JLCPCB SMT Assembly Service

[stoepie2002]

How would you then apply the paste?

[SteveyG]

With a spatula.

[stoepie2002]

An 0805 size spatula 

For those few components that I can't get from JLC, I will apply some solder paste with a syringe, add the components and use my hot air pencil.

[SteveyG]

An 0805 size spatula 

For those few components that I can't get from JLC, I will apply some solder paste with a syringe, add the components and use my hot air pencil.

  The spatula only needs to be smaller than the stencil.

For 0805's you'd just solder them with a soldering iron

[stoepie2002]

Not if other JLC-pre assembled parts are sticking thru the stencil cut-outs you suggested!   

Putting some paste on pads with a syringe, putting the parts in that, and using hot air is way faster and much nicer than soldering by hand.

[SteveyG]

Not if other JLC-pre assembled parts are sticking thru the stencil cut-outs you suggested!   

Can't work out the difficulty, unless you're missing my suggestion 
You can't assemble using a production paste screen squeezee, but this is prototype assembly. 

[stoepie2002]

This was your suggestion:

You could design a stencil with cut-outs around already placed parts.

Parts would be sticking thru the cutouts and you wouldn't be able to put solder on the board using a spatula any bigger than an 0805.

My boards are typically compact and that odd component that's missing would have many surrounding components sticking thru your cut-outs... you still don't understand? If so, I give up.

[SMTech]

His suggestion, crazy tho' I consider it, is I think actually spread across the two near identical threads. For complex parts not fitted by JLC you would cut out the footprints a bit like rework stencils, so a tiny stencil with just that device aperture on it, how feasible this is would depend greatly on layout and how accurate you needed to be, it would certainly be very fiddly.

[juwi]

just tin the pads and apply flux, then put the part on and give it a heat gun warmup. For bgas skip the tinning part if its a enig PCB and just apply flux.
Works perfekt.

Best regards
Julian

[SteveyG]

His suggestion, crazy tho' I consider it, is I think actually spread across the two near identical threads. For complex parts not fitted by JLC you would cut out the footprints a bit like rework stencils, so a tiny stencil with just that device aperture on it, how feasible this is would depend greatly on layout and how accurate you needed to be, it would certainly be very fiddly.

Agreed, it depends entirely on the layout, but these are prototypes anyway - unless it's a BGA device you don't even need a stencil for 95% of typical footprints. They can all be reworked with hot air as you would do on any repair.

[stoepie2002]

Possible yes, but slow, when doing a board with many components. I then prefer a stencil. Or SMT assy service (back on topic).

My boards will arrive on Monday, and I will post a mini review.

[stoepie2002]

My first SMT assembled boards arrived.

The quality is top notch.

I made the names only 25mil, where JLC says they should be 32 mil. Even at 25 it looks great.

[cyberfish]

Thanks for the photos! They do look very good. Did you use any 0402 passives? Also, were you able to figure out the load capacitance for the YXC crystals? Their datasheet says it's encoded in the part number, but without a mapping table.

My first SMT assembled boards arrived.

The quality is top notch.

I made the names only 25mil, where JLC says they should be 32 mil. Even at 25 it looks great.

[stoepie2002]

I did not use any 0402's. They're marginally cheaper than the 0603's I used, but I wanted to be able to easily rework/modify the board if needed. 0603 is easier, IMHO. Also, since I have most values in 0805 and 0603 and nothing in 0402.... (except a reel of 100nF's)

Very good question regarding the load capacitors. It had me wondering as well, and then I decided to throw 22pf's against the C70567 YXC crystal (click).

I already soldered in the missing parts (on just one board) and tested it to find everything works. The exact frequency isn't critical in this application. I will test (temperature cabinet) if it works (starts) over the entire temperature range. I did a quick test (hot air and freeze spray) and am not too worried.

DHL was faster than they said they would be and I didn't mind at all! 

The boards arrived within one week of ordering. I find that truly amazing! 

[DBecker]

I've been trying to work through generating fabrication files from Kicad in a way that can be automated and clearly documented.

Right now the sticking point is an automated way of getting part rotations correct.

At a guess, Kicad footprints generally are using the IPC-7x51 standard.  Parts are oriented so that pin 1 is on the upper left, or on the left for two pin parts.  Rotation from that base orientation is CCW e.g. a +90 degree rotation puts pin one on the lower left.

JLC seems to be using the EIA-481-D standard, or perhaps the IEC61188 standard.

The EIA-481-D standard would explain why my SOT223 parts are 180 degrees wrong,  and my SOIC parts are 90 degrees off, but I'm not sure if it is consistent with my diodes (including LEDs) being reversed.

[SteveyG]

I've been trying to work through generating fabrication files from Kicad in a way that can be automated and clearly documented.

Right now the sticking point is an automated way of getting part rotations correct.

At a guess, Kicad footprints generally are using the IPC-7x51 standard.  Parts are oriented so that pin 1 is on the upper left, or on the left for two pin parts.  Rotation from that base orientation is CCW e.g. a +90 degree rotation puts pin one on the lower left.

JLC seems to be using the EIA-481-D standard, or perhaps the IEC61188 standard.

The EIA-481-D standard would explain why my SOT223 parts are 180 degrees wrong,  and my SOIC parts are 90 degrees off, but I'm not sure if it is consistent with my diodes (including LEDs) being reversed.

I think it may have more to do with how the parts are supplied on tape, but I understand there will be a function soon on the website to allow rotation of parts to be easily managed.

[SMTech]

This and manufacturer specific packages that don't really have a name or using a name for that package that is barely recognizable (looking at you Altium) are the things that make automated quotation and conversion of centroid files to manufacturing data more difficult. There's a lot of work there to cover everything as you essentially need to add all the data manipulation tools the SMT software might add and offer them in your web frontend. Limiting whats on offer to a stocklist certainly helps tho, if they made that list graphical like a mini datasheet, you'd be able to fix your data to match theirs.

[DBecker]

I've been trying to work through generating fabrication files from Kicad in a way that can be automated and clearly documented.

Right now the sticking point is an automated way of getting part rotations correct.

At a guess, Kicad footprints generally are using the IPC-7x51 standard.  Parts are oriented so that pin 1 is on the upper left, or on the left for two pin parts.  Rotation from that base orientation is CCW e.g. a +90 degree rotation puts pin one on the lower left.

JLC seems to be using the EIA-481-D standard, or perhaps the IEC61188 standard.

The EIA-481-D standard would explain why my SOT223 parts are 180 degrees wrong,  and my SOIC parts are 90 degrees off, but I'm not sure if it is consistent with my diodes (including LEDs) being reversed.

I think it may have more to do with how the parts are supplied on tape, but I understand there will be a function soon on the website to allow rotation of parts to be easily managed.

Those standards cover how parts are delivered on tape, and how to consistently describe a difference from the nominal orientation.

I've seen references on datasheets before, but have always ignored them as "someone else's problem".
I did quick survey of a few datasheets and found only EIA STD 481 listed.  Unfortunately that's the standard that is the least predictable e.g. parts changing orientation between SOP and TSOP.

[stoepie2002]

Exactly the reason I put the 'correction table' for orientation inside the ULP to create the BOM and the CPL files.

This enables me to add component corrections (based on value and package type) and to have that available for my future designs.

In my previous job I developed fairly complex electronics products on multilayer boards, and I was also responsible to get the products 'production ready'. IE make sure it can be built, see to proto testing and spec verifications, EMC/CE, vibration test, documentation, alignment procedures, mechanical specifications, etc etc.

Besides all that, we had someone who was was dedicated to source the parts (with procurement) and making sure all parts were properly available in Cadstar and on the SMT assembly line. That was a fulltime job. This bit is never easy, so getting this right between yourself and a Chinese board/SMT house will be work. So be it.

[GreggD]

I have sorted many of the parts in the Excel file to make it much more convenient to use.
Hope you like it. "jlcsmt_parts_library_4.zip" 1.4MB

http://s000.tinyupload.com/?file_id=35614263837265853771

[avion23]

I have sorted many of the parts in the Excel file to make it much more convenient to use.
Hope you like it. "jlcsmt_parts_library_4.zip" 1.4MB

http://s000.tinyupload.com/?file_id=35614263837265853771

Wow thanks! That's actually useful

[n0mjs710]

I realize this thread is a bit stale, but... I've been trying for several days to get even the simplest board to populate in parts in the viewer. I use Eagle 7.7. The Gerbers work fine. I'm to the point of making a test board with 4 x 0805 resistors on it. I've used the (thank you!) excellent ULP script posted here, and hand built the files for my test board... Everything appears to work flawlessly, but now parts show up in the viewer during the final stage.

Has anyone tried to get a board assembled recently? Yesterday, even running the demo on their site resulted in no parts populated in the image.

[vad]

I realize this thread is a bit stale, but... I've been trying for several days to get even the simplest board to populate in parts in the viewer. I use Eagle 7.7. The Gerbers work fine. I'm to the point of making a test board with 4 x 0805 resistors on it. I've used the (thank you!) excellent ULP script posted here, and hand built the files for my test board... Everything appears to work flawlessly, but now parts show up in the viewer during the final stage.

Has anyone tried to get a board assembled recently? Yesterday, even running the demo on their site resulted in no parts populated in the image.

I had the same problem few days ago, and created support ticket. They advised to ignore the placement preview applet, and just submit the order. JLCPCB will take care of validating the placement and rotating components themselves.

Here is the quote:

“Hi Sir,

Many thanks for your email.

Pls note that currently there is some errors with our online applet so no preview of SMT placements.

But the preview is jut for reference only. So pls no worries because for each SMT order our engineer will double check the orientation & polar of each parts making sure they are correct. If there is any issue of them we will let u know immediately. Pls don't worry about the preview and go ahead to submit order.

Best Regards.

Thanks and best regards.

Tony Li“

[n0mjs710]

Thanks vad!

I'd opened a ticket when I first saw it and, well, to be honest, couldn't really understand the answer, but it looked like a boiler plate "do it right and it will work". I tried again this morning and it works now. I guess they were working on it, or your ticket was more persuasive than mine. Either way, glad to have this service and this thread – without what I read here, I'd not be to the point of placing an order today!

[danie1]

Does anyone know if there is any way of managing out of stock parts other than just waiting and retrying?

I've emailed them to ask if it's possible to put the order in and have the boards wait until all parts are in stock or if they can give some indication of when parts will be available but no answer so far.

The parts are available on LCSC so I'm surprised they are out of stock for SMT assembly - I assumed they can go grab them from LCSC stock but who knows what the actual logistics of the service are.

Might be a bit of a faff if you have to wait for all parts in stock on a board with lots of different parts - some of the out of stock parts are just basic resistors but one is an IC that seems to go out of stock on LCSC for weeks at a time