I'm pulling my hair out over here reading these responses. I don't understand the reluctance to find a bigger supply?
Even "IF" it's not the immediate issue, some day you will want to operate the LED's at more than 16%, correct?
You are running the supply on the edge of it's limits. It's not good for this setup.
Grumpy,
the power bank is an inland, proHT power bank, with a 1A USB output. I opened it once and it has a tiny board with the usb and a nimh 3.7V 2600mAH battery.
http://www.frys.com/product/8323447?site=sr:SEARCH:MAIN_RSLT_PG
YellowGTM,
the goal i not to run the setup at full power but to make it preprogramed and as light as possible to be wearable, and to make it last 2 to 4 hours.
for the power drain, yes I ballparked it based on the led brightness values, which I assume have a linear or log scale relationship. My brightest mode should be at 80% max power, the lowest uses a few leds and should last for days. if i need to reduce the brightest mode, what should be the target? Most important, how does that solve the stabilization of the vcc voltage if the leds rely on pwm to adjust brightness?
the above statement of course did not account for the non uniform led power consumption peaks.
So did you already try the inductor > capacitor approach?
A ~100uH low ohms inductor and plenty of capacitance behind it.
Leo..
That link says:-
Get fast charging with 1A max output
The LEDs are no more than 16% max brightness, which means 0.8A average.
That is just wishful thinking. Without measuring you have no idea about the current consumption. If you think you have then you are fooling yourself.
Most important, how does that solve the stabilization of the vcc voltage if the leds rely on pwm to adjust brightness?
Using PWM reduces the average current not the peak current. Bottom line is that your power supply is not good enough and no amount of tinkering will get over that.
I did some research online and it seems the leds pwm at 500hz or 2msec, all of the pixels do it asynchronously with an internal clock. this info is not very well publicized.
In theory, caps should allow me to use the battery by handling the current spikes, no?
Wawa,
i did not, what would the schematics look like? i never worked with inductances and do not know how they'd size. a rlc filter would have the possibility of ringing, no? hence the need to tune it well.
wawa... this one? looks like id need 4 to 8 in parrallel
http://www.frys.com/product/2320939
5v_L___R__ led 5V
| |
= C D
| |
gnd___R____
No, that one wouldn't be able to handle the current.
Don't you have a junk box.
Find a small ringcore ~10-20mm, and put as many turns throught it as possible, with ~1mm laquered wire.
Plenty of them on old PC motherboards, or old supplies.
Or remove one from an old dimmer.
Just connect the inductor between battrery and Arduino.
And add several capacitors to ground on the Arduino side.
Leo..
wawa... this one?
No, that can't handle the current.
Something like this
http://uk.farnell.com/bourns-jw-miller/1130-102k-rc/high-current-choke-1mh-2a-10/dp/1103642?ost=1mH+idductor&categoryId=700000005459
In theory, caps should allow me to use the battery by handling the current spikes, no?
No, not if the current drain is too high, as I said you don't know what it is you are only guessing. When you run close to the average limit of the supply the capacitor ripple current rating is important, that means you need a special kind of capacitor.
i can probably poach a power supply yeah...
so what would be ok for a 1A supply? running at 0.5A average?
I could also put caps along the strip farther to power, closer to remote leds. i'll take measurements though i dont have an oscilloscope. i can probably find a friend who does though...
Old motherboards and PC supplies are great for spare parts.
I use a small blow torch to quickly remove them.
Plenty of 3300uF/6.3volt caps. Don't use the bulging ones though.
Leo..
i've measured that the led strip is about 2ohms after 44 leds, so I tried to put the bigger 470uF at the end of the strip, and it reduces significantly the power supply buzzing to a few clicks, without impacting the nano board's boot. This feelz like the right way to go about it, and I'll investigate Wawa's suggestions.
recap: 2812b led brightness is pwm controlled at anout 500hz on each pixel, imdependently of data line or other leds. therefore each pixel can generate a current peak. worst case would be when leds get sync'd up, causing a 5A draw for 15% duty cycle, which is unlikely.
worst case would be when leds get sync'd up, causing a 5A draw for 15% duty cycle, which is unlikely.
What makes you think this is unlikely, I would have thought it is a racing certainty.
it's not a certainty except for murphy's law.
i stumbled on pages of a guy sutdying those led strips.
he demonstrates the pwm is 500hz on an independant clock for each led, not sync'd by data pin signal. hence leds can't be driven faster than 2ms and see
the display.
adding the cap at the end of the led strip allows the nano board to be away from the r-connection point hence avoid the rc ramp up.
Your big 470uF prevented it starting because an empty capacitor connected to a newly switched on power supply demands theoretically infinite current until it has filled up. In practice you get away with your 220uF filling up before there is a problem, but your 470uF may go over the 6A momentarily. Suppose that you insert about 0.5 Ohm or a long lead between the 220uF and the supply. Can you then add the 470uF if it has a 1Ohm between its + terminal and the power supply?
he demonstrates the pwm is 500hz on an independant clock for each led,
Independent free running clocks of frequencies that are close to each other will tend to synchronise due to electrical noise. That is physics for you.
it reduces significantly the power supply buzzing to a few clicks, without impacting the nano board's boot. This feelz like the right way to go about it
...because a few clicks are better than no clicks?
IOW, the power supply is still overloaded.
I'm just going to hang around and watch until you decide to upgrade the supply.
As I sugested before, use a capacitor bank as supply.
And use the weak supply to top up the capacitor bank through an inductor.
Leo..
wawa,
ideally that's where i plan to go yes, likely spreading the caps on the led strip which has significant resistivity. I have not yet sourced a coil.
YellowGTM
no, i won't. i'll try something else instead of bulking the setup with a nonstandard supply. a few clicks is better than a constant buzzing in terms of fatigue and early failure of the power supply. it is in faxt ok for it to eventually fail since it's easy to replace. ideally i'll have to find an oscilloscope and get an idea of the DC signal shape for a full passive L-C fix.
Grumpy
i'll keep in mind that property.
ad2049q
so you're saying the power supply shut down on short circuit detection or similar. what you say is what i achieved by soldering the caps farther on the led strips.