LED polarity went wrong in PCBA FAbrication

[Mosaic]

Hi all...the 2 SMT 0805 LEDs on the red layer are in parallel but opposing polarities. They have a common 2.5V but are driven by a Low or Hi digital 5V signal to select which one lights. The thru hole LED placement is a DNP item which I have as a bicolor LED option. Now the PCBA manufacturer installed both SMT LEDs in parallel , same polarity so I cannot select which one lights independently. I note the tplace layer shows 2 square dots close to the centre point with orientation info.

So Idk which file in my gerbers to look at so as to locate what went wrong. Can anyone advise?

THis is the PCBA manufacturers response to me:
BTW the red '+' is just the silk screen left over from the bipolar thru hole LED option.

 

[JonSea]

I agree that it's confusing with the plus sign from the through-hole LED adjacent to the "backwards" LED.

My advice is to include detailed notes and even drawings of details like this, and actually almost anyplace orientation matters.

I had boards assembled with through-hole SIP networked resistors. It was obvious to me that the end of the resistor with the mark indicating the common pin should be positioned next to the triangle next to the footprint on the circuit board, but that detail wasn't obvious to the assembler who very consistently positioned both SIPS on every board the other way around.

That created a very strange error. One of the SIPS had pullup resistors for /MCLR (on the opposite end from the common pin), 2 switches and 2 option jumpers. I (stupidly) tested the boards without the option jumpers and everything worked perfectly. I installed one of the jumpers, and shipped the boards to my client.

When my client tried the boards, when he pressed the two switches simultaneously, the board would lock up, not even responding to the reset button. Turns out with the SIP the wrong way around, /MCLR was kind of at the wrong end of a voltage divider, and with 3 of the SIP resistors pulled low, the voltage level on /MCLR was in the disallowed range, hanging up the processor.

 

[Mosaic]

Well a manual assembly is one thing...but for SMT PCBA...is it possible the EAGLECAD model of the LED can identify the polarity and this translates to the GERBER somehow?

Something is curious though..based on the arrows for LED6 and 7 ..it appears that they SHOULD have been mounted with opposite polarities, not so?

 

[JonSea]

The polarity of the surface mount LED is indicated by the small square dot, but it's ambiguous because of the much larger plus sign adjacent to it. Even with the proper pattern on the board, a human programs how the parts are placed in the pick&place machine.

Document everything. Include a PDF of the Gerber file with appropriate layers shown, and add any and all details to make parts placement absolutely clear. I use the tdoc layer to call out details and show parts placement - this doesn't get screened onto the circuit board but providing an assembly drawing with this layer shown clarifies details to the person loading the pick&place.

 

[Mosaic]

Oh, well, I got 200 LEDs to swap around...
I have added explicit silk screened polarities for the LEDs to clarify the matter in the future.
I thought it was an automated GERBER process into the pick and place....didn't realize human was in the way.

 

[JonSea]

Ugh. That's a lot of rework!

 

[kubeek]

How Washington the pick and place program generated? I was under the impression that this is done straight from eagle, without any gerbers?

 

[Wade Hassler]

Agree with reply#7. Does Eagle generate a pick-place list?If so, it could be as easy editing it, if the parts' centers are placed correctly

 

[JonSea]

Eagle does generate a centroids list for pick&place. But I believe some human intervention is needed to go from that list to the actual pick&place code.

For example, a chip might be listed as having a rotation of 270°. But how does pin 1 relate to this orientation? It depends on how the footprint was oriented initially (package drawn vertically or horizontally). As far as I know, there is no universal way to do this.

 

[Mosaic]

Actually, the PCBA manufacturer...has admitted their mistake. They noted the silk screen instead of the actual part orientation when setting up the pick and place. run. They will reimburse for the re-work.

 

[dknguyen]

Actually, the PCBA manufacturer...has admitted their mistake. They noted the silk screen instead of the actual part orientation when setting up the pick and place. run. They will reimburse for the re-work.

Yeah my assembler did the same thing and all the ICs on the PCB were reversed. It seemed like they really really averse to inspecting the pick-and-place data and really try to squeeze everything they could from looking at the the silk-screen. So they used some stray dot on the silkscreen footprint (which in my mind was nowhere near the expected location for a pin 1 indicator. It wasn't even in the corner. I had to stare at the footprints for a good 30 seconds before I found what dot they were talking about.) as the pin 1 indicator. This was even after they asked me for clarification on the orientations and I sent them a diagram.

 

[JonSea]

I have stated including an assembly drawing when getting boards assembled. It includes the Eagle tdoc layer, which adds detail beyond the silk screen, along with any notes I add about orientation. Like I said before, don't assume that what's crystal clear to you will be for anybody else.