Fritsch placeAll retrofit

[Zach]

Hi all,

I just picked up a 2000-era Fritsch placeAll from an auction for $600.  Fritsch is a German company, but the machine design seems to be the same as Autotronik in Europe or Manncorp in the US.  For the price, it seemed like I couldn't go wrong.

It's roughly 1000lbs, came with 64 feeders, three cameras, auto head changer, a pneumatic feeder tray, nice TDK rails, beefy servo motors (with encoders) and pneumatics throughout, all being run by the guts of a 486 PC with a 3.5" floppy.  The machine seems fully functional. Fast and smooth. I can drive it around using the DOS (!?!) software, I just can't get any new files on there because I got rid of my last computer with a 3.5" floppy drive about 15 years ago.  Seems like a perfect candidate for an OpenPNP retrofit. 

Here's my plan and I would love any feedback/advise:

Get a modern windows PC (probably a Newegg referb running Windows 10)
Get a Cohesion3D ReMix (since it doesn't have built in stepper drivers) for the gcode-to-step+direction conversion, reading the end stops, reading the vacuum sensor and triggering the 6 pneumatic valves
Get four G320X gecko drives (way overkill for the pick and rotate, but I don't know of a cheaper/easier alternative) and use the step+direction signals from the ReMix to drive them.
Use an Arduino Mega to drive the existing feeder logic boards (they use DM74LS154 chips to fan out to all the feeders).  Each feeder looks to just get a pulse to advance.
I'll either utilize the analog cameras that are already onboard with a cheap USB capture device or bolt on a USB microscope (though I'd rather not have to try and feed the USB cable back through the already beautifully bundled cable harness) and

Seems pretty straight forward and mainly just figuring out the existing wiring pinouts and configuring OpenPNP.  Am I missing anything?


 
 

[Daniel Dumitru]

buildcoolstuff.com  -  I like domain name !

REgarding machine, if it's working don't destroy something that's extremely complicated !!!

Virus-free. www.avast.com

--
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/36c8b204-54ed-4014-9c18-ad6b7c35b3c4%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Mike Harrison]

>> It's roughly 1000lbs, came with 64 feeders, three cameras, auto head
>> changer, a pneumatic feeder tray, nice TDK rails, beefy servo motors (with
>> encoders) and pneumatics throughout, all being run by the guts of a 486 PC
>> with a 3.5" floppy. The machine seems fully functional. Fast and smooth. I
>> can drive it around using the DOS (!?!) software, I just can't get any new
>> files on there because I got rid of my last computer with a 3.5" floppy
>> drive about 15 years ago. Seems like a perfect candidate for an OpenPNP
>> retrofit.

You know you can get USB floppy drives, right..?

[mojalovaa1]

Can you send some link that I can see where you are buy it , I need some reflowe oven like yours on buck side this image ?

[Zach]

Hahaha... Yes. I know USB floppy drive exist, I even looked at them for a few minutes. However, after thinking about the reasons I want a pick-and-place, I realized the for me and my business, this machine only makes sense to run if I can tune/customize the workflow (which openpnp allows and the DOS app doesn't).

I'm really excited to integrate bar-code scanning (Mouser is really good about bar coding the data on their part bags) and streamlining the process of feeder setup. I also would like to use strip feeders and I couldn't find a way to get the old code to do it. 

I'm not religiously opposed to using the original software, I'm just thinking that with a few extra bucks and a little work, I'd have a modern PnP that I can totally customize.

[Zach]

The photo is from the place holding the auction.  I wanted that oven as well, but I didn't get it.

[mojalovaa1]

Can you send me  private massage  with link to that auction ?

[Marek T.]

@Zach,

I want to warn you that full retrofitting to Openpnp will be not fast and easy way. I have survived it doing the same with some Philips.
Effect that have is realy good but it was not so quick and easy job as I've expected.

The biggest problem that you will have - if you want get the fastness similiar to original: drives for motors and it's proper tuning. All the rest is piece of cake.

What kind of servo motors are there, DC or AC, brushed or brushless? I don't know too much about brushless and AC.
If there are DC - now I know something about this... The most important thing is to use power supply x2 than nominal voltage of motors - if you want get proper fastness and dynamics.

Also think twice about removing original cameras and using some cheap Chinese replacement. It's highly likely that camera which is there is not so bad as you expect and much better than some cheap USB that you will buy.

br
Marek

[Zach]

Thanks Marek!

All the motors are Maxon brushed DC (they spent some $$ on motors).
Giving the drivers proper voltage headroom is great point. I probably have the parts to build up a 48volt linear supply (toroid, rectifier and caps) just to power the motor drivers.
Also great thoughts on keeping the original cameras.  For the few bucks, I'll try out one of the USB capture devices. If they work, great. If not, I'm out the $20 for the USB adapter. 

I'm pretty comfortable tuning motor control systems. :)

[John Kasunich]

On Fri, Feb 9, 2018, at 4:59 PM, Zach wrote:
> Thanks Marek!
>
> All the motors are Maxon brushed DC (they spent some $$ on motors).
> Giving the drivers proper voltage headroom is great point. I probably have
> the parts to build up a 48volt linear supply (toroid, rectifier and caps)
> just to power the motor drivers.

Can't you use the existing power supply?

And maybe the existing motor drives?


--
John Kasunich
jmkas...@fastmail.fm

[Michael Anton]

Likely the machine has support for tray feeders.  You can treat strips of tape the same as a tray.

[Zach]

The existing power supply is only 24volts, which is a little weird considering the nameplate on the X axis motor says it's a 36volt.  I'll probably test everything with the existing supply, but figure I can upgrade that later if needed.  I'm actually not as worried about pushing this machine to it's maximum CPH.

As for the motor controllers, it's using an ISA card for control and external dual half bridge modules (I haven't see those in a long time).  While the Gecko drives aren't cheap, they are typically pretty bullet proof and easy to replace if one dies.

[Zach]

Well, I just couldn't wait... so I pulled the trigger on:

1. DELL Desktop Computer 790 Intel Core i3 2nd Gen 2100 (3.10 GHz) 4 GB 250 GB HDD Windows 10 Pro 64-Bit ($141)
2. 4 Gecko drives ($529)
3. AVerMedia EZMaker 7 video capture adapter, I'll buy a second if the first works ($39)
4. Cohesion3D ReMix ($159)

For a grand total of $868.  If I add what I paid for machine, I'm still way south of $2k for what should be a pretty cool machine. 

Now to start ringing out these cables...

[Anthony Webb]

Should be a fun build to follow.  Keep us posted on how you get along!

--

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/2d378337-a905-4569-852c-b8d3870d1024%40googlegroups.com.

[Marek T.]

So,

1. My motors are 75VDC and I have used drives 80VDC - advised by drives manufacturers that it's best solution. Then surprised why they work  twice worse (slower and anemic) than in original. After that started think why in my machine originaly they were supplied from 160VDC, even manufacturers who designed it for Philips/Yamaha (Sanyo Denki) was not sure of it. So I have followed original solution, changedt drives to powered from 160VDC and it started work as should do, the trick is in fast motors inductance energy reload much more effective if made with higher voltage (of course later proper PWM and current limiting must care to don't burn the motors).
@John Kasunich advise to follow original supplier is good advice.

2. My original drives were fully integrated with motion controller. There was no solution like some motion controller using step/dir to control drives. So not possible to use my original drives. Hard to say how it is at you r machine, but really worth to check it.

3. I don't know Gecko. But what can say, in my new drives there is some 6-7 parameters to tune. And it is not easy to get good effects in every cases:
- short moves with no loosing encoder pulses
- long moves with god dynamics and speed
I may risc to say that with that simple drives it is not possible to get fully good results. Interresting is that in SanyoDenki controller destined for the motors as I have there is some 30 not6 parameters to tune (90% of them can't imagine what are they what for...), this drive costs $800/1motor not $150.
Rather don't expect that you will get so amazing motor control results as in original construction.
Also note that in original construction the motion controller is integrated with drive/encoder and motion controller exactly knows about real position of the axix. In the construction controller+drive, controller send some order to move to the drive, but later the controller doesn't get feedback from the drive what the drive executed with axis. Compromises, compromises...

4. Cheap USB cameras have absolutely shity optics and speed (settle time?). If you will have get perfect results you spend many hours on tuning it. Probably in original camera many that problems are done. Before you will change camera into "better" try use what you have - old doesn't mean not good.

5. Don't know how many C motors you have there. Openpnp expects that each your head has its own C-motor, not common (shared) one for every heads like in many old machines. So you will need modified Openpnp not standard version. Be prepared to digging in java code and compiling personalised soft versions. Cri's is your friend in it ;).

All this is absolutely not to discourage you to play with Openpnp. Only to let you know the pros and cons, and let you know what waits for you.
Good luck and satisfaction! :-)

of the

[Marek T.]

1. Suggest you i5 and 8GB, at least. Java eats realy many resources.
Hard to compare AMD to Intel directly, but my Athlon A8 and 8GB is not too much.
4. I have used Smoothieboard. The limit that I got is the STEP max frequency. To get max motors speed at encoders resolution 4x I would need 200kHz STEPs. Smoothie gives max 100kHz, so I had to resign of 4x and use 2x, compromise speed and resolution. You should calculate what min STEP frequency you need at 4x encoders (best positioning accuracy) mode to get max of your motors rotation speed.
Maybe Cohesion is good with this with it's firmware writing, Smoothie has similiar processor but has STEP limit 100kHz - I'm not sure if it's only because of it's firmware or processor hardware. Check it and check what you really need.

br
Marek

[Zach]

Hey Marek,

Thank you for all the input!  For the record, I don't take any of your comments as discouragement. I love talking through these sorts of projects. :)

I totally agree with you regarding running the motors drivers at a higher voltage. That's why I think it's a little weird that the original design supplied a LOWER voltage to the motors.  The motors are rated to 36V, but the original power supply is 24V.

I've used Geckodrives for quite a few projects over the years and have been pretty happy with them. 

The glass scales have a max output frequency of 50Khz, so that's hopefully the bottleneck and not the Cohesion board. 

I'd like to have as much speed as possible, but I'm totally ok with having slightly lower CPH numbers.  For my business (and I know this is different for everyone), it's less about high throughput and more about having the ability to do quick turn board assemblies.  For years I've relied on hand assembly and/or Screaming Circuits to stuff my standard quantities of between 2 to 50 PCBs.  Being able to do them in-house is highly valuable to me and my customers.  Plus it's just cool to have a pick and place machine to show people when they visit the shop.

I could see the OpenCV chewing up some CPU resources, but unless I'm way off the mark, I'm pretty sure a dedicated 3Ghz i3 should run a Java app generating point to point gcode, which it sends at 115,200bps.  The MCU on the Cohesion board running at 120Mhz seem to actually doing most of the heavy lifting.

I'm usually a bare metal C kind of guy, but I can swing a Java hammer when needed.  Looking forward to digging into the source.

[Bernd Walter]

On Saturday, February 10, 2018 at 2:11:12 AM UTC+1, Zach wrote:

Hey Marek,

Thank you for all the input!  For the record, I don't take any of your comments as discouragement. I love talking through these sorts of projects. :)

I totally agree with you regarding running the motors drivers at a higher voltage. That's why I think it's a little weird that the original design supplied a LOWER voltage to the motors.  The motors are rated to 36V, but the original power supply is 24V.

Seems to be a common thing.
I can only speak about the Siplace feeders, but they use 18V Bühler and Maxon motors and run them with 30V.
In the Schulz type feeders, which are MCU based they run the motors only for a short time and in case of problems they go in failure mode.
But with the older non MCU based feeders they apply power until they reach position, so they can mechanical block under 30V.
The wider non MCU based feeders have a separate motor for the cover tape, which is purely driven by a tension switch.
If the cover tape breaks the motor will run forever at 30V without load.
I can imagine that the machine itself has an overcurrent protection, which kicks in if a feeder blocks, but I don't think they protect
against a free running cover tape motor.
In my control board I've added current measurement with an INA226.

[Marek T.]

Bernd, in feeders it's not important to get max performance of the motors. So not important if you get their 50, 80 or 100%.
If motors are 36V I'd use drives powered 75V and PWM 0-50%, instead of drives 0-90% powered 40V.
If they used 24V for 36V motors, maybe all mechanical construction doesn't allow move axis too fast?? For sure these motors are used out max 50% of their power.

Zach, i5 instead of i3. The problem is not to run the controllers. The problem is to serve the cameras. The same cameras "connected" to Windows will use 5-10% of CPU, but when served by Java - CPU usage will grow up to 60-70% !
People run Openpnp on different toy computer. But my opinion is that computer for this solution should be really solid and strong.

[Marek T.]

120MHz of processor is not warranty of high STEP output frequency. Ask them what they provide for steps. If it's 200kHz, I go to buy it at once :-).

[Zach]

Still dialing things in...  Getting a few miss-picks, some undetected, which means I need to tweak the vac sensor settings. My cut tape holders need some adjustments.  They hold paper tape perfectly, but plastic tape is a little loose and occasionally parts bounce (thinking of putting some fuzzy felt or similar in the channel under the plastic tape to support it).    I'm also getting some parts moving during the trip from the bottom camera to my board. Thinking I may be using the wrong nozzles for certain parts.  However, it functions well enough to stuff a hand full of boards.

For those who are interested, I just shot a quick video of my setup.
https://youtu.be/dWgPDSP78Fw

[Jason von Nieda]

Nice work Zach, looking great!

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/0eecd788-0d14-48e5-975f-e5e97a0259cf%40googlegroups.com.

[Marek T.]

I have modified the code of vacuum checking. Now have vacuum tested (read) in the loop in every 20ms and when the vacuum grown to required level I at once go further with the code. So in fact, I don't need any delay before the vacuum checking.
Plus, it's not good done in Openpnp that Vacuum is tested after PICK_COMMAND but before Z up.
Z DOWN
PICK_COMMAND
VACUUM
Z UP
Problem is that the vacuum can be got proper but while the nozzle goes up the part may be lost. Then in my oppinion the Vacuum should be tested twice while the picking, before and after Z-Up.

W dniu niedziela, 11 marca 2018 17:00:42 UTC+1 użytkownik Jason von Nieda napisał:

On Sun, Mar 11, 2018 at 10:48 AM Zach <za...@buildcoolstuff.com> wrote:

Getting a few miss-picks, some undetected, which means I need to tweak the vac sensor settings.

For those who are interested, I just shot a quick video of my setup.
https://youtu.be/dWgPDSP78Fw

[Zach]

Hey Merek - I was thinking of making a similar change!

[Marek T.]

If you want I may send you tomorrow the code which I use.

[Zach]

Would love to see it.  Thanks!   Also let me know if you try out those GUI changes I made.

[Marek T.]

Sure, I'll try do both tomorrow.

[Marek T.]

Hi Zach,

Attaching my RefferenceNozzle, find modified VACUUM procedures for pick and another for place (129 and 182 lines starting).
Note in PLACE I'm checking vacuum on PUMP_ON after the Z is up (but I have pneumatic heads not motors and don't use Z-up build-in Openpnp bud my configured G-codes in PLACE_COMMAND). It's because on PUMP_OFF I can't see a reason that vacuum may still exist, so no sense to check it as for me. But if part stacks on the nozzle or just nozzle is dirty - with pump ON checking the vacuum I can rate it's ok or not (must be low vacuum, high vacuum means trouble to check). So you need modify It a bit for you, if problem - tell me. Made with help of Cri so you may think they are made rather properly ;).

My thoughts about your GUI:
- few really universal changes
- Discard, BOT-CAM, JobPanel changes, VAC, SaveSet - seems be universal and appreciated by everybody. You could easy make request to Jason's git to merge it (if he find them made in agree with java-art of course, I can't rate it)
- 4mm - for me personaly useless if I don't use that feeders type. I will rather make for myself 90 instead 4 - for convenient rotating.
- TIPs switches - as above till I don't have AutoChanger device.

VAC ON and VAC OFF fire VAC actuator. How you configured them for you machine, what exactly makes ON and OFF for you?

Generally, good job :-). After few modifications it will let me get machine much more friendly :-)

[Zach]

Thanks Merek!

[Marek T.]

Good luck :-).
Pls answer in free time:

[Zach]

Ah sorry... in my gcodedriver, under the VAC actuator's ACTUATE_BOOLEAN_COMMAND, I have:
M80{True:0}{False:1}

This just sends an FET on/off command to my Remix/Smoothie to turn my vacuum valve on/off.

[Marek T.]

Ok, I just was curious only if you have "connected" there only the vacuum turning or something more yet.

[Mike M.]

Anybody knows which type - part No.,.. connector is this on Siemens Siplace feeders?

picture enclosed

Mike

[Bernd Walter]

On Sat, Jan 11, 2020 at 04:03:37AM -0800, Mike M. wrote:
> Anybody knows which type - part No.,.. connector is this on Siemens Siplace
> feeders?

Have to dig up what type of connector it was, but be aware, those had
been listed at mouser for about 30 EUR each.
I decided to replace them.

> picture enclosed
> Mike
>

--
B.Walter <be...@bwct.de> http://www.bwct.de
Modbus/TCP Ethernet I/O Baugruppen, ARM basierte FreeBSD Rechner uvm.

[Mike M.]

Clear - I will replace as well all connectors
I only need 1 female part to test the feeder without taking it apart....
thanks
Mike

[Jed Smith]

I'd guess that its a LEMO brand connector from its appearance. If so, it'll be pricey but they're well made.

[Mike M.]

Bingo - Jed - Big thanks

https://www.ebay.de/itm/Compatible-LEMO-FGB-0B-307-EGB-0B-307-7-Pin-Connector-Male-Plug-Panel-Socket-60/123655333483?_trkparms=ispr%3D1&hash=item1cca6ea66b:m:maE7Crq1tNPICI2RahDVg1A&enc=AQAEAAACUBPxNw%2BVj6nta7CKEs3N0qXTPEpLEz7cdWYiXHTyFvKef62Xo6g8qcAITkjI%2F8eBUlIluEGjKPEI%2Fap5UWXarSJKG7ByF%2F%2BJC32Fdaguh3Z0skgLC0EpuBNu0vzVLRMP%2FktnlXleXiuXETv4STcsT%2BJBDQCJ2h9fwTbj9A7lfRDJD6tYpzS4e74b0o%2BAo2LxyBRUC2bJNvOSikzyz7Blg6Df0sIU0GMe%2FQTciDCCkGSk3guhJyLXqBdWrptvEAAvLcfuBGimTsSi%2B%2BxFERO7J79cftsDx9QPVI0xz8aeXMILLhxm9MlXUochNIKQ6%2Fh2rBcoeqnMDHAhjGkkmG8DMJRJYRyPUUoaitHvFB0rkXCzlRWV7eTVsKDgMci9VGnHZsnMz7WdGAmEvfjXgJVm8qP7BbogT9iLGQ4gXUMDN94cEeMaYUzI6SCTSt1JnCKWMd72XlUghk9PjLhqF3FAchhFTz1NngAyoZ%2FdzqBeUvGs1qKUDaFeYM64Em0ZgxYPIgmePEozKUk98JWKmQIqUwdMkspl256RN49K2PhM96W9aG%2FGwApr%2FVTCJmU8%2Fs34ekI7P6%2FS3wKMd5A0Ey6%2FP0zMvyPyw3Xz7NLMZPv8Uq7QG9l2hFWqyZmZBj0vPR7XxkNJ4FSMmcvplTFeG7v4AkoKP83uvKIS9uhcAKd7ZSXBcEOH5xs1YQzrX0Ihigjige8AUGVU12WLqnTaY3vlIAHk8BJdLDOujJy1sBQzRz098KSYFdFTPHSttUrUi4RX9lSnLHN779hYXYGv1kODKck%3D&checksum=1236553334839066dd1f65614766a33a410ae17d10cf

Let me know the adress to send a beer!

I will survive!

Cheers Mike

[Bernd Walter]

On Sat, Jan 11, 2020 at 09:13:14AM -0800, Mike M. wrote:
> Bingo - Jed - Big thanks

Well - it says so on the connector (at least on mine):
LEMO FGG.1B

Be aware: there are a few different shelf sizes.

> --
> 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 view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/5941f5c8-d5ef-44de-b805-a904e9240aac%40googlegroups.com.

[Mike M.]

Thanks for warning, it’s a 7 pin - nothing written on my old feeders ...
mike