Scanning Deep Pockets based plastic parts using ATOS 5 Scanner & GOM Software 2021

[[Im...]]

Hello Members ,

I have a plastic parts with deep pockets feature to be scanned ( You can check the parts pictures below ).

                

 

I know there is options to scan deep pockets features utilizing GOM Software 

Can anyone suggest or guide me how to approach to scan the parts especially the deep pocket region ( With what setting & tweaks we need to do in software to scan those features )

I tried using multiple scans got the overall good result but in case of deep pockets, its not upto the mark. 

Kindly share your experience in this regards ( If you can share with any examples or snapshots it would be really helpful )

I am using ATOS 5 with GOM Software 2021 

Thanks in advance 

[[Ja...]]

I think maybe there is some confusion , the 'deep pockets' terminology i associate with the t-scan hawk harware , nota software function.

Looking at the pictures this could be very challenging to get everything.  There are multiple factors at play here

• Line of sight - aspect ratio makes it tricky to get visual access
• Angle onto walls shallow - there is a limit.  Could attempt to allow long triangles in scanning settings
• Part is dark and shiny - this in combination with shallow angle makes life very difficult.   More exposure times and ' more points' option might help, otherwise digitising spray 

Beyond this i think youll need some direct help from your distributor , so they can work with you to confirm the limits.

[[Im...]]

Please sign in to view this quote.

Dear James,

Thanks for the reply , i really appreciate your response.

I am very much aware about deep pockets association with particular t-scan hawk hardware.

What i meant here by deep pockets is about deep depth features in particularly , there are certain ways using single camera we can scan particular area with better scanning details . Since ATOS 5 have triple scan functionality i think it could be done , obviously not full scan but as you know much more scanned portion much more better it will be to reverse engineered.

I want to know  how to approach those setting so that it can be done in better ways. There are so many member in this forum who must had tried those setting or may be have face similar challenge .

I am also looking forward for GOM experts response ,what they had to suggest .

Regards

[[Ch...]]

Another option when spray is not relevant - is to try out the "reflection detection" (this functionality is more for automated/scanbox applications)
but it can be applied in a manual way as well.

[[No...]]

We scan a lot of parts like this, but never without spraying. We use an ATOS Capsule with two different MVs and find it next to imposiible to scan something like this (black and shiny) without spraying. And I'm talking about the part as a whole here, not just deep pockets. Maybe newer scanners are somehow better in this respect (though I can't imagine how)?

In my experience deep pockets scan better with a smaller MV. Although the distance between camera(s) and projector basically stays the same, smaller MVs seem to work a bit better. You should use a rotary table and try to scan the pocket so that both cameras have the widest possible viewing angle as often as possible. Also make both camera images visible and try to find a part position where the area to be scanned is visible in both images. That gives you the best chances to get as much data as possible.

By the way, a thing you probably won't be able to prevent: The narrower a deep pocket gets, the more you'll experience an "orange-peel" effect on the surfaces scanned. This seems to be related to light reflected by the pocket walls and seemingly can't be avoided.

[[Im...]]

Please sign in to view this quote.

Dear

Please sign in to view this username.

Thanks for the response, definitely i will use scanning spray  .

FYI i just shared the pictures without spray for better viewing so that people can can see the deep pockets .

I will give a try with reflection detection (manually method )

Regards

 

[[Im...]]

Please sign in to view this quote.

Dear

Please sign in to view this username.

Thanks for the response, really appreciate .

As I mentioned in above comments definitely i will use scanning spray  .

FYI i just shared the pictures without spray for better viewing so that people can can see the deep pockets .

Regarding the smaller MV ( We have 170 & 320 let me try with that & i will get back to you )

Till now tried with bigger lenses of MV 700 & result is not upto mark( As u can see in the attached snaps below )

The deep features details is needed to make accurate RE parts especially the rib & others 

Regards

[[Na...]]

Hello Imran,

it looks like the sensor is able to scan the surface. Even in the pockets is some scan data. Did you try to align the channel with both cameras? The view from the sensor's position would look like on your last screenshot. Then tilt the sensor a bit, so it is not looking perpendicular on the plane inside the channels. 
Maybe you can get some more data when you edit the creation parameters of the measurement -> Scan points -> Quality and increase the max. angle Sensor/Surface to 85°.

Regards,
Nanno

[[Ja...]]

Hey imran,

Glad response was useful,  i think everyone is giving much the same advise here.  No one likes using spray and you can use it less these days but there are times when it cant be avoided.

what i would say is that switching lenses is not going to help much here .   Hard to see scale but looking at result at the moment mv700 seems appropriate for level of detail.

smaller lenses do not improve the  line of sight and by definition have shorter depth of field which doesnt help for deep pockets.

i would say you should be able to get the plane of the bottom of the channel with the right angle, but no chance youll get the wall base plane radius transition.

I have a lot of experience with the systems and parts like this are definitely amongst hardest to approach for maximising coverage efficiently

P.s. Thanks Nanno , the ' long triangles' i mentioned is relating to the max viewing length.   

[[Da...]]

i would like to use CMM/ profile or cut it for this because whenever the results may look good, but can be trusted ? 

but we have to try 🙂it seems to me dots 1.5mm ? looks small 

MV70 is good, i would like to have 40, i do not belive in succes on MV320 

beyond methods above i would try something what is very time consuming and because of it, that may be bad method ;]

set some good looking position and scan with rotation table for about 24 positions and try if it will find something looking good - its easy way 

if yes find good positions viewing them all on scan section and try use simillar for scan more, if it not worked change angle of head and again, again few times. 

also sometimes i'm using scan with like 60 degrees on rotating table and about 6-12 positions for deep scans. 

 

playing manually - hardest part but biggest fun ^^ and may give best results -  may let you catch it, and do not worry if you see everything good in camera mode but nothing on scans.  try try try try try .... it will be boring be strong

sometimes changing head angle for 2-5 degrees can give good results

sometimes changing distance of head from object will work, 

few times i deleted my good working projects and i was not able to scan again inside because of bad angle 

few times i deleted my bad scan projects and next time i "accidently" made good scans of some holes previously been sure its not poseable

• changing head positions to some weird angles may help, sometimes position looks perfect - its not if it not works
• also try to use camera > exposure times > 2 > and set exposure to be looking good inside, while outside everything will be red. Here try different setting more or les brighter. This will make bad scans outside but software will calculate outside scans good (if you already have some good scans of outside) and then play again. 
• combination of angle and brightness is really hardest part because till you catch something inside - you never know does its poseable or its just wasting time. 

 

this here i worked about four four hours for scanning this exact hole, and its really good i think, how do you think ? 

I hope in future g suite will have option to define exact point/area on model, max time for search, and will use some processor power to automatically find this hard places instead of me. If i can do it, your software also can do it, just do it, waiting for g suite 2024 😉

[[No...]]

Please sign in to view this quote.

Technically speaking you're right of course. But nonetheless, I usually get more data out of features like the channels with our smaller MV (we have 320 and 120). It's of course a matter of part size as well. I didn't mention that, assuming everybody here has been trained and knows the recommended part / MV size ratio.  

Looking at the scans above I'd say you may be able to get a bit more data from the channels if you try really hard, but overall you won't get a lot better than this.

Edited January 13, 2023