Part request to apa102-2020


#22

Ok, I think the original part is indeed incorrect, for reasons I don’t understand the pcb seems to be rotating on import, and without a pin1 marker it was difficult to see the original part is flipped and won’t line up properly with a real chip (the pins would be reversed, connector0 / pin1 on the right instead of the left as it should be). For this corrected (I hope!) part I grabbed @Sublimeartistry 's latest footprint, added a silkscreen group (it was just the rectangle) and added two lines to mark pin 1 (although there doesn’t appear to be any way to tell on the chip that I can see in the data sheet) and changed the pin numbering to match that on the data sheet (with the extra ground as connector6 / pin7 and the extra VCC as connector7 / Pin 8 as they aren’t numbered on the data sheet). In schematic I changed the pin numbering to match the new scheme (which puts the pin numbers out of order, but matches the chip and bb layout) and corrected the names to CLKI (instead of 1) and CLKO (instead of 0) as they should be. On bb I rotated the image by 90 degrees to align it properly and moved the input pins to the left and output pins to the right as is standard. Again the pin numbers are out of order, but they match the physical layout of the chip which I think is the best choice. I changed the variant to 2 in the fpz file and corrected the label on one of the grounds (which was set to VCC), but I don’t think a straight replacement is likely to work as the pin numbers have changed. This one is a new part and should coexist with the others. If anybody knows how to stop pcb from rotating I’d like to hear it (because is annoying having to rotate it to the correct orientation).

edit: there looks to be an error in the data sheet and this part is incorrect. Deleted til we can make
a proper part.

Peter


#23

did I started a crazy topic? :lol

not yet too far yet, 40copys chained

I’ll upload my working prototype to show the directions of the part.
it’s kinda hard to tell which directions’ in/out from the front side (and back) of the chip

but yeah, it works anyway

also, there seems to be different 2020 packages on the market, with different footprint
i haven’t tried or seen any of them besides mine,


#24

Here is my working prototype. With my "pads part"
Footprint, one can find the board from the APA-102-2020 - the ultimate smd addressable led.
The package is distinguishable from top,but can’t from bottom
There’s IC implanted in right side of the chip,
And leds in the left.

The density is 4LED/cm
Hopefully could be as dense as 5/cm, with specialized footprints
But there are power pins on the side of the package so I’m afraid of shortage

edit: from observations, the Din/Cin and Dout/Cout were inverted in your footprints,
just need to switch label Din<->Dout, and Cin<->Cout


#25

yes,yes
using W7,too , sometimes W10
not sure if Chinese Big5 OS language is possibly a problem?
much crashes on xml editting,

and yes, the board designed, shrink in height direction but not width.
so I need a few try, and even txt-edit the .svg for fitting.
thanks, will try to look into default unit next time.

@sublimeartistry
my next wish is that Fritzing can do 4-layer or more layer boards,
so that power line and gnd can easily hidden,
but 2020 for me, there is still plenty of room for power lines on board.


#26

@vanepp

Actually every single part we have made so far is wrong. I wish they had a good datasheet available.
If we look at this picture from the datasheet we can see that when viewed from above the top left is vcc and the top right is gnd which is opposite of all the versions we have made so far (and opposite of the diagram next to the pad layout in the datasheet). As well as the previously mentioned In and Outs being on the wrong sides.

Basically so far we have only actually got 2 of 8 pins correct :joy:
I am going to delete all of my above uploads as to prevent anyone from trying to use them in the future.


#27

Yep I had just come to the same conclusion, based on the working board the data sheet appears to be wrong, and thus so are all the parts. I agree we should delete them all while we regroup :slight_smile: .

edit: I think your latest footprint should be fine (as our understanding of the pin numbers seems to be the error) so I’ll start from that and the original working footprint and try again …

Peter


#28

That is possible, but it may also be a bad setting in preferences because I use xml edit a lot and other than the group issue (which is mostly my own fault) haven’t had any crashes. There are so many settings available its possible something is set wrong. I have found that if you hit a wrong key Inkscape will happily update preferences and remain screwed up. I generally keep a copy of the

C:\Users\Owner\AppData\Roaming\inkscape\preferences.xml

around and replace it when I screw something up. That may be worth trying (if you can get a copy from a default install somewhere).

Peter


#29

I think this time I may have it. The data sheet is in fact not wrong, just not standard. They are showing the bottom of the LED so they showed bottom view (i.e. horizontally swapped from a conventional footprint layout) where the other two views are conventional top down view and didn’t note that. I assumed the normal top view which isn’t correct in this case. In this new (and I hope correct :slight_smile: ) part, I changed the fpz so variant is back to 1, added a note to the description about which exact part this is and that there are different footprints so check yours, and corrected the pin numbers and bus definition to match the corrected pin numbers. Breadboard, changed the pin numbers to match the new ones, same with schematic. PCB rotated the image by 90 degrees so pin 1 is on the bottom left like an IC (that is why Fritzing is rotating the footprint), it was otherwise correct (and is the latest better routing one).

APA102-2020_fixed.fzpz (13.1 KB)

and a test sketch that indicates it may be correct this time (but please check!)

fixed.fzz (20.5 KB)

Peter


#30

not sure if there’s wrong with the datasheet or not,
by far that’s the best one I can find. and that matches the ones i bought.
(apa102-5050 or 2020 have bad/ugly datasheet when manufactured in China)
(and datasheets for ants, with so tiny and blur fonts)

@vanepp
yes
there’s a bit tricky when reading the datasheet, mostly when flipping top/bottom view of the chip
though the datasheet didn’t specific mention the views, the top/bot view is obvious if didn’t consider transparency
Its months ago, but I’m sure it was correct or otherwise I can’t make a working prototype, which means my pin assuming is justified, in the first try export with pad :confused:

and I also found that, the pins of 2020, were same directions of apa102 - 5050,
just with extra power/gnd pads beneath.

not much datasheet I’ve read, so not sure how other sheets flip a part.

and No, nothing to do with Chinese, I read them, but It tells nothing important in the datasheet. :frowning:


#31

not sure if apa102-2020 only for serious maker or not, :fire:
the 2020 package makes it hard(oh, impossible i mean) to hand solder , needs paste and air gun at least?

there are plenty of I.A. LED products(ws2812in5050/3535) apa102-5050 that sold in LED strips

if they made 2020 a strip product, i wouldn’t have to draw my own :cry:


#32

They do make it in strips already. I contacted them the other day and got a price. It has to be the most expensive strip I have ever seen and it only comes in 1m strips.

http://neon-world.com/uploads/soft/20161130/1480498721.pdf


#33

hell yeah, I admit I’m happy for one second.

looked into the link and found that they’re not very dense, 100LEDs/m and 200LEDs/m i see?
still not fit my need, but better
its not possible to left cut/connection pads when making super dense one

it’s possible to gap-less connect strips together so you get a 5-m roll
the 144LED / m apa102-5050 strip is about ~25USD/m , and they all came in 1m strips
the more the LED the greater the price.

add edit:
I can’t find a product of 2020 strip in public (alibaba/taobao)
must be a rare, or fail using product?


#34

I would agree that they could do something closer to 400led/m but once you get the price from them for 100led/M and 200led/M you may understand why they aren’t commonly available and maybe why they haven’t considered a 400led/M strip.

You can get 25 - 100mm x 4mm strips out of a single panelized 100mm x 100mm x 0.6mm pcb and you can get 5 of those for $5 so that makes the pcb only $0.62/M and with the LEDs being $0.07/ea when you buy 1000 you could make a 400led/M strip for under $29.00/M. I understand that flexible PCBs cost a lot more but for my use 100mm long semi-flexible strips would be fine and sooooooo much cheaper.


#35

I wonder where to buy those I.A. LEDs at 0.07 each
or maybe just need quantity, dont they?

good to hear that, since POV product uses a lot.


#36

http://www.rgbledcolor.com/sale-8995234-full-color-apa102-2020-integrated-rgb-led.html

It says right on the page.


#37

at first it looks messed, but a turn fixed it.
suddenly I found that it’s corrupted by the old version.

the sketch looks fine and should work.
still need tries to makes sure the unbalanced pads dont drag the chip too much when soldering


#38

I hadn’t even thought of that as an issue :slight_smile: , I can switch back to the original footprint if it turns out to be a problem. In fact if you like I can create a new part with the original footprint so you can try both on a single board if you like.

Peter


#39

@vanepp @AlanChen

I would consider making the one last long pad in the middle short like the others. It will still allow the better routing while being balanced. By better routing I mean either routing under the chip or even putting a single via under it for the ground as well as pulling heat from the chip better than without the via.


#40

Perhaps both? One with the recommended foot print, and one with both extra legs shortened for better routing as you say. Then the user can choose which suits their application best.

Peter


#41

while chaining the new chips, i found it better(?) to use copper text underline “_” :champagne:
for some wire, that intensively connected to the same destination (power lines),

also, copy_rename with correct number irritates :sob:
the copy part with auto-sorted numbers are not sorted
better hide the part-label and write my own every few chips.