Author Topic: JLCPCB x-ray report interpretation help  (Read 1952 times)

0 Members and 1 Guest are viewing this topic.

Offline davegravyTopic starter

  • Regular Contributor
  • *
  • Posts: 238
  • Country: ca
JLCPCB x-ray report interpretation help
« on: November 07, 2024, 10:41:20 pm »
Hi I used JLCPCB for assembly of my board with a Ublox LTE module. They sent me x-rays of the module and I'm concerned because I see circular/bubble-shaped voids or lighter areas over most of the pads. I've never reviewed an x-ray before so I hope I'm worrying needlessly.

I attached a sample x-ray and drew red arrows to examples of these voids. It's a bit hard to see because the module's subcomponents are overlaid.

The module integration manual is linked here, page 101 shows the pad layout:

https://content.u-blox.com/sites/default/files/SARA-R5_SysIntegrManual_UBX-19041356.pdf#page=101
« Last Edit: November 08, 2024, 03:00:09 am by davegravy »
 

Offline thm_w

  • Super Contributor
  • ***
  • Posts: 7345
  • Country: ca
  • Non-expert
Re: JLCPCB x-ray report interpretation help
« Reply #1 on: November 08, 2024, 08:59:11 pm »
Profile -> Modify profile -> Look and Layout ->  Don't show users' signatures
 
The following users thanked this post: Kean

Offline jayx

  • Contributor
  • Posts: 45
  • Country: gb
Re: JLCPCB x-ray report interpretation help
« Reply #2 on: November 08, 2024, 10:34:12 pm »
These voidings are massive, I suspect the problem may be lack of solder. Ask JLCPCB about stencil gerber and check apertures size, also ask what stencil thickness was used (ublox datasheet recommend quite thick 150 µm stencil). Also I'm curious what x-ray brand they're using, perhaps you have non-cropped screenshot?
 
The following users thanked this post: davegravy

Offline davegravyTopic starter

  • Regular Contributor
  • *
  • Posts: 238
  • Country: ca
Re: JLCPCB x-ray report interpretation help
« Reply #3 on: November 09, 2024, 02:19:28 pm »
These voidings are massive, I suspect the problem may be lack of solder. Ask JLCPCB about stencil gerber and check apertures size, also ask what stencil thickness was used (ublox datasheet recommend quite thick 150 µm stencil). Also I'm curious what x-ray brand they're using, perhaps you have non-cropped screenshot?

Thanks for confirming my suspicion. I don't have non-cropped x-rays, seems to be a limitation of their automation or equipment.

I messaged them a couple days ago to notify of the problem and was told to expect an email reply. When I get a one I'll request the items you mentioned.

Is this an expected part of the development process or is this a major error on their part... or both? I don't see any means to specify the stencil thickness in their ordering interface except for the "other/comment" field provided.

I have fine pitch packages (e.g. UQFN) on the same PCB - I wonder if this maybe drove them to use a stencil thickness that's less than appropriate for the larger pitch Ublox module? I just learned about step stencils, is this something they'd likely use to accommodate varying component pitch?

The order stopped progressing in their online tracking, usually it would have shipped by now so they may be investigating. I'm not sure what a reasonable resolution to this looks like, or what I should push for besides information and suggested design tweaks. The modules aren't reworkable (I've tried and failed miserably on past revisions) but if they have enough contact to function I can still extract some value from validating aspects of the design. It would have been nice to do thermal cycling tests on this revision but that seems a bit fruitless now.
« Last Edit: November 09, 2024, 02:43:00 pm by davegravy »
 

Offline davegravyTopic starter

  • Regular Contributor
  • *
  • Posts: 238
  • Country: ca
Re: JLCPCB x-ray report interpretation help
« Reply #4 on: November 10, 2024, 02:19:28 pm »
I just read the below in another thread relating to JLC PCBA

I got a message from JLC that they don't use the customer's paste gerber file, and create the stencil from the component footprints


I'd assumed they were using my gerber, now I undertand why jayx suggested requesting the gerber.

The Ublox module I'm using wasn't available in their library, so I had to request that they add it. From what I gather, this process involved them designing the stencil from the module's datasheet. Unfortunately, there was no opportunity for me to verify or review the stencil design before proceeding.

There was a mistake when they added the part, which resulted in the unit cost being incorrectly set to $0, and I couldn't submit the order until they fixed it. That speaks to their QC when adding new parts.

I'm hoping the best-case scenario is that the stencil design has some minor issues, like too small aperture, that could be corrected once they address it. I would rather it be something adjustable, like the stencil, than an issue with the paste or reflow profile, which might be more difficult to rectify.

Has anyone else run into this issue with Global Sourcing, or have tips on how to handle such stencil discrepancies (short of switching providers)?
« Last Edit: November 10, 2024, 02:39:37 pm by davegravy »
 

Offline jayx

  • Contributor
  • Posts: 45
  • Country: gb
Re: JLCPCB x-ray report interpretation help
« Reply #5 on: November 11, 2024, 09:06:34 pm »
You've mentioned UQFN so I think they may have used thinner stencil than recommended 150µm. For fine pitch QFNs and for example (imp) 0201 size components it's usually recommended to use 100µm or less. Step stencils could be used in this case, but they are much more expensive and create some issues during printing. It may be easier to top up the paste amount using jet printing or dispensing, but I wouldn't expect it from a cheap manufacturer.

As for non-cropped x-ray pictures, I've meant pictures which show the whole user interface rather than x-ray itself. I thought I may be able to figure out the machine brand.
 

Offline davegravyTopic starter

  • Regular Contributor
  • *
  • Posts: 238
  • Country: ca
Re: JLCPCB x-ray report interpretation help
« Reply #6 on: November 11, 2024, 09:50:51 pm »
As for non-cropped x-ray pictures, I've meant pictures which show the whole user interface rather than x-ray itself. I thought I may be able to figure out the machine brand.

Yeah that's all I got unfortunately.

No reply still from JLC, but I did get a response from Ublox support who said they've sent some resources to past customers who have had voiding issues with the SARA packages. Attaching in case anyone else ends up here and finds them helpful.

These modules are pricey, $90CAD a pop in prototyping quantities. If it takes multiple revisions and working with different manufacturers to prove out a reliable process, costs could get ugly. I asked Ublox if they have an inexpensive "dummy" package that's useful for this exercise - I also wonder if just making my own bare pcb with matching footprint would be a good enough stand-in or if all the module subcomponents have a significant impact on the thermal behavior during reflow.

 

Offline davegravyTopic starter

  • Regular Contributor
  • *
  • Posts: 238
  • Country: ca
Re: JLCPCB x-ray report interpretation help
« Reply #7 on: November 12, 2024, 02:45:33 pm »
Other info I now realize may be relevant:

I selected Economic PCBA which given the number of extended components on my PCB isn't actually much cheaper than Standard and I think means the stencil is shared with other customers on the same panel. Perhaps Standard would allow their engineer more flexibility in the stencil? I asked this to JLC.

I selected HASL finish, would ENIG potentially reduce voiding?
 

Offline hanakp

  • Regular Contributor
  • *
  • Posts: 116
  • Country: cz
Re: JLCPCB x-ray report interpretation help
« Reply #8 on: November 12, 2024, 04:04:28 pm »
Yep, those voids are definitely bad, I've seen them many times on "my" X-ray machine. They form underneath these PCB-like modules soldered on a larger PCBs quite often. It happens when gasses from flux in solder paste become trapped between the pads (module pads and PCB pads) and cannot escape out. Unfortunately, it depends on many factors (solder paste composition, reflow profile etc.), so here is no universal answer how to avoid them. If JLCPCB won't solve it somehow, just go to a different assembly house.
 

Offline davegravyTopic starter

  • Regular Contributor
  • *
  • Posts: 238
  • Country: ca
Re: JLCPCB x-ray report interpretation help
« Reply #9 on: November 13, 2024, 01:12:56 pm »
JLC's reply:

Quote
1-We double checked the X-ray inspection result and it seems the soldering should not be OK. Sorry for that. It is some careless on our side. Could you please kindly try some reworking on your side? We would like to apply some compensation for you.

2-For the stencil option I mentioned in the last reply, we support solder paste for unpopulated pads, special dimension of stencil openings, special thickness of stencil and step stencil (on one stencil, some area is thinner and some area is thicker). If any request is needed, you can select the option and remark clearly, so we will have some evaluation based on your design.

3-For the soldering of such parts, we sincerely suggest customer using our standard assembly service to ensure better quality. The stencil used in Standard assembly would also be more accurate so the solder paste printing is better. Hope this information is helpful to you.

Look forward to your reply.

Rework
This is extremely difficult if not impossible, and Ublox doesn't recommend it. I haven't found a way to successfully desolder these modules without destroying them (of two attempts I destroyed both). The wafer material they are built on is extremely thin (~0.7mm), there's no good lift point. To get tweezers to grab and lift them from the wafer edge instead of the metal RF shield while solder is molten isn't possible with any tweezers or tool I own. Dislodging the RF shield inevitably dislodges subcomponents beneath and then it's a write-off. I think doing this requires some kind of specialty precision tweezers with a wide mouth and wide grip, that I haven't found for sale (I looked high and low).

I can't simply reflow the boards in oven because of through hole headers with plastic that will melt. At best I could do targeted reflow using hot air over&under the module. But I think someone mentioned earlier that's unlikely to help.

Step Stencil
This is still unclear. I *think* they're saying the step stencil option doesn't apply ONLY to unpopulated pads but I'm still not certain. No response to my request for the stencil thickness that was used on my recent order.

« Last Edit: November 13, 2024, 02:02:58 pm by davegravy »
 

Offline 48X24X48X

  • Frequent Contributor
  • **
  • Posts: 524
  • Country: my
    • Rocket Scream
Re: JLCPCB x-ray report interpretation help
« Reply #10 on: November 14, 2024, 01:43:47 pm »
I have been designing products with LTE modules for the past couple of years. I would say getting these LGA module to have void solder paste is harder to happen than shorting them. In nearly all of my design uses JLCPCB stencil. Going for larger volume, I have a PCBA that handles them and there's when they are shorted as these guys tend to modify the paste layer. For all my JLCPCB stencil that has LGA modules, I tend to follow the manufacturer's recommendations and ask JLCPCB not to touch that part (I'm okay with them touching the rest). I also don't recommend running long tracks with huge vias in between pads under these module as recently I have a board that wasn't designed by me, it was shorted all over under the LGA when assembled. My wild guess for your case would be, the temperature wasn't hot enough as JLCPCB's assembly usually consist of many other projects as well. Instead of using your board that consist of a heavy module as the temperature of reference, it probably just use an average of the rest.
 
The following users thanked this post: davegravy

Offline davegravyTopic starter

  • Regular Contributor
  • *
  • Posts: 238
  • Country: ca
Re: JLCPCB x-ray report interpretation help
« Reply #11 on: November 14, 2024, 02:04:33 pm »
I have been designing products with LTE modules for the past couple of years. I would say getting these LGA module to have void solder paste is harder to happen than shorting them. In nearly all of my design uses JLCPCB stencil. Going for larger volume, I have a PCBA that handles them and there's when they are shorted as these guys tend to modify the paste layer. For all my JLCPCB stencil that has LGA modules, I tend to follow the manufacturer's recommendations and ask JLCPCB not to touch that part (I'm okay with them touching the rest). I also don't recommend running long tracks with huge vias in between pads under these module as recently I have a board that wasn't designed by me, it was shorted all over under the LGA when assembled. My wild guess for your case would be, the temperature wasn't hot enough as JLCPCB's assembly usually consist of many other projects as well. Instead of using your board that consist of a heavy module as the temperature of reference, it probably just use an average of the rest.

Interesting, thanks for the insight.

Good news I received the order and at least two of the five boards function.

JLC responded that the stencil thickness used was 120um, so a bit thinner than the manufacturer-recommended 150um. I'm thinking my next run I will select the step stencil option and ask them to ensure the LTE module has 150um thickness and the fine pitch components have whatever is recommended. I wonder if I panelize if I would not share space with other projects and then could ask that the module be used as temperature reference.

If the issue is indeed insufficient temperature shouldn't hot air rework help? I don't have an x-ray to confirm :(

My production runs will only be 25-50 units so I'd really like if I can make this work with JLC instead of bringing this to another PCBA where the unit costs will be much higher.
« Last Edit: November 14, 2024, 03:02:11 pm by davegravy »
 

Offline jayx

  • Contributor
  • Posts: 45
  • Country: gb
Re: JLCPCB x-ray report interpretation help
« Reply #12 on: November 14, 2024, 09:41:20 pm »
I'd expect all of them to work fine. Large voids may cause a problem in the long term, likely depends on temperature swing. 120µm is not that much different from 150µm so likely it's some other problem.
Did you actually enquire some other CEM? Is JLCPCB cheaper because they provide everything (PCBs, components, assembly)?
« Last Edit: November 14, 2024, 09:43:49 pm by jayx »
 

Offline thm_w

  • Super Contributor
  • ***
  • Posts: 7345
  • Country: ca
  • Non-expert
Re: JLCPCB x-ray report interpretation help
« Reply #13 on: November 15, 2024, 09:27:29 pm »
I'd expect all of them to work fine. Large voids may cause a problem in the long term, likely depends on temperature swing. 120µm is not that much different from 150µm so likely it's some other problem.
Did you actually enquire some other CEM? Is JLCPCB cheaper because they provide everything (PCBs, components, assembly)?

As per the other thread, OP is using these outdoors and expecting them to be highly reliable.

JLC is going to be the cheapest for small batch stuff like this as NRE is nothing, but, why risk it... just asking for trouble.
Profile -> Modify profile -> Look and Layout ->  Don't show users' signatures
 

Offline davegravyTopic starter

  • Regular Contributor
  • *
  • Posts: 238
  • Country: ca
Re: JLCPCB x-ray report interpretation help
« Reply #14 on: November 16, 2024, 02:20:33 pm »
I'd expect all of them to work fine. Large voids may cause a problem in the long term, likely depends on temperature swing. 120µm is not that much different from 150µm so likely it's some other problem.
Did you actually enquire some other CEM? Is JLCPCB cheaper because they provide everything (PCBs, components, assembly)?

As per the other thread, OP is using these outdoors and expecting them to be highly reliable.

JLC is going to be the cheapest for small batch stuff like this as NRE is nothing, but, why risk it... just asking for trouble.

If JLC is actually a risk then I agree it's not worth it. I want to give them a decent chance because there's a potential I'll need to make a lot of these (in small batches) over the years and if it's just a matter of selecting the right ordering options to get good yield then the longterm savings is significant.

Why not just order a large volume from a better PCBA?  That's venturing into the realm of business decisions where I have a limited voice  ;)
 

Offline hanakp

  • Regular Contributor
  • *
  • Posts: 116
  • Country: cz
Re: JLCPCB x-ray report interpretation help
« Reply #15 on: November 17, 2024, 05:47:00 pm »
This is extremely difficult if not impossible, and Ublox doesn't recommend it. I haven't found a way to successfully desolder these modules without destroying them (of two attempts I destroyed both).

It won't help you with the current batch, but I have. In fact, I've successfully removed and resoldered one uBlox module 2 times when they were hard to get due to Covid lockdowns. The key is to use old good lead solder paste, which melts at around 190 deg. C. The module's metal lid is soldered with lead-free solder, which melts at around 215 deg. C. By finely controlling temperature around 190 deg. C, you can lift off the module before the lid (or other components) fall off. Of course, you'll need proper equipment for that, I did it on Ersa IR/PL 650 rework station.
 
The following users thanked this post: thm_w


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf