Pick and Place Machine TVM802A / TVM802B

[mikeselectricstuff]

Does anyone have ideas on how to remove lock-tight on up facing camera lens?

I hooked the camera's up to openPNP and tried the part recognition pipeline. Seems to work very well but because the part is not focused the alignment is out by about 0.1mm on one side.

Surely there's a setting to determine the Z height when holding parts over the camera?

[protoneer]

Does anyone have ideas on how to remove lock-tight on up facing camera lens?

I hooked the camera's up to openPNP and tried the part recognition pipeline. Seems to work very well but because the part is not focused the alignment is out by about 0.1mm on one side.

Surely there's a setting to determine the Z height when holding parts over the camera?

Good  point...  The machine head is Cam based and I should be able to move the inspected nozzle a few mm in the negative direction(Part needs to farther away) to focus it.

[ar__systems]

I removed on mine. The stuff is tough, but I'm tough too. I used long nosed pliers and screwdrivers.

[ar__systems]

Surely there's a setting to determine the Z height when holding parts over the camera?

Yes there is but the focus point was still too high. It could not focus on taller (taller than 0603 that is) parts at the highest nozzle position.

[mikeselectricstuff]

Is the part height set correctly? This should be being used to adjust the focus distance for different part sizes

[vonnieda]

Is the part height set correctly? This should be being used to adjust the focus distance for different part sizes

Yep, check part height and camera height in the settings. When OpenPnP wants to image a part with the bottom camera it uses the camera's Z location (set in camera properties) and the part's height to calculate where it should place the nozzle. When you set the camera Z, it should be with no part on the nozzle and with the nozzle bottom in focus over the camera. Then when OpenPnP images it adds the part height to this value.

[tmf]

Is the part height set correctly? This should be being used to adjust the focus distance for different part sizes

Yep, check part height and camera height in the settings. When OpenPnP wants to image a part with the bottom camera it uses the camera's Z location (set in camera properties) and the part's height to calculate where it should place the nozzle. When you set the camera Z, it should be with no part on the nozzle and with the nozzle bottom in focus over the camera. Then when OpenPnP images it adds the part height to this value.

I wonder if the stock software does this??

RichardS

[ar__systems]

Today I tested out our own feeder and tape stripping system on the TVM802. The stripping works independently of the TVM, so it works fine with the stock s/w. I made a completely new feeder block that solves multiple issues of the original one. And don't even get me started on the TVM's stripping mechanism...

So far we only ran test jobs. Placed over 500 of 0603 caps and close to 100 of SOT23. I think there was one discarded components in total.

My feeder block packs tapes tighter (3 mm spacing instead of 5 of the original), so more lanes can fit into the same space. It has 18 of 8mm lanes and some space for configurable lanes, which allows setting up an arbitrary combination of tapes of any width. (That part is not tested yet.) If you fill configurable space with 8mm tapes only it gives 11 more of 8mm tapes. So L side can host 29 of 8mm tapes in total.

My stripping mechanism provides constant pulling force and it supposedly will never require any adjustments. Well maybe once when you install a new reel with a different peel force. However suitable peel force for the tapes that I checked is in fairly wide range, so I don't think it will be ever required.

[ar__systems]

I wonder if the stock software does this??

Yes it does

[tmf]

Nice! Any images?

Today I tested out our own feeder and tape stripping system on the TVM802. The stripping works independently of the TVM, so it works fine with the stock s/w. I made a completely new feeder block that solves multiple issues of the original one. And don't even get me started on the TVM's stripping mechanism...

So far we only ran test jobs. Placed over 500 of 0603 caps and close to 100 of SOT23. I think there was one discarded components in total.

My feeder block packs tapes tighter (3 mm spacing instead of 5 of the original), so more lanes can fit into the same space. It has 18 of 8mm lanes and some space for configurable lanes, which allows setting up an arbitrary combination of tapes of any width. (That part is not tested yet.) If you fill configurable space with 8mm tapes only it gives 11 more of 8mm tapes. So L side can host 29 of 8mm tapes in total.

My stripping mechanism provides constant pulling force and it supposedly will never require any adjustments. Well maybe once when you install a new reel with a different peel force. However suitable peel force for the tapes that I checked is in fairly wide range, so I don't think it will be ever required.

[Sugrob]

Hi All,
I got my TVM802A a month ago and except all known bugs I got one very specific. I can't fix it and want to ask community about it.

After 1-2 weeks of using my TVM802A I found it's hit nozzle into stack bolt. I don't see this crash but hear it and when I turned around, I found that nozzle is broken.
Ok, I order 10 nozzles ))))
After that I continue to use this machine but collisions occurs regular.. more and more frequently. At the begining I guess it's all because PC and Windows... I found information here that TVM802A may skip commands from PC (W10) by ethernet. But I use Win7. In any way I tried reinstall Win7 and even more - I tried several different PC and even one with Win XP. XP don't work correctly at all. So I return to Win7.
As I already wrote, collisions happens more and more frequently. Now it happens every days ((( Todays it was 2 collisions but I finally see it and understand one important thing - It not seems only like escaping of command. It's looked like something absolutely wrong happens.
First collision:
1) nozzle pick up chip 0805 (right action)
2) nozzle coming to camera  (right action)
3) recognize chip (right action)
4) move nozzle down by Z-axis (under camera plastic box level) (why ?)
5) than it's move and hit nozzle to camera box side.

Second collision:
1) head come to stack (right action)
2) move prick down  (right action)
3) machine move head to camera (.....why ?)

So first time machine make wrong action, which can't be at all. Next time machine must move prick by X-axis, than pick up chip and only after that move head to camera.
Anybody have ideas what happens and how to fix it?

[huebsch]

Hi!
I have crashes as well but eventually once a week using it for 2-3 hours a day. What SW version are you using. My experience is that one of the SW updates reduced the number ob mismovements.

Do you have any other activity on that PC during usage?
-AH-

[ar__systems]

In the first sequence it can still be skipped commands. I.e. after the camera it was supposed to move to the component location and then nozzle down, but it just went straight to nozzle down.

I'm using XP SP3. Also, I recommend not to have any other equipment on that network. I have only PC and machine connected to a switch.

[Sugrob]

I have crashes as well but eventually once a week using it for 2-3 hours a day.

It's also frequently.
I use is every day 4-6hours

My experience is that one of the SW updates reduced the number ob mismovements.
Do you have any other activity on that PC during usage?

Sorry, SW.. what do you mean? SurfaceMounter app version? I will check it tomorrow.
And of course I don't run any other applications on PC. It's stay alone with fresh Win7 without any other applications. I even switch off WiFi )))

In the first sequence it can still be skipped commands.

I don't think so, because after camera it's tried go to the PCB, like nothing had happens.

[ar__systems]

what version of s/w do you use?

[Sugrob]

ar__systems, sorry, but english is not my native language and I don't sure what exactly you mean when ask about s/w, but I guess software.
I check QiHe Vision Mounter version is V2.30

[harry4516]

I am using the TVM802 since 9 month and never had any crash or wrong movement.
My software is version 2.12, I will not update it since it works perfectly.

your problem may be caused by missing commands over the network, like the well known Win10 problem. However, I cannot understand how a command can get lost in a network. This can only be an issue of the PC software or the firmware in the machine itself.

So we need to compare the firmware version too, but I have no idea how to read this version number.

[Sugrob]

harry4516, yes I also think about machine controller. When spring festival in China will be over, I will try to order another controller from qihe but if it will not help..
In this case only one way I see is order another machine like neoden4 to etc.

Do you know some one who has successfull experience of using this machine with free software??

[ar__systems]

I am using the TVM802 since 9 month and never had any crash or wrong movement.
My software is version 2.12, I will not update it since it works perfectly.

your problem may be caused by missing commands over the network, like the well known Win10 problem. However, I cannot understand how a command can get lost in a network.

I can tell you how. Normally two separate commands arrive in separate packets. But W10 can be merging some packets if they are issued close enough. Then the machine does not recognize them. I've confirmed that with WireShark when I worked on reversing protocol.

[Sugrob]

I can tell you how. Normally two separate commands arrive in separate packets. But W10 can be merging some packets if they are issued close enough. Then the machine does not recognize them. I've confirmed that with WireShark when I worked on reversing protocol.

Do you successfully start machine under free software as a result of reversing protocol ?

[harry4516]

I can tell you how. Normally two separate commands arrive in separate packets. But W10 can be merging some packets if they are issued close enough. Then the machine does not recognize them. I've confirmed that with WireShark when I worked on reversing protocol.

interesting fact. So it is an issue of the controller firmware which cannot separate the packets. Maybe QiHE should be able to solve this problem.
Many people use TCP as a packet oriented protocol (like UDP), but it is stream oriented (the application is responsible to find the start and end of an information). This makes problems if packets are sent too close.
Looks like I am  lucky, because I use one of my oldest PCs for the TVM802, this thing is too slow to make this kind of problem.

[ar__systems]

I can tell you how. Normally two separate commands arrive in separate packets. But W10 can be merging some packets if they are issued close enough. Then the machine does not recognize them. I've confirmed that with WireShark when I worked on reversing protocol.

Do you successfully start machine under free software as a result of reversing protocol ?

No. I've worked out the protocol and made basic application that allows to control everything, but not past that. In the hindsight, the machine has limitations that go beyond the crappy software, so even porting it to OpenPNP would not be sufficient to make it useful.

In particular, the stepper controller is not smooth enough. It creates too much vibration. As a result parts, especially ceramic caps dance on the nozzle which results in frequent discards and/or misplaced parts.

[Sugrob]

No. I've worked out the protocol and made basic application that allows to control everything, but not past that. In the hindsight, the machine has limitations that go beyond the crappy software, so even porting it to OpenPNP would not be sufficient to make it useful.

In particular, the stepper controller is not smooth enough. It creates too much vibration. As a result parts, especially ceramic caps dance on the nozzle which results in frequent discards and/or misplaced parts.

Bad stepper drivers and nozzle vibration is the smallest problem. I will be happy if machine will stop to destroy itself and nozzles.
I tried yesterday to reduce machine speed to 60% and every components speed to 60%. So all movements seems very slow, but finally after 4 hours of working machine again broke nozzle. I hate it (

[ar__systems]

maybe you should ask the manufacturer.

[Sugrob]

I already ask support but they don't know what happens and don't give me any recommendations (just only useless like check program step by step and etc). It seems like my case not interesting for them. I will send them videos after 6.FEB but I not sure that it will help me.