Soldering / circuit / LED strip advice

Forums › Home Workshop DIY › Soldering / circuit / LED strip advice

mdf

3566 posts

Uber Geek
+1 received by user: 1519

Trusted

#248281 18-Mar-2019 12:48

I am playing around with RGB LED strips. Right now, I'm trying to make a seven-segment digit (ultimate goal is something like this but I'm a long way off that right now). This is very much learning/prototyping stage. I'm using WS2812B strips from ebay, so only three wires (+5V, GND and Data). I checked before I cut and everything lit up as it should.

The strips are cut to size, laid out, and soldered back together in a figure 8. But it's my first time soldering something this fine solo, and I've clearly done something wrong as only a couple of segments light up. I've managed to probe with a multi meter and can see I have (or at least seem to have) 5V all around the circuit/figure 8. I think the issue is therefore with the data wire between segment 2 (working) and segment 3 (not working).

I've done the obvious (or at least, I think the obvious) and reheated and disconnected that connection, cleaned the pads with solder wick, tested the wire, and resoldered but still no dice.

Would appreciate any bright ideas or guidance to figure out what has gone wrong. I might have burned out the pad? Shorted something out when I tinned the pads? Is there a way of figuring out whether the issue is at the exit or entrance of a segment?

My soldering isn't great, but there aren't obviously great blobs of solder shorting out the pads.

1 | 2

gordy7
2001 posts

Uber Geek
+1 received by user: 505

ID Verified

Lifetime subscriber

#2200761 18-Mar-2019 13:10

Do a continuity check - ohmeter check - from your wire and long each LED strip at the pads to make sure the strips copper tracks are still intact.

I have helped out someone with a similar LED strip. Too much heat with the soldering iron could make the pads lift or move on the substrate and break the copper connection between the pad and the track.

Gordy

My first ever AM radio network connection was with a 1MHz AM crystal(OA91) radio receiver.

mdf

3566 posts

Uber Geek
+1 received by user: 1519

Trusted

#2200769 18-Mar-2019 13:19

Gordy7:

Do a continuity check - ohmeter check - from your wire and long each LED strip at the pads to make sure the strips copper tracks are still intact.

I have helped out someone with a similar LED strip. Too much heat with the soldering iron could make the pads lift or move on the substrate and break the copper connection between the pad and the track.

Thanks. I've just rechecked and don't get any continuity at all on any of the data pads, even the ones I know are working. Each of LEDs is individually addressable via a little circuit and I assume something about that prevents checking continuity.

Gordy7

gordy7
2001 posts

Uber Geek
+1 received by user: 505

ID Verified

Lifetime subscriber

#2200776 18-Mar-2019 13:35

It is possible the manufacturer has given all the pads a clear coating that you will have to scratch though.

Gordy

My first ever AM radio network connection was with a 1MHz AM crystal(OA91) radio receiver.

1101

3141 posts

Uber Geek
+1 received by user: 1143

#2200789 18-Mar-2019 13:52

throw in the bin and start again . Honestly the easiest solution.
Make sure you test the complete strip before cutting it up.

If you want to fix it, remove the problem section from the fig 8 & test it separately.
Multimeter Continuity test, on switchable diode circuit & associated components.... :-)

Try tracing +ve & -ve voltage through the circuit .
Then trace the switching pads through the circuit , that will depend on how they switch (how they are triggered ).

mdf

3566 posts

Uber Geek
+1 received by user: 1519

Trusted

#2200933 18-Mar-2019 17:42

Further investigations carried out.

Each of the segments (8 LEDs each) work fine on their own (and worked fine prior to cutting). +5V and GND definitely work as a continuous loop all the way around the circuit. If I connect Data In of each segement directly to the Arduino (i.e. bypass the previous segment), each LED will light up correctly. So I think that means that I've got an issue with the Data Out side of things. It would be quite the coincidence if I've managed to solder every +5V, GND and Data In correctly, but all the Data Outs incorrectly.

I suspect that I've soldered one or more badly enough to introduce noise on the circuit and it attenuates the data signal sufficiently to no longer work. Does that sound plausible?

Fred99

13684 posts

Uber Geek
+1 received by user: 10018

#2200978 18-Mar-2019 18:46

Or heat or voltage leakage from the soldering iron has fried the data chip on the individual LEDs. Could try earthing and heat sinking the chip by using an earthed clip of some kind across the contacts as you solder them. Cut out one of the still working sections of a borked strip and give it a try.

Lenovo

Shop now for Lenovo laptops and other devices (affiliate link).

richms

29097 posts

Uber Geek
+1 received by user: 10205

Trusted

Lifetime subscriber

#2200980 18-Mar-2019 18:50

There are fake WS2812 LEDs around - I got some in one lot and about 4 failed, just refused to pass data out their out pin.

Also flexible strips are terrible for the connections cracking to the LEDs. That is why there are the newer ws2813 LEDs that have a dual data wire to bypass a single failed connection.

Do you have an oscilloscope to probe the output of that strip for data?

Also check you are outputting enough data for that many LEDs. I was caught out by that with a multiplication mistake when putting some White/amber LED pixels into some shelving.

Richard rich.ms

mdf

3566 posts

Uber Geek
+1 received by user: 1519

Trusted

#2201079 18-Mar-2019 20:24

Cheers guys. It's possible I have completely fried something along the way, but I'm not quite ready to throw the whole thing out or re-cut. I'm reasonably confident all the individual chips/LEDs are working:

Everything worked as it should prior to cutting, so I don't think it is a chip DOA scenario.
I've checked the code on another (un-cut) strip so I'm confident that is working.
I can make each segment individually behave as expected when that segment is directly attached to the arduino.

It is just when they are all chained together I can't get beyond a segment or two.

I'll continue to experiment. Ideally I want to make six of these. Could be quite a lot of troubleshooting before I'm done. Youtube made it look so easy too!

geocom

597 posts

Ultimate Geek
+1 received by user: 143

Subscriber

#2201159 18-Mar-2019 21:11

The data pins are not commoned. They pass the data through the chip the first 24 bits are dropped and then the rest is sent to the next chip and the process continues till the end of the string, so a dead chip will stop at that point and not continue.

Check that all the connections are going from DO to DI.

Geoff E

richms

29097 posts

Uber Geek
+1 received by user: 10205

Trusted

Lifetime subscriber

#2201176 18-Mar-2019 21:23

pictures plse

Richard rich.ms

frankv

5705 posts

Uber Geek
+1 received by user: 3666

Lifetime subscriber

#2201177 18-Mar-2019 21:23

If you overheat a chip when soldering you will kill it. If the first chip in a series of WS2812s is dead, the rest of the strip won't light up.

Also, check that you don't have the dead strip wired up back to front... i.e. DO from one strip going to DO on the next.

You can't introduce noise by bad soldering. But you can have intermittent/failed connections. Try pressing down with your finger on the connections going into the failed section, or out of the last working section. With a bit of luck, it may burst into life.

Also check for solder bridges (i.e. zero resistance) between +5 or GND and the DI/DO lines.

Dyson

Shop now for Dyson appliances (affiliate link).

Tracer

343 posts

Ultimate Geek
+1 received by user: 151

#2201178 18-Mar-2019 21:23

You haven't got some of the strips backwards (i.e. DO connected to DO on the next strip)?

mdf

3566 posts

Uber Geek
+1 received by user: 1519

Trusted

#2201223 18-Mar-2019 22:07

Oh yes, was very alive to the possibility of being a complete bonehead and definitely checked data in/out twice before posting the first time!

Please excuse the recycled cardboard backing but this is what I'm trying to achieve:

Figure 8, 8 leds to a segment. In this image, data and power enter from the right (3 o'clock), go in a loop clockwise back to 3 o'clock, then turn more sharply for the middle segment before exiting to the left (9 o'clock).

As you can see from the above image, the first two segments (bottom right and bottom) light when connected to the Arduino and powered on. Stops at the junction between bottom and bottom left. Apologies, soldered from the bottom so it's still stuck down that way.

However, I can get all seven segments/all LEDs to light if connecting their data in pin directly to the Arduino:

This also works if I wire the data out pin of the previous strip directly to the Arduino, i.e. the data is getting "in" through my manky soldering, but not getting out from the data out pin. You can see in the middle left (9 o'clock again) one of the green data wires has been disconnected. If I attach that directly to the Arduino, it lights as it should.

It's not just the bottom/bottom left junction. If I light the bottom left segment by directly connecting to the arduino data pin, I then need to do the top right, top, top right etc. directly to the data pin too.

I've checked all the way around, and am still getting 5V between positive and ground at the exit point. I will need to check for solder bridges tomorrow in the better light.

Close up of my soldering. Some are better, some are worse:

And the code I'm using to test. Pretty basic:

[code]

#include <FastLED.h>

#define NUM_LEDS 56

#define DATA_PIN 3

#define LED_TYPE WS2812B

#define COLOUR_ORDER GRB

#define BRIGHTNESS 100

#define COLOUR Blue

CRGB leds[NUM_LEDS];

void setup() {

FastLED.addLeds<LED_TYPE, DATA_PIN, COLOUR_ORDER>(leds, NUM_LEDS);

FastLED.setBrightness(BRIGHTNESS);

FastLED.clear();

}

void loop() {

fill_solid(leds, NUM_LEDS, CRGB::COLOUR);

FastLED.show();

}

[/code]

I've confirmed that this will make the first 56 leds on another strip (not cut) light up blue as expected.

richms

29097 posts

Uber Geek
+1 received by user: 10205

Trusted

Lifetime subscriber

#2201256 18-Mar-2019 22:45

Hmm, ive never soldered to the back of the strips before. Possibly the plating thru on the data pins isnt that great and its not getting contact? perhaps try the front pads instead since that is the side that trace is actually on.

Richard rich.ms

frankv

5705 posts

Uber Geek
+1 received by user: 3666

Lifetime subscriber

#2201364 19-Mar-2019 09:08

However, I can get all seven segments/all LEDs to light if connecting their data in pin directly to the Arduino:

The default state for WS2812 LEDs is on (white). If they're all on, then there is NO data connection (or you've turned them all on). I guess this does verify that the LEDs themselves work and the power circuitry is good.

A thought... what version of Arduino are you using? If you have a 3.3V one (e.g. Nano), then it can't reliably drive WS2812s directly. You need a 5V (e.g. Uno) Arduino (or a level shifter).

It's not just the bottom/bottom left junction. If I light the bottom left segment by directly connecting to the arduino data pin, I then need to do the top right, top, top right etc. directly to the data pin too.

So, no matter where you connect the Arduino to your hardware, you always get just the first 16 LEDs lit? That sounds like a software problem to me. I'm not familiar with the fastLEDS library, but at first glance your code looks good. Is fastLEDS widely used? Maybe try the NeoPixel library.

But this

I've confirmed that this will make the first 56 leds on another strip (not cut) light up blue as expected.

sounds like hardware.

1 | 2