Hi all,
I’m attempting to power a single 4w neopixel plus a pico and a level shifter using a single 18650 lion battery with a boost converter.
The neopixel in question: Ultra Bright 4 Watt Chainable RGBW NeoPixel LED - Cool White - ~6000K | Buy in Australia | ADA5162 | Adafruit | Core Electronics
The proposed battery: Lithium Ion Battery, 18650 Cell (3.7V 3400mAh, positive cap) | Buy in Australia | CE10144 | Core Electronics
The proposed booster: 5V Step-Up Voltage Regulator U3V40F5 | Buy in Australia | POLOLU-4012 | Pololu | Core Electronics
According to that booster converter, I should be expecting around 90% efficiency at 3.7V. So at 4W with 3.7 volts with 90% efficiency, I need to supply some 1.2A which seems well within the maximum of a high-drain 18650 and within the capacity of the boost converter listed here.
I had previously tried (and failed to read the product description, evidently) with this booster:
and this battery from jaycar https://www.jaycar.com.au/18650-2600mah-li-ion-protected-battery/p/SB2299
but the voltage sags to 3V when getting to the neo pixel so im not getting a huge amount of light.
Are my calculations and product selections better this time? Is there anything else I should know before throwing a high-drain 18650 into my project?
Thanks in advance,
J
Addendum - sorry I should probably be more specific about a few things. The end goal here is to run TWO of these high-powered neopixels and I have space for 2 x 18650s in my enclosure. I dont want to run the two batteries in series as Ive heard this is dangerous without proper BMS, so i need to boost from 3.7 to 5v. So the options I have are to attempt to run one battery per neopixel with one boost per, or run both batteries in parallel and use a boost converter which will handle the extra amps (which the one listed above looks like it will)
1 Like
Hi James
Your previous regulator could be a bit light as suggested by this graph.
4W @ 5V equates to 0.8A which is as you see is nearly max current ( a bit over 1A) so the voltage could well have been sagging depending on exact circumstances.
The other graph of interest
Would suggest an efficiency of about 85% at this 0.8A.
Your proposed converter has a much greater capability of about 2.5A + @ 3.7V input
Which is much more robust and allows plenty of head room (or fudge factor as we used to call it)
The efficiency is far better too @ about 93% @ 0.8A
All up Yes I think this one should do the job far better than your original.
Cheers Bob
2 Likes
Hi James
Yes you are correct. 2 cells in series will require a BMS so 2 in parallel would be the go. Might take a bit longer to recharge but as the saying goes about having cake and eating it, there are compromises.
According to the charts you should be able to provide about 2.A at an input of 3V (worst case. You should shut the supply down when the battery gets to about 3V or a little above)
Whatever, leave yourself a bit of extra. Don’t operate right up to the THEORETICAL limit or as things change a bit (with age ??) you will find yourself with problems. So if the graphs are a bit optimistic you could find yourself running out of current and volts. Remember due to component tolerances, ageing etc no 2 devices will be the same so as a result these published numbers must be some sort of compromise and are not the holy grail. But if it is a quality product made with quality components they should be pretty close.
Cheers Bob
PS: Regarding current supply, I personally like to go about 50% up on what I think I will require and usually stay out of trouble that way. But at the end of the day it is up to you.
An example. I have some Xmas lights that probably require up to 5A or so @ 5V. Depending on what the lights are doing at the time. I use a 5V 14A supply. Some might say it is an overkill but hey, I don’t have any foreseeable problems and am confident on adding some more without any hassle. All good.
2 Likes
Thanks for the detailed answer, Bob.
Unfortunately I tend to pick projects where size and weight restrictions are very tight. Where I do have some wiggle room is with brightness / duration - I’ll tend to run the unit for only 5 mins at a time and rarely at full brightness.
The new regulator has arrived and I’m having very similar results to the last one, which surprises me as the math should suggest that I have at least an amp’s worth of headroom - and that’s at full brightness.
Testing scenario and assumptions are as follows:
- One rgbw 4w led neopixel, set to R=0, G=0, B=0, A=255.
- With 255 set to all channels, I would consume 8w so 0.8A @ 5v.
- So just the white channel should pull 200ma.
- Testing with 1 or 2 18650s @ ~3.7V.
Observations:
- When running from a power supply, I do indeed see 5v 200ma at the neopixel (this of course bypasses the regulator)
- When using a battery (18650 with PCM from Jaycar) with no load I see 5v from the regulator output.
- When using a battery (18650 with PCM from Jaycar) with the Neopixel load I see ~3v from the regulator output at around 30ma (pretty close to the result of the last regular I was testing with). I see around 50ma getting pulled from the battery.
- Adding another 18650 in parallel does not appear to change how much current is pulled in total (it just equalises across both batteries).
So I’m at a bit of a loss at the moment. If the 18650s with PCM were only capable of delivering a small number of ma each (which I doubt??) then I don’t see why adding another in parallel would do nothing. My load is only 200ma which should be well within that which can be comfortably supplied by the booster and the battery once efficiency is taken into account.
The only other thing I can think of is that I’m using wires (maybe ~600mm) to the neopixel, but surely that’s not adding so much resistance that can’t be compensated for elsewhere in the system?
Hi James
Maybe it is not the resistance but the inductance of that length of I suppose fairly thin wires running a PWM signal. An easy experiment would be to try again with about 150mm wire.
The difference could be that your power supply (bench supply ??) would be a very low output impedance where the regulator is anybody’g guess without measuring it. If the load is reactive (inductance) this would effectively be like a resistor in series with the 5V so show up as about 3V at the load.
As a point of interest what is the voltage as measured at the regulator output when the load shows 3V.
Cheers Bob
Ok just tried again with ~150mm to the neopixel, same result.
With no load connected the booster shows 5v on the nose. With the led, one channel w=255 I see about 3.3v straight out of the booster.
This isn’t something bizarre with the EN pin on that regulator that I’m missing? 3.3v is close to the voltage supplied by the battery - maybe I’ve done something to set the booster to low power mode?
Hi James.
Just extracted this from the device description text
The regulator is enabled by default: a 30 kO pull-up resistor on the board connects the EN pin to VIN. The enable pin can be driven low (under 0.4 V) to disable the regulator and put the board into a low-power state. However, please note that due to their standard boost regulator topology, the U3V40Fx family of regulators has no way of disconnecting power from the load, so the input voltage will pass directly through to VOUT when the regulator is disabled. The quiescent current draw is typically under 2 mA with no load (see the quiescent current graph below).
Sounds like the regulator is DISABLED so passing battery voltage directly to the output pin. It is supposed to be enabled by default by that pull up resistor. Maybe that is not happening.
Measure resistance between VIN and EN with battery disconnected. Should read the 30kΩ.
It looks like it is working OK until you draw some current, like connecting the load. Then the regulator shuts down and just passes battery current. Strange and as usual no schematic available.
You could try replacing the battery with your power supply set to about 3.5V and see what happens. It could be the VIN sags momentarily when the load connected and disables the regulator. I don’t see any explanation to reset the regulator without removing VIN.
You could also try a resistor of about 20-30k between VIN and EN in case there is something amiss with the fitted resistor.
All very strange especially as this is the second regulator (I think, please confirm). This first regulator you tried does not have an EN pin. It might be a good idea th re connect this and make sure that when it drops voltage it is indeed the battery voltage being passed through.
How long and what size is the wire from battery to VIN. Also the ground wire as this is essentially part of the circuit.
Could you post a schematic of EXACTLY what you have set up. There must be something missing.
Cheers Bob
1 Like
Will do.
For what its worth, just did another test. Shortened the wires from battery so they plugged straight into the regulartor as opposed to the power rail on the breadboard, then into the regulator. Slowly ramped up the colour from a=0 to a=255, and saw the voltage from the regulator slowly drop from 5v to ~4 where the neopixel was r=0, g=0, b=0, a=255. Given this was a smooth drop and not a snap from 5v to 3.3 my guess is that the regulator is working as intended.
Nevertheless, I’ll go through the steps you have outlined and report back.
Thanks again for all your help so far.
Hi Bob, I think you’re right about impedance. I added a single 10v 1000uF cap right next to the neopixel lead and managed to get 4.5v out of the previous test. Placing another 10v 1000uF cap did on the regulator IN did not make any difference. So it sounds like either my cables / breadboard setup or the booster doesn’t like having to deal with PWM.
Sorry just cleaning up my setup and diagram…
1 Like
Hi James
Having a virtual rats nest around a breadboard does not help when dealing with PWM. Cleaning up is a good idea.
When dealing with PWM there are so many variables. Firstly you have the wire resistance which plays a significant role when dealing with larger currents. Especially if your starting voltage is only 5V or even the horrible 3.3V. If you start to lose a bit due to ohmic resistance you don’t have to go far to find yourself running out of available volts.
The other bogeyman is the inductance of the wire itself when dealing with the sometimes fast rise time pulses. The fix here is to use wire as thick and as short as possible to reduce both the resistance and also inductance.
A practical example:
Some 15 or more years ago I used to fix broken Golf buggy speed controllers for friends at the local Golf Club. The earlier ones were simple PWM generators driving a N channel Mosfet as a low side switch with a flyback diode across the motor. Simple and worked pretty well. These were required to switch up to about 20A.
I had my own variable load for test purposes. This consisted of 2 buggy motors coupled together. The idea was to drive one and use the other as a generator with a system of 12V auto globes for an adjustable load. Worked well.
Sometimes (read mostly) it was not convenient to use the buggy battery so I had a bench supply connected via an extension cable of about 1.5metres. I point out here that the existing battery wires are very short and difficult to juggle the connection while trying to trouble shoot or any meaningful work.Hence the power supply and cable extension.
My first problem came while I was trying to set up a low battery cut out bit of circuitry to an existing controller. When the controller was under some load the voltage measured at the supply actually went UP so when I tried to reduce the power supply voltage to emulate a lower voltage what I was reading was meaningless.
The cause… The inductance of my extension cable was causing a charge pump effect and the generated voltage added to the power supply making any adjustment efforts futile.
The fix… 10000µF electrolytic across the supply right at the controller and motor input. I learned at this stage that this is a common problem when using longer leads and some controller boards actually have provision for this if required. I might point out that with the very short battery leads in this application this cap is not required. It was my extension that caused a problem.
Cheers Bob
Hi Bob,
Just closing the loop on this. After shortening wires to 150mm and switching to 22 awg, 1 x 18650 is now running 1 x 4x neopixel with full rgba with no perceiveable sag in voltage. Rock solid at 5v. Thanks for all your help.
