Siemens SIPLACE 80 F3 to OpenPNP Conversion

[Mark Harris]

I had a full semi-trailer of toys arrive today.. so this is going to be my place to post progress and reverse engineering information for the Siemens SIPLACE 80 F3 machines, their wafflepack, and feeders. No guarantees on success.. no guarantees on speed this will be done at it's mostly going to be an after hours project for me, so I can use my limited budget to pay for PCBs and components to convert the machines rather than pay for my labour.

I'm hoping to document all the internals of the machines, connectors, cables, etc so people who are building a DIY machine will get some benefits out of seeing how the machines work and are put together - I made a start on that with the video's I've already posted. Now the machines are on site I'll be doing more of that once they make it out of the machine shop and up to the lab (this is planned for monday, as 8 pallets are taking up much of the loading bay)

We have two machines - one will be kept as is, with a line computer running it. The other will be reverse engineered and hopefully converted to OpenPNP, this however is going to take awhile. The working machine will be used to check how/where IO runs and works, as well as being used for some small assembly jobs in the meantime. We have students getting involved with this work too - so it's a great learning opportunity for the automation technologies kids.

Obligatory delivery shots.

[Mark Harris]

Well those attachments went well.

[Mark Harris]

On Friday, November 18, 2016 at 5:02:26 PM UTC-7, Mark Harris wrote:

[Jason von Nieda]

Really nice Mark, can't wait to hear more about these machines. I would love to see OpenPnP running one of these monsters :)

Jason

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/b6d12fb5-3fce-437b-aeb3-8b562da8d8eb%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Mark Harris]

It's going to be a big challenge to get OpenPNP running - the big ones are going to be building our own motor controllers (it's not servo or stepper driven), the sheer number of IO (so custom IO boards), and the starhead (that's the biggest one). I'm not worried about any of the items - the controllers are all things i've designed and built before, just a bit different this time :)

The fact it's a chipshooter with the star head and IC head, the in head camera (starhead) plus up camera for the ic head worry me the most.

The feeders are going to be really easy to drive, even if i do just replace the controllers in them - not much in the way of an encoder to worry about, they way they work is very clever.

I'm hoping the waffle pack feeder can be used for cut tape, thats also going to be quite a bit of work to get running, but isn't a huge priority.

On 18 November 2016 at 22:34, Jason von Nieda <ja...@vonnieda.org> wrote:

Really nice Mark, can't wait to hear more about these machines. I would love to see OpenPnP running one of these monsters :)

Jason

On Fri, Nov 18, 2016 at 6:38 PM Mark Harris <ma...@rris.com.au> wrote:

On Friday, November 18, 2016 at 5:02:26 PM UTC-7, Mark Harris wrote:

I had a full semi-trailer of toys arrive today.. so this is going to be my place to post progress and reverse engineering information for the Siemens SIPLACE 80 F3 machines, their wafflepack, and feeders. No guarantees on success.. no guarantees on speed this will be done at it's mostly going to be an after hours project for me, so I can use my limited budget to pay for PCBs and components to convert the machines rather than pay for my labour.

I'm hoping to document all the internals of the machines, connectors, cables, etc so people who are building a DIY machine will get some benefits out of seeing how the machines work and are put together - I made a start on that with the video's I've already posted. Now the machines are on site I'll be doing more of that once they make it out of the machine shop and up to the lab (this is planned for monday, as 8 pallets are taking up much of the loading bay)

We have two machines - one will be kept as is, with a line computer running it. The other will be reverse engineered and hopefully converted to OpenPNP, this however is going to take awhile. The working machine will be used to check how/where IO runs and works, as well as being used for some small assembly jobs in the meantime. We have students getting involved with this work too - so it's a great learning opportunity for the automation technologies kids.

Obligatory delivery shots.

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

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

To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/b6d12fb5-3fce-437b-aeb3-8b562da8d8eb%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CA%2BQw0jwBnT%3D0MkQwHWT-pejft3%3DPq0UCxs%2BU46RrCDBQiB%2B7%2Bg%40mail.gmail.com.

[Mark Harris]

I went down to take a look at the machines today - we're going to be rearranging the office tomorrow to fit them upstairs. Unfortunately I have meetings off campus from 1pm on so I won't get to see their exodus across campus (same building, but about 600m outside to get from our loading bay to one of the other loading bays that have a big roller door into the building's corridors/freight elevator). At least there is no snow on the ground or forecast :)

The equipment looks to be in good condition. These things would make a wonderful treasure trove of parts for any CNC application - so many huge rails, motors, belts and sensors everywhere. Everytime I look a little further at how something works I find more motors and pulleys. The motor moving the gantry on the machine slated for conversion is dated 2006, so I guess it was replaced a decade ago. These are US$3k new and still manufactured. There's certainly some parts of the machine that at the moment you look at and have to ask yourself what on earth it does - it's going to be a great learning experience figuring all this out.

Hopefully Wednesday night I'll be staying back to get some photos and perhaps video of how bits and pieces work.

[Mark Harris]

Well, its wednesday - so time for pictures.

http://www.markhphoto.com.au/Siemens-Siplace-80-F3/

I'm not going to bother trying to upload to google groups. It didnt like my earlier pictures :(

[Mark Harris]

I also have the full service manual as a PDF if anyone is interested, it's 14mb though so I can't upload it - email me directly.

[John Socha-Leialoha]

Thanks for posting the photos! I have some questions about the feeders. I noticed there are some registration pins along the front and back, under the feeders. Can you tell me the spacing of these pins?

Do the 8 mm feeders come as a bank of two?

Can you measure the widths of the feeders for different tape widths?

I'm curious about these dimensions as I've been working on a design for injection molded tape strip holders, and one of the last things I need to finish is a way to hold them in place, and a registration system.

Thanks,

  -- John

[Mark Harris]

Im not in the office today so I will have to check tomorrow for you. These feeders reach weigh a few kg each, they basically a solid block of metal. I'll get pictures tomorrow for you.

The feeders mount on a feeder table that weighs several hundred kg, we had to lift it onto the machine with a forklift. The table has a small registration pin at the front which the feeder slides onto (front is a v shape) and just behind that is a magnetic strip which pulls the feeder down quite strongly. The rear pin is just a pocket so it locks the feeder square and in as far as it needs to be. Nothing but it's own mass and the magnetic strip holds it in place.

There are no spring contacts for the feeder, there is a cable that goes down under the feeders to the feeder control computer. All I know about it so far is that it has 7pins, and a constant 30v dc on one pin (according to the service manual).

The feeders are indeed dual lane for 8mm and can move take along either 2mm or 4mm per component. I need to make a video showing how they work as it is incredibly simple and reliable. No fancy encoders required or much gear reduction. The encoder has one pulse per 2mm movement!

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+unsubscribe@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/97c0829b-e39f-42b6-b872-6903d8bb7a74%40googlegroups.com.

[Mark Harris]

Sorry for the late reply, I haven't been in to work. I made a quick video for you John showing the feeders and how they mount - as well as a few photos which are linked in the video description:

https://youtu.be/S-kT5envoLE

[Mark Harris]

Had a little bit of time to poke around the machines today. 

Here's a look at the motor controller PCBs and some other rambling:

https://www.youtube.com/watch?v=ssowPOaZqp0

Here's the X axis motor breakout and the magnetic encoders:

https://www.youtube.com/watch?v=ptLdYD3QuQQ

If there's anything specific you'd like to see, I am probably going to be in a position to record some more video/photos on Friday, work is closed between christmas and new year so I wont be able to do anything in that time.

[Mark Harris]

To follow that video up, looking at the maintenance manual, regarding the Y0004 board that is the "Conversion Board, Large Axis", which is next to the X axis motor.

Connectors X5 and X6 are "END POS. Y AXIS BERO 1" and BERO 2. BERO is Siemens generic name for their whole line of sensors of all types, so not overly helpful.

X8 and X9 (2 pin) are "LIMIT SWITCH X AXIS".

X4 which has many pins and goes to the encoder is "Y SCALE". The encoder strip has an accuracy of 0.0025mm

Other "Conversion Boards" include the Gantry and two Small Axis.

This is the X Axis motor:

[Mark Harris]

For future reference on the Conversion Board - Small Axis (Y0005), since the manual is hard to search.

(1) 'Small Axis' conversion board Y0005

X1 Plug for the CRDL measurement board (option)

X9 Ribbon cable to connector X1 on the revolver head

X10 Ribbon cable to connector X2 on the revolver head

X13 Ribbon cable to vacuum measurement board Y0007

X15 Cable - ’Motor/tacho/BERO z axis’

X16 Cable - ’z axis track’

X17 Cable - ’Screwdriver 2’

X18 Cable - ’Screwdriver 1’

X20 Cable - ’Forced air valve, placement circuit’

X21 Cable - ’Track signals, X axis’

X22 Cable - ’Illumination control, PCB camera’

X23 Cable - ’PCB camera’

X24 Cable - ’Illumination control, component camera’

X25 Cable - ’Component camera’

X26 Cable - ’BERO end position, X axis’

X27 Cable - ’Forced air valve, reject circuit’

CRDL is a component value tester. Logican said they tried it and it was useless. Slowed down production and didn't add anything much in the way of quality/reliability improvements.

(1) Vacuum generator block

(2) Quick-release couplings for the compressed air lines.

[Mark Harris]

The X axis motor is made by parker, the tachogenerator may be possible to use as an encoder, but not a quadrature. I'll take a look with the oscilloscope on Friday if I'm in.

The motors are here: https://inverterdrive.com/group/Servo-Motors-DC/RS330E-R1100-Servo-Motor-Parvex-ABB-Alsthom-GEC/ - datasheets are on that page too.

There is a quadrature encoder module for this motor, the motor has a rear shaft it bolts onto, available here: https://inverterdrive.com/group/Encoder-Tachogenerator-Parvex-Resolver/Parker-SSD-Parvex-C2-Encoder-Codeur/

The Y Axis motor does not have a quadrature encoder available, only the tachogenerator that is built into it. Both axes do have the "scale" encoder though, it will be interesting to look at the output/input on the pins of the header (X4 on the Conversion Board, Large Axis) to see what that "BERO" is doing.

[Mark Harris]

I dropped by work to pick up some stuff to work on over the christmas break, and had a quick play with the SIPLACE.

Linear Axis Encoder:: https://www.youtube.com/watch?v=N6mH-cIOEaQ

Motor Tacho Generator: https://www.youtube.com/watch?v=SSll8LpYNJ0

The tacho gen doesn't look to be of much use.

I know i'm not getting replies on this thread - but i'm guessing people are still interested as i'm getting a few views for the videos.

There won't be any more work on the PNP until next year - our facility is closed and locked until January 2nd.

Having two machines is very handy, I was able to have one powered up and the other off so I can probe to check voltages/signals as well as probe where pins go very quickly!

[Anthony Webb]

Still as jealous as ever, thanks for the updates!  Have a nice holiday break Mark!

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyQv8QvTD64rLv2PrfNbf6-RCVqOckU%3D-%3DiNMWXpOo%2BMyQ%40mail.gmail.com.

[Graeme Bridge]

Watching reverse engineering is always good 

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

To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/97c0829b-e39f-42b6-b872-6903d8bb7a74%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

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

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

To post to this group, send email to ope...@googlegroups.com.

[Jason von Nieda]

Hey Mark, definitely keep the posts coming! Really enjoying watching the videos and reading about your adventures with these machines. I don't have much to say besides "Cool!" :)

Jason

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/2331fec6-c132-4e6b-b993-2bca460fc5f1%40googlegroups.com.

[Mark Harris]

Thanks guys.

Just to be clear, i wasn't asking for replies, only justifying my spam haha. 

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

To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/2331fec6-c132-4e6b-b993-2bca460fc5f1%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CA%2BQw0jwQuV86dHNdPJbkhKOG%3DbH7KMr-kccM8byzT2uv8678-g%40mail.gmail.com.

[Mark Harris]

Back at work today - anything anyone would like to see more of on the Siplace? Or, anything anyone thinks I should dig into next? :)

Requests are welcome, otherwise I just go randomly poking through stuff hehe.

[Mark Harris]

So today when talking to a colleague and standing near the pick and place machines... I had an idea of how to make this conversion go a whole lot easier.

Basically, rip the 13 heads off and replace them with two. Get the machine working like that using Gino Magarotto's Tarocco code (https://www.crowdsupply.com/citrus-cnc/tarocco) on a higher voltage, higher amperage board (85v motors...)  This is then just a 1500kg typical OpenPNP machine :) It just happens to have a huge work area and DC servos for the X/Y (and possibly Z too!)

Then, the IC head will be reinstalled after I get it reverse engineered. Then the starhead, some time much later this year once openpnp can run it, and i've got the firmware/hardware to run it.

The Conversion Plan: https://www.youtube.com/watch?v=d0LGr6BsCX0

So, off with the IC head:

https://www.youtube.com/watch?v=QMmR8aeJuI0

And then a lot of bumbling around figuring out how to pull the Starhead off: https://www.youtube.com/watch?v=ylE5XYXqbKE (not really worth watching IMO)

And finally, The Conversion Plan Part Deux. https://www.youtube.com/watch?v=8PF9zup9vs8

Now I just need to get some measurements, some CAD, some waterjetting... bob's your mother's brother.

I will be doing a teardown of the feeders some time soon, I really want to get to this - they are so very simple yet fast and accurate - it's brilliant.

[Glen English]

Mark, are you putting an intermediate layer between the SiPlace and OpenPNP?

That's what I did for the TVM920- OpenPNP was a bit too high level for driving the TVM920.

In fact there are two extra levels- the TVM920 STM32 based motion and IO controller, (unmodified)  and a helper application on windows that turns high level OpenPNP commands like Move_to into all the motion control stuff to tell the STM32 where it must go and what speed and what accel and process all the bitwise GPIOs into high level commands  and status for Open PNP.

My help application deals with all the stuff like figuring out when the machine has finished movements etc and reporting that to OpenPNP.

Happy to provide ideas, suggestions...

[Mark Harris]

The SIPLACE uses a HPIB signal and expects a line computer to be controlling it... i will not be using any of the siplace electronics at all.

My plan for getting this machine going in a hurry (plan is 4 weeks!) is to run it on a smoothie board hahaha.

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+unsubscribe@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/7cb1cb8e-202b-4d09-a51b-3232f2a88c09%40googlegroups.com.

[Mark Harris]

For my own future reference - I suspect the feeder is using 1B series connectors:

http://www.digikey.com/products/en/connectors-interconnects/circular-connectors/436?k=&pkeyword=&pv88=15&FV=ffe001b4%2Cffec7872&mnonly=0&newproducts=0&ColumnSort=0&page=1&stock=1&quantity=0&ptm=0&fid=0&pageSize=500

They are visually identical for both sides, I'll need to take some measurements to confirm.

[Mark Harris]

The Backplane Connectors are DIN 41612.

The motor controllers are Type F, 32 pin - http://www.digikey.com/product-detail/en/harting/09061486901/1195-1306-ND/3179889

The Y axis motor uses Molex MLX™ 42021/2 connectors to the motor controller.

[Mark Harris]

Replacement X/Y Motor controllers are designed. I have to thank Gino Magarotto who designed the firmware to run these:

https://www.crowdsupply.com/citrus-cnc/tarocco - this is his project if you're looking for some motors and drivers for your machine. DC servos are very cheap and with closed loop control, very accurate vs open loop steppers.

These should slot right into the existing backplane. If that fails, there are barrier blocks for connecting power and motor - as well as the terminal blocks at the end. The board is very large and empty because I'm replacing the existing board and have about 500 less passives than the original (no joke).

The MOSFETS on this board are 5.9mOhm RDSon... the original boards were 85mOhm (and state of the art!). So no heatsink required, especially considering that these are right above the rack fans. I have provisions for a 60mm fan should I decide to use this design for other projects which need higher amperage. The mounting holes are also for if i decide to use it for other projects :)

The original board does not provide Step/Direction inputs obviously... so I'm going to hijack these at the backplane. There is a ribbon connector that provides the logic control to this board which I'll be taking over.

Pinouts for the connectors on the backplane board.

Backplane board.

I'm actually keeping the wiring to the motor. This allows me to use the very large inductors they are running the motor wires through (guessing for dumping the field collapse into), as well keep the existing brake boards! The brake controller is switched on and off with a 12V relay which makes life super simple for me. Both the X and Y motor controllers have identical pinouts, the X axis motor controller has bigger fets however.

They are not exactly the cheapest motor controllers, but I didnt design them to be (I used a CA$9 fet for reverse polarity protection...). But they should be good for at least 150V and the board is likely to be just fine with over 120A of current on it. The connectors are only rated for 30A, but whatever :) The largest motor in the machine is 85v/15A - about 1.2kW.

These will be ordered monday or tuesday - hopefully I'll get them in 2 weeks before China shuts down for spring festival.

There are many photos of the  backplane wiring here: http://www.markhphoto.com.au/Siemens-Siplace-80-F3/

Next up will be new feeder controllers. Hopefully I can get them in too before spring festival!

[Rich Obermeyer]

Mark, How much are those boards from China?

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyTGJzg0jU%3DuA8cYuWWYAQLdGzTAU%3DpEhWFKy%3D6Zpij-Yw%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

--

Rich

[Mark Harris]

No clue, havent gotten that far yet! Just setting up the output job now. I'll probably get them from PCBCart.com - more expensive than other chinese vendors for just a couple of boards (due to tooling) but *very* nice quality. I figure boards that are going to be running 85v and throwing around the gantry on these machines are worth spending a few more dollars on.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAOvy4macRVAb%3DPdAh4LR_LcKjG00Ua6ua-R1vJpnnF1v%2BE1w4A%40mail.gmail.com.

[Rich Obermeyer]

Try OSHpark.com and compare prices.  They are US based and don't shut down for a month each year.  Top notch from the boards I have got.  Much better than supporting China for no good reason.

Let us know what you find.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTySbXDa14r0cQYJBDtXNwJDXa4g%2B16qrbzZTzE-UC1_VYA%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

--

Rich

[Mark Harris]

Honestly, I haven't been a fan of oshpark. Turnaround, options, file checking and quality have not compared to my chinese suppliers - plus they are not exactly cheap.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAOvy4mZRd%2B7p53z94BWQYWKZUtLRT6eQkq4kf%2B%2B6tBJ%3D0dVbqg%40mail.gmail.com.

[Paul Jones]

Try http://www.circuitlabs.co.nz/ if you haven’t already. Price is reasonable (not china price though) and I turnaround is as fast as you want. Being in the sameish timezone is also a plus! (I take it you are in Aus?)

 

Paul.

 

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

To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/7cb1cb8e-202b-4d09-a51b-3232f2a88c09%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyTGJzg0jU%3DuA8cYuWWYAQLdGzTAU%3DpEhWFKy%3D6Zpij-Yw%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

 

--

Rich

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAOvy4macRVAb%3DPdAh4LR_LcKjG00Ua6ua-R1vJpnnF1v%2BE1w4A%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

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

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

To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTySbXDa14r0cQYJBDtXNwJDXa4g%2B16qrbzZTzE-UC1_VYA%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

 

--

Rich

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

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

To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAOvy4mZRd%2B7p53z94BWQYWKZUtLRT6eQkq4kf%2B%2B6tBJ%3D0dVbqg%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyQuE4eyoOzKy-8GU6zoLa5XSHgBm94TZV8CN71-%2BUOW4g%40mail.gmail.com.

[Mark Harris]

Heh actually i'm in Canada. I used to live in Western Australia a few years ago

APCircuits in town (calgary) do same day turnaround: http://www.apcircuits.com/ - they boards look like crap though (they function.. but look like crap). They are also not cheap.

I use Dirty PCBs for prototypes: http://dirtypcbs.com/store/pcbs

For breakouts and such for dev use I use the Voltera V-One ;)

For 6+ layer boards with blind vias and such I tend to use Ray Ming PCB: http://www.raypcb.com/ - they are not actually a manufactuer, more like PCB cart in they act as the 'english' interface between the pcb factory and the client.. I use english to lightly describe the language they use. The comments from the PCB engineers can be truly baffling and hilarious. But the prices are great for really complex boards, and the quality is fantastic.

As mentioned, general prototypes/small production batches go through PCBCart, their english is fantastic, their prices are good - i'm happy to pay tooling for the level of attention they give the board when checking it over before manufacturing.

I've used Amitron in the USA for overnight 4 layer boards (on Ray K's suggestion) - their boards were beautiful, expensive and had great service. But for overnight boards, they were 1/4 the price of 4PCB.com (Advanced Circuits)... staff were very helpful over the phone too.

I've used OSHStencils a number of times for stencils (typically I just order them with PCBCart order for $20) - beautiful work from them - but not much in the way of post processing. They wont modify the apertures for large paste areas like PCBCart will. The quality of the laser finish and steel is far superior though.

At work we don't make anything at all - I'm basically a contract engineer using government money to do client jobs. So everything I do is a one off, the most i've made of something so far is 10 - soon to be greatly surpassed by 50 units for an internal job. I make around 50 prototypes a year.. so I've tried many PCB services (plus I've been doing this sort of thing for about 8yrs), typically money is not so much a consideration as time and quality. A bad board or a mistake on my part in submitting files costs far more in lost time or wasted effort than a tooling charge :)

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

To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/7cb1cb8e-202b-4d09-a51b-3232f2a88c09%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyTGJzg0jU%3DuA8cYuWWYAQLdGzTAU%3DpEhWFKy%3D6Zpij-Yw%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

 

--

Rich

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAOvy4macRVAb%3DPdAh4LR_LcKjG00Ua6ua-R1vJpnnF1v%2BE1w4A%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

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

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

To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTySbXDa14r0cQYJBDtXNwJDXa4g%2B16qrbzZTzE-UC1_VYA%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

 

--

Rich

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

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

To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAOvy4mZRd%2B7p53z94BWQYWKZUtLRT6eQkq4kf%2B%2B6tBJ%3D0dVbqg%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyQuE4eyoOzKy-8GU6zoLa5XSHgBm94TZV8CN71-%2BUOW4g%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/SY3PR01MB0971CB11BF6CB5DA62184A009E640%40SY3PR01MB0971.ausprd01.prod.outlook.com.

[Mark Harris]

Interestingly, PCBCart are $2 cheaper than DirtyPCBs ($78) for the prototype grade boards. Production boards only $140 for 10 - $55 for tooling on a 2 layer (I just spent $600 on 10 prototypes at PCBCart on Friday.. so $150 is awesome).

Considering these have $85 in reel volumes of parts on them, meaning more like $100-120 each for 2 boards... $140 for boards is no big deal :) I'd prefer to pay the extra for thicker mask to insulate (voltage) the boards.

[Mark Harris]

Controller  board files are on Github: https://github.com/issus/Heavy-Duty-Tarocco - PDF schematic added for non-altium users.

[Michael Anton]

Have you ever tried http://www.smart-prototyping.com/?  I use them for pretty much all of my boards, and the quality has always been excellent.  They seem to be the same folks that Seeedstudio use.

Mike

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

To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/7cb1cb8e-202b-4d09-a51b-3232f2a88c09%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyTGJzg0jU%3DuA8cYuWWYAQLdGzTAU%3DpEhWFKy%3D6Zpij-Yw%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

 

--

Rich

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAOvy4macRVAb%3DPdAh4LR_LcKjG00Ua6ua-R1vJpnnF1v%2BE1w4A%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

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

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

To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTySbXDa14r0cQYJBDtXNwJDXa4g%2B16qrbzZTzE-UC1_VYA%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

 

--

Rich

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

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

To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAOvy4mZRd%2B7p53z94BWQYWKZUtLRT6eQkq4kf%2B%2B6tBJ%3D0dVbqg%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyQuE4eyoOzKy-8GU6zoLa5XSHgBm94TZV8CN71-%2BUOW4g%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

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

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

To post to this group, send email to ope...@googlegroups.com.

[Mark Harris]

I have not tried them - but I have ordered boards from Seeed at work, experience was not pleasant. Our file names didn't *exactly* match their required format (mine are just the default altium names), and they couldn't be arsed looking at what they were. 

Forget submitting Gerber X2 files to them - made that mistake once. Your order will just sit there for 3 weeks as you ask what the heck is happening with it every 3 days. Only to get the same, useless canned response every time. That really screwed up the project timeline and then every other project timeline - cost us thousands of dollars. Will never use Seeed again, even if they give me boards for free.

[Mark Harris]

So on the feeders - confirmed the voltage on the connectors is 30v. Motors are 30v surprise surprise. Solenoid is 18v which I find interesting, so I'll just PWM that I think.

The light gates are omron parts: https://www.omron.com/ecb/products/pdf/en-ee_sx1088.pdf

[Michael Anton]

I can't say I've had those sorts of problems, though I deal with Smart-Prototyping directly, rather than going through Seeed like I used to.  I usually find the communication with them to be pretty good, and they will produce boards that don't fit in with their online ordering system, if you ask them via email.  I'll admit that I do submit the files as they want them, as it seems to solve a bunch of problems.  I use PCAD, so I can set the names to be whatever I want, so I don't find this to be a big deal.  One thing to note, if you ever do want to try them, they don't read readme files, so if you want something special done, put in a note as part of the order process.  Also if you want some holes unplated, submit separate drill files for plated and unplated holes, and tell them in the note.  Sometimes they get it right regardless, but I've had them mess up before, and since I switched to two drill files, they always do it correctly.

[Mark Harris]

Sounds like it's just worth spending the extra money at pcb cart :)

--

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/48516465-9ec1-477a-9803-294e6f19bcc2%40googlegroups.com.

[Andrew Frazer]

I'd give pcbway.com a go.  They made a LOT of boards for me in the last 12 months and they are reliable and good value. And you can 

On Tuesday, January 10, 2017 at 9:00:12 PM UTC+13, Mark Harris wrote:

Sounds like it's just worth spending the extra money at pcb cart :)

On Jan 10, 2017 00:58, "Michael Anton" <3d.m...@gmail.com> wrote:

I can't say I've had those sorts of problems, though I deal with Smart-Prototyping directly, rather than going through Seeed like I used to.  I usually find the communication with them to be pretty good, and they will produce boards that don't fit in with their online ordering system, if you ask them via email.  I'll admit that I do submit the files as they want them, as it seems to solve a bunch of problems.  I use PCAD, so I can set the names to be whatever I want, so I don't find this to be a big deal.  One thing to note, if you ever do want to try them, they don't read readme files, so if you want something special done, put in a note as part of the order process.  Also if you want some holes unplated, submit separate drill files for plated and unplated holes, and tell them in the note.  Sometimes they get it right regardless, but I've had them mess up before, and since I switched to two drill files, they always do it correctly.

On Monday, January 9, 2017 at 11:04:25 PM UTC-7, Mark Harris wrote:

I have not tried them - but I have ordered boards from Seeed at work, experience was not pleasant. Our file names didn't *exactly* match their required format (mine are just the default altium names), and they couldn't be arsed looking at what they were. 

Forget submitting Gerber X2 files to them - made that mistake once. Your order will just sit there for 3 weeks as you ask what the heck is happening with it every 3 days. Only to get the same, useless canned response every time. That really screwed up the project timeline and then every other project timeline - cost us thousands of dollars. Will never use Seeed again, even if they give me boards for free.

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

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

[Glen English]

Agreed with Andrew on PCBWAY. good product.

watch out for their board quote calculator it can be a bit odd....

I found a small 8 layer PCB ( 50x75) was more much expensive than a 150x150 or 150 x 200

then I found I went smaller to 40x70, it got cheaper again, IE something amiss -   but just be aware of it if the size v cost looks non monotonic.

[Michael Anton]

Nah, now that I have it all sorted out, they are great to deal with...  For many of the boards I run, it just doesn't matter.  It is only once in awhile that I actually care if they do something odd.

On Tuesday, January 10, 2017 at 1:00:12 AM UTC-7, Mark Harris wrote:

Sounds like it's just worth spending the extra money at pcb cart :)

On Jan 10, 2017 00:58, "Michael Anton" <3d.m...@gmail.com> wrote:

I can't say I've had those sorts of problems, though I deal with Smart-Prototyping directly, rather than going through Seeed like I used to.  I usually find the communication with them to be pretty good, and they will produce boards that don't fit in with their online ordering system, if you ask them via email.  I'll admit that I do submit the files as they want them, as it seems to solve a bunch of problems.  I use PCAD, so I can set the names to be whatever I want, so I don't find this to be a big deal.  One thing to note, if you ever do want to try them, they don't read readme files, so if you want something special done, put in a note as part of the order process.  Also if you want some holes unplated, submit separate drill files for plated and unplated holes, and tell them in the note.  Sometimes they get it right regardless, but I've had them mess up before, and since I switched to two drill files, they always do it correctly.

On Monday, January 9, 2017 at 11:04:25 PM UTC-7, Mark Harris wrote:

I have not tried them - but I have ordered boards from Seeed at work, experience was not pleasant. Our file names didn't *exactly* match their required format (mine are just the default altium names), and they couldn't be arsed looking at what they were. 

Forget submitting Gerber X2 files to them - made that mistake once. Your order will just sit there for 3 weeks as you ask what the heck is happening with it every 3 days. Only to get the same, useless canned response every time. That really screwed up the project timeline and then every other project timeline - cost us thousands of dollars. Will never use Seeed again, even if they give me boards for free.

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

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

[Michael Anton]

I'll keep them in mind since you recommend them.  Smart-Prototyping has extra charges for panelized boards of the same design, which just doesn't make much sense, so it would be worthwhile looking at someone else if that is needed.

[Mark Harris]

More progress.

Dual lane feeder controller that can go in an existing double 8mm siplace feeder. RS485 or digital io control. 4 motor drivers, two solenoid drivers, 4 photointerrupter inputs, 4 button/2led connections. All on a 2 layer board.

PCB order going in now :)

The motor controllers have been confirmed and are scheduled to ship at some point next week.

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/88221fe1-acd2-4878-8406-e11fbdde051e%40googlegroups.com.

[Jason von Nieda]

Looks great Mark! What motor drivers are you using?

Jason

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/88221fe1-acd2-4878-8406-e11fbdde051e%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyTyiT3KcrGrxUjaLYpXwxM2C8QB9RRdbwVWm6_8De-2zQ%40mail.gmail.com.

[Mark Harris]

2x A4954 integrated h-bridges. The motors run on 30v, but are very low amperage, stalling at 0.75A and pushing tape through at full speed using 0.2A.

The design files are on github: https://github.com/issus/Dual-Lane-Feeder-Controller

Altium files though... so you might not be able to do much with them unless you have altium.

The bit motor controllers are up there too: https://github.com/issus/Heavy-Duty-Tarocco

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/88221fe1-acd2-4878-8406-e11fbdde051e%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyTyiT3KcrGrxUjaLYpXwxM2C8QB9RRdbwVWm6_8De-2zQ%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CA%2BQw0jy0%2BYW60UYNqbs7mGUbpH18DK_G3Femv8uzROiRjbQNvA%40mail.gmail.com.

[Anthony Webb]

Great looking boards Mark!

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyRyfVshh_0ABaqT0rog0zrCSLspR_b1PKv-6Wv3NtiU9w%40mail.gmail.com.

[Mark Harris]

Motor controller parts arrived today, boards shipped last night - should be here on Monday.

[Anthony Webb]

Thats pretty exciting, how many of them are you building?

On Fri, Jan 20, 2017 at 12:26 PM, Mark Harris <ma...@rris.com.au> wrote:

Motor controller parts arrived today, boards shipped last night - should be here on Monday.

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+unsubscribe@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTySRbTG_%3DS6x2mgYW_0tiAi-QKS-asNeBfF3TbN-G%3DGrEQ%40mail.gmail.com.

[Mark Harris]

I'll build 1, if it doesn't blow up, I'll build 1 more. I have parts for 3 of them, and 10 boards :) At the moment I just want to get the X and Y of one machine moving, the other machine is going to stay as delivered for reference usage. It's going to take me a bit of time to figure out where the linear encoder signals terminate within the machine, and how power gets up to the small motor - I've started that process but am short on free time right now.

The feeder boards are due to ship out next Tuesday, so I'll get the parts for those next week. I have enough boards for 25 feeders (50x 8mm lanes). If these work out, I'll make a separate board for the wider feeders at the point I need them. I'll probably only convert about 5 feeders over for now, since that's all I'll need for reels.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CALsNZy3qoAYFSh-5S1gPU74ziLxSGyck1RUMxMgfsR0Q4QCUoA%40mail.gmail.com.

[Andrew Frazer]

This converting big machines to run openpnp is taking a mind of its own.

Jason pointed out that image processing is going to be the real challenge.

On Saturday, January 21, 2017 at 8:41:04 AM UTC+13, Mark Harris wrote:

I'll build 1, if it doesn't blow up, I'll build 1 more. I have parts for 3 of them, and 10 boards :) At the moment I just want to get the X and Y of one machine moving, the other machine is going to stay as delivered for reference usage. It's going to take me a bit of time to figure out where the linear encoder signals terminate within the machine, and how power gets up to the small motor - I've started that process but am short on free time right now.

The feeder boards are due to ship out next Tuesday, so I'll get the parts for those next week. I have enough boards for 25 feeders (50x 8mm lanes). If these work out, I'll make a separate board for the wider feeders at the point I need them. I'll probably only convert about 5 feeders over for now, since that's all I'll need for reels.

On 20 January 2017 at 12:30, Anthony Webb <anthon...@gmail.com> wrote:

Thats pretty exciting, how many of them are you building?

On Fri, Jan 20, 2017 at 12:26 PM, Mark Harris <ma...@rris.com.au> wrote:

Motor controller parts arrived today, boards shipped last night - should be here on Monday.

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTySRbTG_%3DS6x2mgYW_0tiAi-QKS-asNeBfF3TbN-G%3DGrEQ%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

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

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

To post to this group, send email to ope...@googlegroups.com.

[Mark Harris]

As in speed or getting the cameras? I'm just going to mount ELP cams on mine for now haha. I'd like to run some USB 3 global shutter cams, especially when I get the starhead running later in the year.

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/0b3b71f5-0348-45f1-a0d1-17c8f44be540%40googlegroups.com.

[Andrew Frazer]

I think it is speed that you'll be able to capture the image.. This might be solved with with global shutter.. I don't know.. somethign that will need to be considered.

[Mark Harris]

For now the ELPs are going to be fine, running a 12,000-18,000CPH chip shooter on two little heads at more like 1500CPH is kinda criminal, but its a stepping stone. There are 8 motors with 6 encoders in the star head I believe, with two cameras (one down cam, one component cam - the equivalent of openpnp's up cam). The stock machine does vision in under 70ms per part - so 840ms to pick 12 parts, vision them and have them ready for placement. 

At 30FPS you have one image frame, maybe two depending on timing to get a shot of the part as it flies by - so i'm thinking I'll run 60fps global shutter cameras to give openpnp some hope of getting a picture of the part at full speed. I suspect it's going to be a bit of an iterative process with Jason's help to get this machine anywhere near its stock speed... but I don't need it anywhere near its stock speed - i'd be happy with 100cph out of it for now, 2-3hrs to build a prototype board that takes me longer than that (90% of my time is spent looking for where the damn part is on the board) and I can be off doing other things while it runs.

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/a57f31a5-1e79-444f-9109-34c1e0b5f52b%40googlegroups.com.

[Paul Kelly]

Nb, this is mostly my meandering continuation of a long running intermittent conversation with Mark. There is a sliver of relevance to the list so….

 

but I don't need it anywhere near its stock speed - i'd be happy with 100cph out of it for now, 2-3hrs to build a prototype board that takes me longer than that (90% of my time is spent looking for where the damn part is on the board) and I can be off doing other things while it runs.

 

Last week I would have agreed with you… 

We’re taking on a couple of board fab jobs for other peoples projects (because the machines sit there 95% of the time turned off and I’m trying to justify another hire).. Anyhow, having worked all through Christmas and new years (as usual) I took a couple of slack days and did a sly time and motion study of this run of 100 boards (about 80 placements per PCB). What was clear was that the CPH (It looked like the machines were running around 1200-1500) was a real bottleneck…

 

PK

 

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/a57f31a5-1e79-444f-9109-34c1e0b5f52b%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

 

--

You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyQWHzNCauVq2bjErf8QD7AXTVQzWOfjszOPrb0Gpg_Ytw%40mail.gmail.com.

[Mark Harris]

1200-1500 isn't really that awful for the age and style of machines you have - they are not chipshooters... but neither do you (or I!) need them to be. You have board magazines and conveyors right? 

Most of my stuff is one off prototypes, 1-3 boards a week, sometimes I get a gap of 2-3 weeks between boards which is nice. The big push for me to get one of these machines running sooner rather than later is a run of 50 boards for an internal project due some time in feb, at which point I don't want to be hand placing them (there is neither budget nor scheduling for that much labour) and the run is too small to hand off to a local assembler.

But my typical use for these is going to be boards with 80-400 parts in 50-150 line items. If hand assembling boards 20-30 times, I can get my speed up quite a bit.. I never do that though. But finding 400 placements on an assembly drawing/bom is far more time consuming than placing the actual parts. The SIPLACE software setup isn't really designed at all for one offs, its 100% a line machine and would take me 3-4hrs to setup a job on... so no point (again, Paul and I have had this discussion several times over the years). Getting it over to openpnp should drop that time dramatically... such that actual labour time on the assembly for me is a fraction of what it is for hand assembly, even like I say, if it is 100cph - the machine wouldn't be doing anything else in that time, and I sure will be!

I'm hoping that almost 2kW of motor power will overcome dropping to two heads, and give me something like the CPH of a PPM Quad or the Topaz/similar standalone machines.

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

To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/a57f31a5-1e79-444f-9109-34c1e0b5f52b%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyQWHzNCauVq2bjErf8QD7AXTVQzWOfjszOPrb0Gpg_Ytw%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/001f01d273d4%24e3ef0880%24abcd1980%24%40caswa.com.

[Matt Brocklehurst]

How do the commercial machines (esp. the older ones) do vision so fast ? I've said before I've played a lot on a iPulse M3 which is a few years old now and it's blisteringly fast with vision too

Is OpenCv just really slow? I'm surprised on modern cpus compared to what will be in the iPulse

Sent from my iPhone

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyTTVxjyNRtp9kXs9ia3y9hs-wnQpZCPV4qdrbBRf7QOoA%40mail.gmail.com.

[Jason von Nieda]

OpenCV is fast, or at least fast enough. More than fast enough. Any slowness is going to be on the part of OpenPnP.

The situation is basically that no one has really tried to use OpenPnP with a machine this fast, so there are unknowns. The biggest hurdle is making sure you get a clear image. If the camera is capable of doing that with the machine not fully settled, or if the machine doesn't actually need any settle time, it shouldn't be a problem. 

Most of the hardware people use with OpenPnP is slow (compared to commercial) and not very rigid. In addition, we use low end cameras that have frame delays and buffering and all kinds of issues. So, to overcome all that, when we image we completely stop the machine, wait a while and then grab an image. On my machine, for instance, that "settle time" is 250ms, which is probably more than it took the camera to grab all four images in that video above.

In addition, OpenPnP has *never* had any speed optimization done. There's just been no need for it. As we start to move onto machines that are much faster than what we've been using there will likely need to be some changes to keep up, but this should be pretty easy. It's not like we're running a bunch of highly optimized code that we will only be able to tweak here and there; there are likely huge gains to be made very easily all over the system.

So, bottom line: Try it, see where it needs improvement, ask questions, make improvements. There's no reason we can't go that fast, or faster.

Jason

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/a57f31a5-1e79-444f-9109-34c1e0b5f52b%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

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

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

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyQWHzNCauVq2bjErf8QD7AXTVQzWOfjszOPrb0Gpg_Ytw%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

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

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

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/001f01d273d4%24e3ef0880%24abcd1980%24%40caswa.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyTTVxjyNRtp9kXs9ia3y9hs-wnQpZCPV4qdrbBRf7QOoA%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--

You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/01010159c1ab4835-86852521-7330-4df4-a6bd-92881cc29473-000000%40us-west-2.amazonses.com.

[Mark Harris]

The original machine has dedicated vision hardware. There is no way an 80386 running dos is going to do vision at any speed while doing other stuff hehe. It also has didicated motion control hardware.

My machines are running ICOS hardware which has since been acquired by: https://www.kla-tencor.com/products.html

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

To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/a57f31a5-1e79-444f-9109-34c1e0b5f52b%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyQWHzNCauVq2bjErf8QD7AXTVQzWOfjszOPrb0Gpg_Ytw%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/001f01d273d4%24e3ef0880%24abcd1980%24%40caswa.com.

For more options, visit https://groups.google.com/d/optout.

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

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

To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyTTVxjyNRtp9kXs9ia3y9hs-wnQpZCPV4qdrbBRf7QOoA%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

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

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

To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/01010159c1ab4835-86852521-7330-4df4-a6bd-92881cc29473-000000%40us-west-2.amazonses.com.
For more options, visit https://groups.google.com/d/optout.

--

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CA%2BQw0jzmoRNApHpakTxZ%3Dm7fUs2hAXmYy9Mp9Djfhy398m4uig%40mail.gmail.com.

[Glen English]

on vision speed. 1st there is nothing slow about OpenCV.

The cameras are the first bottleneck. The commercial machines 1990-2003 at least for Yamaha  have high speed composite out video cameras (>60 fps), and dedicated vision hardware. My YVL88 (which is a bit newer than the YV(L)100) has a dedicated vision processor board of 6 x large ALtera FPGAs. Which is alot.

The YVL88 does not stop (and YV100 to my knowledge)  when it does the vision.

*************

In OpenPNP there is a capture delay, default 250mS

Now, also, the low cost USB cameras, in uncompressed mode , 2 mega pixel top out at about 5 frames per second, 

So, you need to stop and wait until at least the last frame (before you stopped) is int he trash, then take the next one.

Worst case for an image like that is 2 x frame period  that is about 400 mS for a capture.....

So ! 250mS default delay is probably reliably NOT ENOUGH !

I bought a Point Grey camera, global triggerable shutter, and strobed the exposure with shutter open, you could get  1-2mS for a vision..

Yes 1 to 2 miliiseconds. But that sort of timing has be be handled a bit closer to the hardware, it is not within a PC OS environment timing.

(That is a USB3 camera)

So, for the cheap ELP USB2 cameras, for faster acquisition, you want to keep the resolution low to get the frames-per-second up.

Here is the table of frames per second rates:

https://www.aliexpress.com/store/product/Black-And-White-2MP-1080P-1-2-7-CMOS-OV2710-usb-webcam-camera-Android-Linux-Windows/416848_32362682348.html?spm=2114.12010612.0.0.rVtZ8O

1920 x 1080 : YUY2 : 6 fps

1280 x 1024: YUY2 : 6fps

1280 x 720 :  9fps

1024 x 768 : : 9fps

800x 600 : 21 fps

so it is clear

- ALWAYS select linear YUY2 NOT any compressed video modes......

if you want speed you are going to have to spend the big bucks >USD300 on a real imager, plus lens.

some of the bigger 5MP EP webcam imagers do a little better on fps, but not much. USB2 is the bottleneck.

for a B&W imager, 1920 x 1080 8 bit samples that is 16.5 mega bits ..... 

ogle.com/d/optout.

[Mark Harris]

Have you had your camera working with openpnp yet Glen?

--

You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+unsubscribe@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/fd07b880-4afc-4a09-a02e-9de46d745451%40googlegroups.com.

[Graeme Bridge]

If the bottleneck is the USB interface what about the IP cameras? a lot more bandwidth available on an ethernet connection

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

To post to this group, send email to ope...@googlegroups.com.

[Mark Harris]

USB3.0 has more bandwidth than Ethernet :)

There's a really good writeup on the different types of sensor/shutter here: http://www.edmundoptics.com/resources/application-notes/imaging/understanding-camera-sensors-for-machine-vision-applications/

for those who are interested in reading more.

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/e852d898-4f2b-49e0-ad72-ecba6b6cfa74%40googlegroups.com.

[Graeme Bridge]

so something like this would be more up to the task

http://www.mouser.com/ds/2/233/Leopard_LI-USB30-V034_datasheet-493193.pdf

[Mark Harris]

I'd be looking at something more like this: http://www.edmundoptics.com/cameras/usb-cameras/point-grey-chameleon-usb-3-0-cameras/33163/ as a "cheap" global shutter USB3.0 cam. With 5hbit and a raw 8 bit stream it will take about 6ms to transfer the picture.

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/0fe18d91-c467-4030-88fe-1a55cf2e2a0d%40googlegroups.com.

[Glen English]

Hi Mark

The ELP USB cameras work fine in open PNP, but I found that occasionally it would vision wrong after the move, and I found that to be still having the previous frame in the buffers- so increasing the delay to >400mS  (2 x frames) fixed that (frame is about (1/6 seconds))

*******

do you want me to test my Point Grey USB3 Camera in Open PNP? 

Easy to buy direct from Point Grey. Their E-store is efficient.

glen

[Anthony Webb]

I would be interested in the point grey cameras, if you have one and could test that would be great glen. Curious how they compare to the elp and if they are better suited for downward or upward (or both)

Thanks.  

Sent from my iPhone

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/5b3613e1-9b41-4704-9e49-543a7a91bae3%40googlegroups.com.

[Andrew Frazer]

On Sunday, January 22, 2017 at 4:35:46 AM UTC+13, Matt Brocklehurst wrote:

How do the commercial machines (esp. the older ones) do vision so fast ? I've said before I've played a lot on a iPulse M3 which is a few years old now and it's blisteringly fast with vision too

Theres a large stack of boards that have a stack of old FPGA's and some other stuff on them.. I've not dived in to have a look.. But judging by the age of these machines, new FGPAS woudl be do the same level fo processing in teh space of a postage stamp..  My YV100's have 486's in them ( so much more modern than Marks 386 ...<chuckle> )  and ther5e is no way they could do the image processing in teh CPU.

Is OpenCv just really slow? I'm surprised on modern cpus compared to what will be in the iPulse

No, but its the old hardware vs Software issue reall is'nt it.

[Mark Harris]

The main reason I havent bought the cameras yet is not being sure about how they will work with OpenPNP. I'm adding an IO board to the machine anyway, so triggering the camera per pick is no big deal signal wise, especially if that gives a better image :)

I'd love to see how your camera works in OpenPNP (even if its just hand held). What model do you have?

On 21 January 2017 at 15:13, Anthony Webb <anthon...@gmail.com> wrote:

I would be interested in the point grey cameras, if you have one and could test that would be great glen. Curious how they compare to the elp and if they are better suited for downward or upward (or both)

Thanks.  

Sent from my iPhone

On Jan 21, 2017, at 3:01 PM, Glen English <gleneng...@gmail.com> wrote:

Hi Mark

The ELP USB cameras work fine in open PNP, but I found that occasionally it would vision wrong after the move, and I found that to be still having the previous frame in the buffers- so increasing the delay to >400mS  (2 x frames) fixed that (frame is about (1/6 seconds))

*******

do you want me to test my Point Grey USB3 Camera in Open PNP? 

Easy to buy direct from Point Grey. Their E-store is efficient.

glen

On Sunday, January 22, 2017 at 8:27:11 AM UTC+11, Mark Harris wrote:

Have you had your camera working with openpnp yet Glen?

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

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

To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/5b3613e1-9b41-4704-9e49-543a7a91bae3%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CD7AA5AF-FD6B-44FB-9E0C-A170CA649BEB%40gmail.com.

[Mark Harris]

My main concern with the industrial vision cams is the max cable length of 5M.. I suspect I need a longer cable than that.

[Andrew Frazer]

>>The YVL88 does not stop (and YV100 to my knowledge)  when it does the vision.

My intial thought was that it did' stop, but in fact it does'nt..    Its vision is done on the fly, at least for the upwards looking multivision camera.   As yet i've not used the single vision camera and the downfacing camera does seem to stop, but that maybe because its running to a position to find a fiducal and it is just giving the illusion of a stop.

 

[Andrew Frazer]

Maybe we should stop hijacking yoru thread about your conversion.. But this vision conversation is really interesting.

[Mark Harris]

Doesnt seem to be a hijack to me at all... its all relevant to stuff i'm looking at right now for the machine...

However, I suspect a lot of people don't read this thread because I've been posting a lot of reverse engineering notes for myself, so they've tuned it out. It might be worth starting another thread for industrial vision cams when Glen tries out his camera :)

--

You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+unsubscribe@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/ccc291d3-abff-4963-b6ac-acf45a63be72%40googlegroups.com.

[Glen English]

Thept grey casmeras are not just high end USB cameras

They are sophisticated pro cameras .. they are quite good value for money. Their $300 camera I use makes the $50 ELP cameras look like a piece of junk

They are the camera you put on your spacecraft for the mission to mars.

Everything from one shot triggerable shutter, configurable on board frame buffers, hardware cropping bit depth reduction you nam eit, it has it

Once you get FlyCapture-2 their control app up, (install the whole SDK) you see what you really have.

So, in the high speed application , my tests were to use the external GPIO interface pins on the camera... shutter. With my Rigol AWG,  I could open the shutter (pin high), then I would strobe my LEDs (700uS) , and then close the shutter (pin low) . you can also config those GPIO for things like sync outputs, active line / pixel outputs etc

I think USB3 is the way to go. But Gig ethernet is a very close 2nd.....There are lots of white papers on their site on the  subject. 

I have a Chameleon3 CM3-U3-13S2M

a monochrome, 1288 x 960 pixel device. 1/3", CS lens. I like 12 or 16mm lens. Mostly I like 12mm for this job

However to use with 'standard capture apps' you need USB3VISION firmware installed, and this is only supported in the "Grasshopper" Pt Grey cameras

not the chameleon. For the chameleon like I have I need to interface to their capture driver that you are suggested to statically link to (LIB), or use the DLL.

There are numerous examples for numerous OS and languages (c#, C, c++, .net fw) . If you download the Flycapture2 SDK or Spinnaker SDK, all the whitepapers and appnotes come down with it.

But USB3Vision with Grasshopper might be easier, or use the GigEthernet camera

[Mark Harris]

Reading through the ptgrey website, it seems as though everything they have made since 2013 confirms to USB3 Vision standard. I guess it's more a matter of how to get USB3 Vision supported in OpenPNP, as I'm not expecting them to show up as a USB webcam in windows.

Looking through the pt grey website they say you can use Active USB cables (monoprice have them pretty cheap - i use a couple at home)  for upto 20m which seems more than enough for me. I guess i'd need 6-10M of cable to get from the head to a PC. The up cam would be fine on a 5m cable.

--

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/5014b160-005a-49f2-ab5d-065241302df9%40googlegroups.com.

[Glen English]

And IE OpenPNP couldn't use the camera with the standard firmware loaded (that talks to their driver)

Grasshopper3 Cameras are $$$ compared to my Chameleon

So I will ask support what can be done for a USB3vision firmware

Another thing about Pt Grey, they have driver binaries for ARM..... the Pt Grey worked on my ODROID !

Looking at the SDKs, it looks super simple to load the PtGrey DLL and get an image, set params, trigger etc in C or a C derivative language.

Maybe I will get one and send to Jason. But I think Jason said he's pretty well done now. The use / intended use of pro machines is stretching some of the goals I suspect of OpenPNP .

Probably easiest would be for OpenPNP to interface to the C / C++ Pt Grey DLL

Of course, you could reduce your camera resolution to say 800x600, and run a slightly different lens and auto variable focus etc (like the big boys) and have high frame rate count (30 fps ) using a ELP camera. That sort of res is still more than most of the 10+ YO machines effectively  have......

That would be my approach from scratch for a low end machine...

[Mark Harris]

I think the main thing with beating the old analogue cameras isnt resolution, but:

a) global shutter, so you can vision on the fly and/or very quickly when stationary (no settle time).

b) time get get a frame out for vision (a shutter triggered from openpnp would be fantastic)

c) image quality - the old cameras despite having low res had very clear images with little noise. The ELPs despite higher resolution have a lot of noise UNLESS given a huge amount of light.

See my Octoring Up Cam light thread for A/B comparisons of ambient light (single 60W globe on a roof approx 2m away) vs a bunch of high CRI leds:

https://groups.google.com/forum/#!msg/openpnp/QbqwNTamX_Y/VzM-u9d1EgAJ

Ambient is very blotchy as the gain (ISO) goes up on the camera, the "throw tons of light at it" picture is much better with  the camera on the lowest ISO.

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/5fedde1f-e61e-4784-8d45-84fc491ae7ae%40googlegroups.com.

[Glen English]

Agreed.

I had a read of your Octorlight . Neat !

Some of the guys I know are experimenting with the PWM drive serial loaded LEDS so you could light up as many as you wanted, or , for example, light up one side...

like APB102s etc

https://www.adafruit.com/product/2686

one self clocking clock+data wire between all of them..

-glen

On Sunday, January 22, 2017 at 11:21:59 AM UTC+11, Mark Harris wrote:

I think the main thing with beating the old analogue cameras isnt resolution, but:

a) global shutter, so you can vision on the fly and/or very quickly when stationary (no settle time).

b) time get get a frame out for vision (a shutter triggered from openpnp would be fantastic)

c) image quality - the old cameras despite having low res had very clear images with little noise. The ELPs despite higher resolution have a lot of noise UNLESS given a huge amount of light.

See my Octoring Up Cam light thread for A/B comparisons of ambient light (single 60W globe on a roof approx 2m away) vs a bunch of high CRI leds:

https://groups.google.com/forum/#!msg/openpnp/QbqwNTamX_Y/VzM-u9d1EgAJ

Ambient is very blotchy as the gain (ISO) goes up on the camera, the "throw tons of light at it" picture is much better with  the camera on the lowest ISO.

On 21 January 2017 at 17:14, Glen English <gleneng...@gmail.com> wrote:

And IE OpenPNP couldn't use the camera with the standard firmware loaded (that talks to their driver)

Grasshopper3 Cameras are $$$ compared to my Chameleon

So I will ask support what can be done for a USB3vision firmware

Another thing about Pt Grey, they have driver binaries for ARM..... the Pt Grey worked on my ODROID !

Looking at the SDKs, it looks super simple to load the PtGrey DLL and get an image, set params, trigger etc in C or a C derivative language.

Maybe I will get one and send to Jason. But I think Jason said he's pretty well done now. The use / intended use of pro machines is stretching some of the goals I suspect of OpenPNP .

Probably easiest would be for OpenPNP to interface to the C / C++ Pt Grey DLL

Of course, you could reduce your camera resolution to say 800x600, and run a slightly different lens and auto variable focus etc (like the big boys) and have high frame rate count (30 fps ) using a ELP camera. That sort of res is still more than most of the 10+ YO machines effectively  have......

That would be my approach from scratch for a low end machine...

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

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

To post to this group, send email to ope...@googlegroups.com.

[Mark Harris]

The PWM speed is too slow for reliable vision applications IMO.

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/9c04ec59-a1ad-4b21-9a52-c41e7054890e%40googlegroups.com.

[Jason von Nieda]

NeoPixels eat camera images for lunch. Avoid them. At least on all the webcams I've tested the PWM frequency is so low that it causes banding in the images.

If you want to PWM your LED lighting you need to use a high frequency controller like those used in LCD backlights.

Jason

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/9c04ec59-a1ad-4b21-9a52-c41e7054890e%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyS5_RFndO7oPChwj%3DeM2iLcxNFjKBPnnJiUQ44%2BOs%2BxQQ%40mail.gmail.com.

[Glen English]

Well, for 1 MHz clock the APA102s PWM rate is about 4000 Hz (0.25mS). so for 1mS exposure, that would be four cycles

if the shutter is open for 1mS, I would have thought that would have been OK.

I think it would be possible to run n addressable  single shot cycle 

Jason, I guess you are talking about progressive shutter webcams, not global shutter machines.

-glen

[Jason von Nieda]

You're right, maybe not an issue on a global shutter.

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/ab840957-3b80-4150-9363-1a74ac2f8167%40googlegroups.com.

[Glen English]

It's enough of a potential problem that you would need to think about it.. 

"neglegas, ad periculum tuum"

Some of those LEDS only run their clock at 300-400 kHz, some will run up to 2 MHz internally

In general, for the 256 steps, PWM period will be clock / nsteps.....

[Andrew Frazer]

With a 700uS flash,  at a movement of about 4m/s ,  that would be 2,8mm, which in teh context of a 0603 passive is a large bit of motion.,.

I'm asusmign your still trying to image when the component is still.

1/700 flash seems really long when i think about it..  

[Andrew Frazer]

The camera goes down to an exposure much faster than that.. what am i missing...   ( abouty 1/20,000th )

That would give you a blur of around .2mm  about 14x less?

You might need to get the light level up.. and maybe still bring the speed down.

[Glen English]

No problem with de blurring the image is velocity is constant.

if velocity is not constant that is *much* more difficult.

[Glen English]

I've been thinking a bit more about the use of the $$$$ cameras

for most Open PNP people  they probably dont care about every last millisecond, unlike say an Essemtec Cobra.

Large view are required to image large component with high res (like 40mm FOV, with 50uM pixels) still that is only 40/0.05 = 800 pixels

plus maybe 25% lost at the edges/ unusable - 1066 pixels does it.

so a 1920x1080 camera does it, but at 5 fps ish (200mS per frame, x2 for tr timing uncertainty). really 400mS isnt too bad for imaging an uncommon component.

but for faster imaging of smaller parts using the same camera at 800x600 linear  gives you 30 fps . 33mS per snap, 66mS after ambiguities.

and 66 is alot faster than 400. you are going to have to slow the gantry down and stop and that will probably take that long.

so , 800x600, with say 10mm FOV, 30uM pixels,  (IE at least 16 pixels for the short end of an 0402 ) 10/0.03 = 300, add 25% = 400 pixels

which delivers using a 800x600 grab. 

BUT, same camera would need a different lens, or different distances etc

So I think for the most part, it is more economical, and really just as good to use TWO cameras for UP

One with lens suited for 40-50 mm FOV- for imaging large components, doesnt matter if it takes time.

the other suited for 10mm FOV  running much smaller pixel count for speed. for 1/4" sensor ,  16mm is highest cheap common unless you buy one of those var-focal ones that are like 6-20mm adjustable,

- those vari-focal zoom ones also have adjustable iris which is useful to stop it down a bit which gives you a little more depth of field- or at least optimise it.

for a 1/3" sensor which I am running really wants a 20mm lens (or just use less of the sensor with a 16)

g

[Mark Harris]

Today is a good day for parts!

First off, we started the day with mouser. Microcontrollers and dev boards (mostly just to use as a JTAG).. somehow we ended up ordering double of each dev kit. No matter, devkits are always handy to have!

Then DHL came, and I had to walk allll the way over campus to central distribution to get the boards. Since I'm too impatient to wait for internal delivery tomorrow hehe.

Boards look good, even if my photos do not.

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/35be866d-0607-4388-bbc1-b68caea1c96b%40googlegroups.com.

[Mark Harris]

Gino told me I should assemble these tonight rather than wait till later in the week. Sooo...

Ready to assemble! Got my drawing, board and stencil.

Stencil setup isn't fancy, but it works great for one offs.

Ready to Party

[Mark Harris]

Gino told me I had to assemble the boards tonight rather than wait until later in the week.

So I have my stencil, board and assembly drawing all set to go.

Stencil setup isn't fancy, but it does the job.

Ready to Party

All the parts on the board set for reflow:

[Mark Harris]

Through hole parts placed!

New vs Old

[Jason von Nieda]

Looking good Mark! Intending to fire them up tonight?

Jason

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/fa50a06e-515f-4878-b9b3-b4724e6989fe%40googlegroups.com.

[Mark Harris]

No, I'm already home. Gino said he wanted to make some firmware changes before I do anything to them, so I'm going to wait for those :)

I also want to have someone else with me who can turn the power switch off if I need to. 1.2kW of motor is a scary prospect when the power switch is on the opposite corner of the machine than I'll be on.

I also need to trace the quadrature output back through the ribbon cables so I can get access to it near the machine. Still a few hours worth of work to go before I can fire it up and see what lets the magic smoke out :)

[Mark Harris]

https://www.youtube.com/watch?v=xos2MnVxe-c

I thought I had messed up the switchmode powersupply that generates the 12v, as nothing was coming online. In a way I did, the undervoltage protection was set to cut it off at about 40VDC... my bench powersupply only goes to 30v haha. Removing the bottom leg of the resistive divider however, everything lights up. Still no firmware programmed, thats all I had time for tonight - but 5V and 12V rails are online.

For testing, I've decided to go a different route, I have a CIM 12v robotics motor (First Robotics competition sort of thing) from another project, I bought a CUI encoder which I can drive with the motor, so I'll be able to test it out on the bench with no load. If all goes well with Gino's code, then we'll try it on the the biiiiig motors.

[Jason von Nieda]

Pretty exciting Mark! Bench testing sounds like a very, very good idea :)

Jason

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyQwfuzLwKQ-b5dcMpFghMRo0Yj5SBjX9CG0bMNbBT-%3DgQ%40mail.gmail.com.

[Mark Harris]

Feeder controller boards should be in tomorrow. Despite tracking saying end of day today, its after end of day today and still hasn't made it to north america haha. Ordered the parts from them on digikey today, so those will get in tomorrow too. Not sure I'll have time to assemble them tomorrow, but its starting to come together.

On 24 January 2017 at 17:15, Jason von Nieda <ja...@vonnieda.org> wrote:

Pretty exciting Mark! Bench testing sounds like a very, very good idea :)

Jason

On Tue, Jan 24, 2017 at 6:09 PM Mark Harris <ma...@rris.com.au> wrote:

https://www.youtube.com/watch?v=xos2MnVxe-c

I thought I had messed up the switchmode powersupply that generates the 12v, as nothing was coming online. In a way I did, the undervoltage protection was set to cut it off at about 40VDC... my bench powersupply only goes to 30v haha. Removing the bottom leg of the resistive divider however, everything lights up. Still no firmware programmed, thats all I had time for tonight - but 5V and 12V rails are online.

For testing, I've decided to go a different route, I have a CIM 12v robotics motor (First Robotics competition sort of thing) from another project, I bought a CUI encoder which I can drive with the motor, so I'll be able to test it out on the bench with no load. If all goes well with Gino's code, then we'll try it on the the biiiiig motors.

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CAJMBTyQwfuzLwKQ-b5dcMpFghMRo0Yj5SBjX9CG0bMNbBT-%3DgQ%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--

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

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

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CA%2BQw0jxRoU1N_Aoj%3D%3D7AERr91cxAWkpY1-nwPBgbq82oPsxBMA%40mail.gmail.com.

[Mark Harris]

So, I was going to go home, but then I got distracted by the PNP...

I found something interesting on the waffle pack feeder:

The motor is *almost* the same part number as the short axis on the gantry, but what most intrigues me is the metal spacer and then extra bit after the tacho... it appears to be a quadrature encoder. I've mentioned before that the motors have a specific quadrature encoder that bolts on... this is not it. They've used a spacer and then bolted some other quadrature encoder on instead. Its a 1000 line, so it could be interesting to see if I can mount that on the gantry motor instead. I have two wafflepack feeders and will only require one... so this one could be a nice parts bin.

So, I found out what was in the box with the two grills on either end that sits under the conveyor. It was in my way, so i pulled it apart.

Its a little mini rack! This board gets full AC voltage given to it, the board closest to the camera is an AC conversion board, behind it is a board with I think an 8051 controller - it looks an awful lot like the one in the feeder controllers, but has a sticker on it that I haven't pulled off yet, so haven't quite looked at it yet. I'm guessing this is the controller for the conveyor. The board all by itself there is the motor driver board for the conveyor, it has a bunch of h bridges (made up of discrete fets) on it. The ribbon cable all goes to a board that is under the Z plate. Getting that Z plate out is going to be MASSIVE pain, so I just ripped all the wires off it to get to my main object for the night.

Note the huge pin array on the right - thats where ALL the gantry assembly wires terminate. The sonars, solenoid ram thing, motors, limit switches. Those all seem to go off to the controller.

The box in the foreground with the D connector is the up camera. I yoinked the top off (lighting module) to get access. That part is meant to be pulled off, and must be removed for mounting the wafflepack. Interestingly, it has lights way down next to the camera lens, and then diffused lights right up the top too. 

[Mark Harris]

The main event for tonight however was pulling the conveyor itself out. First, starting with the end most bit that can be separated from the rest... huge huge pain to get this thing out.

Then the rest came out. I will be putting it back in later, but I need feeders under control first, and I need to make a board to take all the sensor inputs for the conveyor, plus figure out the width adjustment motor (could be a stepper motor, looks a bit like an old one rather than a DC servo). For now, the plan is to get the machine working - so I'm probably just going to throw some MDF in the router and make a platform the same height as the feeders, and use it for cut strip and the boards themselves. It's a travesty I know, MDF bed and dual head nozzle in a chip shooter!

The machine looks very strange without it in there.

[Mark Harris]

The sonar units are all Siemens SONAR BERO 3RG63 42-3AB01 units. These are still manufactured and available today... exactly the same wire colour on the cable as an inductive prox sensor:

Brown = 18-35v

Blue = Ground

Black = Output

It consumes about 20mA of power running at 24v, output with no pull down resistor is 9.3v or so with nothing in the sensors range, and 23.4v when in range. Super simple.

The conveyor motors run pretty nicely on 30v, pulling around 300mA, 400 if I load it up a bit. Simple geared brushed DC motors, no encoders. They are all Faulhaber Schonaich 2862W028C.. .made in Germany of course. They only seem to appear online at surplus places, cant find any other references to them.