I deal with these guys http://www.tyrialight.com directly, on large orders for my instalation work. I have asked about this specific issue, but i have never nutted out a resolution. They said that the strips can be laid on a thicker base but the cost then goes up(by about 10%) and they are much less flexible(thats not a big deal). I have ordered some test strips from them which im waiting on that are made on thia thicker base(but i only have a 1/2 meter sample so testing isnt going to show anything). Most of my projects are over 1500 x apa102 pixels and i can confirm the maximum data rate i can get is about 6-7mhz
@Leon_Yuhanov I do not know if you are setup for this but you can probably just run a somewhat high DC current through and measure DC volt drops on both strips of similar length and see if you note a difference.
strip base thickness may or may not imply thicker copper which is what would help.
@JP_Roy this issue is not about power delivery. Full stop!
@Leon_Yuhanov Cool…that is good info. Thanks for confirming the speed rates. I’ll keep those guys bookmarked for future orders (I usually get mine from Ali Express)
@Leon_Yuhanov I am sorry but power delivery is definitely a factor in any circuit design when the very highest speeds are desired. Full stop !!
@JP_Roy This issue has been discuss here very often. I have spent weeks testing and troubleshooting it with my projects. You are not wrong, power delivery does play a huge role in keeping high data rates for the APA102 pixel. However all my tests have ruled out power delivery as being the cause of this issue. I would love to se a solution where a strand of 1000 pixels cold be driven at or above the 10Mhz rate. I would LOVE to see you prove me wrong with an example, because it would make my work so much easier! Im currently in talks with my supplier to see if they can make a thicker base strip for testing
Hi again @Leon_Yuhanov , I agree that the issue has been brought up within this forum often but no real solution has emerged yet.
I traced back to Daniel Garcia comments where he mentions a setup of 768 APA102 LEDs run at 24Mhz…
https://plus.google.com/106349087113242416534/posts/Ffe52KxqSck
So it proves that the APA102 by itself is capable of much more than the 6~8 Mhz that you are getting using some strips.
In all your tests, you most likely never really tackled what seems to be the root cause, the strips themselves !
When I say improve power distribution, the addition of decoupling capacitors, distributed along the strip is, in my humble opinion, the most likely improvement to the strips power distribution that could help achieve higher clock rates over longer lengths. Beefing up the traces on the strips could help but, I think, to a lesser degree.
I would love, not to prove you wrong, but help you and others achieve higher clock rates and unfortunately, at this time all I have is a sound theory backed up by years of experience in digital designs !
@Leon_Yuhanov You already have contacts with a strip manufacturer, I would really appreciate it if you bring the specific issue of distributed decoupling capacitors with them. They are present in the WS2812b strips even at 144/meter density. I understand that the routing of clk and data, ins and outs made the inclusion of decoupling caps problematic but I am guessing that a strip manufacturer that is able to produce a strip proven to tolerate longer lengths at higher speeds might benefit from that.
@JP_Roy That post you linked to, Dan says he has the LEDs on a board, not strips…this has to be a major reason. The APA chipset is a solid platform, the problem is directly related to strip manufacturing.
@Corbin_Dunn I know you say that power is NOT an issue but I am curious on what will happen if you vary the voltage levels to the strip. Does the APA need precision voltages for longer runs?
@JP_Roy Daniels test board is a physical board, its not made of strip, APreciate your input on this, the more points of attack the better
@Leon_Yuhanov He mentioned that he got that board from Aliexpress. Here’s a link to a similar board, if not the same …
http://www.aliexpress.com/item/P10-APA102-full-color-RGB-indoor-LED-Display-unit-panel-28-28pixels-280mm-280mm-Aluminum-plate/1998114424.html?spm=2114.30010308.3.59.34N92f&ws_ab_test=searchweb201556_6,searchweb201602_2_10034_10033_507_10032_10020_9912_10017_10005_10006_10021_10022_401_10018_10019,searchweb201603_9&btsid=ff3941ed-1708-41ab-a862-fef1ad38bb3e
This is actually lower density 100 LEDs/meter than the strip that has 144 LEDs/meter but I assume that the +5v and GND occupy almost full copper layers and have superior power distribution to that of strips.
@Jon_Burroughs I’m guessing that power isn’t an issue, based on two things: my actual project has power supplied at both ends of my LED strip section, and in my breadboard tests (with some other strips at 60/m) they have the same problem even when powered at both ends. I don’t have a way of testing varying voltages (I need a power supply – right now I use a 120v -> 5v supply on my breadboard).
@Corbin_Dunn Side note: That upgraded metal tube bender is awesome! Is that a pipe threader modified for the application? LOL Genius!
@Jon_Burroughs Thanks! I wish I thought of that idea, but they are hacks that “Swag Offroad” came up with…
@Corbin_Dunn do see that jitter directly on the Teensy outputs before the LEDs?