RGB Pixel and Decoder Testing

ThaiWay

1500 C9's are not green
Joined
May 10, 2010
Messages
98
Location
Bangkok
I received a string of 12 RGB Nodes from RightLEDs and an LT-DMX-6803 DMX512 Decoder from LTech today. These are samples I requested. They were shipped DHL Saturday and delivered this (Monday) morning.

I forgot to put my ruler in the picture, but the little decoder is 10.5 cm wide (a little more than 4".) Fortunately two 3-pin XLR connectors were included, and the first thing I did was to make cable adapters for XLR > RJ-45 (in the DLA Wiki thanks to Phil http://diylightanimation.com/wiki/index.php?title=Enttec_Pro_Connections )

Then I connected my Lynx Dongle to the DMX IN port and connected the DMX OUT to an Original Lynx I had on the bench. I verified that the decoder passed DMX with no additional connections.

RGB Setup and Test 001 resize.png

The Nodes themselves (part no. RFK12-1RGBD) are 33mm top to bottom and are encased in a solid silicone cylinder. The "bulb" end also has some silicone sealant around its base, as do the two 4-wire leads at the bottom. The leads are 24AWG ribbon. The specs say that the nodes are "waterproof" with an IP68 rating. (more: http://en.wikipedia.org/wiki/IP_Code ) I don't think I would want to try a submersion test until I'm ready to potentially waste these samples. Nor would I want to put much of any strain on the leads themselves; they can easily be pulled apart at the bottom of the node.

RGB Setup and Test 003 resize.png

Next I pulled out the Computer PSU I use for MR16's and made a +5VDC/GND connection (easy with Phil's Tiger ATX PSU adapter board) to power the decoder and the nodes. Yes, these nodes are 5VDC. The 4-wire Nodes connect to the decoder as +5 to Red, GND to Blu, Data to Yel, and Clock to Grn.

Then I set the decoder's start channel to 1, easy by flipping a dip switch. Started Vixen, 512ch Test Sequence. Turned on power supply, nodes flashed briefly, then off. Then chose various channels with Test Channel Function from 1-36 (12 Nodes X 3 ch) and verified that the decoder did indeed cascade up the individual Pixel addresses from the start channel of 1 on the first node in the string. Oila!

RGB Setup and Test 005 resize.png

The Nodes are quite bright. There is a half-wave 110V C6 string for comparison in the top of the picture. The viewing angle is high also, probably close to 120deg. They've been on for an hour at least, and don't feel the least bit warm to the touch. Still have to start up LSP and configure the decoder as a controller and do some RGB testing, but it looks very promising.

Some sample 12V RGB pancakes with three 5050's in each module also came in the package, and I'll test them too.

John[/attach]
 

fasteddy

I have C.L.A.P
Global moderator
Joined
Apr 26, 2010
Messages
6,645
Location
Albion Park NSW
Looks very good, great review, So the DMX controller appears to be the same 360 channel controller that Ray Wu sells.
The pixel lights also look similar to what Ray sells.
Do you have a price on this gear from these suppliers?
 

AussiePhil

Dedicated elf
Administrator
Joined
Jun 20, 2009
Messages
1,386
Location
Canberra, ACT, Australia
(easy with Phil's Tiger ATX PSU adapter board)
Wonder if i should revive this little board?

Interesting and promising, the immediate comparison that i would draw with JEC/MPH pixels is the use of coloured ribbon cable that may not be UV stable and the use of silicon to seal the wire end.
The JEC/MPH pixels have a epoxy resin filled silicon tube, the epoxy resin actually bounds with the plastic wire covering ensuring long term water proofing.
Standard silicon sealant can seperate away from the wire and seep water into the electronics.

Cheers
Phil
 

fasteddy

I have C.L.A.P
Global moderator
Joined
Apr 26, 2010
Messages
6,645
Location
Albion Park NSW
AussiePhil said:
Interesting and promising, the immediate comparison that i would draw with JEC/MPH pixels is the use of coloured ribbon cable that may not be UV stable and the use of silicon to seal the wire end.
The JEC/MPH pixels have a epoxy resin filled silicon tube, the epoxy resin actually bounds with the plastic wire covering ensuring long term water proofing.
Standard silicon sealant can separate away from the wire and seep water into the electronics.

Cheers
Phil
I wouldn't expect the same quality as the MPH/JEC pixels, you pay for what you get, I would tend to think the the MPH/JEC pixels are of the commercial quality standard and the Ray Wus would be more of the basic quality that may need some extra silicon to ensure they stay sealed.
 

ThaiWay

1500 C9's are not green
Joined
May 10, 2010
Messages
98
Location
Bangkok
Eddy... I was more interested in wrangling free goods for testing out of the manufacturers than in haggling over price. I suspect for small quantities Ray's prices include a bit of margin for him.

Phil... that little board is nice. I just wish the LED would light. And you saw the MPH/JEC gear first hand, so you know the kind of quality there. Sure, you get what you pay for. Yes, the silicone seal job is one of the first things I checked. It's iffy at best

Perhaps someone should deal with the Node factory direct and try to negotiate better quality and factory pricing to make a more DIY-friendly product available.

Sure is fun!
John
 

AussiePhil

Dedicated elf
Administrator
Joined
Jun 20, 2009
Messages
1,386
Location
Canberra, ACT, Australia
ThaiWay said:
Phil... that little board is nice. I just wish the LED would light. And you saw the MPH/JEC gear first hand, so you know the kind of quality there. Sure, you get what you pay for. Yes, the silicone seal job is one of the first things I checked. It's iffy at best
Sure is fun!
John
So John, turn the LED around, works on my board :)

Yeah the epoxy is a positive, but if we find the correct silicon that will bind to the plastic around the wire we have an alternative.

Phil
 

ThaiWay

1500 C9's are not green
Joined
May 10, 2010
Messages
98
Location
Bangkok
Uhhh... Shenzehn, we have a problem.

Simple transitions in LSP exhibit a terrible flickering of the Nodes with this Decoder. So in Test Controller Hardware, everything looks good at Full On Level 255, but from 245 down to Full Off at Level 11 the flicker is very noticeable. Same result using Test Channels in Vixen. And the flickering is progressive. With all 12 Nodes at Level 255 (for any color or mix) and then gradually decreasing the Level, Node #1 starts flickering. Lower the level more and #2 also starts flickering. More low and #3 starts too, progressively down the line. The problem is not LSP nor Vixen, because dimming on the controller after the decoder functions normally when fading up or down.

One of the preset functions on the decoder, set by a dipswitch: SWITCH9 IS ON: 7 COLOR SMOOTH (8 speed steps) fades very nicely through seven color transitions RGBYPCW with no flicker.

I have no clue what to do next.
 

dmoore

Senior elf
Joined
Apr 26, 2010
Messages
586
Location
Houston, TX, USA
AussiePhil said:
Yeah the epoxy is a positive, but if we find the correct silicon that will bind to the plastic around the wire we have an alternative.
Here is a close-up photo of the same pixel strings I tested: http://diylightanimation.com/index.php?action=dlattach;topic=2842.0;attach=2916;image

I'll agree that the very back of the pixels where the wire comes in isn't 100%

THe MPH nodes are close to the same design, just that they have seperated the wires at the base which allows for a better seal:
http://response-box.com/rgb/wp-content/uploads/2010/04/IMG_7931.jpg
 

dmoore

Senior elf
Joined
Apr 26, 2010
Messages
586
Location
Houston, TX, USA
ThaiWay said:
I have no clue what to do next.
I'd email Ray Wu (alpha_bill@yahoo.cn) and tell him what you found. He can then translate it (he speaks and writes fairly good english) and then send it to the vendor (he seems to know most of these companies). See if he can get them to troubleshoot the problem.

This is really promising - when compared against the othre current pixel based solutions, it "looks" like a winner on cost:

* LOR CCR: $250 for 50 pixels = $5 per pixel (3x 5050s)
* d-Light FireFli (non-DMX): $170 for 16 pixels = $10.50 per pixel (3x 5050s with dome)
* Generic 6803 pixels: .54 x 100 ($54) + $40 (controller) $94 = $.94 per pixel (1x10mm LED)
 

ThaiWay

1500 C9's are not green
Joined
May 10, 2010
Messages
98
Location
Bangkok
David, thanks for your input. Your posts and emails have been a huge motivation for me to get involved with RGB in the first place. Your picture of the Nodes is much clearer than mine. You can see that the MPH Nodes are of a much better quality build.

Another comparison:
  • MPH/JEC Nodes: $389 (4x42 pixels=168) + $185 (T3 Controller) = $574 = $3.42/pixel

That's a big motivation for me to explore this promising option.

We did some troubleshooting on the decoder last night in chat. Changed the DMX IN/OUT connection; tried a sequence instead of test channels; pulled the pic and inspected all the usual suspects, all to no avail. Junk DMX code was suggested as the culprit. Waiting for others to receive their decoders and test.

Inside LT6803 resize.png

Funny... I just chatted on MSN with Lydia at LTech:
"lydia says:
using the "FUN" setting #9 everything looks good
Friendof says:
correct
lydia says:
when the decoder connect the console,the 6803 driving IC just at level 32
Friendof says:
i connect the decoder to my PC through an ENTEC DMX PRO
lydia says:
the chip LPD6803 itself just at level 32
Friendof says:
i don't know what you mean "at level 32"
lydia says:
the chip itself decide it
Friendof says:
the LPD6803 has the capability of 256 levels, right?
from 0 to 255
lydia says:
no.has the capability of 32 levels
when connect the console
sorry.i must go now"

... and she's off to some Trade Show until Sunday.

More to come.

John[/attach]
 

AussiePhil

Dedicated elf
Administrator
Joined
Jun 20, 2009
Messages
1,386
Location
Canberra, ACT, Australia
The 32 levels makes sense

Each chip receives a 16 bit value based on the data sheet and this would split 3 x 5 bit plus a spare bit, a 5 bit binary = 32 step values, this = 32,000 colours.

Phil
 

ThaiWay

1500 C9's are not green
Joined
May 10, 2010
Messages
98
Location
Bangkok
In the course of chatting with another sales manager at Carmbo Lighting, I received this picture which illustrates the difference in waterproofing David noted:
Waterproof Pixels Carmbo.png

He also mentioned that his factory could not provide the black color lead type "because too many mistake in soldering" LOL
 

dmoore

Senior elf
Joined
Apr 26, 2010
Messages
586
Location
Houston, TX, USA
AussiePhil said:
The 32 levels makes sense

Each chip receives a 16 bit value based on the data sheet and this would split 3 x 5 bit plus a spare bit, a 5 bit binary = 32 step values, this = 32,000 colours.

Phil
So does this mean that if he flips it around and goes from 0 to 31 for the intensity, that it should work? 32 levels is still better than the poor fading of most LED strings hanging off an A/C controller.
 

dmoore

Senior elf
Joined
Apr 26, 2010
Messages
586
Location
Houston, TX, USA
In the course of chatting with another sales manager at Carmbo Lighting, I received this picture which illustrates the difference in waterproofing David noted:
View attachment 75

He also mentioned that his factory could not provide the black color lead type "because too many mistake in soldering" LOL

That's it! Would it kill them to seperate the wires at the base when the mold them? The wires are soldered on each side so it's possible.[/attach][/quote]
 
Top