Author Topic: Issues with TVM920 pick and place machine aligning components with small pitch  (Read 1139 times)

0 Members and 1 Guest are viewing this topic.

Offline dimbmw

  • Regular Contributor
  • *
  • Posts: 78
  • Country: us
Hi everyone.

I am wondering if anyone had similar issues.

Our TVM920 machine seems to have issues with aligning of small pitched components.

It looks like it endlessly trying to turn the IC CW and CCW, and fails to get it aligned - it jumps over the correct position every time.

Below is the video and the screenshot.

Our engineer have tested the 360 degrees rotation of the component.
As we know at 0.00° and at 360.00° the IC must be in the same position.

So he tested it at 0.00°, 0.23° and 360.00°.

The respective angles the machine "reads" are:

set angle: 0.00° - read angle: 3.90°
set angle 0.23° - read angle 4.21°
set angle 360.00° - read angle 4.21°

Conclusion:

When we order the nozzle to rotate by 360.00° the IC rotates by 360.23°.

Note that 0.23° seems to be one microstep of the stepper motor that rotates the nozzle. Typical steppers are 200 steps/rev. With microstepping = 8 it gives 1600 steps per rev, and as a result 360°/1600=0.23°

So this seems to be a bug in the software - they seem to have have the stepper make extra step when crossing 0 degrees set angle. Just a wild guess - but it very much looks like this!



Anyone had that?   I contacted Huang in QiHe  tech support but no solution yet.

Thanks!

Video of the experiment with 360 degrees rotation:



Video of the TMV920 locked up trying to align the IC, going CW->CCW->CW back and forth:


« Last Edit: July 31, 2018, 03:27:28 am by dimbmw »
 

Offline dimbmw

  • Regular Contributor
  • *
  • Posts: 78
  • Country: us
Not too many QiHe customers here? Come on guys, I need your help  |O
 

Offline ar__systems

  • Frequent Contributor
  • **
  • Posts: 460
  • Country: ca
Well, what can we say... The visual system on TVM920 is crap...
 

Offline dimbmw

  • Regular Contributor
  • *
  • Posts: 78
  • Country: us
Well, what can we say... The visual system on TVM920 is crap...

I agree it sucks. But it is a software bug we are dealing with here... i can tell looking at the video that the chip is going back and force by more than 0.23 degrees.

QiHe support Huang said he will bring this issue up to their software guy - I am keeping fingers crossed - hopefully it is an easy fix.
 

Offline dimbmw

  • Regular Contributor
  • *
  • Posts: 78
  • Country: us
QiHe sent me the latest version of their software 1.41beta
The problem is still there unfortunately.
It makes the machine unusable if you have ICs with the pitch of 0.5mm or smaller.
 

Offline Smallsmt

  • Frequent Contributor
  • **
  • Posts: 449
  • Country: de
Quote
It makes the machine unusable if you have ICs with the pitch of 0.5mm or smaller.

Why do you think it's possible to place fine pitch IC < 0.5mm using these kind of belt machine?
The precision can only reached when using a position feed back on X/Y axis.
Do you think your machine has?

You bought a simple machine able to place components from 0402 down to >= 0.65mm pitch.
I don't know the angle resolution of the A axis and belt tension on X/Y axis has influence too.

You are near the limits these kind of machines can do.
 

Offline Dubbie

  • Supporter
  • ****
  • Posts: 854
This is not a software problem, it is a hardware problem. Whatever drives the rotation has enough backlash and slop that it cannot hit the set point.
 

Offline dimbmw

  • Regular Contributor
  • *
  • Posts: 78
  • Country: us
This is not a software problem, it is a hardware problem. Whatever drives the rotation has enough backlash and slop that it cannot hit the set point.

I respectfully disagree. In this machine the nozzle is attached directly to a shaft of a stepper motor that provides rotation, so there are no screws and nuts to cause backlash.
 

Offline Dubbie

  • Supporter
  • ****
  • Posts: 854
Ok, fair comment.

Maybe the stepper isn’t being driven properly?
 

Offline dimbmw

  • Regular Contributor
  • *
  • Posts: 78
  • Country: us
Ok, fair comment.

Maybe the stepper isn’t being driven properly?

Well, I don’t know... the easiest fix would be to double the number of microsteps from 8 to 16 to increase the angular resolution. But I guess even if we will switch the physical hardware driver of the machine to 16 microsteps somehow we will still need to account for that in the software.. so we will still need help from QiHe... And it seems to be tricky to even explain the problem to them in a way that they understand it :(
 

Offline nisma

  • Contributor
  • Posts: 31
  • Country: it
You can check if with more illumination it works better. As you maybe have noted, the upper edge enclosing rectangle changes at the two angles relative to pins.
Try to put additional two leds as example to that area (pin should reflect that to camera, you must check it on video when installing it) or just use desk lamp or
torch. If that solves the problem, you could try to adjust illumination/camera shutter settings, and if that don't work, make better illumination.

To the driven angle and read/real angles, it could be possible that the actual/last fiducial angle correction was -3.90 degree or something like this and that was just
cal in , or that the machine have this offset itself when traveling.
« Last Edit: August 05, 2018, 02:41:16 am by nisma »
 

Offline ar__systems

  • Frequent Contributor
  • **
  • Posts: 460
  • Country: ca
Why do you think it's possible to place fine pitch IC < 0.5mm using these kind of belt machine?
The precision can only reached when using a position feed back on X/Y axis.
Do you think your machine has?

Yes it does. So easy to check, no need to guess. It is perfectly accurate. One physical issue is nozzle mount that relies on o-ring. It creates to much play and passes vibration on A axis that make the part rotate randomly.
 

Offline pl116

  • Contributor
  • Posts: 12
  • Country: ru
dimbmw, do you have this issue only with large packages or with small too ?
 

Offline dtf

  • Regular Contributor
  • *
  • Posts: 77
  • Country: ca
Quote
It makes the machine unusable if you have ICs with the pitch of 0.5mm or smaller.

Why do you think it's possible to place fine pitch IC < 0.5mm using these kind of belt machine?
The precision can only reached when using a position feed back on X/Y axis.
Do you think your machine has?

You bought a simple machine able to place components from 0402 down to >= 0.65mm pitch.
I don't know the angle resolution of the A axis and belt tension on X/Y axis has influence too.

You are near the limits these kind of machines can do.

I am placing 0.3mm just fine, you are selling a competing machine, and clearly have very poor ethics to trash talk other companies machines like that. go away
 

Offline jedas

  • Contributor
  • Posts: 16
  • Country: lt
Not sure if it's the same problem or not, but we've seen even more aggressive turning: https://www.eevblog.com/forum/manufacture/pick-place-machinetvm920/msg1302585/#msg1302585

At the time it looked like that it depends on the feeder it takes part of. I was suspecting some kind of interference between particular valve control and stepper driver. It doesn't happen much anymore, but we do get occasional rotation by a degree or so. Still i have no idea if this is software or hardware problem, happens pretty randomly.
 

Offline Smallsmt

  • Frequent Contributor
  • **
  • Posts: 449
  • Country: de
Quote
It makes the machine unusable if you have ICs with the pitch of 0.5mm or smaller.

Why do you think it's possible to place fine pitch IC < 0.5mm using these kind of belt machine?
The precision can only reached when using a position feed back on X/Y axis.
Do you think your machine has?

You bought a simple machine able to place components from 0402 down to >= 0.65mm pitch.
I don't know the angle resolution of the A axis and belt tension on X/Y axis has influence too.

You are near the limits these kind of machines can do.

I am placing 0.3mm just fine, you are selling a competing machine, and clearly have very poor ethics to trash talk other companies machines like that. go away

@DTF Trash talk that's funny!
If you placed 0.3mm pitch components on a TVM920 machine congratulations for your results.
It would be useful to see a video and final result placing 0.3mm pitch parts on your machine.

I never said this machine is bad! Only if you buy a VW Golf you can't expect a Mercedes!
The only reason why I posted the message was to stop the useless dicussion why can't a cheap machine not reach the precision of other much more expensive machines.
It's easy to understand different machine classes have different results.

I talked about a belt machine and ball srew spindle machine receive different results.
I don't know why we need to use AC servos plus optical encoders and ball screws to reach the precision to place 0.4mm pin pitch parts.

The TVM920 has an improved new design the TVM925 and we are selling too!


Best Regards
Michael
 

Offline dimbmw

  • Regular Contributor
  • *
  • Posts: 78
  • Country: us
dimbmw, do you have this issue only with large packages or with small too ?

Angular errors mostly affects larger components as any angular error multiplies by the radius of the component and ultimately leads to a significant linear pad placement error.
 

Offline dimbmw

  • Regular Contributor
  • *
  • Posts: 78
  • Country: us
Not sure if it's the same problem or not, but we've seen even more aggressive turning: https://www.eevblog.com/forum/manufacture/pick-place-machinetvm920/msg1302585/#msg1302585

At the time it looked like that it depends on the feeder it takes part of. I was suspecting some kind of interference between particular valve control and stepper driver. It doesn't happen much anymore, but we do get occasional rotation by a degree or so. Still i have no idea if this is software or hardware problem, happens pretty randomly.

I’d check my compressor if I had smth like this.
 

Offline dimbmw

  • Regular Contributor
  • *
  • Posts: 78
  • Country: us
Quote
It makes the machine unusable if you have ICs with the pitch of 0.5mm or smaller.

Why do you think it's possible to place fine pitch IC < 0.5mm using these kind of belt machine?
The precision can only reached when using a position feed back on X/Y axis.
Do you think your machine has?

You bought a simple machine able to place components from 0402 down to >= 0.65mm pitch.
I don't know the angle resolution of the A axis and belt tension on X/Y axis has influence too.

You are near the limits these kind of machines can do.


This is BS. Plenty of machines I have at home (3D printers, CNC mills) work fine without feedback and are very precise.
 

Offline dimbmw

  • Regular Contributor
  • *
  • Posts: 78
  • Country: us
You can check if with more illumination it works better. As you maybe have noted, the upper edge enclosing rectangle changes at the two angles relative to pins.
Try to put additional two leds as example to that area (pin should reflect that to camera, you must check it on video when installing it) or just use desk lamp or
torch. If that solves the problem, you could try to adjust illumination/camera shutter settings, and if that don't work, make better illumination.

To the driven angle and read/real angles, it could be possible that the actual/last fiducial angle correction was -3.90 degree or something like this and that was just
cal in , or that the machine have this offset itself when traveling.

We have a custom illumination, the original one sucked big time. It was the first thing we improved.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf