Pick and Place Machine TVM802A / TVM802B

[tmf]

There is also probably nothing that can not be "tweak" in code if there is something inherently wrong with the controller, hacking the controller code at this point is not what I am sure a lot of 802 users want to do... the current software works for me very decently, there are some tiny issues I would like fixed and OpenPNP will probably be able to address them better....

Even if the controller does has a drift issue due to bad math, we can understand the error and account (calculate) for it in the new control software.

For me stock 802A is the way to go....

I am curious what controller is in mine, but I do not have time to rip it apart... :-)

RichardS

[nisma]

What are this issues, apart tall components, top vision for resistors or caps , nozzle toolchanger , missing autocalibration using vision and
to improve vision image preprocessing.

 

[glenenglish]

tmf: If the LEDs came off tape, how on earth did it put it in 180 deg ?

I would not be confident of the code in the micro....
I don't agree it is  irreversibly modifying their machine . If you can put new code in, you can likely (as I doubt they have jtag read fuses set) pull the old code out...
I guess though, on the other hand someone has t actually DO IT ! :-)

g

[protoneer]

Yes... It is likely that the firmware can be retrieved and re-flashed if needed.

I have a few of the T962 re-flow ovens and the original firmware could not even do proper time tracking. I found a solution online that showed how to flash it with better firmware and now these little ovens are worth double the money I paid for them.

If someone with the skills and time wants to take it apart and build better firmware I am all for it. The better we understand these machines the better they will perform in the future. For now I am putting time and effort into the openPNP driver that will use the existing firmware. (I use my machine for production work and will only risk new firmware updates once its well tested or I have a second machine)

[nisma]

Just tested, c2j translation works to some extend.
How should openpnp driver handle buttons, LEDs, vacuum and prick sensors?

[glenenglish]

Abstraction layer needs to be thought out.

This is probably off topic.  QiHe now have been monitoring these threads and we (I) am filling them full of junk.

[ar__systems]

tmf: If the LEDs came off tape, how on earth did it put it in 180 deg ?

Parts rotate on the nozzle in movement due to vibration.

[ar__systems]

There is also probably nothing that can not be "tweak" in code if there is something inherently wrong with the controller, hacking the controller code at this point is not what I am sure a lot of 802 users want to do... the current software works for me very decently, there are some tiny issues I would like fixed and OpenPNP will probably be able to address them better....

Even if the controller does has a drift issue due to bad math, we can understand the error and account (calculate) for it in the new control software.

For me stock 802A is the way to go....

Controller does not have a drift. All math errors are in S/W. However I'm now thinking I'd be better off with a new driver/controller entirely. The way original 802 controller works, it creates too much vibration.

[tmf]

There is also probably nothing that can not be "tweak" in code if there is something inherently wrong with the controller, hacking the controller code at this point is not what I am sure a lot of 802 users want to do... the current software works for me very decently, there are some tiny issues I would like fixed and OpenPNP will probably be able to address them better....

Even if the controller does has a drift issue due to bad math, we can understand the error and account (calculate) for it in the new control software.

For me stock 802A is the way to go....

Controller does not have a drift. All math errors are in S/W. However I'm now thinking I'd be better off with a new driver/controller entirely. The way original 802 controller works, it creates too much vibration.

I would say one step at a time, yes it creates vibrations, but we can first get OpenPNP to work with the current commands, then someone can tackle firmware and use the same commands to smooth out the vibrations.... or swap out the hardware and use a known hardware to OpenPNP...

RichardS

[harry4516]

.... However I'm now thinking I'd be better off with a new driver/controller entirely. The way original 802 controller works, it creates too much vibration.

I think thats the right way to go. Buildig a new and better controller costs less time than hacking the existing one.

[glenenglish]

anyone got any photos, pseudo schematics posted anywhere of its insides ?
how similar to TVM920 ? (microcontroller talking to ATOM PC on ethernet), expansion IO board on ribbon talking to misc single IOs and feeder solendoids, closed loop stepper interface...
g

[mrpackethead]

Anyone got a Lasagne recipe?

English-glen. that is offtopic

[nisma]

Ar__systems, as you surely have tested different acceleration and speed settings,
do you think this is because of simplified stepper accel/decel algorithm or because of
some meccanical stickness or slight disallinement?

[ar__systems]

Yeah, from what I read, the driver does not have current control and as a result applies to much torque at lower speeds. So the stepper completes full transition from one state to another faster than necessary for the given speed. In other word, it moves in small jerks instead of rotating continuously.

There is a bit of play in the nozzle lowering mechanism, but that does not seem to have significant effect.

[huebsch]

Folks!
Does anyone have a hint for me: I'm playing with the camera parameters, and found no solution. I'm using the upward camera to look on a SOT23. Sometimes it takes 1-3 corrections and the machine places the part. Next time it requires several minutes i.e. 100'reds of retries to identify it as OK and next it places it correctly. I'm using SW version 2.29
 
Could anyone post the configuration of the camera page?
Thanks
-AH-

[tmf]

I have also found SOT23 to be troublesome, when it messes with it adjusting the "100's" of times, sometimes it flips it 90 and 180 out in its attempt.....

RichardS

[huebsch]

I have also found SOT23 to be troublesome, when it messes with it adjusting the "100's" of times, sometimes it flips it 90 and 180 out in its attempt.....

I played a lot with the parameters so it does not rotate any more but it slows down speed dramatically. So again has anybody found a good configuration combination to recommend?
Thanks!
-AH-

[harry4516]

yes, I found a solution, but you will not like it:
I set the cam to quick mode and reduce the speed to 20% (for SOT23 only)
This is very slow but works for me.

The problem with SOT23 are vibrations. Nozzle suction is too weak for the vibrations, so the SOT23 are dancing and flipping.

After replacing the built in vac pump with an external one, its much (!) better. Now I can run SOT23 with 100% speed
using 504 nozzles.

[huebsch]

The problem with SOT23 are vibrations. Nozzle suction is too weak for the vibrations, so the SOT23 are dancing and flipping.

Thanks for the recommendation I'll try it.

Regarding the vibrations, I have seen that in an earlier posting. I have not that experience on my machine. I do MELF or SO23 with 80-100% speed and I have no problem loosing a component. I played around with the nozzles to find a component / nozzle combination that works fine. There are vibrations visible if the nozzle is too small, that the component dances around as you described. With a suitable nozzle even round ones like MELF work fine.
I had to readjust the prick offset for some strings for some reason. Quite interesting that this happens only for specific feeders. I had to redo that several times, only +/- 0,5 to 1mm
-AH-

[tmf]

The 90 and 180 are not vibration for me, there are exactly out 90 and 180 due to camera error.

If the quick takes a picture and in that picture it can not outline the pins correctly, (due to lighting and contrast) its not going to work either, at any speed.

RichardS

[ar__systems]

The 90 and 180 are not vibration for me, there are exactly out 90 and 180 due to camera error.

I doubt your assessment is correct. AFAIK the s/w is not capable of making visual corrections of more than 45 deg.

More likely the initial rotation was due to incorrect pick or the package bounced in the pocket.

[tmf]

The 90 and 180 are not vibration for me, there are exactly out 90 and 180 due to camera error.

I doubt your assessment is correct. AFAIK the s/w is not capable of making visual corrections of more than 45 deg.

More likely the initial rotation was due to incorrect pick or the package bounced in the pocket.

Interesting... I will keep my eye on this and try and see why then, does not happen that often so could be a bit before I see it again.....

RichardS

[laurianus]

Can someone please send to me the latest software of the TVM802A PNP?

Thanks

[huebsch]

Can someone please send to me the latest software of the TVM802A PNP?

Just contact Gera directly via mail. I got the SW within hours and got asked not to share it with other or put it publicly on the net. Current version is 2.29
-AH-

[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.