RGB strands with the 9813 IC

[gregf]

Hi All,
I'm new to this site so please go easy on me if I break some protocols. I'm interested in buying a large number of stranded pixels for a large full-wall matrix I'm planning to make. I'm hoping I can make a 16x32 display with 3" between the pixels diffused by some sort of frosted lighting panel typically used for ceiling CFLs. My main question is whether the 9813 ICs as available here:


http://www.aliexpress.com/fm-store/701799/209889132-428214395/led-pixel-module-9813IC-DC5V-input-IP68-50pcs-a-string.html


are a good choice for this purpose or whether I should stick with the WS2801 IC? I've gleaned from reading many posts here that the WS2801 pixels have advantages like 8bit color, constant current, and pretty fast bus speeds (the ws2801 strip I have allows 2Mhz), but they do require power injection every 50 pixels which shouldn't be a problem in my matrix layout. The 9813 seems to be 8bit as well but otherwise I can't seem to find much information except that it's cheaper. So that's really why I'm wondering how it compares.


Any other feedback on my design or in general is welcome.


Thanks!


Greg

 

[kane]

Some of the others will come back with more of the technical differences, but I suppose one key question - what are you going to drive the pixels with?

The gear that most of us are using to run pixels (pixad8, p12r from j1sys or the e680/e681 from sandevices) don't support the 9813 - so you would have to find something else that can drive it (which may be something you've already thought of!)

Cheers

Kane

 

[gregf]

Good question. I'm doing all Arduino stuff so hopefully I can find a library somewhere that helps or I'll have to decipher the datasheet.


-Greg

 

[Mike]

From what i have found so far on this chip:

They are 8 bit
Constant Current
There are 2 variations of the Chip , LPD9813 and WS9813
I seen somewhere they are 1.5mhz, dont know what variation that was or if its both types

Email Ray and ask for the data sheet, he might be able to help out somewhat

 

[gregf]

Thanks, that helps alot. I've emailed Ray but haven't heard back yet. I found a link to this description of the protocol, which appears relevant: http://www.seeedstudio.com/wiki/Grove_-_Chainable_RGB_LED

 

[David_AVD]

The 9813 uses a protocol that's a mix between the 2801 and 6803 chips. Here's a thread I started a while ago:

http://auschristmaslighting.com/forums/index.php/topic,1261.msg10378.html#msg10378

 

[random]

Similar protocol except the colours are in a different order, and 8 bit rather than 5 bit.


One thing to bear in mind with this chip is that you have additional check data - this allows the chip to detect corruption however not enough to correct for it. I'm not sure whether it would drop that pixel worth of data and then be ready to receive the rest of the string or whether it would ignore all further data until a reset sequence gets set. If it's the latter it would be useful in preventing corrupt data being sent to the lights but you will end up with a glitch regardless.


The downside is that the check bits create a 33% overhead in the protocol, as it's redundant data. WS2801 is more efficient at the expense of error detection. I have no idea what the maximum clock rate on a given length of lead-in is, for each chip, but I don't really see a point in detecting errors you can't correct for. In the middle of a string the pixels are close together anyway so there won't be any errors, it's just the lead-in you have to worry about and I'd take the more efficient WS2801 protocol in that case.

 

[jstjohnz]

Some of the others will come back with more of the technical differences, but I suppose one key question - what are you going to drive the pixels with?

The gear that most of us are using to run pixels (pixad8, p12r from j1sys or the e680/e681 from sandevices) don't support the 9813 - so you would have to find something else that can drive it (which may be something you've already thought of!)

Cheers

Kane

I have added code to support the LPD880x and P981x series pixels on the E68x controllers. The 880x is tested and working, but I don't have any 981x pixels on hand to test that code yet.

 

[fasteddy]

Thats great news giving everyone extra choice in pixels by supporting these chips