Todays blog post is about the effects that power draw has on RGB lights. We've taked about it before in our blog post about how wiring affects power distribution and in our demonstration example house where we lined it with pixels and handled the issue of power consumption.
This video is pretty simple - it shows 160 pixels (480 DMX channels) connected to a single power supply and our HolidayCoro pixel controller. It is intended to show the effects that the gauge of the power wire and the power consumption of the pixels themselves have on the actual light output.
(open in YouTube with High Def for better viewing)
Post any questions below.
Thanks,
David
HolidayCoro.com
Great informative video.
ReplyDeleteI've known a bit about the BASICS of power draw since I started this hobby last year, but this video is a great way to actually see, in black-and-white so to speak, how draw and drop really effects your lights.
Some hopefully helpful comments from an electrical engineer who was just checking out the site. I know it's quite lengthy, but hopefully useful to someone starting out.
ReplyDeleteThere is defiantly a voltage drop problem in the demo. Measuring the VOLTAGE (not CURRENT as mentioned in the video) at different places is a good place to look to see if you have this kind of problem. Increasing the wire size will also help reduce VOLTAGE drop.
If you choose to inject power at different places, it is best to NOT create a loop. If you want to put voltage at both ends that is fine, except you should cut the positive wire some place around the middle. Keeping the ground wire is ok, and is actually needed for the serial data bus to use the same ground through the whole string. Another way to inject voltage is in the middle of the strand of LED's. Keep the ground wire and data lines connected through, but cut the positive wire. Now connect your power supply to the ground, and the wire cut from the rest of the LED's. It can come from either the same power supply or a separate one.
Voltage drop is also caused by the resistance of each connector. One way to reduce this voltage drop is to solder the wires from one strand to the next. While it does make it more of a pain in the neck when connecting strands, with some pre-planning before setting up your display it can help reduce that voltage drop. Another good idea (as mentioned in the video) is to increase the size (or gauge) of the wire, especially the wire from the power supply to the string of LED's. If you have a long distance from the power supply to the LED's (or in this case the display controller), you have already dropped the voltage before it has had the chance to be acted on by the LED's.
Another problem that is most likely affecting this demo is the size of the power supply. Expressed in either amps or watts, most likely the power draw of the lights is greater then the size of the power supply. Let's say that each string of lights is 2 amps for each red, green and blue (unless the LED's are really bright, this is probably too high, but it keeps the math easy). Each sting would therefore take 6 amps of CURRENT. For the 9 stings of lights in the demo, that would be 54 amps total. Watts is voltage times amps. So the power supply would need to be rated for 54 (or larger) amps at 12 volts, or 648 watts at 12 volts.
In a normal power supply, the larger the current draw, the lower the voltage will be. That is why he mentioned measuring the VOLTAGE of the power supply before the lights are connected, and then after the LED's were connected. That's the "No Load" and "Load" measurements. If the current draw of the power supply is exceeded, the voltage will drop down very quick, and most likely the power supply will get very hot, then turn off, or burn up. You can get a larger power supply then you need in terms of amps or watts, but using lower will always give you problems.
In the setup in the demo, if you didn't have a single 55 amp or larger power supply, you could use two smaller ones if they add up to 55 amps or larger. One power supply would be connected to the front of the string, and the other connected to either the end (with the power connection cut some where in the middle), or connect the second to the middle (cutting the power wire to prevent a "loop" in the front of the string, and of course make sure the ground of the power supply is connected as well in the middle).
Thanks for your feedback. I should have posted additional technical details on the items used in this demo.
DeleteThe strings in this video are 47ma per pixel (all white) see (http://www.holidaycoro.com/Smart-Pixel-Square-RGB-Module-p/657.htm) , so the current draw on 160 pixels should be (excluding other losses) around 7.5 amps at 12v, so the power supply (see http://www.holidaycoro.com/350w-Dual-Output-Power-Supply-p/49.htm) at 29 amps at 12v should be sufficient to drive the connected pixels, including losses.
Yes, your 29 amp power supply should have no problem running that string. So I remove that from my list of possible problems in my post above. I knew my numbers were over, just not by that much. As I stated, they were actually chosen to keep the math simple and easy to follow.
ReplyDeleteI would be interested to know the voltage drop of those connectors. Perhaps a test where the voltage was measured at the start of the string, and then voltage at the end of the string using the plug connectors. And then compare that to the same string where all the strings were soldered to each other. Find the voltage drop on the 2nd test (soldered), and subtract the voltage drop from the 1st test (plug in connectors). Divide the difference in voltage drop by 8 (number of plugs bypassed when soldering) and you will have the voltage drop per connector.
This could be done with a single connector, but I expect that there will be such a small drop, the meter may not give a good reading. By replacing several connectors and getting a average per connector would give a better average voltage drop per connector.
I believe the biggest source of power drop is the combination of the wire (20 AWG) and the heat generated by the CV pixels. I believe just this discussion shows how important actual testing of lights, power supplies and controllers is before building a system.
ReplyDeleteI'm planning a display using a centralized power supply (the same one in the video) and the 27 channel dmx controller. I need to make a few "extension" power/data cables to reach certain parts of the display. I have some of the 20g waterproof connectors and some 18g 4 conductor wire laying around that I was planning on using. Each extension would only supply a max of 30 dumb rgb's. How much would it affect the voltage going from 20g connector to 18g wire back to 20g connector? At least one of the extensions will be close to 30'!
ReplyDeleteI really don't think you'll have a problem and 18ga cable even on the 30ft run. Built it and then test it to be sure. You might also want to read this article: http://blog.holidaycoro.com/2013/03/a-technical-guide-to-rgb-wire-selection.html
DeleteI can certainly attest to the issues being displayed. I have 8 of the HC Candy Canes each with 16 rectangular (3 RGB) LED in each. As I began to put them together I found out real quick that after about 3 candy canes the whites were off and the reds were pinkish. I talked with David and several others and corrected the issue by injecting a fresh 12V feed between cane 3 and 4 and 6 and 7. (deatils can be seen on our website)
ReplyDeleteThis resolved all the issues. I did not cut the 12v at the injection points as suggested by the other poster. I didnt see the requirement to do so. (perhaps someone would elaborate on why..) This is the same method I would use with a long model train set when injecting power to the tracks to compensate for the losses.
Great post David.. keep them coming ..
Bob Moody
Http://moodychristmas.weebly.com
Wondering if the voltage from the controller is actually a PWM signal, not a pure DC voltage. Does your meter account for this signal? Most inexpensive DVM's can't handle RMS on frequencies above 3000 hz I believe, especially a square wave or PWM waveform. Although the point is that a voltage drop does exist, but I'm curious if your actually measuring true RMS.
ReplyDeleteYou are completely correct - I use a Fluke with RMS and it results in an averaged voltage from the PWM signal. Though the testing was performed at output of 255 on all three colors, so PWM should effectively be "null" - it is more or less a constant power output as if it wasn't modulated.
DeleteFor other readers wondering what this RMS stuff is about, see: http://wiki.answers.com/Q/What_are_the_peak_and_RMS_values_of_the_voltage_of_a_pulse-width-modulated_signal
You can estimate the voltage and current parameters of a pixel string, and in particular the voltage drop, using the calculator at http://blinkyflashy.info/calcs/pixpower.php
ReplyDeleteThis has a simple model - one or two leads between the power supply and the first pixel, and constant spacing between the pixels. (It allows 2 leads to cover the case of, say, a 24" lead built into the string, along with a user supplied 12' power lead from the power supply to the built in lead, each with separate wire gauges. It allows you to check the effects of different wire gauges, pixel spacings, string length, etc. Remember that the pixel spacing input is for the wire length between pixels, which is usually a bit longer than the physical spacing between the LEDs in bullet style pixel strings. You could use it for pixel strips if you can estimate the effective wire gauge.
Anonymous electrical engineer - why specifically would it be bad to have a "loop" created by powering both ends of a string from the same power supply? It's pretty low impedance so I'm not imagining that pickup of RF signals would be much problem. There would be some point in the string where current drops to near zero in the pixel string wiring - on one side of this location the net current wold flow from one end, and on the other side it would flow from the other end. Where that point is located depends on how bright the various pixels are set (and on the resistance of the leads going to either end). However I'm not seeing how this would cause a problem (again, using a single power supply to there is no fighting between two supplies).
ReplyDeleteTo reduce the likelihood of a short, the "hot" end of plugs should always be female. If you loop both male and female will be hot.
Delete