Question about Non-Linear LED Loading

[TechGuy2016]

I often create LED arrays, and have encountered a common issue where, as the LED load is increased, the total current requirement is not a linear factor as each branch (LED strip) is added to the circuit.

It is common to have many branches of LED strips on a 12VDC 60W power supply.

Power source: 12VDC 60W power supply

Each led strip is rated and measured at 3W, individually.

Here is the phenomenon in question:

As each LED strip (branch) is added to the circuit, the load is not a factor of the each branches power (3W) by the number of branches. The diagram shows 24 branches as an example. If one branch is connected, then it is in fact, measured to be 3W. Using a simple linear calculation, 24 branches x 3W is 72W. However, the actual 24 branch circuit draws only 3.88A which is 46.56W (at 12V).

I connected and measured from 1 to 58 strips and plotted the results. The resulting function is definitely very non-linear. I could place 58 of the LED strips on the circuit before I reached the 60W limit.

Does anyone know the Theorem or equation explaining this non-linear loading phenomenon?

I would greatly appreciate anyone's feedback on this. Thank you!

 

[anorlunda]

Can you tell us more about the 12V 60w suppply, and about the wiring. One suspect is that all or some of the 3w strips are not getting the full 12v as total load is increased.

 

[Algr]

Check for resistance in the wiring leading up to the strips, and from strips to ground.

Alternatively, If strips dim as others are added, then you are approaching the limits of your power supply.

 

[TechGuy2016]

The power supply is a 12VDC 60W (5A max), commercial grade, constant voltage power supply. It is made specifically as an LED driver.
The wiring is 20awg. The length between each LED strip is about three inches. The wire from the power supply is 20AWG and is eight inches. Note that I kept the lengths short for testing purposes to minimize resistance from the wire.

 

[jim hardy]

EDIT i see others have posted, so this may be redundant...

it is in fact, measured to be 3W.

It's important to understand one's test equipment lest he be fooled by it.

What is the nature of your power supply ? Link to its manual ?
What is the nature of your Watt measurement ? Link to the instructions for the instrument you use ?
Can you scan and upload your plot ?

Otherwise your question is well stated, so probably it's half answered..

24 branches x 3W is 72W. However, the actual 24 branch circuit draws only 3.88A which is 46.56W (at 12V).

You expected 6 amps and only got 3.88 ?

Just a U_nS_cientificW_ildA**G_uess here -
3.88/6 = 0.647
That's pretty close to the ratio of peak to average for sinewaves (0.636). Is your supply well filtered?
Could be as simple as a bad filter cap in your supply.
To check ---
Measure current or voltage (or both) with your DMM set to AC instead of DC
and plot AC reading as % of DC reading with 1 string, 12 strings and 24 strings .

 

[TechGuy2016]

Thanks for the responses.

This phenomenon happens with all power supplies such as many clean (Agilent) lab power supplies and with LED drivers. Also, I have tried to use large 16AWG gauge wire from the power supply, but still get the same results. I must use 20AWG to each strip because that is as large as I can put on the contacts. Being such, I have assumed the phenomenon is a characteristic of the load, not the equipment.

Consider that I have measured with multiple type of equipment from precise Fluke meters through lab bench Agilent DMM's.

Also, please consider that the power is "clean" DC power as I see virtually no ripple (evaluated on an o-scope) on many of the power supplies I have tried.

 

[berkeman]

What is the nature of the 3W LED strips? Are they just series LEDs with a single current-defining resistor? Or do they contain DC-DC converters to drive the LEDs?

If they contain DC-DC conterters, then as the input voltage droops (because of the increased load on the power supply), they will draw more current at that lower input voltage to supply the same output power.

 

[TechGuy2016]

Oh, right. Sorry, I did not mention the composition of the LED strips.

The Strips are three LED's in series with a ballast resistor per segment. There are 36 LED's (12 segments) per strip. Power is 3W (250mA @ 12VDC) per strip of 36 LED's.

 

[berkeman]

I connected and measured from 1 to 58 strips and plotted the results.

Can you upload the plot? And sorry if you mentioned it already, but was the 12V supplied to each LED strip steady at 12V as you added more and more strips?

 

[TechGuy2016]

Here is the plot.

As for the voltage, there is a voltage drop that increases as each strip is added to the circuit.

 

[berkeman]

As for the voltage, there is a voltage drop that increases as each strip is added to the circuit.

Can you plot the distribution voltage on the same graph? Just add it as another column (if you are using Excel) and re-plot, and then click on the voltage plot and right-click, Properties and select "separate axis".

 

[TechGuy2016]

Unfortunately, I do not have the voltage data. I made the set-up a little while back and stored the data in an excel file. If I remember correctly, the total voltage drop was down to about 10.5V, measured on any of the strips. Does that help?

 

[berkeman]

Unfortunately, I do not have the voltage data. I made the set-up a little while back and stored the data in an excel file. If I remember correctly, the total voltage drop was down to about 10.5V, measured on any of the strips. Does that help?

Did you notice whether they were dimmer? They should have been if the input voltage was drooping that much. The voltage drop across the LEDs wouldn't change much as the current varied, but the voltage drop across the current-setting resistors was changing, so the total current per strip was dropping. If you were able to re-take the data and record the strip voltages, I'm guessing that you would be able to explain this effect.

 

[jim hardy]

If I remember correctly, the total voltage drop was down to about 10.5V, measured on any of the strips. Does that help?

well, YEAH. Berkeman explained it with instructions how to put a number on it.

LED's have a roughly constant voltage drop
so the voltage across your current setting resistor dropped by a larger percentage of power supply voltage than you think.

https://en.wikipedia.org/wiki/LED_circuit
LED arrays
Strings of multiple LEDs are normally connected in series. In one configuration, the source voltage may be greater than or equal to the sum of the individual LED voltages; typically the LED voltages add up to around two-thirds of the supply voltage. A single current-limiting resistor may be used for each string.

Just assume 1.9V per led. Red is about there, white may be 3V.

1.9 X 3 = 5,7 volts across red LED's, 12 - 5.7 = 6.3 volts across current setting resistor.
When supply drops to 10.5 you have only 10.5 - 5.7 = 4.8 across current setting resistor
so current drops to 4.8/6.3 = 76% of what it was . Power decreases even more because it's V X I and both went down..

Flimsy power supply, as others have noticed, explains your symptom.

 

[TechGuy2016]

Well, if I leave the array as a simple daisy-chain, then each strip is progressively more dim due to the stacked voltage drop. What I do to manage this, is connect both the leading end and the trailing wire ends of the LED strip array to the power supply lead wire so as to balance the load. That makes all LED's have perceptibly equal output.

 

[TechGuy2016]

Jim,

Thank you for the detailed response. However, I do not understand what you mean about a sub-standard power supply. Please consider that this phenomenon happens will all power supplies, including robust lab grade power supplies. (I have about 15 different lab supplies, including some well respected ones such as Agilent, B&K, etc.)

Considering this, it is hard to imagine that it is a sub-standard power supply. Clearly you are very talented in this field. Do you have any other thoughts as to why this is occurring, aside from the power supply?

Berkeman, if you have any thoughts as well, I would appreciate them.

 

[berkeman]

However, I do not understand what you mean about a sub-standard power supply. Please consider that this phenomenon happens will all power supplies, including robust lab grade power supplies. (I have about 15 different lab supplies, including some well respected ones such as Agilent, B&K, etc.)

A lab-grade power supply will not droop its output from 12V to 10.5V if the output power spec is being met. If you try to draw more current from it than it is rated for, you will likely get a drop in output voltage (or a full current-limit behavior, depending on the type of power supply). You are saying that all of the supplies in your lab were not able to deliver their rated output current without the 12V output drooping to 10.5V?

 

[jim hardy]

As for the voltage, there is a voltage drop that increases as each strip is added to the circuit.

I remember correctly, the total voltage drop was down to about 10.5V, measured on any of the strips.

Attention to wording is the price of clear communication.
A voltmeter has two probes. It reports voltage between them.
10.5 volts measured between what two points "on any of the strips" ? What was voltage between power supply output terminals ?

it is hard to imagine that it is a sub-standard power supply.

Well, you said you saw 10.5 volts. Where did you see it ? Why was it not 12 ?

Instead of applying observation to the things we wished to know, we have chosen rather to imagine them. Advancing from one ill founded supposition to another, we have at last bewildered ourselves amidst a multitude of errors.
https://web.lemoyne.edu/giunta/ea/LAVPREFann.HTML
Not bad for 1789, eh ?

Go back and take a look. One observation is worth a thousand expert opinions.

 

[TechGuy2016]

Well, if I leave the array as a simple daisy-chain, then each strip is progressively more dim due to the stacked voltage drop. What I do to manage this, is connect both the leading end and the trailing wire ends of the LED strip array to the power supply lead wire so as to balance the load. That makes all LED's have perceptably equal output.

 

[TechGuy2016]

Berkeman,

Thank you. Clearly, I need to observe more about the circuit. (It is a lot of work as you can imagine, but I need to get this phenomenon figured out, so I will make the configuration again.) Sometime early next week will be my next chance to set it up and observe. Once I do so, I will put my findings here.

Jim and Berkeman, thank you much for your persistence it trying to get this figured out.

Have a good weekend.

 

[jim hardy]

Sometime early next week will be my next chance to set it up and observe. Once I do so, I will put my findings here.

Good on you !
Sorry if i sounded unfriendly.
You cannot imagine how many times i have wished i'd been more meticulous in recording my observations , it's a good habit to form early in your experimenting.
Let's explore the droopy supply hypothesis
Here's how one approaches LED strings...
Basically, the LED voltage is related to the energy of the photons emitted. Planck's Constant i think ..
http://www.seos-project.eu/modules/Earth'spectra/Earth'spectra-c01-p11.html an interesting article by the way...

Since a white LED is like a fluorescent lamp, a uv emitter with phosphors in it that absorb then re-emit the energy as visible, it ought to have voltage drop of 3.5V or a little more. UV is beyond blue.

Here's a chart of "typical" LED forward voltage drop. Everyplace on the net has a slightly different set of curves, see if your supplier has one.

here's a nice table from a kind hearted manufacturer
http://www.oksolar.com/led/led_color_chart.htm

So if a white LED drops ~3.6 volts at 20 ma
three of them in series will drop 10.8 volts
and with your 12 volt supply
that leaves 1.2 volts across your ballast resistor and the interconnecting wires..
Now to your chart

Lavoisier also cautions to be precise in wording so let me get this straight
as i understand
Each "strip" (not 'array' as i mistakenly wrote on chart) has twelve "strings" of 3 LED's , each "string" with its own ballast resistor ?

From your chart
For the first two LED strips , 239 ma is 19.92ma per string, real close to 20 ma per string
three LED's at 3.6 V each = 10.8 volts as above
leaving 1.2 volts for the ballast resistor
1.2 volts / 19.92 ma suggests ~60 ohm ballast resistors

now out there at 54th and 55th strips we see only 14.9ma change per strip not 239
yet we have 4.8 amps shared among 55 strips = 87.2 ma per strip ,
87.2 is way down from 239 but a lot more than 14.9
which means the other strips must have donated some of their current for the newcomers
and for their current to decrease the supply voltage arriving at their terminals must have decreased...

87.2 ma per strip= 7.3ma per string
suggesting 0.0073 A X 60 Ω = 0.438 V across ballast resistors.
If you saw 10.5 volts at that condition what was voltage across LED's ?
10.5 V_supply - 0.438 V_ballast = 10.06 across LED's, 3.354 per LED
which is a decrease of only ~¼ volt per LED .
Look at the colored LED curves up above. White should have changed more than ¼ volt for a current change from 20 to 7 ma .
But it looks about right for blue...

I'm guessing that your LED's are either white or blue based on your observations, and that your power supply voltage droops. When you get better data you'll pinpoint your trouble.

Suggestions
First check every one of my steps, not only am i awkward but my Windows calculator has taken to doing strange things when i hit the "negate" key...

Your sag from 12 to 10.5 volts at 3.88 amps is only 0.39 ohms someplace. If it's in a bad connection it should be getting hot.
55 strips at 1/4 amp each is 13.75 amps , are your supplies up to that ?Have fun and let us know what you find.

old jim

 

[Baluncore]

Power source: 12VDC 60W power supply
Each led strip is rated and measured at 3W, individually.

I connected and measured from 1 to 58 strips and plotted the results. The resulting function is definitely very non-linear. I could place 58 of the LED strips on the circuit before I reached the 60W limit.

You expected quantity 60 * 1 W = 60 W.
60 W / 12 V = 5 A. That would be OK with that power supply.

But you claimed you measure the strips individually as 3 watt, quantity 60 * 3 W = 180 W. 180 W / 12 V = 15 A.
It is not surprising then that the supply voltage is being reduced to 10.5 V to limit the current to 5 A.

Is the 3 watt rating the power per strip or per metre ?
Are your cut strips 1 W each or 3 W each ?
Are all the LED strips the same colour ?

 

[Tom.G]

Ref. Post #19:

Well, if I leave the array as a simple daisy-chain, then each strip is progressively more dim due to the stacked voltage drop.

Instead of daisy-chaining the strips connect each one directly to the power supply. Apparently the thru-wiring in the strips is small gauge wire creating a large voltage drop by the time you get to the last LED strip.

 

[jim hardy]

Now there's some good thinking !

Apparently the thru-wiring in the strips is small gauge wire creating a large voltage drop

Likely it's a printed circuit board trace ?

Your sag from 12 to 10.5 volts at 3.88 amps is only 0.39 ohms someplace.

Trace on a 1 oz copper pcb is 1.37 mils thick http://www.pcbuniverse.com/pcbu-tech-tips.php?a=4
this trace resistance calculator will estimate resistance from dimensions
http://referencedesigner.com/cal/cal_05.php
It returns a trace 0.00137 " thick by .040 " wide as 12 milliohms per inch, and that should get plenty warm at ten amps.