strictly speaking of larger projects that require more than 50 watts of power to drive the LEDs, ones with hundreds of LEDs, like matrix displays.
I've been thinking about this one for a while, Why don't people use 20V, 30V, or any multiple of 5V supplies for powering their 5V ARGB strip projects? Wherever I see, I see them use a 5V high-amp power supply instead of connecting their LED strips in series. Even Adafruit instructs to use a 5V supply Power | 32x16 and 32x32 RGB LED Matrix | Adafruit Learning System. I'm in no sense keenly knowledgeable in electronics, but my basic understanding is higher voltage is easier to manage ( and get! ) than higher current.
So why, in this case, where sometimes 10s of amps are required, ( in my current on-hold project, which has 288*60mA per LED = 17.28 Amp @ 5V ) recommend using a 5V supply?
It's significantly harder to find a good quality 5V 10A or 20A supply that can reliably perform.
Oh god the circuit traces are a nightmare -_-, too many power injection in between to keep the circuit traces reasonable and within safe operating temps.
on the other hand, finding a 20V,30V supply is a breeze, one that can reliably output say, 5 Amps. I can always use a step-down converter to power the peripherals. Heck, maybe even a good quality laptop power supply at 19.5 V, about 4*5V strips should get about 4.8V ideally.
I feel like I am missing something VERY important that stops me from doing this. One thing that pops in my mind is that addressable LEDs can be switched of individually, so would turning off one strip in the say 4 strips in series connected to a 20V supply affect the other strips voltage?
PS. It's a given that I'm not in-fact too knowledgeable in terms of electronics, just bits and pieces here and there. So.. go easy maybe?
Things in series all get the same current, in the case of LED strips you want to be able to control them, which means they draw different amounts of current; that alone is enough reason for not putting them in series.
You'd also face problems with getting control signals to them at various different voltages.
Or put another way: all the signalling is 5V relative to a single ground and far too fast for
standard optocouplers, making level-shifting the control signals very tedious.
But basically you can't power something designed for constant voltage from a current supply.
There is actually an issue with lots of LED strips all at 5V, and that's the thickness of wiring
needed at high current and low voltage.
This is solved by running all the power at 24V to the strips and giving each strip a buck converter
down to 5V locally - known as point-of-load supply regulation. All the strips can share ground,
yet the power system is higher voltage - best of both worlds. Even though fairly small and
cheap buck-converters are available, this can be a problem aesthetically.
For instance say you want no more than 0.25V drop along some 5V supply wiring carrying upto
10A - that means the max resistance of the total wiring must be 25 milliohms, necessitating very thick wires for long runs.
With 24V supply you'd only need about 2A in the wires for the same 50W, and with individual
regulation can afford 15V or so drop in the wiring, so the max resistance of wiring is actually 7.5
ohms, 300 times larger - or put another way long runs of thin wire will be fine. (Well the ground
wiring still has an issue if you want to keep the grounds common enough in voltage, so its not
a total panacea)
PerryBebbington:
Things in series all get the same current, in the case of LED strips you want to be able to control them, which means they draw different amounts of current; that alone is enough reason for not putting them in series.
You'd also face problems with getting control signals to them at various different voltages.
The ws2812b strips all have a separate data line, so would that still be a problem?
PerryBebbington:
Things in series all get the same current, in the case of LED strips you want to be able to control them, which means they draw different amounts of current; that alone is enough reason for not putting them in series.
MarkT:
Or put another way: all the signalling is 5V relative to a single ground and far too fast for
standard optocouplers, making level-shifting the control signals very tedious.
But basically you can't power something designed for constant voltage from a current supply.
There is actually an issue with lots of LED strips all at 5V, and that's the thickness of wiring
needed at high current and low voltage.
This is solved by running all the power at 24V to the strips and giving each strip a buck converter
down to 5V locally - known as point-of-load supply regulation. All the strips can share ground,
yet the power system is higher voltage - best of both worlds. Even though fairly small and
cheap buck-converters are available, this can be a problem aesthetically.
For instance say you want no more than 0.25V drop along some 5V supply wiring carrying upto
10A - that means the max resistance of the total wiring must be 25 milliohms, necessitating very thick wires for long runs.
With 24V supply you'd only need about 2A in the wires for the same 50W, and with individual
regulation can afford 15V or so drop in the wiring, so the max resistance of wiring is actually 7.5
ohms, 300 times larger - or put another way long runs of thin wire will be fine. (Well the ground
wiring still has an issue if you want to keep the grounds common enough in voltage, so its not
a total panacea)
I cant visualize the control signal issue.. as far as I know, Ws2812 chips receive 5V data, subtract a portion, and then send it on anew, as another 5V signal.
So as long as the ws2812b chips themselves are powered with 5V, why would i need to level shift the data? This is from a hackaday post i saw once, where the first led had a few mechanisms to adapt the voltage. According to Cheating At 5V WS2812 Control To Use 3.3V Data | Hackaday this hackaday post, the 2nd led and so on gets 5V signal, because the 1st led regenerated or modified the signal. In the higher voltage case, the only signal that should matter is the first 5V signal. from the arduino, shouldnt it?
The separate buck converter idea is really good, I hadn't thought of that.. But anyways, It would still be cost-inefficient, like in my case, where there are 8 strips of 36 LEDs, so about 2.2Amp per strip. I wouldn't put some very low-cost converters when they could possibly draw more than 2 Amps. This local one is dirt cheap, but i doubt it can continuously manage to output 2 amps.
kaseftamjid:
The ws2812b strips all have a separate data line, so would that still be a problem?
You can drive them with different data lines, or the same one, however it suits your purpose.
The point is that all data lines are referenced to the ground line on the Arduino so for each strip you must be able to connect its negative line to the ground line on the Arduino, so all strips must have their negative ("GND") line connected to the same point. This immediately prevents the power being connected in series.
kaseftamjid:
Could you please elaborate a bit?
How much current a strip draws depends entirely on how many LEDs (and colours) are lit, and at what brightness. Unless every strip is displaying exactly the same pattern, they will each be drawing a different current.
Hi,
Can you please post a diagram of how you would invisage your idea of strips in series and how you would connect them to an Arduino controller?
Please include power supplies.
I think you have a misunderstanding about how the strip uses the signal and the signal voltage reference points for each of the strips and the signal reference point of the Arduino.
The other point is that if the current flowing through the first RGBLED of the top strip is going to then flow through the first RGBLED of the next lower strip and then through the first RGBLED of the following strip in series, then the current sharing will not allow each of the RGBLEDS to operate properly independently.
Paul__B:
You can drive them with different data lines, or the same one, however it suits your purpose.
The point is that all data lines are referenced to the ground line on the Arduino so for each strip you must be able to connect its negative line to the ground line on the Arduino, so all strips must have their negative ("GND") line connected to the same point. This immediately prevents the power being connected in series.
How much current a strip draws depends entirely on how many LEDs (and colours) are lit, and at what brightness. Unless every strip is displaying exactly the same pattern, they will each be drawing a different current.
TomGeorge:
Hi,
Can you please post a diagram of how you would invisage your idea of strips in series and how you would connect them to an Arduino controller?
Please include power supplies.
I think you have a misunderstanding about how the strip uses the signal and the signal voltage reference points for each of the strips and the signal reference point of the Arduino.
The other point is that if the current flowing through the first RGBLED of the top strip is going to then flow through the first RGBLED of the next lower strip and then through the first RGBLED of the following strip in series, then the current sharing will not allow each of the RGBLEDS to operate properly independently.
Thanks.. Tom..
Ahhh i get it now.. I feel kinda stupid now..
So basically, In series, the voltage DIFFERENCE is 5V for each strip, but the supply and ground is not same.
If the first Led power input has 20V, the ground of that strip is 15V, the next one has 15V supply and 10V at the ground.
The generated signal is 5V, relative to the ground of the said strip, So the first signal would seem like 10V instead of 5V for the next strip and so on and so forth. So to make this thing work, level shifting is required at every step ( for every led that is in series).
In all your technical expertise, which one would you choose and why? is there a one size fits all option in turns of complexity and cost? ( basically the pros and cons of each one..)
Use a high amp power supply at 5V.
Higher input voltage feeds to multiple buck converters, each for a single strip.
kaseftamjid:
In all your technical expertise, which one would you choose and why? is there a one size fits all option in turns of complexity and cost? ( basically the pros and cons of each one..)
Use a high amp power supply at 5V.
Higher input voltage feeds to multiple buck converters, each for a single strip.
Depends what you are doing and how many strips there are, you have to compare the 2 methods for the particular job you are doing, in some situations one will be best, in other situations the other will be best.
"So basically, In series, the voltage DIFFERENCE is 5V for each strip, but the supply and ground is not same.
If the first Led power input has 20V, the ground of that strip is 15V, the next one has 15V supply and 10V at the ground.
The generated signal is 5V, relative to the ground of the said strip, So the first signal would seem like 10V instead of 5V for the next strip and so on and so forth. So to make this thing work, level shifting is required at every step ( for every led that is in series)."
No. All the WS2812Bs need a 5V power source and Gnd. All the way down the line. You put 5V in at one end of the strip, and if the strip is not too long, you get 5V at the other end. If the strip is too long, you connect 5V and Gnd at both ends, and maybe even in the middle of the strip. But each chip/LED needs to see 5V/Gnd.
The data is buffered and passed along chip to chip, each chip retaining the last 3 bytes it go to drive its LEDs.
A 74AHCT (such as 74AHCT541) part may be used to receive a 3.3V signal and buffer it up to 5V when powered from 5V for the first WS2812B in a string; it will accept as low as 2V as a valid high, and up to 0.8V as valid low.
Other AHCT parts will do the same, I mention the '541 as I know it is still available in DIP.
There are other WS28xx variants that can run from 12V. The same applies there, 12V is needed by all the chips.
PerryBebbington:
Depends what you are doing and how many strips there are, you have to compare the 2 methods for the particular job you are doing, in some situations one will be best, in other situations the other will be best.
Paul__B:
It is going to be a cost comparison between buck converters and heavy cable.
So we are entirely clear then - the idea of putting strips in series is a dead duck!
I was more concerned about the safety aspect.. But I think overkill wires should be safer than separate buck converters..
CrossRoads:
"So basically, In series, the voltage DIFFERENCE is 5V for each strip, but the supply and ground is not same.
If the first Led power input has 20V, the ground of that strip is 15V, the next one has 15V supply and 10V at the ground.
The generated signal is 5V, relative to the ground of the said strip, So the first signal would seem like 10V instead of 5V for the next strip and so on and so forth. So to make this thing work, level shifting is required at every step ( for every led that is in series)."
No. All the WS2812Bs need a 5V power source and Gnd. All the way down the line. You put 5V in at one end of the strip, and if the strip is not too long, you get 5V at the other end. If the strip is too long, you connect 5V and Gnd at both ends, and maybe even in the middle of the strip. But each chip/LED needs to see 5V/Gnd.
The data is buffered and passed along chip to chip, each chip retaining the last 3 bytes it go to drive its LEDs.
A 74AHCT (such as 74AHCT541) part may be used to receive a 3.3V signal and buffer it up to 5V when powered from 5V for the first WS2812B in a string; it will accept as low as 2V as a valid high, and up to 0.8V as valid low.
Other AHCT parts will do the same, I mention the '541 as I know it is still available in DIP.
There are other WS28xx variants that can run from 12V. The same applies there, 12V is needed by all the chips.
Isnt it a bit of a waste to use a whole octal buffer? I mean, cant i use a npn high side switch, or cascade 2 npn switching circuit to deliver 5V signal? the 2n2222a says max switching speed upto 2 Mhz, where the strips are at best 800Khz.
btw, the strip works fine with a 3.3V signal, I messed around and checked the things, and seems like they dont mind the signal at all. as you said, 2V as high and .8V as low
CrossRoads:
A 74AHCT (such as 74AHCT541) part may be used to receive a 3.3V signal and buffer it up to 5V when powered from 5V for the first WS2812B in a string; it will accept as low as 2V as a valid high, and up to 0.8V as valid low.
Other AHCT parts will do the same, I mention the '541 as I know it is still available in DIP.
Why use a high speed family?, the 74HCTxx family is fine for this purpose and generates less RFI,
normally the "A" families are used only when the speed is required as the EMI compliance gets
harder.
Someone commented:
Isnt it a bit of a waste to use a whole octal buffer? I mean, cant i use a npn high side switch, or cascade 2 npn switching circuit to deliver 5V signal?
These are jelly-bean logic chips, what's the issue? Buffers are packaged 6 or 8 at a time, you aren't
expected to have a multiple of 6 or 8 lines to buffer(!) - BTW you can parallel several buffers in a package
for more current handling if you want too.
I doubt you could make a suitably fast logic edges using discrete BJT transistors, CMOS chips expect
crisp ~3--10ns edges, slow edges can lead to multiple transistions.
And one chip is much simpler than 2 transistors + associated resistors!
I suggest the AHCT vs HCT as HCT has a high minimum Vin high of 3.15V with VCC = 4.5V.
With Vcc = 5V, min Vin high is closer to 3.5V, see attached.
Vout high for a 3.3V device may not be sufficient, as it also has a transistor in it's output which will have some voltage drop across it.
