OpenPnP TVM802 and TVM920 Support

[Jason von Nieda]

Paging Richard Sloan and Bertus Kruger...

I wanted to start a thread to give us a place to discuss getting OpenPnP working for the TVM series of machines. I thought that first Bertus could let us know how things are going with his API and perhaps Richard can see if he can get anyone else interested in helping with this effort to join the thread.

For my part, I am more than happy to help with the effort. If Bertus completes his API that will make things very easy. I'll be happy to use the API to write a driver for the machine. The real tricky part is testing. It would be helpful to have access to a machine while developing and testing.

So, first off, is there anyone with a hundred miles or so of Kansas City, MO who has a TVM machine that would be willing to let me come play with it?

Alternately, if anyone (or a group of someones) would be willing to donate a machine to me, I will commit to making OpenPnP work with it.

Anyway, just some thoughts to get this discussion going.

Jason

[Ray Kholodovsky]

Jason, is the idea here that we just need to get you $3-4k so that you can get one of these Chinese PnP's and hack it?
I want to think that it would be possible to raise at least a part of this amount via a donations campaign thru the people in this group - a bunch of $50 contributions perhaps?

Also since Robotdigg carries the same types of machines albeit under different model numbers, and have used the Openpnp name generously, perhaps it would be worth contacting them to see if they will sponsor a unit or at least offer a discount?

[Dave Hunt]

I will send a note off to Richard.

Dave

[Jason von Nieda]

Ray,

Yep, that's the idea. Specifically, I have heard a lot of interest in getting the TVM802A/B machines working. There seems to be a lot of folks over on EEVblog's forum buying these machines and wanting to use OpenPnP, so I thought I'd throw this out there. There has been talk of a campaign to pay someone to write the driver for it and I figure I'm probably the best qualified to do that, except I need a machine.

So, the idea is that if I can get my hands on the machine, I will write the driver for it :)

Richard Sloan pinged me last night about this, so I thought he might be interested in coordinating something.

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/e0dc9059-78a8-4dff-924d-95ff8e65de7b%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Cri S]

The gcode driver based on ar-system is working.
Actually the FSM gives a bit a problem inserting a vision stage independently of feeder implementation between preFeed and postFeed.
Testbed is a tvm802B with exchanged hollow shaft stepper motors and stepper driven rotary table to simulate feeder. We want be sure before cutting the base.
Actually we testing the possibility to monitoring the peel tape tension with microswitch and wire that in series with the solenoid microswitch. A lot of ideas about improve ring things. Further we want test performance differences between real grbl driver and this controller. Nozzle changer is actually not present as it must go where actually the camera is mounted in order to retrain the possibility to have sliding tray over fixed tray.
The other remaining possibility disallow that.
The actual problem is, openpnp is to slow at the moment under 2kcph where qhi sw do more as double the performance even doing really slow up looking vision. Another thing, we use the nozzle calibration offset. It have several bugs but performs better as not using it. Before this machine is working, that must be fixed.

[Jason von Nieda]

Hi Cri,

Are you saying you've written a custom driver, or are you using the built in GcodeDriver? If the GcodeDriver, I assume you must be using a serial to TCP gateway of some type? Can you share more information about your project and your plans for it?

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/88e526db-2796-4ac1-9c27-b0f70e3d6eed%40googlegroups.com.

[Richard Sloan]

Hello, I made it.... lets see how this pans out with people here and take it from there.... The price right now is $3.2K delivered (minus taxes) to USA, we should talk to Robotdigg yes, I think they are Chinese Company and they are hard to reason with about this type of thing.... tried this many many times on projects over the years! The reason chinese software always is bad, they do not listen.... :-)

On Monday, October 31, 2016 at 9:32:43 PM UTC-4, Jason von Nieda wrote:

[Cri S]

I use the modified grbl driver with socat.
Rs232 over TCP/IP. For windows com2net and other exists. Eventually its possible to merge the linuxcnc driver for direct network sending.
C program listen on network for gcode,
parse that and send out command sequences over network to tvm802. The decoded protocol from ar-system with some api errors removed.
Solenoid errors or vacuum error are send as error to grbl driver in the case vacuum error is enabled.
As driver have softlimits the limit error is not existent.

[Richard Sloan]

Not sure if possible, but would be cool to use a GRBL like board with USB and ETHERNET (if it exists), connect USB to computer, and ethernet to 802, then run OpenPNP on the USB and use the controller as a bridge, this would also allow the controller board to also be used for other items like tool change etc.... or even upgraded vacuum detectors?? just a thought.... 

[Jason von Nieda]

Is there any reason not to just use the built in controller? Adding tcp support to the gcodedriver in openpnp is trivial. No need for a gateway.

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/f496e0c1-07f2-4218-8b48-8624372141c4%40googlegroups.com.

[Cri S]

The gateway is just a program, no real hardware.
The code started from here:
https://github.com/TVM802/TMV802-Protocol
Not really from here I have received the code as mail attachment.
And the timing requirement with the many threads inside java VM in addition to the fact that the code is already in C have ported to this solution.
I know, I can just copy the code from linuxcnc driver and don't need the rs232 to network program.
Having rs232 allow use gcode programs to test several things, like camera delay as example.

[Bertus Kruger]

HI All.

Looks like we are making progress. I am almost done with the API. Getting the machine state is complete and on/off functionality like pump/solinoids is also ready.

All that is left is actual movement. I discovered that the machine controller can deal with multiple connections. So the original software can run while my API is connected and listening. It also helps with movement testing as the original software will deal with the endstop while its not fully implemented in the API.

The API now also has a test page to test functions.

Documentation is on the lite side and care must be taken when testing.
https://github.com/TVM802/TVM802-WEB-API

[rayk...@gmail.com]

Side note - these machines do not come with any cameras, correct?

Is there a plan to physically implement vision onto the machine as well?

Ray

> --
> You received this message because you are subscribed to a topic in the Google Groups "OpenPnP" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/topic/openpnp/735aZn-GFRw/unsubscribe.
> To unsubscribe from this group and all its topics, 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/25989350-d601-4076-a531-460fba470cb9%40googlegroups.com.

[Kruger Creations]

They have 2 usb webcame like cameras. :)

The 920 we still need specs....

On 2/11/2016 6:50 am, <rayk...@gmail.com> wrote:

Side note - these machines do not come with any cameras, correct?

Is there a plan to physically implement vision onto the machine as well?

Ray

> On Nov 1, 2016, at 1:43 PM, Bertus Kruger <krugerc...@gmail.com> wrote:
>
> HI All.
>
> Looks like we are making progress. I am almost done with the API. Getting the machine state is complete and on/off functionality like pump/solinoids is also ready.
>
> All that is left is actual movement. I discovered that the machine controller can deal with multiple connections. So the original software can run while my API is connected and listening. It also helps with movement testing as the original software will deal with the endstop while its not fully implemented in the API.
>
> The API now also has a test page to test functions.
>
> Documentation is on the lite side and care must be taken when testing.
> https://github.com/TVM802/TVM802-WEB-API
>
> --
> You received this message because you are subscribed to a topic in the Google Groups "OpenPnP" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/topic/openpnp/735aZn-GFRw/unsubscribe.

> To unsubscribe from this group and all its topics, 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/25989350-d601-4076-a531-460fba470cb9%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to a topic in the Google Groups "OpenPnP" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/openpnp/735aZn-GFRw/unsubscribe.

To unsubscribe from this group and all its topics, 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/BADAE249-7570-4332-815D-755A3EBD277C%40gmail.com.

[Richard Sloan]

again just thinking out loud... :-)

[Cri S]

As the xbox cam work well, are cheap and have low latency, I don't think on 960 there are using other cameras. Specs suggest that there have preferred staying at 0.5mm pitch limit and for smaller pitch relay on the body case instead of doing it right just to allow larger connectors tobbe populated instead.

[Jason von Nieda]

So, current summary:

* Bertus has an API that is coming along nicely. Sounds like it's not far off from being able to control the machine.

* Cri S has a driver plus gateway software that sounds like it's working. More details are needed. And it sounds like there may be some limitations in OpenPnP related to performance.

It seems like it might make sense for Cri S and Bertus to combine their efforts, if they are amenable to that.

At the moment, without a machine, I can't really help very much but I am certainly willing to help keep things on track and offer guidance where needed. 

Let's use this thread to coordinate efforts, and if there is something that I can do to help, please let me know.

Jason

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/f5467971-c8ae-4bbb-91e4-fa7250ede8f4%40googlegroups.com.

[Robert Walter]

I have a TVM920 and would be happy to help getting a native driver for openPNP going. Just let me know what you need, and I will see what I can do. Jason, if you want, I am happy to let you stop by and tinker. I am a couple hours away from you up in Vancouver, BC, but probably the closest machine to you currently.

R.

[Jason von Nieda]

Robert,

Thanks for your offer, but I've recently moved to Kansas City, MO, so it's a little more than a few hours for me now :)

Please keep an eye on this thread, though, as we'll definitely want to support the 920, as well, if it's similar to the 802.

Jason

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/822b86f9-82ac-44dd-8a51-b8b7cab346c7%40googlegroups.com.

[Richard Sloan]

920 has a PC inside.... and 4 nozzles, I suspect its very different.....

RichardS

[Robert Walter]

920 does have a PC inside, but still uses a updated ethernet based control board that appears to be based on the 802. So small single board atom PC, connected via ethernet to the main control board. The main control board seems to be all step / direction, with a daughter board connected via ribbon cable, with 7 x stepper drivers attached. Another 3 x closed loop discrete stepper drivers control the X, Y1+Y2 motors

The feeders are controlled from the main control board using 4 x serially connected driver boards. After a quick look at the driver boards, they are just HC595 shift registers connected to relay driver IC's. They are all daisy chained and the main control boards provides clock / data (SPI) to the feeders. Simple but it works.

All other IO, including pushbuttons, indicator lights, limit switches, etc go to the main control board.

Up / Down cameras are composite video, and each connects directly to a 2 channel video input PCI card on the internal ATOM PC.

I loaded openPNP onto the ATOM, and it started up without issue, however a little slow due to the low RAM installed on the ATOM (2GB I think). A little more RAM would be a good idea. Cameras were able to be seen in openPNP after a little configuration. Not sure on stability of openPNP as I didn't have much time playing with it. But it was a promising start.

R.

[Cri S]

The 6 stepper motors have surely the same protocol. For the missing part network sniffing is needed. Either installing sw or do hw sniffing. It is hovewer questionable if
the effort justify it. Can you do a closer look at the S curve drivers or controller?
Eventually adding arduino mega clone interfaced with existing interfaces is all what is needed.
Sure the acceleration settings are to find out.

The use of composite cameras explain something.
What is the reason you want to exchange the software? How many feeder you actually have?

[Bertus Kruger]

It is likely that the protocol will be similar. I extensively use WireShark for decoding the protocol(Very slow on an Atom PC). From AR__Systems's code it looks like Acceleration and Deceleration can be controlled for each axis. I will soon know if it actually works or not....

@Cri S have you used machine speed in your implementation?

Personally I don't care to much about speed. My concerns are with accuracy and not having to baby sit the machine.

Playing with the machine data I can already see that there might be some rounding issues with the 802 software. 

Currently the API already allows for step/mm configuration that will allow us to better tune the machine. Homing might also be an issue with the standard software.

[Richard Sloan]

On the 802A/B we could calculte the video latency buy placing the camera on say a fiducial and then instruct the camera to move and see how long it takes to see the movement.... 

[Cri S]

Yes, that works, including homing.
I have not experimented with different accel setting,
simple used the 6 speed setting from reference
Code.
I have measured 80ms camera delay.

[Cri S]

Speaking for homing issue, several homing dots resolve the problem nicely.

[Richard Sloan]

Yes it would everything can be based on the zero of the camera :-)

[Bertus Kruger]

Do you have an A or B model? I have an A model and the code from AR_Systems did not work on my machine. (I know he has a B)

I just get the feeling we might be dealing with different versions of firmware as well.

[Cri S]

B

[Bertus Kruger]

Cri S.

Can you tell us what bit was slow with openPNP?

[Bertus Kruger]

And how you would suggest to make it better?

[Paul Jones]

Tip on using wireshark: setup a bridge between the atom pc and the controller board using two Ethernet connections on a normal pc, then you can use wireshark on the good pc. Even better if you happen to have a managed Ethernet switch that has port mirroring.

 

Paul.

 

--

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/d37cde0e-7d02-48da-83fa-e7f549f67314%40googlegroups.com.

[Richard Sloan]

We may want to have a different thread for 920 :-) I suspect its very different from 802A/B

[Cri S]

Can you check the 3 xy drivers (chips). 

In order to know if these are just drivers or controllers

and what interfaces there have.

If this are standalone S curve controllers, then I see no reason to reverse engeenering and use arduino mega in the machine for controlling everything and a up looking webcam with adjusted optics offset. 

--

You received this message because you are subscribed to a topic in the Google Groups "OpenPnP" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/openpnp/735aZn-GFRw/unsubscribe.
To unsubscribe from this group and all its topics, 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/021cb029-3728-469c-a807-8d3be962549b%40googlegroups.com.

[Jason von Nieda]

The benefit of reverse engineering the machine is that we can have a solution that works out of the box for people who own these machines. It sounds like the vast majority has already been reverse engineered. Why give up?

Jason

On Wed, Nov 2, 2016 at 8:06 AM Cri S <phon...@gmail.com> wrote:

Can you check the 3 xy drivers (chips). 

In order to know if these are just drivers or controllers

and what interfaces there have.

If this are standalone S curve controllers, then I see no reason to reverse engeenering and use arduino mega in the machine for controlling everything and a up looking webcam with adjusted optics offset. 

Il mercoledì 2 novembre 2016, Richard Sloan <themind...@gmail.com> ha scritto:

We may want to have a different thread for 920 :-) I suspect its very different from 802A/B

On Wednesday, November 2, 2016 at 7:33:29 AM UTC-4, Paul Jones wrote:

Tip on using wireshark: setup a bridge between the atom pc and the controller board using two Ethernet connections on a normal pc, then you can use wireshark on the good pc. Even better if you happen to have a managed Ethernet switch that has port mirroring.

 

Paul.

--

You received this message because you are subscribed to a topic in the Google Groups "OpenPnP" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/openpnp/735aZn-GFRw/unsubscribe.

To unsubscribe from this group and all its topics, 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/021cb029-3728-469c-a807-8d3be962549b%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/CAJGcfUjni91Hkj9Bq5KVGjJSvd%3Dh954wfe6jv5v08Tu3R08fSg%40mail.gmail.com.

[Cri S]

Can you check the 3 xy drivers (chips). 

In order to know if these are just drivers or controllers

and what interfaces there have.

If this are standalone S curve controllers, then I see no reason to reverse engeenering and use arduino mega in the machine for controlling everything and a up looking webcam with adjusted optics offset. 

Il mercoledì 2 novembre 2016, Richard Sloan <themind...@gmail.com> ha scritto:

--

You received this message because you are subscribed to a topic in the Google Groups "OpenPnP" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/openpnp/735aZn-GFRw/unsubscribe.

To unsubscribe from this group and all its topics, 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/021cb029-3728-469c-a807-8d3be962549b%40googlegroups.com.

[Bertus Kruger]

I managed to get the machine moving last night.

Next on the list is to do homing and I still need to find a non-blocking way to move and stop when the end stops are triggered.

[Cri S]

Are there differences about at implementation and your for moving?

[Bertus Kruger]

The data going to the machine seems to be similar....

The hard part is porting it from C++ to JS. What I like about Node is that the byte buffers that I send to the machine looks exactly like the bytes being recorded on WireShark.

Where as the C++ bytes are in reversed order and gets padded. Example --> CommandQueue.push_back(0x20 00 08); actually sends 0x08 00 20 00

To make things interesting the positions data are stored in Big-endian format over 4bytes(32bits)

Example: Machine data --> 0x68103200  
Reversed Big-endian  -->0x00321068
Converted to Decimal steps --> 3'281'000
Converted to MM Float(steps/mm = 32808ish) -> 100.00mm

[Bertus Kruger]

Hi Jason,

Looking at:

void moveTo(ReferenceHeadMountable hm,
            Location location,
            double speed)

The TVM802's uses a single stepper to push down both nozzles and I am just thinking of how my movement API needs to look like...

At the moment the endpoint will look like this  /API/move?X=10.00&Y=10.00&Z=10.00&A1=10.00&A2=10.00

Z is the odd one as it will need to control two things. Maybe not a big issue as negative values can apply to the second nozzle.

[Jason von Nieda]

Hi Bertus,

This type of system is what I refer to as a cam based dual Z. OpenPnP has explicit support for this type of system. One Z value is all that is required. Negative values will be sent for the second nozzle. 

When configuring this, you will set your Z axis to use the CamTransform, which does all the math required to convert the two linear Z axes into a single angular Z axis value.

More information here: https://github.com/openpnp/openpnp/blob/develop/CHANGES.md#2016-10-17

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/58a9e719-ec2e-4d9f-a423-4cb69b55a2cf%40googlegroups.com.

[Richard Sloan]

JS?? Javascript, thought OPnP was Java....

[Richard Sloan]

So then is the coding going to try and keep a running error and adjust on the fly? and not use a "fudge factor" scaling like the OEM software?

On Wednesday, November 2, 2016 at 8:51:49 PM UTC-4, Bertus Kruger wrote:

[Jason von Nieda]

OpenPnP is Java, Bertus is writing a TVM API layer in JavaScript that runs separately from OpenPnP. When he finishes, we'll write a driver in Java that talks to his JavaScript API. 

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/6f1350d6-84b4-42cb-9fea-c46ef3bc0e7c%40googlegroups.com.

[Bertus Kruger]

The API will be configurable with settings like steps per mm. This allow you to actually calibrate the machine rather than scaling to "Fix" a wrongly calculated movement.

Jason, do you have some tips on how to test calibration settings? I was thinking of some sort of PCB that has accurate 200mm x 200mm lines. This might also point out alignment issues.

[Richard Sloan]

I placed my calipers opened to 250mm and used the camera to accurately calculate the scaling factor, this is good in one direction, but not for any rotational measurements.

[Bertus Kruger]

Ho much did you have to adjust it? (I am still running old software that does not have the scaling option)

[Richard Sloan]

So I was running 300x300mm PCB and there was at least 0.8-1.0mm difference at the far X,Y, now there is zero.

0.9974 X 0.9992 Y

Side note: lots of parameters must be stored in the PNP machine, if it can not communicate lots of values are zero.....

RichardS

[Richard Sloan]

3-4X more error accumulated in the X than the Y.... must be due to some steps per mm/inch rounding thing....

[Robert Walter]

I had long discussions with Qihe regarding scaling calibration on the X + Y axis'. The 920 I received was one of the first few off the production line, and it very nicely built, but did have a few bugs in the software (both PC based, and Controller firmware). However, Qihe was very responsive in sending me a new control board with updated firmware, as well as about 4 or 5 updates to the PC software. Every time a few little bugs get fixed, and a new option or so gets added. So far I am really happy with the purchase. A openPNP port would be fantastic though, as having the ability to add features, and gain granular control of vision operations would be the greatest benefit. Also, the ability to add vision based strip feeder functionality, and better BOM management would also be nice. Other than that, the build quality on the 920 is really, really good.

[Robert Walter]

Oh, also, what I found with the X/Y positional errors, is not seeming to be software, but the lack of tolerance in the drive pulleys / belts. In the original qihe firmware, pulley diameter was considered a constant, and was not adjustable to allow for dimensional errors in the aluminum pulleys. After they added the calibration adjustment options in the configuration window, it took very little time to get the machine calibrated and operating very precisely.

[Bertus Kruger]

I observed the same thing. Could be lots of things but to me its a positive sign as it can only get better... ;)

[Bertus Kruger]

I agree, the 802 I have is built really well and once we upgrade the software it will only get better.
I do envy the 920's feeders. :)

[Richard Sloan]

Yes the only think I do not like are the feeders on 802A, other than that its a nice machine, but for my purpose they are fine, and so far have not had trouble with them, just not nice to load and unload, but I keep mine loaded all the time so I am good..... 920 is 2X+ the price and takes lots of space!!!!

[Richard Sloan]

I am sure one of us will come up with a design for those feeders on a 802 once the OpenPNP is done, because then we can support them with new controllers working with OpenPNP, I am considering it myself.....

[Bertus Kruger]

I agree... I have seen the feeder banks for sale. They are also very easy to operate. Serial to RS485 with some RJ45 connectors would make for a simple solution.

[Robert Walter]

I can totally agree about the 802 type feeders. I had virtually identical feeders on a Yushengtech machine, and despite working (in theory) they are problematic at best. Constant fiddling / adjusting to make work. Better on paper tapes, really bad on plastic tapes, especially with LED / tall components.

The Yamaha CL  type (clone) feeders on the 920 are really good. I did not order mine from Qihe. I ordered them directly from a china based manufacturer for around $60 a piece delivered (70+ feeders ordered). The CL feeders are pneumatic, but reliable, inexpensive, and in great supply both in the used and new markets.

Clean, dry air is important to long lifespan, so a good dryer, or at minimum a quality water / oil trap is essential, but that goes for the rest of the machine anyhow. A little oil in the air to lubricate all the valves / pistons is good every so often, but quality air supply is a must for any machine to operate well over a liffe time.

I think once a rock solid driver is written for the 920, it will be the machine to beat on the market. If someone were to rewrite the firmware on the control board to be more openPNP friendly, with a few added features, it would be incredible. Essentially turning the 920 into an klller open pick / place hardware platform.

R.

[Bertus Kruger]

Hi Jason,

How well does openPNP's part alignment work? I few people were mentioning that SOT23 is an issue on the 802... 

Would it be helpful if I provide captured images before alignment for you to test with?

[Jason von Nieda]

Hi Bertus,

I don't really have good data to tell you how well it works. The best I can say is that I've found it to be very stable for my tests and demos and I haven't heard of any issues from people using it in the wild. Not very helpful, I know.

If you can send me some example images that would be great. I will run them through the pipeline and show you the results. It will also give me a chance to tune the pipeline for the machine. This is usually necessary to make sure we are able to properly subtract out any background objects.

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/3e1f18ee-9243-4409-8400-85e110795dbb%40googlegroups.com.

[Glen English]

protocol interface reverse engineering started...

http://www.eevblog.com/forum/manufacture/pick-place-machinetvm920/msg1089592/#msg1089592

there is really not alot to it. it doesnt use fixed length command packets  though, nor replies

IE do not start from a assumption that it is sensible.

[Glen English]

DONE !

http://www.eevblog.com/forum/manufacture/pick-place-machinetvm920/msg1090928/#msg1090928

see that address to download the info.

[Jason von Nieda]

Nicely done Glen! You made quick work of that! Can't wait to see this machine under OpenPnP control!

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/1c1c22d4-8f32-4be1-a968-a744ecdbbbd5%40googlegroups.com.

[Andrew Frazer]

Quite transformational. 

Glens sent the software out to a few others and they seem to be reporting good results.

[Richard Sloan]

So it seems this is 920 based, does this work on 802A?

[Glen English]

Hi Richard

seem my eevblog post - I dont know, how similar... but they have the same engineer , so I would expect similarity

I intend to buy one in the next 30-45 days and port it.  so, stay tuned. 

I want to expand the use of loose parts pickup to compensate for lack of feeders, so I want facility to say, tear 100 x 0603 caps into a small tray and have the thing sort it out and pick  

[Richard Sloan]

So I return, and can not find any new info possibly for 802? Is this true? Was hoping something would have come up by now :-)

[Kruger Creations]

No... I have stopped working on a port for the TVM802. I am going to upgrade to a new machine later this year. Looking at the new SmallSMT machine with propper feeders.

--
You received this message because you are subscribed to a topic in the Google Groups "OpenPnP" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/openpnp/735aZn-GFRw/unsubscribe.
To unsubscribe from this group and all its topics, 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/aee6a6a7-0e62-40f8-b089-822f7264eb58%40googlegroups.com.

[Richard Sloan]

So got a msg from bertus, that he is not working on the tvm802 project anymore do you know anyone else that is??

[Bionik Mana]

Hi,
Do you know if someone made a teardown of the TVM802 to see what hardware inside?
Badr

[AIR]

TVM802B doesn't stop then go home

Hi. Such problem TVM802B doesn't stop then go home. We found out that the controller has a broken input from connector J4 - pin 4 (resistance less than 1 Om). We bought STM32F107. Using the ST-Link programmer connected to the J1 connector, we were able to read the contents of the Flash memory controller  and uploaded it to new one. So now the mashine doesn't work but I can ping it. Then We returned old controller - the mashine begins to work, doesn't stop then go home. Can you say me how can we get firmware and upload it to controller?

[Marek T.]

Do you really think that manufacturer didn't protect the controller against to read it's flash??
You rather need try to buy programmed processor from manufacturer.

[Michael Grossman]

The exact same thing happened to me. Were you able to get the machine working again?