The new Raspberry Pi B Model B+ is here - Pick one up in the store and check our detailed guide!

LPD8806 - "injecting" power midstream

by MattM on Fri Jan 25, 2013 2:05 pm

Hi, I have a project that uses a whole mess 'o LPD8806 LED strips -- probably 20-30 meters. I have the strips distributed in a pretty organized fashion from the controller & power supply, and the setup is working mostly pretty well. Since each strip has its own connection back to the controller & power supply (I guess that'd make it a "star" architecture), the brightness of one lane has no effect on the brightness of the others. I've got about 900W of power supplies, balanced in such a way that the maximum draw that would happen if the strips were run wide open will not cause any problems.

So far so good.

Here's the problem. A couple of the longer runs are dimmer than I'd like -- not gradually dimmer toward the end of the run, but just dimmer overall. Since this particular strip is about 25 feet of cable away from the power supply, and is a very long run (about 140 pixels; at 32leds/m, figure about 4 meters worth), I certainly understand why.

There are all kinds of mentions all over Adafruit, both in the tutorial and in these forums, about splitting off power and applying it again every meter. Questions:

1. Is there a schematic for how to do this? In particular, I would assume that the grounds need to be tied together in some manner, but I'm not clear exactly how. I haven't seen a diagram anywhere showing what leads need to be tied together in order to accomplish this. Since these strips are very expensive, I'm nervous about frying them trying to get them to be a little brighter than they currently are.

2. Can you use a separate power supply to apply power farther down the line? I've seen some tutorials on the web where the instructor applies one set of the power supply's leads directly to a strip, and then connects a second set of leads to some lamp wire, which are in turn used to power the second strip in the chain further down the line. In my case, I do not have the ability to run another wire from the power supply to where the strips live, but I DO have AC power available near where the strips live and could place a second power supply on the strip side of things in order to boost the power getting through these strings. But the whole idea of having two separate grounds and +5Vs concerns me because it seems like there's a really good opportunity for a huge short circuit (or a really long fried strip) if I don't tie these in correctly.

Thanks for the help!
Matt
MattM
 
Posts: 9
Joined: Wed Jan 25, 2012 12:47 am

Re: LPD8806 - "injecting" power midstream

by adafruit_support_bill on Fri Jan 25, 2013 2:18 pm

You can apply power at any point. The 5v and ground conductors run the full length of the strip.

If you are powering from multiple supplies, tie all the grounds together but cut the 5v connections between the separately powered sections.
User avatar
adafruit_support_bill
 
Posts: 27989
Joined: Sat Feb 07, 2009 9:11 am

Re: LPD8806 - "injecting" power midstream

by MattM on Tue May 14, 2013 11:16 am

Hi Bill -- thanks for the reply. I haven't been working on this for a couple of months, so I apologize for the late follow up to my earlier post.

I have been experimenting with injecting additional 5V power into my LPD8806 strips, with unexpected results.

First of all, the "before" scenario... in my setup, as I mentioned, a couple of the particularly long stretches of pixels aren't lighting correctly. I have read elsewhere that when there isn't enough voltage being delivered to one of these strips, you'll see the color drop off towards the end of the run. In my case, that's not what's happening at all -- the entire run of about 20 feet is off. If I program RGB white (255R, 255G, 255B) for the whole strip, I get a very dim brownish red through the entire strip. I metered the voltage on the strip, and it's only reading 2.6V, so it's very clear what the problem is.

So as I asked below, I took the suggestion to try injecting some additional power down the strip. In your message below, you mention that if I'm powering from two different power supplies (rather than just feeding another line from the same power supply to the opposite end of the strip, for example), I should cut the +5V connection running between the two segments of the strip, such that the data, ground, and clock feeds are shared between the two segments, but the first segment gets +5V from one power supply, and the second segment gets +5V from the other.

When I read that, I assumed it had something to do with what would happen if you ran the power supplies in series -- rather than delivering 5V to the strip, the combined power supplies would be delivering 10V, which would destroy the strips.

Here's the thing, though...I tested the above scenario on a short, expendable piece of strip to see what would happen: I provided power, data and clock from the front end of the strip, and then added an ADDITIONAL +5V AND ground from a second power source. The strip immediately got a lot brighter, which was expected, although I wondered if the strip would wink out from being over-driven.

Surprise, though...when I metered the strip, though, it still showed +5V -- exactly what I'm shooting for. The strip seemed to be doing exactly what it was supposed -- the colors REALLY popped and it seemed to be responding to its color information exactly as I'd programmed.

The only thing I noticed was that my LED controller (a SANDevices E682) is reporting that there's pixel power present, even with the power supply feeding the controller cut off. So the 5V is leaking backwards to the controller, but it doesn't seem to care about that.

So, the question:

in your message below, you mention cutting the +5v connection between strips when powering from two different power supplies. From my testing, at least, it does *not* appear that having two power supplies are resulting in an over-volt condition. I do notice that power is leaking back into the controller, such that the LED controller is reporting the presence of +5V even when the main power supply is disconnected, but is that really a problem? What is the concern we're trying to avoid by cutting the +5V connection between strips?

I have tried running a second run from my main power supply down to the other end of the strip (which, presumably, doesn't require cutting any connections between the strips). It DOES work, but the strip only meters +3.4V and isn't as bright, obviously, as it would be at +5V. So I hope you see my dilemma -- I'm giving up a pretty significant amount of brightness by not using a secondary power supply. If I'm going for brightness (and I am), I'd like to do what's going to get the maximum bang for the buck, but I can't tell if there's another consideration that I'm missing.

Thanks for your help!
Matt
MattM
 
Posts: 9
Joined: Wed Jan 25, 2012 12:47 am

Re: LPD8806 - "injecting" power midstream

by adafruit_support_bill on Tue May 14, 2013 1:09 pm

The reason you want to cut the 5v line between 2 sections that are powered by different supplies is that it is generally not a good idea to connect power supplies in parallel.

Due to natural variability, one may be trying to regulate the output at 5.01v while the other is trying to regulate at 4.99. In that case, the higher output one will be taking on more than its share of the load.
User avatar
adafruit_support_bill
 
Posts: 27989
Joined: Sat Feb 07, 2009 9:11 am

Re: LPD8806 - "injecting" power midstream

by MattM on Tue May 14, 2013 1:34 pm

Thanks Bill.

To explore this a little further -- if one power supply is regulating at 5.01V, and the other is at 4.99V, I understand that that means one supply is taking on more of the load than the other.

But why does that matter?

As this branch of my overall system currently stands, I have a 600W rated power supply driving pixels that collectively draw 580 watts. The main power supply (the one feeding the LED controller at the front end of the system), in other words, is rated to handle the whole load by itself. The issue isn't that the one power supply can't deliver enough watts -- the problem stems from the length of the cabling feeding this run and the voltage drop being caused by the length of the pixel string itself.

If I add a second power supply, am I not simply taking some of the load off of the primary supply? Is the 5.01V vs. 4.99V putting some sort of stress on the power supply's regulator circuitry?

Is there another factor in play that makes it not a good idea to run these supplies in parallel?

Thanks for indulging me. I understand the "how" now, just trying to figure out the "why."

Matt
MattM
 
Posts: 9
Joined: Wed Jan 25, 2012 12:47 am

Re: LPD8806 - "injecting" power midstream

by adafruit_support_bill on Tue May 14, 2013 1:47 pm

Your original supply was not able to power the whole strip. So what if that is the one that is trying to regulate to 5.01v? Your new 600W supply will be coasting along comfortably while the old one will still be running at max output trying to push that last 0.01v out there.

In this case, the resistance of the strand will tend to limit the effect of any mismatch. But many things that 'work' are still not recommended practice.
User avatar
adafruit_support_bill
 
Posts: 27989
Joined: Sat Feb 07, 2009 9:11 am

Re: LPD8806 - "injecting" power midstream

by MattM on Tue May 14, 2013 1:55 pm

Ah, got it. So in the one-power-supply scenario, because of the voltage drop, the power supply sees 2.4V coming back to it and says "I'd better turn up the voltage" in an attempt to get the voltage back up to the regulated 5.01V. It doesn't get there but the needle is nevertheless pinned all the way to "max voltage" in order to try.

In a two-power-supply scenario, were they run in parallel, since the two supplies may not exactly agree on the regulated voltage they're striving for, one supply is still going to have the "needle pinned all the way to the max" all the time, even though the difference between the two supplies is only 0.02V?
MattM
 
Posts: 9
Joined: Wed Jan 25, 2012 12:47 am

Re: LPD8806 - "injecting" power midstream

by adafruit_support_bill on Tue May 14, 2013 3:11 pm

That's the basic idea. Neither supply knows that the other one is out there. They just do whatever they can to keep the voltage at the level they think it should be.
User avatar
adafruit_support_bill
 
Posts: 27989
Joined: Sat Feb 07, 2009 9:11 am

Re: LPD8806 - "injecting" power midstream

by MattM on Thu Jun 27, 2013 2:30 pm

Hi Bill,

Got a new but related question.

On the section of strips I described previously in this thread (4 meters of strip in a rectangular frame with a separate strip along each edge, connected to each other using the JST 4-wire connectors), I have now done some pretty extensive re-wiring, such that the controller (a SanDevices E682) is still feeding data & clock via CAT5 cable from about 50 feet away, but power is being delivered locally. I have also re-wired the connections between each section of strip such that power is being re-injected at each corner of the display.

When I fire the whole thing up with my power supply (a 300 watt Mean Well HRPG-300), I am getting a more reasonable color/brightness out of the strips now (no more brownouts, at least), but the strips are still metering between 3.0 and 3.4 volts when run wide open. I am still leaving a lot of brightness on the table here.

The Mean Well power supplies have a voltage adjust pot, and I've cranked mine up to its maximum, 5.8 volts. That helps a little, but I'm still not even hitting 4 volts anywhere on the strip when the strip is run wide open.

Two questions, then:

1) Given our earlier conversation about regulated power supplies having to "work harder" to keep voltage at a particular setting, in a single-supply arrangement like this, is the power supply working harder to attain a voltage that it's not coming within more than a volt of hitting?

2) Mean Well also makes a 7.5 volt version of the HRPG series that can be potted DOWN to 6.8 volts. I know from the way I originally set up this run (power supply located more than 50 feet of CAT 5 away from a 4-meter run of strip) that I was losing nearly half the output of a 5V supply when I tried to run the strips wide open. Hypothetically, if I were to revert to this earlier arrangement, but replaced the 5V supply with, effectively, a 6.8 V supply, it would seem that the end result would be a voltage much closer to the optimal 5V throughout the system. It's my understanding that the strips will burn out if you feed them 6V, but when the power supply is that far away, they will never see that much no matter how many or how few pixels are lit at any given moment.

BUT -- going back to question 1 above -- does compensating for the voltage drop across the entire system in this way -- by intentionally using an over-voltage power supply -- put undue stress on the power supply itself?

Thoughts?

Thanks as always for your help!
Matt
MattM
 
Posts: 9
Joined: Wed Jan 25, 2012 12:47 am

Re: LPD8806 - "injecting" power midstream

by adafruit_support_bill on Fri Jun 28, 2013 5:02 am

going back to question 1 above -- does compensating for the voltage drop across the entire system in this way -- by intentionally using an over-voltage power supply -- put undue stress on the power supply itself?

That is a wrong and dangerous approach. You will destroy your strips that way. If you try to compensate with a higher voltage supply, the strip when will be subjected to the full higher voltage when it is idle. These strips will not tolerate a voltage higher than 6v.

The HRPG-300-5 should be capable of supplying 60A which would be more than enough to drive these strips. Not sure where the voltage drop is coming from. What kind of wire are you using to connect to the strips? How long are the power leads? Is any of the wiring getting warm in operation?
User avatar
adafruit_support_bill
 
Posts: 27989
Joined: Sat Feb 07, 2009 9:11 am

Re: LPD8806 - "injecting" power midstream

by MattM on Fri Jun 28, 2013 10:05 am

Thanks for your warning, Bill. It's well taken and greatly appreciated.

Believe it or not, I essentially answered my own question shortly after I wrote that post. I had my existing 5V Mean Well HRPG-300 "overclocked" to about 5.9V -- very close to burning out the strips if the whole amount was getting through from the power supply. Given the voltage drops I was seeing I didn't think that was a problem, but then it occurred to me that if I was seeing only 2.6V across the strip when the strip was being run wide open, but wasn't otherwise noticing these sorts of brownouts when less white-intensive programs were running, it would probably be wise for me to see what kind of voltage was running through the strips when they were simply dark. Sure enough, when the strips are dark they show most of that voltage getting through -- I was seeing readings of around 5.8V on an otherwise dark and idle strip. Yikes.

So compensating for voltage drop with extra voltage is a bad idea -- got it. Scratch that from the playbook. I didn't quite earn my "blue smoke" patch but it sounds like I have been pushing my luck. I think I'm going to just resign myself to the idea that these strips are better suited to color patterns rather than being used for general area lighting, and leave it at that for now. There are programmatic ways to ensure that the overall color is even and consistent (i.e. not pushing them to DMX maximum), and I think I'm going to pursue that approach now that I'm re-injecting power throughout the system as has been suggested.

To answer your question about wiring, I am using CAT5 wiring throughout the system, with the exception of the initial feed coming in from the power supply to the controller. For that, I am using 18 gauge lamp wire. The SANDevices E682 (which is a GREAT little controller and soldering project for anyone who's into that sort of thing, by the way -- http://www.sandevices.com/) is capable of supporting 16 separate channels of LED strips (I am not using all of them because of DMX limitations), so I am running channel pairs on each CAT5 cable. The controller feeds a 12-port ethernet patch panel (to facilitate connections rather than having to continuously plug and unplug the Phoenix connectors on the E682), which in turn feeds about 25-50' of CAT5 cable between the controller and the strips. On the other end I have something of a Frankenpatch, where I've connected a JST 4-pin cable to the back of a CAT5 port, which in turn feeds the strips themselves.

Do they get warm? I can't say I've noticed the cables, JST connectors or the strips getting warm at all, even after being on all day. The body of the CAT5 ports on the strip ends get _slightly_ warm, maybe a couple of degrees warmer than ambient. It's really subtle and because this rig is up near ceiling height I'm not always sure if it's just my imagination or they are generating a little heat.

In any case, having now thought about this more, and discovered how much voltage flows through the strips when they are idle, it seems reasonably clear to me that the voltage drop is coming from the inherent nature of these strips, not the system wiring. That 1) when all or most of the pixels are idle, nearly all the voltage gets through and we're only looking at a drop of .1V or so from the level measured at the power supply's outputs some 50' of cable away, and 2) when all or most of the pixels are showing full-value RGB white (255R, 255G, 255B), the combined draw of all the pixels is causing the voltage to drop by, in my case, nearly 3V. I would also note that the data and clock lines run on the same cable at the lower logic voltage of 3.3V, and am not seeing any sort of data or clocking problems at all. If this were a wiring problem wouldn't we expect to see data/clocking problems if those signals weren't properly arriving to the first pixel in the strip (after which, they are regenerated)?

I *AM* running some very long runs of CAT5 cable -- 50 feet or more -- but given the measured voltages when the strips are idle it seems pretty clear to me that the wiring is not the problem. Would you agree? Is the slightly elevated temperature of the connectors something I need to be worried about?

Rig picture, for the curious:

https://www.facebook.com/photo.php?fbid=537591049606230&set=a.379752268723443.90952.188797797818892&type=1

Thanks!
Matt
MattM
 
Posts: 9
Joined: Wed Jan 25, 2012 12:47 am

Re: LPD8806 - "injecting" power midstream

by adafruit_support_bill on Fri Jun 28, 2013 10:40 am

I am using CAT5 wiring throughout the system, with the exception of the initial feed coming in from the power supply to the controller....

...which in turn feeds about 25-50' of CAT5 cable between the controller and the strips.

So you are sending power over 25-30' of CAT5? That is probably where the voltage drop is. Cat5 is usually 26AWG wire. You are probably looking at a resistance of at least an ohm for that distance. 1 ohm would give you 1 volt drop at 1A.

I *AM* running some very long runs of CAT5 cable -- 50 feet or more -- but given the measured voltages when the strips are idle it seems pretty clear to me that the wiring is not the problem.

Actually, it seems clear that it IS the problem. By law (Ohm's law) the voltage drop due to the resistance of the wire will be proportional to the current. At idle, the voltage drop will be nil.

Nice looking installation. :D
User avatar
adafruit_support_bill
 
Posts: 27989
Joined: Sat Feb 07, 2009 9:11 am