMC01/02 -> motors

Now all the other chatter and upgrades:

2560 -> Smoothie  (well in progress beta boards due at the end of the month) 

+ smoother path movements and on card Cartesian to rotary conversion. 

+ code working demos videos available 

+ code pull accepted by the base smoothie project

+ custom board survive plug and play

+ USB / Ethernet / SD card support

- more cost not included in original kits.  

PC + 2560 -> machinkit running on a BBB or RPi2(with additional hardware)

+ self contained package and full linuxCNC  support

+ no external PC needed with all the trimmings (Ethernet, SD, KB, mouse, video)

- more cost not included in original kits.  

- new custom cards needed to be designed 

- no new news on this in quite a few weeks. Not sure if anyone is still working on it yet. 

Yep, that's a great summary, Peter.   Smoothie could easily hook up to either a PC, laptop, single-board computer (RasPi, BBB, BananaPi, that new $9 one on kickstarter, etc).  As long as it runs Debian Linux, Windows, or OSX, and has some USB ports,  it will run OpenPnP just fine.

[ric...@hornbaker.com]

Awesome, thanks Neil and Peter for the quick and detailed feedback.

Yes, coming into this completely green, there has been a lot to wrap my brain around.  I'm accustomed to hacking my way through code, low-level electronics, and programming a spectrum from Perl scripts to AVR assembler - but 3D printing robotics is its own world of packages, hardware, problems, and terms to grasp.  If you need a noob to test out your documentation, I'd be a good candidate.

As I go about reinventing the wheel, at least I have an idea where we are today.  I'll try to get functional with off-the-shelf parts before I go nuts with tweaking. ;-)

I completely appreciate the role of the PC now.  It seemed there was a lot of potential duplicity / overlap between it and the RPi (or whatever its successor is eventually).  It certainly makes sense to go for rapid development, then shrinkage.

Cheers,

Richard

[Malte R.]

Hi Paul,

great job - I'll buy one. Just transferred the money via PayPal :-)

Thanks and regards
Malte

[Charles Steinkuehler]

On 6/15/2015 8:52 PM, Neil Jansen wrote:
> On Mon, Jun 15, 2015 at 9:26 PM, <ric...@hornbaker.com> wrote:
>
>> MachineKit seems to be highly regarded as an end goal, but it's software
>> not hardware. Would it run on MotionBoard, or...?
>
> Charles might be able to give a better explanation. MachineKit is an
> attempt to get one board (BeagleBone Black) to run the application
> processor and the real-time motion control. This is accomplished because
> the BBB has PRUs (programmable realtime units), little bits of silicon that
> allow real-time stuff to be done while the application process is off doing
> non-real time stuff like file and network IO, and user-level stuff. The
> code for these PRUs aren't easy to write (they're assembly-based IIRC and
> rather terse), unlike an Arduino or the Smoothie dev environment.

The BBB and it's PRUs make a great platform for motion control, but
the real benefit to using Machinekit IMHO goes beyond that. Since
Machinekit runs on Linux, it's possible to (easily) migrate to
different platforms. Since the real-time system includes a flexible
HAL layer, it is easy to do things like switch between steppers and
true servo motors.

In addition to running on the BBB, I think it might make sense to
migrate the motion control to an FPGA card in an x86 system, so
everything runs on one machine (you can get low-end FPGA motion cards
for about the price of a Smoothie or BBB). This might not be needed
for the maker crowd, but could be a great option for turnkey systems.

I also think there's a lot of potential on the software side if more
of the code (vision, trajectory planning, and motor control) is
running on the same platform.

--
Charles Steinkuehler
cha...@steinkuehler.net

[Karl]

I would be fascinated to see PH stuff in FPGA. PH calcs are compute intensive. But that's getting way ahead of myself and assumes success of current FireStep efforts.

 

In addition to running on the BBB, I think it might make sense to
migrate the motion control to an FPGA card in an x86 system, so
everything runs on one machine (you can get low-end FPGA motion cards
for about the price of a Smoothie or BBB).  This might not be needed
for the maker crowd, but could be a great option for turnkey systems.

Charles Steinkuehler
cha...@steinkuehler.net

[Charles Steinkuehler]

I don't think you'd need to push the PH calculation into the FPGA. A
system like the BBB or Pi can probably handle the task (still TBD),
but if not a cheap x86 box has a *LOT* of computing power. If it
turns out control computer bandwidth is a limiting factor, once you're
running on an OS (like Machinekit running on Linux), it's pretty easy
to move to get more number crunching power by spending a bit more money.

--
Charles Steinkuehler
cha...@steinkuehler.net

[greg....@kutu.com.au]

Hi Karl and Charles,
   If you're going to use a FPGA, a Zynq is the only way to go.  It's the same as the BBB with the FPGA built into the chip.  The chip is $50 which is a bit pricey for makers, but the real advantage of it is apart from the dual-core A9, there are 80 MAC blocks which run at 400Mhz, so it's good for speeding up the video.  This part is a lot of work though.  The other thing is you can hook up as many cameras as you need.
I use one of these at work as a basic HDMI written in logic, and it runs either Ubuntu or Xubuntu.  The HDMI output signals are at 1485Mhz so the I/O is very capable.
The secret to using them is to use the micro for the complex part, and the FPGA for the timing related and repetitive parts.  Controlling a stepper with a FPGA is much eaiser than a micro, but the calculations are better done on the micro.

Karl, I just finished working through the delta kinematics trigonometry from first principles.  I find it's the only way I can really understand the code.  But I did get the equations out :)
While your working on FireStep, I'm going to use the Arduino to do some basic movements using GRBL as a base.  This is what smoothie used as a base, but it's in C, so that makes it easier for me.  I've looked at the approximations in Marlin, and I don't really like the idea.

regards,
Greg

[Charles Steinkuehler]

On 6/16/2015 6:20 PM, greg....@kutu.com.au wrote:
> Hi Karl and Charles,
> If you're going to use a FPGA, a Zynq is the only way to go. It's the
> same as the BBB with the FPGA built into the chip. The chip is $50 which
> is a bit pricey for makers, but the real advantage of it is apart from the
> dual-core A9, there are 80 MAC blocks which run at 400Mhz, so it's good for
> speeding up the video. This part is a lot of work though. The other thing
> is you can hook up as many cameras as you need.
> I use one of these at work as a basic HDMI written in logic, and it runs
> either Ubuntu or Xubuntu. The HDMI output signals are at 1485Mhz so the
> I/O is very capable.
> The secret to using them is to use the micro for the complex part, and the
> FPGA for the timing related and repetitive parts. Controlling a stepper
> with a FPGA is much eaiser than a micro, but the calculations are better
> done on the micro.

The Zynq parts are nice, but there's the Altera SoC family as well
(which I've been playing with for my day job). I like the new $99 dev
board just released by Terasic:

http://www.terasic.com.tw/cgi-bin/page/archive.pl?Language=English&CategoryNo=163&No=941&PartNo=1

...but realistically I don't think this or the Zynq is probably that
beneficial to the FPD project. If the FPD needs more horsepower it
will likely be related to the vision software, and even low-end x86
machines are still much better in absolute processing power than the
ARM parts. Add a GPU and you can *REALLY* start crunching on pixel
streams (that's my day job, working at NewTek designing HD video
editing hardware). The ARM parts really shine when you need low-power
processing (lots of MIPS per W), but I don't see the FPD as a mobile
battery-powered device, so that's not too much of a concern.

:)

--
Charles Steinkuehler
cha...@steinkuehler.net

[Peter Shabino]

You mean the Intel SoCs :) (they just bought out Altera) 

Going to be interesting when the first batch of FPGAs with Intel CPU cores hit the market. 

Later,

Peter

My projects:
http://www.wire2wire.org/

> Date: Tue, 16 Jun 2015 19:29:37 -0500
> From: cha...@steinkuehler.net

> To: fire...@googlegroups.com
> Subject: Re: [FirePick] Introducing MotionBoard!
>

[Charles Steinkuehler]

On 6/16/2015 7:42 PM, Peter Shabino wrote:
> You mean the Intel SoCs :) (they just bought out Altera)
> Going to be interesting when the first batch of FPGAs with Intel CPU cores hit the market.

:)

Yeah, I'm looking forward to that, but I suspect it's probably going
to be a few years. Intel does have some CPUs that include FPGA
fabric, but it's not something that's generally available (yet).

--
Charles Steinkuehler
cha...@steinkuehler.net

[Douglas Pearless]

I have been looking at Samsung's new Atrik platform, and I am wait listed for their Alpha programme https://www.artik.io

Their 10 module is very interesting, USD$100, real time Linux etc; If we had this board, and EMCXX with a USB lowered Hub, then I believe we could migrate everything onto this environment:

Key Module Components

APPLICATION PROCESSOR

ARM A1...@1.3GHz + A7...@1.0GHz

MEMORY

2GB LPDDR3 + 16GB eMMC

HW + SW SECURITY

• HW Embedded Security Element (eSE) + TLS (DTLS)
• Data Encryption
• Device Authentication
• FW Security & Update, and TEE

DISPLAY

TBD

PMIC

9 Buck converters + 38 LDOs

WIFI

802.11 b/g/n 

BLUETOOTH

BT/BLE

ZIGBEE

Thread planned

External Interface

GPIO

51 (2 PWM output including)

ANALOG CAPTURE

6 (0 – 1.8V range)

UART

3 ( FLOW CONTROL 1 PORT)

I2C

6

SPI

1

I2S

1

USB

1 USB 2.0 HOST + 1 USB 3.0 

MIPI

1 DSI (2 LANE) + 1 CSI (2 LANE)

CLOCK OUT

1 (24MHz) + 1 (32.768KHz) 

Power

INPUT POWER

3.4V – 5V

OUTPUT POWER

1.8V/100mA + 2.4V/100mA

POWER CONSUMPTION

TBD

Software 

OS

Yocto 1.6 (Fedora)

DEVELOPMENT ENVIRONMENTS

• Arduino® IDE
• Samsung SDK
• C/C++
• Java
• Groovy

Physical

DIMENSIONS

39 x 29 x 3.5 mm

CONNECTORS

• 80 pins (2 Panasonic AXT480124) 
• 40 pins (1 Panasonic AXT440124)
• (0.4 mm pin pitch, mating height 1.5 or 2.5mm)

OPERATING TEMPERATURE

TBD

Multimedia / Network

HW AUDIO CODEC

HW 5.1 Channel I2S + TDM up to 8 Channels + HW mixer 

HW VIDEO CODEC

1080p@120fps H.263/H.264/ MPEG-4/VP8 + MPEG-2/VC1 decoding

GPU

• Mali T628 MP6 GPU
• Open GL ES 1.1/2.0/3.0
• OpenCL 1.1
• OpenVG 1.0.1
• DirectX 11
• Google Renderscript

NETWORK

• LW M2M
• CoAP
• MQTT
• 6LoWPAN
• IPv6
• Audio/Video Codec
• mDNS for WiFi
• OpenHAB based Framework
• OpenStack(Swift) Framework

[Roy Nielsen]

Hello,

The pcduino arches is somewhat similar - but not exactly the same specs.  It has a A15x4 + A7x4 arm chip as well.

They've released their schematics.

http://www.linksprite.com/?page_id=826

one can purchase it here:

http://store.linksprite.com/linksprite-arches-single-board-computer-pcduino8-beta/

Regards,
-Roy

[Douglas Pearless]

Hi,

I looked at that board a while ago, my concerns were:

(1) only 14 I/O pins (not enough)

(2) no real-time OS

Cheers

Douglas

[The Guru]

I think I saw this at the national maker fair. Looks interesting but still quite expensive.

Tom

[Peter Shabino]

Paul any updates on the MotionBoards? Just wondering where your at on the build.

Later,
Peter

My projects:
http://www.wire2wire.org/

---

From: pa...@pauljones.id.au
To: fire...@googlegroups.com
Subject: [FirePick] Introducing MotionBoard!
Date: Fri, 12 Jun 2015 01:37:39 +0000

Hi All,

 

I’ve finally finished it – and given it another new name. MotionBoard.

This is a motion processor module with an Arduino footprint that runs Smoothie. The pinout is compatible with both Firepick EMC02 and Ramps 1.4 boards.

Douglas Pearless has the original one and has his Firepick running on it.

 

Currently it has an LPC1769 Arm Cortex M3 processor. NXP is releasing a Cortex M4 later in the year (sometime between July and Nov I think??) that has a very similar footprint to the M3 chip, so I’ve designed the board so it works with both.

 

Design file are available at: https://github.com/PeeJay/MotionBoard

 

Price is $55 + $25 postage (USD)

If you want more than one contact me for a postage quote.

Payments can be made via PayPal to pa...@pauljones.id.au

 

ETA is boards being sent last week of June/First week of July.

 

Special thanks go to Douglas Pearless for implementing the RotaryDelta arm solution in Smoothie, and of course our sleepless leader Neil Jansen for creating such an awesome open source project!

 

Cheers,

Paul.

 

 

 

[Paul Jones]

Yes, I got the boards in on Friday and made the first 3 last night. Sort of, I’ve misplaced a reel of 100n caps… I’m going to go pinch some from work and I’ll get another 8 or so made today, so they should all be shipping this week.

 

 

Paul.

[Douglas Pearless]

Excellent, as soon as I have mine I will prep the changes for the smoothie firmware and get those to my repo

Sent from my iPhone

[Peter Shabino]

Sweet. Finishing up the wiring on my FPD now so will be ready to fire it up by next week :) will be nice to get the smoothie in there from the get go. 

Thanks again, 
Peter

My projects:
http://www.wire2wire.org/

---

From: pa...@pauljones.id.au
To: fire...@googlegroups.com

Subject: RE: [FirePick] Introducing MotionBoard!
Date: Sun, 28 Jun 2015 00:45:10 +0000

[Douglas Pearless]

I have also started work on 

(1) Porting the OpenPnP changes required to Smoothie

(2) Adding my own code for a pressure sensor that sites between the vacuum pump and the SMT head to detect if a part falls off, this can be queries via an M code at the end of he travel to enable OpenPnP to determine if the part made it :-)

Cheers

Douglas

[Paul Jones]

RE pressure sensor: You can achieve the same thing by monitoring the current to the vacuum pump. I just tested it with one of the pumps I have here (not the firepick one), it draws 300 mA with no part and 220 mA with a part. Could be simpler if it’s reliable.

 

Paul.

 

From: fire...@googlegroups.com [mailto:fire...@googlegroups.com] On Behalf Of Douglas Pearless
Sent: Sunday, 28 June 2015 1:54 PM
To: fire...@googlegroups.com
Subject: Re: [FirePick] Introducing MotionBoard!

 

I have also started work on 

[Douglas Pearless]

The only current monitor is not unique to a particular tool so other things changing the overall current draw would confuse the FPD.

I have a simple 4mm T-Barb the sends a feed to a MPXV7007 sensor, then a LM311 which I can set the trigger point (I will have a potentiometer to set the threshold); currently the bits are on my desk; need to complete the circuit diagram and build it on strip-board :-)

Need to complete the circuit and test it; it has the advantage that Smoothie simply needs to check an I/O pin, 1 = vacuum on, 0 = vacuum lost/off

Cheers

Douglas

[Michael Anton]

You might find that it would be better to have a pressure sensor that will sense over a wider range.  A range of +-7 kPa is pretty small, and just the pressure drop of the nozzle may be enough to max out the sensor (depending on where you are measuring the pressure).  The sensors I've seen used have a measurement range of something like 60-80kPa.

Ideally, you would want to measure this with an A/D channel, as the pressure target may change with nozzle size.

Mike

[Douglas Pearless]

I have been reviewing the OpenPnP implementation on Smoothie, and it looks like it is a working solution with a Smoothie X5 board.

There needs to be further analysis to figure out what is required to get OpenPnP working on the Smoothie, (I have also been reviewing the Marlin code too) but if *may* be rather simple (based on my mapping to date.

Looking at the code, I notice OpenPnP uses 2 x solenoid valves per nozzle; is FPD going to require these solenoids for its nozzle?

Cheers

Dougls

[Peter Shabino]

If I had to guess one is for suck the other it for puff.  That allows for a faster pick and drop cycle than just suction as you don't have to wait for the pump to stop and the air to bleed back in. 

That said I bet they have a way to set the dwell on the puff stage. To use their existing framework with just suck control connect the pump to the suck channel and leave blow disconnected. Then set the blow dwell time to however long it takes the pump to stop and bleed down. 

I am working on a dual ported solenoid valve that will allow the FPD to go from suck to blow. Current plan is just a single coil that toggles between them. But if I get it working and folk want it could be modified  easily to dual coils. 

Later, 

Peter

My projects:
http://www.wire2wire.org/

---

From: douglas....@gmail.com

Subject: Re: [FirePick] Introducing MotionBoard!

Date: Mon, 29 Jun 2015 00:34:55 +1200
To: fire...@googlegroups.com

[Jason von Nieda]

No need to guess - just ask :) 

The second solenoid is used as exhaust. I found puff is not necessary, but opening the vacuum channel to atmosphere gets the job done. 

A little background may be helpful:

The current OpenPnP "reference" machine is what I've been calling OpenPnP OpenBuilds. It's an OpenBuilds v-slot based machine which is documented at https://github.com/openpnp/openpnp-openbuilds

This is the machine I had with me at Maker Faire, and it's relatively complete. It's not perfect, but it's working and reproducible.

It uses a Smoothie X5, as you've seen, and the OpenBuilds driver included in OpenPnP works with Smoothie. The driver is a little more complex than a basic Smoothie driver because it needs to handle my dual Z cam arrangement. You can see the driver code at:

https://github.com/openpnp/openpnp/blob/develop/src/main/java/org/openpnp/machine/openbuilds/OpenBuildsDriver.java

The build instructions for the machine are at https://github.com/openpnp/openpnp-openbuilds/wiki/Build-Instructions and they might give you some insight into what a working OpenPnP build looks like.

As I said above, if you have any questions about OpenPnP, just ask. I'm monitoring this list and happy to answer any questions. If it wasn't clear, I'm the author of OpenPnP :)

Jason

[Douglas Pearless]

Hi Jason,

Yep I joined the dots, OpenPnP is rather awesome (as well as FPD ! ).

I have ordered http://www.aliexpress.com/item/Two-position-three-way-mini-12V-solenoid-valve-DC-electronic-control-solenoid-exhaust-snuffle-valve-Free/1934656266.html to try instead of two valves as I simply need ‘ON’ = Nozzle gets vacuum from pump, ‘OFF’ = nozzle connected to fresh air.

I am sure I will have quite a few questions, my goal is to try and have my FPD working on rev A of the MotionBoard by next week at the current level of OpenPnP; and soon after when rev B arrives :-)

Note: I am still considering using 4 x Murata micro-blowers mounted on the to of the rotatable stepper on the PnP head to see what suction it can meaningfully generate.

Cheers

Douglas

[Jason von Nieda]

Thanks Douglas.

I think those solenoids will work okay if you are talking about a single nozzle, so you are probably fine on a FPD. They wouldn't work for me with dual nozzles because the vacuum path would always be open to atmosphere, and it would be impossible to pull a vacuum. 

As for the Muratas, others have been down this path before. You might find some useful information in this thread: https://groups.google.com/forum/#!searchin/openpnp/murata/openpnp/hRXVB-2RtPo/Sak6g4OTFwwJ

As I said earlier, happy to help in anyway I can with OpenPnP and FPD. You can find me on this list, the OpenPnP IRC channel or the OpenPnP mailing list. Lots of contact options available at http://openpnp.org/

Jason

[Douglas Pearless]

NXP have released a pin compatible Cortex M4 with FPU chip, it is the LPC4078FBD100 cortex M4 (LQPF100), I have spent a bit of time mapping the pins and there are minor differences but these should not affect the operation of the chip (to do with grounding the analog reference).

I have made enquiries about getting one of these chips and if successful will replace the LP1629 on the rev A board and enable single precision hardware floating point for Smoothie… (Note the M7 is required for the next level of performance, but NXP have yet to release their version).

Cheers

Douglas

On 12/06/2015, at 1:37 pm, Paul Jones <pa...@pauljones.id.au> wrote:

Hi All,

 

I’ve finally finished it – and given it another new name. MotionBoard.

This is a motion processor module with an Arduino footprint that runs Smoothie. The pinout is compatible with both Firepick EMC02 and Ramps 1.4 boards.

Douglas Pearless has the original one and has his Firepick running on it.

 

Currently it has an LPC1769 Arm Cortex M3 processor. NXP is releasing a Cortex M4 later in the year (sometime between July and Nov I think??) that has a very similar footprint to the M3 chip, so I’ve designed the board so it works with both.

 

Design file are available at: https://github.com/PeeJay/MotionBoard

 

Price is $55 + $25 postage (USD)

If you want more than one contact me for a postage quote.

Payments can be made via PayPal to pa...@pauljones.id.au

 

ETA is boards being sent last week of June/First week of July.

 

Special thanks go to Douglas Pearless for implementing the RotaryDelta arm solution in Smoothie, and of course our sleepless leader Neil Jansen for creating such an awesome open source project!

 

Cheers,

Paul.

 

 

 

-- 
You received this message because you are subscribed to the Google Groups "FirePick" group.
To unsubscribe from this group and stop receiving emails from it, send an email to firepick+u...@googlegroups.com.
To post to this group, send email to fire...@googlegroups.com.
Visit this group at http://groups.google.com/group/firepick.
For more options, visit https://groups.google.com/d/optout.

<front.jpg><back.jpg>

[Paul Jones]

I’ve been trying to get my hands on one as well. It will be a drop-in replacement on Rev B MotionBoards. You’ve got the only existing Rev A board, so just wire pin 13 to ground and the rest is the same.

I’ll do you a deal – You get me a sample or two of the M4 and I’ll send you another MotionBoard with the MCU unpopulated J

 

Paul.

 

From: fire...@googlegroups.com [mailto:fire...@googlegroups.com] On Behalf Of Douglas Pearless
Sent: Friday, 3 July 2015 12:38 PM
To: fire...@googlegroups.com
Subject: Re: [FirePick] Introducing MotionBoard!

 

NXP have released a pin compatible Cortex M4 with FPU chip, it is the LPC4078FBD100 cortex M4 (LQPF100), I have spent a bit of time mapping the pins and there are minor differences but these should not affect the operation of the chip (to do with grounding the analog reference).

[Ristola]

Sweet !

Cant wait to receive the New board. !,  Nice Job !

[Ristola]

Mouser has 246 in stock ready to ship.

On Thursday, July 2, 2015 at 10:38:15 PM UTC-4, Douglas Pearless wrote:

[Ristola]

I can place an order and ship a couple if you can test and populate my board prior to shipping to me.

[Douglas Pearless]

You might want to stick with the LPC1769 for a wee it longer as I have to change the compiler settings and re-test the firmware and make sure it all works properly.

If you have a hot air rework iron available to you, it is relatively easy to replace the CPU when I have it all working.

Cheers

Douglas

[Ristola]

Yes, I have a re-work station, and it would be no problem replacing myself.

Was just offering if you were ready / wanted to test.

[Paul Jones]

Woot!! Last time I checked the delivery date was Nov!.

 

From: fire...@googlegroups.com [mailto:fire...@googlegroups.com] On Behalf Of Ristola
Sent: Friday, 3 July 2015 12:58 PM
To: fire...@googlegroups.com
Subject: Re: [FirePick] Introducing MotionBoard!

 

Mouser has 246 in stock ready to ship.

[Paul Jones]

Yep, I’d wait if you want it to work! There shouldn’t be anything major to change, but who knows what weird stuff will show up J

 

Paul.

 

From: fire...@googlegroups.com [mailto:fire...@googlegroups.com] On Behalf Of Ristola
Sent: Friday, 3 July 2015 1:06 PM
To: fire...@googlegroups.com
Subject: Re: [FirePick] Introducing MotionBoard!

 

Yes, I have a re-work station, and it would be no problem replacing myself.

[Peter Shabino]

so will the existing orders get upgraded (ie did you buy the parts and build them already)?

Later,

Peter

My projects:
http://www.wire2wire.org/

---

From: pa...@pauljones.id.au
To: fire...@googlegroups.com

Subject: RE: [FirePick] Introducing MotionBoard!

Date: Fri, 3 Jul 2015 03:15:13 +0000

[Paul Jones]

I’ve bought the parts already, a few orders have shipped and the rest will follow over the coming week. Not to mention the new chips are more than twice the cost!

The existing Cortex M3 seems to have plenty of horsepower left to spare so I don’t know that it would make any practical difference at this point.

 

Paul.

[Peter Shabino]

np was just curious. Are you sending out messages when you send folks stuff out? (just so I know when it is coming) 

Thanks,

Peter

My projects:
http://www.wire2wire.org/

---

From: pa...@pauljones.id.au
To: fire...@googlegroups.com
Subject: RE: [FirePick] Introducing MotionBoard!

Date: Fri, 3 Jul 2015 03:25:45 +0000

[Paul Jones]

Yes, I will. You’ll get a tracking number.

[Douglas Pearless]

Hi, 

Perhaps if you want to ship three to Paul (assuming you have purchased one from Paul), he can put one on your board and send it on to you, one one his board for his testing and one on mine to continue the firmware development and we can accelerate the the migration to fully floating point.

Or is that a bit too much to ask?

Cheers

Douglas