Known issues with SSCv3 and TM1804 square pixels?

[Steve Gase]

Building my 3rd coroflake, and having massive problems with flickering with the SSCv3.
A couple years ago I built 2 coroflakes with similar square modules, but they had 3 LEDs in each module.  With the SSCv3 I have no problems.  The new flake has similar square modules BUT they have 4 LEDs instead of 3.
Both old and new square modules have TM1804 chips.
 
Each flake has 50 modules -- so I shouldn't be coming up against a power draw problem (since the SSC should support 120 modules).
Using a SSCv3 with the new flake, I'm seeing a LOT of flickering using the test firmware.  It even seems to get stuck after one pass through the test sequence.
I used the same SSCv3 against the old coroflake and have no issue. 
I tried a Ray Wu decoder instead of the SSC and have no problems with the new flake.
 
Each flake has about 6' of wire going from the SSCv3 to the flake.
 
Questions:
Is 6-foot after the SSCv3 "pushing it"?
Will SSCv4 make any difference?
Has anyone test the 4-LED square modules SUCCESSFULLY with SSCv3?
 
I'm going to try some of my V1 and V2 kits to see if they behave any differently.

[keitha43]

zwiller had this issue here You are not allowed to view links. Register or Login he reverted to the 1st v3 firmware. You might try contacting him and see if it works for you. I have no problem with pixels. And during beta testing of the v4 SSC he didn't have the problem. You might reduce the distance to the first node to 2 feet to see if it makes a difference.

[rdebolt]

 
Each flake has about 6' of wire going from the SSCv3 to the flake.
 
Questions:
Is 6-foot after the SSCv3 "pushing it"?

Yes

Will SSCv4 make any difference?

Yes,  On certain strings I could go as high 10', but 6' worked on all my strings

Has anyone test the 4-LED square modules SUCCESSFULLY with SSCv3?

I have not
 

On my testing I had no flicker at all on very long slow fades using Xlights and LSP. When I got too far from the SSC to the first node I had lack of color control and random lights (rainbows) along with no lights at time. I got my longest distances with flex strips and the shortest with regular nodes.

[Steve Gase]

I shortened the lead wire and it works MUCH better.
With extra wire for the pigtails and the internal wire connecting to the pixels I probably had 7-8ft after the SSC to the first pixel.  Now, I'm about 3 feet and I think it might be reliable.

[rdebolt]

Glad to hear that it works now.

[pk]

Steve -

You should check (i.e. measure) the power you are using.  If each flake has 50 modules you have 200 LEDs.  If they are all turned on to make white, that is a lot of current!!

[Steve Gase]

Steve -

You should check (i.e. measure) the power you are using.  If each flake has 50 modules you have 200 LEDs.  If they are all turned on to make white, that is a lot of current!!

I assumed that the power would be a non-issue...  RJ originally qualified the 3-LED square pixels for a length of 120 nodes -- or 360 LEDs.


I am using 4-LED pixels for 50 nodes -- or 200 LEDs.

[tbone321]

More power being switched on and off equates to more induced noise on the signal lines. 

[combustionmark]

Using 4 led pixels might cause a problem. Even 4 led's in a dumb string.

More power being switched on and off equates to more induced noise on the signal lines.

The more led's that are on will draw more current, this will cause a voltage drop, even before the first pixel. Power = Voltage * current, there is only so much power on a pixel string, go over and you will have problems. May not be just noise on the signal line.

Warning my rambling follows!!

As I remember each led requires voltage and current to turn on. When I was making led strings I could only use up to 3 reds, 3 greens, or 5 reds on a 12 volt string.

I found that red led's had to have 2.0 volts to light up, green and blue had to have 3.3 volts.

Now if you have 4 green or blue the string would have to be 13.2 volts, 6 reds only 12 volts.

The strings would light up at 12 volt, just didn't light up evenly and some even flickered. I just thought that was part my fault for using cheep led's.

This has just kept me from using any pixels that uses 4 on 12.

Strings I used had 3 or 5 led's and a 100ohm resister, all in series, on 12 volt with no problems. The resistor has a voltage drop based on how much current the led's draw, and is sized to keep everything happy. As the string voltage drops, less current in the string, resistor's voltage drop decreases, led's dim.

So I stick with this when doing pixels. No more than 3 led's in a pixel on a 12 volt string.

YMMV

OK!!! You can hit me now......

AOR

[pk]

I am interpreting the table in the Wiki differently than you.  Maybe someone can set me straight.

[tbone321]

That really depends on how the node is configured.  If all of the LED's are connected in series and just using a resistor for current control then that may be a possibility but .... we are only talking about a few feet of wire.  If removing 3 feet of wire is correcting the issues, then I really doubt that voltage is the issue.  Also, if it was the added LED and a voltage drop, then removing wire from either side of the SSC should correct the issue but I bet that even removing many feet of wire from the connection between the hub and the controller will have no effect. 

We also need to remember that we are talking about two very different types of cable here.  The CAT5 being used between the hub and first controller is twisted pair and is highly resistant to induced noise, even over long lengths.  The wire between the controller and the string is in most cases 3 conductor ribbon cable which is highly susceptible to induced noise, especially between the conductors in the cable itself.  The longer this cable gets, the worse the noise gets and the more current being drawn also increases the noise being created.  This is why shortening this wire here is so effective and why these 4 LED strings are having issues. 

Here is an experiment to try.  I would put the long wire between the controller and the first node and see if the problem returns.  If it does (and it probably will) then I would insert a pull down resistor between the signal line and ground and see what happens.  I bet that the nodes will return back to normal operations.  470 ohms should do it and you probably could go a little lower but not so low as to overload the controller.  You also need to remember that there is a current limiting resistor on the signal output and you don't want your pull down resistor to be lower than that one.

[zwiller]

I ran some numbers with chrisatpsu using calcs based on the resistors of the module and we came up with 80 (4 led) modules per SSC as a safe figure.  Pic attached.  Maybe you techies can confirm this. 

I will offer additional comments AFTER Steve posts a picture of the new snowflake.   

[combustionmark]

WARNING More rambling follows!! Skip this one if you value your time.

Warning  my rambling follows!!

As I remember each led requires voltage and current to turn on. When I was making led strings I could only use up to 3 reds, 3 greens, or 5 reds on a 12 volt string. yack yack yack

I did goof a little, I was referring to dumb single color led's. I was only referring to the voltages required for a single color led, and not to a smart string. Inside a smart string pixel is at least 3 led's 1 red, 1 blue, and 1 green inside a single package. These are not in series. Some pixels have, what is called, 3 or 4 led's. In my ramblings this would have been 1 string of red, 1 string of blue, and 1 string of green, consisting of 3 or 4 led's in series. Smart string pixels calling out 4 led really have 12 leds.

OK if I have 12 volts how many blue led's can I light up? We will power a string of blue led's with 17.3mA from 12 volt power. 17.3mA X 12Volts = .2076 Watts. This will give us .2076 Watts of total power for a string of blue led's. Each blue led at 17.3mA will require 3.3 Volts, or .0571 Watts.

A string of 1 blue led's will require .0571 Watts of power, 1 X .0571 = .0571 This is less than our available power of .2076 Watts. OK. The extra power of .1505 Watts will be used to keep things happy.

A string of 3 blue led's will require .1713 Watts of power, 3 X .0571 = .1713 This is less than our available power of .2076 Watts. OK. The extra power of .0363 Watts will be used to keep things happy.

A string of 4 blue led's will require .2284 Watts of power, 4 X .0571 = .2284 This is more than our available power of .2076 Watts. No extra power. Problem. The string may act up. This means that you require more voltage or more current. Increasing the current will blow the led's. If you increased the voltage to a smart string, you may blow the pixels.

And yes these numbers are in the wiki. This is even the numbers used for the pixel squares. "But they don't match!" Yes they do. There are 3 colors in 3 led strings of 3 led's each, for each pixel. 17.3mA for red, 17.3mA for green, and 17.3 mA for blue. 17.3mA X 3 = 52mA. We have a match.

All of this is the same for the green led's. The red is a different story. Red will be ok at 4 led's.

The extra power from each string of led's keeps everything happy and stable.
1 led's per color string pixel has .1505+.1505+.1730=.4740 Watts, all 3 colors happy with 76.1% Extra power available.
3 led's per color string pixel has .0363+.0363+.1038=.1764 Watts, all 3 colors happy with 28.3% Extra power available.
4 led's per color string pixel has -.0208-.0208+.0692 =.0276 Watts, 2 colors may act up, only red happy, with 4.4% Extra power.

The simplest way I know to figure this all out is, power supply voltage divided by led required voltage = max leds in series.
example
12 volt / 3.3 volt = 3.6 led's max. I can't have a .6 led, so I round down. The 3 led's will require 9.9 volts, leaving 2.1 volts to keep things happy.

The wiki shows 52mA per pixel, 80 pixels max for a smart string 80 X 52mA = 4.16 amps for a full string. With all pixels on full white, there will be a voltage losses on the pixel string flat wire, in the smart string controller, the cat5 cable, hub, and power supply. If it is just .2 volt total loss, we have lost 1.7% of the power available for the pixels to use.

All of this is with pixels just full on, no dimming, no pixelnet data, no tm1804 data. The power gets marginal. Throw in pixelnet, and 1804 data, pixels dimming, now you have noise on a marginal power system. Non of this is good. When things are marginal problems are likely to show up. What fixed it 1 time may not the next. Every thing may be fine on the bench, during tests. do you really want to be troubleshooting things in the yard, at night, in the dark, and the weather.

Lets stick with things that work. For me that is 3 leds per color, per pixel on 12 volt power max. Or No more than 3 led's in a pixel on a 12 volt string.

Of course that's just me.

My apologies for this being such a long read, I hate trying to explain things without seeing feed back. I tried to keep it short the first time, and I fear this may make thing worse. I tried.

[combustionmark]

I ran some numbers with chrisatpsu using calcs based on the resistors of the module and we came up with 80 (4 led) modules per SSC as a safe figure.  Pic attached.  Maybe you techies can confirm this. 

Look at the mA. as long as the total mA is below 4.16A, it should be OK. If they work use them.

My ramblings are not meant to discourage any one, or tell them something is wrong. Just my thoughts on the subject. My only point is just to watch out for modules like that one.

I have not try that module, nor do I have any idea of what is in it. I like to stick with things known to work. I do cross the line and try other unknowns out.

There may be a different style of a led die that operates on a lower voltage, and 4 are the way to go. After all, whoever designed it knows a lot more about it than I do.

Sticking with the tm1804 is a definite plus.

[combustionmark]

Is this the module?
You are not allowed to view links. Register or Login

The page states 1.44 Watts

If this is it, it is 120mA per pixel, and 50 of them will be 6 amps. you may have to cut back to 34. If not do you have a link to it?