36mm Square 12V Digital RGB LED Pixels (Strand of 20) - WS28

EL Wire/Tape/Panels, LEDs, pixels and strips, LCDs and TFTs, etc products from Adafruit

Moderators: adafruit_support_bill, adafruit

Please be positive and constructive with your questions and comments.
Locked
Madmax81
 
Posts: 4
Joined: Fri Aug 23, 2013 4:40 pm

36mm Square 12V Digital RGB LED Pixels (Strand of 20) - WS28

Post by Madmax81 »

Hello, Adafruit. I have many problems with 36mm Square 12V Digital RGB LED Pixels (Strand of 20) I bought from you. It's very unhappy that I bought many of them. I tried to use private messaging but every link to technical support brings me to this forum, so I write it here.

1. In the description on the product page I saw that "The LEDs are constant-current driven", actually it's NOT. 4, 5, 6 pins of WS2801 are connected to ground, so current is limited only by resistors in series with each LED pairs. It's a big trouble for me: about one half of my power supply's power is wasted in heat and I have no guarantee that LED current is stable over temperature, time and power supply voltage. Stabilizing of LED current is actually needed in my application: I need to have much longer than 20 pixel strips, I need 80 in one strip.

2. In the description on the product page I saw that "Each strand has two JST SM 3-pin connectors so you can connect multiple strands in a row, as many as you wish, just watch for how much current they want." Ha-ha-ha. If I connect only TWO strands and switch all LEDs to maximum I measure voltage at the beginning an at the end of the strand. The voltage drop across 2 strands was 2.2 volts, so last LED was not as bright as the first, color is also different because LEDs current is not stabilized. Measurement of power lines resistance gives me poor result: summary resistance of positive and ground wire is 0.29 Ohms per meter of the strand, or 0.67 Ohms per 20 pixels. Too much for strand which can draw 2.4 A at 12V, how do you think?

3. In each pixel in the strand there are small capacitors on data inputs and outputs. I cannot use data clock frequency declared in WS2801 datasheet: if I try to use frequency higher than 1.5 MHz the strand don't work correctly. I didn't find any information about it in the product description page.

4. Now I have 4 strands: 3 from one batch and 1 from an other. The LEDs in those batches are connected with differences: in first batch pins 10, 9, 8 of WS2801 IC are connected to R, G, B LEDs respectively, in second - to B, R, G respectively. I wasted some time to find it and ruined two pixels, one from each batch, before I ruined pixels I tried to find a mistake in programming with no result, heh. Also I find that current-limiting resistors in those two batches are slightly different and colors are different too.
Resistors between positive power rail and Zener diode are different in those two batches too: 510 Ohms and 2.0 kiloOhms, so current consumption with zero brightness differs in 4 times!!

So, I want to have pixel strands which work exactly as described: with current stabilization and all the pixels should work equal. Is it possible? If not, it's a subject to refund.

User avatar
adafruit_support_rick
 
Posts: 35092
Joined: Tue Mar 15, 2011 11:42 am

Re: 36mm Square 12V Digital RGB LED Pixels (Strand of 20) -

Post by adafruit_support_rick »

Please email [email protected] with a link to this thread to arrange a refund.

User avatar
adafruit_support_mike
 
Posts: 67454
Joined: Thu Feb 11, 2010 2:51 pm

Re: 36mm Square 12V Digital RGB LED Pixels (Strand of 20) -

Post by adafruit_support_mike »

WRT point 1: while your statement about resistors on pins 4, 5, and 6 is technically correct, I'm not sure you understand how a constant-current driver works.

If you look at the diagram on page 4/15 of the WS2801 datasheet: http://www.adafruit.com/datasheets/WS2801.pdf , you'll see that those pins have the labels RFB, GFB, and BFB, where the 'FB' stands for 'feedback'. The LEDs connect to pins 10, 9, and 8, and current flows from those through op-amp-controlled mosfets to the feedback pins. Each op amp receives its reference voltage from the WS2801's internal bandgap reference, and adjusts the mosfet to make the voltage at the xFB pin match the reference voltage. The table on page 2/15 lists the "Constant Current Feedback Voltage Vxfb" at 0.6v. Using a 20 ohm resistor in the feedback position puts 0.6v across 20 ohms, which requires 30mA of current.

Yes, there's a resistor in the path, but the voltage across the resistor has no correlation to the supply voltage or the LED's forward voltage.

WRT point 2: Line losses for power are common to all power distribution systems. The text on the product page means you don't need a separate controller for each strand of pixels. Whenever powering multiple strands of LED strips, we suggest putting a power connection between each pair of strands. The best possible solution is to have a separate power supply for each strip with isolated VCC rails for each strip, and the VCC connection tapped into the center of the strand.

WRT point 3: Any limits the capacitors impose on the signal rate will be negligible compared to the limits imposed by the cabling. The chip may be capable of 25MHz, but passing megahertz signals through a long wire requires special techniques.. twisted pairs, impedance matching, etc.

The Adafruit library that drives the LED Pixels sets the SPI rate at 1MHz, and contains a comment to the effect that going above 1MHz causes flicker. https://github.com/adafruit/Adafruit-WS ... WS2801.cpp , line 112. Granted that isn't easy to find, but we don't usually get questions about the data rate.

WRT point 4: the zener diodes have no effect on current consumption through the LEDs. The zener acts as a simple voltage regulator for the WS2801, as shown on page 8/15 of the datasheet. The WS2801 uses about 1.5mA internally, and the zener needs at least 1mA of reverse current to hit its target voltage.

The choice of resistor represents a tradeoff between efficency and flexibility. Assuming a 5.1v zener:

- a 2000 ohm resistor will allow about 3.5mA to reach the WS2801 and zener when the supply voltage is 12v.
- a 510 ohm resistor will allow about 3.5mA to reach the WS2801 and zener when the supply voltage is 7v.

Using a 2000 ohm resistor with a 7v supply would allow less than 1mA to reach the WS2801 and zener, so that configuration is more efficient but less flexible. Using a 510 ohm resistor with a 12v supply would allow about 13mA to reach the WS2801 and zener, about 10mA of which would count as overhead. That configuration is less power efficient, but more flexible.

Madmax81
 
Posts: 4
Joined: Fri Aug 23, 2013 4:40 pm

Re: 36mm Square 12V Digital RGB LED Pixels (Strand of 20) -

Post by Madmax81 »

Point 1.

Yes, my statement about resistors is correct, you can take a pixel and see that there is NO any current feedback resistors. The circuit inside these pixels is deadly simplified, as Chinese like to do.

Image

Yes, you can be sure that I know how constant-current driver consisting of operational amplifier and MOSFET works: I designed circuits for driving medical lasers. It's very simple until you need to change the current several million times each second.

In Adafruit pixels voltage across resistors connected in series with LEDs is depended on supply voltage.

Point 2. Ok, but if we use constant-current driver the problem of voltage drop will not occure until voltage drop reaches (MOSFET+current sense resistor) voltage drop. In this circuit it's around 5-6 volts (depends on LED color). In actual Adafruit pixels I and my friends see color differece between first and last pixel of 20-pixel strand when voltage drop between first and last pixel is 0.6-0.8 volts. If we use constant-current driver we will not see any difference with voltage drop of 5 volts! Feel the difference! [Here might be long talk about thermal stability and so on...]


Point 3. I used wire about 10 cm (4 inches) long. Only wire which comes with adafruit pixels, no extra wires soldered. If maximum data bus frequency is limited the pixel supplier must note it instead of giving datasheet for IC used, agree? And I will be glad to see 100 Ohms resistor inside the pixel for twisted pair impedance matching, where is it?

Point 4. Yes, I know. I only want to buy pixels that are equal. And I don't want to waste power of my power supplies. Possibility of powering of those pixels from 7V power supply is nice only if constant-current driver works properly, otherwise it's only waste of power. I want to say, for your information, that there are many voltage regulator ICs that draws only few microamperes and it's very c_h_e_a_p. MCP1702 is good example.

adafruit
 
Posts: 12151
Joined: Thu Apr 06, 2006 4:21 pm

Re: 36mm Square 12V Digital RGB LED Pixels (Strand of 20) -

Post by adafruit »

please re-read the comments from our support staff, you are not correct. if you'd like a refund for the items, please return them within 30 days and contact [email protected]

Locked
Please be positive and constructive with your questions and comments.

Return to “Glowy things (LCD, LED, TFT, EL) purchased at Adafruit”