Parallel power to 30MM dots, ground loop?

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
User avatar
uckticoonox
 
Posts: 22
Joined: Sat Dec 21, 2013 12:08 am

Parallel power to 30MM dots, ground loop?

Post by uckticoonox »

I have a long strip of 30mm Dots (https://www.adafruit.com/products/1547), about 22 strips, 440 lights total. If I connect power in parallel throughout the strip, the colors flicker like crazy after the first couple strips. And the more places I connect power, the more it flickers. If I only power it from the start, the flickering goes away (but, of course, the lights at the end are very dim).

Is hooking power in in parallel causing a ground loop? Or some other sort of problem? Any ideas how to resolve this?

User avatar
adafruit_support_bill
 
Posts: 88098
Joined: Sat Feb 07, 2009 10:11 am

Re: Parallel power to 30MM dots, ground loop?

Post by adafruit_support_bill »

When connecting multiple supplies to a strip, you need to isolate the sections. Connect just GND and signal wires between the sections. Keep the 12v bus sections isolated so that the power supplies don't fight each other.

User avatar
uckticoonox
 
Posts: 22
Joined: Sat Dec 21, 2013 12:08 am

Re: Parallel power to 30MM dots, ground loop?

Post by uckticoonox »

In this case, I'm hooking a single power supply in at multiple points throughout the strip. Should I still isolate the 12v line in each section?

User avatar
adafruit_support_bill
 
Posts: 88098
Joined: Sat Feb 07, 2009 10:11 am

Re: Parallel power to 30MM dots, ground loop?

Post by adafruit_support_bill »

If it is a single supply with multiple feeds, you shouldn''t need to isolate anything. Not sure what would be causing the flicker then. If you post photos or some more details about your installation, we might find a clue.

User avatar
uckticoonox
 
Posts: 22
Joined: Sat Dec 21, 2013 12:08 am

Re: Parallel power to 30MM dots, ground loop?

Post by uckticoonox »

Alright, here goes... Here's an idea of the project, lining a gazebo with lights:
Image

The box that powers this has one 12v power input, a bunch of electronics (I'll get to that), and several outputs:
Image

There's a single 12v 5A power supply going into the box.
One of the 4-wire cables coming out of the box (https://www.adafruit.com/products/744) has 3 lines: Ground, SPI Clock, SPI Data.
The other 4-wire cable has 2 lines: +12v and Ground.

The connector from the box that says "Table" hooks into ~450 Neopixel LEDs, powered by an independent 5v 10A power supply. The only 2 lines hooked up to that connector are ground and the signal. The 5v 10A power supply hooks +5v and ground into the Neopixels, at the table. All the Neopixels on the table are working fine. No flickering.

The 4 DC outputs carry power to various points throughout the gazebo, and they hook in as shown here:

Image

Normally, each string of LEDs has 2 4-pin connectors: signal to signal, power to power. In this case, I hook the incoming power line in the middle of the power connection, so it's supplying power in parallel to what's running down the line.

Now, in the box:

Image

The 12V power in connects to the bus bar on the right. From there, it goes to the 4 DC outputs, and up to the LM2596S at the top. That converts the 12v to 5v, which is then passed along to the Raspberry Pi in the lower left.

The signal going to the 5v Neopixels goes from the Pi to the 3.3v to 5v level shifter (http://www.digikey.com/short/t8rvc7), then out to the table.

At the moment, the SPI signal for the LED Dots is going from the Pi's USB to the FT232H breakout board up top (https://www.adafruit.com/products/2264), to convert to SPI. Then it goes out to the lights. The SPI signal wire gets its ground wire from the 3.3v to 5v board, which gets it from one of the pins on the Pi.

Initially, I was sending the SPI signal from the Raspberry Pi's GPIO pins through the level shifter (3.3v to 5v, in the picture), then directly to the lights. The FT232H was an attempt to get rid of its occasional flickering, in case the SPI signal directly from the Pi's pins was having problems.

If the SPI signal goes through the USB -> FT232H -> Dots, and I connect power in parallel, then the Dots after about the second or third strip start flickering like crazy. It's almost like I'm sending random data to them, but it sometimes shows the right thing. If I only power the strip from the beginning, they don't go crazy, but they'll flicker briefly ever 5-15 seconds.
If the SPI signal goes directly from the Pi to the level shifter to the Dots, it's mostly fine, but every 5-15 seconds, it'll flicker briefly. Again, only the lights after the 2nd or 3rd strip flicker, and the first ones are fine.

I know this is a ton of info. I really appreciate any suggestions you may have. And I'm happy to answer questions about how everything's hooked up.

Thanks for your help.

User avatar
adafruit_support_bill
 
Posts: 88098
Joined: Sat Feb 07, 2009 10:11 am

Re: Parallel power to 30MM dots, ground loop?

Post by adafruit_support_bill »

If the SPI signal goes directly from the Pi to the level shifter to the Dots, it's mostly fine, but every 5-15 seconds, it'll flicker briefly. Again, only the lights after the 2nd or 3rd strip flicker, and the first ones are fine.
Does the power wiring make any difference in the Pi->level-shifter case?

User avatar
uckticoonox
 
Posts: 22
Joined: Sat Dec 21, 2013 12:08 am

Re: Parallel power to 30MM dots, ground loop?

Post by uckticoonox »

Hooking power in parallel only affects the brightness, not the occasional flicker, in that case.

User avatar
adafruit_support_bill
 
Posts: 88098
Joined: Sat Feb 07, 2009 10:11 am

Re: Parallel power to 30MM dots, ground loop?

Post by adafruit_support_bill »

So it sounds like you are better off with direct Pi-SPI via the level shifter. But you are still stuck with an intermittent flicker in the downstream strips. Not sure how the power feed situation would affect the flicker in either case. I referred this to some of the other engineers to see if they have any ideas.

User avatar
adafruit2
 
Posts: 22149
Joined: Fri Mar 11, 2005 7:36 pm

Re: Parallel power to 30MM dots, ground loop?

Post by adafruit2 »

ive seen similar before, possibly its because your signal ground is 'lifting' and eventually the signal isnt well referenced. a pocket oscillscope will help tons in verifying this.

User avatar
uckticoonox
 
Posts: 22
Joined: Sat Dec 21, 2013 12:08 am

Re: Parallel power to 30MM dots, ground loop?

Post by uckticoonox »

Good insight.

I went down the line watching the voltage on the ground line. By the very end of the strip, ground fluctuates up to about 0.8v compared to the ground at the start. Interesting, but that doesn't sound like enough to interfere with the signal, on its own.

However, when I switched over to the clock and data signals, I found something surprising. While the data line maintains 5v down to the end of the very last strip, the clock line slowly degrades. After strip 13, its max voltage is around 3.9v. By the time it reaches the end of the last strip, the clock signal is only showing around 2.8v. I'm surprised it's registering at all, if it's dropped that far.

But I'm not sure that explains why the flicker starts around the end of the 3rd-4th strip, though. Any other ideas, or any more details I could provide that would be helpful?

User avatar
adafruit_support_bill
 
Posts: 88098
Joined: Sat Feb 07, 2009 10:11 am

Re: Parallel power to 30MM dots, ground loop?

Post by adafruit_support_bill »

By the very end of the strip, ground fluctuates up to about 0.8v compared to the ground at the start. Interesting, but that doesn't sound like enough to interfere with the signal, on its own.
Not sure about the WS2801s, but that is more than enough to mess up a WS2811.
But I'm not sure that explains why the flicker starts around the end of the 3rd-4th strip, though. Any other ideas, or any more details I could provide that would be helpful?
It sounds like both the clock signal and the ground reference are degrading. Those effects would be additive. What is the situation at the 3rd-4th strip? What do the clock and data waveforms look like?

User avatar
uckticoonox
 
Posts: 22
Joined: Sat Dec 21, 2013 12:08 am

Re: Parallel power to 30MM dots, ground loop?

Post by uckticoonox »

Here's ground (compared to the power supply), clock, and data right at the start:
Image
Ground is between .02v and 0.08v.
Clock goes up to 4.7v.
Signal goes up to 5.2v.

Here's ground, clock, and data after 3 strips:
Image
Ground is between -0.2v and 0.3v.
Clock goes up to 5.0v.
Signal goes up to 5.6v.

Here's after 5:
Image
Ground is between -0.2v and 0.6v.
Clock goes up to 5.2v.
Signal goes up to 5.7v.

I had to pull things apart, and this is without power being hooked in parallel, so the degradation is more extreme than it would normally be... But it seems pretty clear the degradation in the ground line is a big part of the problem, and hooking power in more frequently seems to help. Unless you have any other recommendations, I'm going to pick up a bunch more power lines and hook it in more frequently, once every couple strips, so every strip has power on one end or the other.

User avatar
adafruit_support_bill
 
Posts: 88098
Joined: Sat Feb 07, 2009 10:11 am

Re: Parallel power to 30MM dots, ground loop?

Post by adafruit_support_bill »

Yea, that ground looks pretty rough. I'd expect that more power & ground feeds should improve things.

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

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