Need cheap USB power bank with "always on" instead of "auto off"

[Peabody]

I've looked at the little single-18650 $1-3 power banks (battery not included), but all seem to have a "feature" that I don't want.  If the load drops below some value, say 50mA, they shut down.  I assume that's because the boost converter would run down the battery if allowed to keep running.  But I would like to use one of these for very low current projects, such as Arduino or MSP430 stuff.  So I don't want it to turn off.  Instead, I want an On/Off switch on the battery's positive terminal.

I found an always-on power bank on Amazon, but it's $25.  Does anyone know of a product that costs more like $5?
Or is there another solution to this problem that I should look at?

If it matters, what I want to do is have a device that can be connected via USB to a computer for power and for firmware updates, just like an Arduino, but which can also be detached from the computer and powered through the same USB connector from a USB power bank.  But I need that power to stay on, even during sleep.

[sokoloff]

If you're willing to burn a 1/4 Watt anytime the switch is on, you could just wire a 100 ohm resistor across the USB +5V and GND pins.

[kripton2035]

what about this kind of device ?
https://www.aliexpress.com/item/32870411748.html

[darrellg]

Here's one with a 20 mA cutoff: https://www.sparkfun.com/products/15204.

[Peabody]

I've found one that may work.  Julian Ilett did a video on it five years ago:



and that Ebay listing is no longer up.  But this looks like the same package:

https://www.ebay.com/itm/Aluminum-DIY-USB-Power-Bank-Case-Only-18650-Battery-Charger-2600mAh-Cover-Kit/361693405448

Of course in five years the innards may have been completely changed out, but if not, the video indicates that there's no auto-off, and it has an On/Off button.

what about this kind of device ?
https://www.aliexpress.com/item/32870411748.html

I think that one is good in not having an auto-off, but the switch doesn't switch off the battery.  It just turns off power to the output USB connector.  And it has no proper "load sharing" circuit to allow charging while powering the load.  There is some chance the first device does have that, but there's no way to tell without testing it.

[westfw]

If you're willing to burn a 1/4 Watt anytime the switch is on, you could just wire a 100 ohm resistor across the USB +5V and GND pins.

Does anyone happen to know I_q (current drawn from battery when load is 0) for the common Li-ion "power packs"?For example, is it comparable to the "turn off" current?

[Peabody]

Provided actually running out of power isn't a concern, and you can deal with 3.0-4.2V, you could just wire your load in parallel with the battery and charger/regulator.

I don't understand what you're suggesting.  A load sharing circuit lets USB directly power a 5V load in parallel with the charger/battery/regulator, all inside the power bank.  Are you suggesting something different?   

[Peabody]

If you're willing to burn a 1/4 Watt anytime the switch is on, you could just wire a 100 ohm resistor across the USB +5V and GND pins.

Does anyone happen to know I_q (current drawn from battery when load is 0) for the common Li-ion "power packs"?For example, is it comparable to the "turn off" current?

I haven't seen much information on I_q for power banks, I guess because they are intended for the retail market where such information woudn't be understood.  But a few of the reviews on Youtube measure that, and I think it's typically tens to a few hundred µA.  That's after the regulator has been shut down.  In contrast, the current below which it automatically turns off is usually 20-50mA, which just means they are completely unsuited for things like Arduino projects which are often put to sleep.

Apparently these power banks are pretty much specifically intended for recharging batteries, and I guess the theory is that if current has dropped to 50mA, that battery has been charged, and it's time to shut down.

[soldar]

Can the power bank be taking charge while delivering current? Because in the case you could have a (correctly isolated) power converter taking current from the output and sending the power right back around to the input. If you are lucky you might get more than unity and become immensely rich in which case please remember who gave you the idea.

[Peabody]

Yeah, with something like that we don't need any of that fusion rubbish.

Sounds like we need  a new Kickstarter for this.  Then Dave can "review" it.

[Peabody]

I just wanted to report that after two months, and a refund from the seller, I finally received the single 18650-based power bank I referred to above that looked like the one Julian Ilett reviewed.  But my suspicions were confirmed that the circuit has been completely changed, and most of the parts have been replaced by a single chip that does both the charging and the boost conversion, and that chip has "auto-off" built in.  The datasheet doesn't say what the minimum stay-on current is, but I stopped testing at 20ma, which produced a shutdown after 18 seconds.  So that's a fail.

So I'm still looking for an inexpensive power bank or module that could be used as a UPS for an Arduino project, preferably 18650-based, one that stays on no matter how little current the powered device draws.  So it would have a micro- or mini- USB power input, type-A USB output, and would include the charger circuit, a boost converter, and a load-sharing circuit that allows powering the load while charging the battery.  If anyone knows of one, please let me know.

The circuit would basically look something like the attached.

[sokoloff]

Could you just drive your Atmega328p directly from the Li-Ion battery (3.7V nominal)? The part is 1.8-5.5V qualified, so it depends on your specific peripherals, but for two Halloweens in a row, I ran our costumes directly off a single 18650 for several hours of being "LED stick figures". I saw no unusual behavior across 6 people and probably 6+ hours of runtime per person.

We didn't charge the batteries while wearing the costumes, but I see no reason to think that couldn't also be done.

[Peabody]

Yes, that would work, but only at 8 MHz, and any peripherals used in the circuit would also need to tolerate that voltage swing.  I also wanted to be able to power it from an AC USB charger if available.

[elektrolitr]

If you don't mind a caseless solution

either go with this module which combines battery charging and step-up:
4056 module
This module doesn't have a battery protection so it's a good idea to pair with protected battery. I like to use cell-phone batteries which I salvage across my relatives, but you can as well buy a li-po battery with protection circuit on ali or ebay.

I've just checked and it stays on for long (>10 min) with only multimeter connected, seems not to have power off feature at all

For unprotected batteries, go for this charging/protection module:
charging and protection module
and add a step-up of your choice (like this one). Separate step-up modules tend not to have a "green mode", they just deliver the output as long as input is in spec.

[kjr18]

This TP4056 module is probably based on MT3608 boost IC or something like that. Those chips have enable pin, but they are hardwired to enable position. With this one is one possibility of overcharging. TP4056 might have problems with termination of charging. I did module like this some time ago, I added one p channel mosfet, one gate resistor and one lowdrop diode. I made it that way: when there is no external voltage source connected, transistor is open and boost is powered through it. When external source is connected, gate of mostet is pulled high and boost is powered through diode. That way battery is disconnected from boost, and charger can terminate properly.


And about powering straight from battery: once I tested one of my small projects: neopixel based random colour flashing leds. My arduino pro mini was happily running @16MHz and about 2,5V.

[sokoloff]

Similar experience here, though I tried to only draw them down to 3.5V before changing.

A Li-Ion battery at a 5-hour discharge rate or lower (0.2C) is going to stay above 3.6V for about 90% of the discharge curve.

[Peabody]

@elektrolitr, I was hoping to find a power bank that already had both USB connectors in a nice case.  But at this point I only know of one like that which is also "always on", and it's $30.  If I can't do better than that, I'll have to resort to modules.

But I'm surprised that so far I have not even found one single-module solution that's a combination LIPO charger and boost converter with load sharing.  It seems the module kings of the Far East could build a combination module like the one you referenced, but using the MCP73871, which has load sharing built in, instead of the TP4056, which doesn't.  Or use the TP4056 but add the mosfet, diode and resistor needed for load sharing.  I don't understand why there is no such module available on Ebay.

With respect to just using a battery, I too have done that for an IR remote control using an 8MHz Pro Mini.  But that won't work for all projects.  In addition to the processor speed issue, there are many peripherals such as sensors, SD cards, and so forth that need to operate at a specific voltage - some at 3.3V, some at 5V, but many may not work at 3.5-4.2V.  So powering directly from a LIPO may work fine for some projects, but not for others.

Anyway, I still think the world needs an "Arduino powerbank" for a lot less than $30.

[sokoloff]

You might be able to run the AVR directly off the LiPo and use the micro to enable/disable an LDO that drives a stable 3.3 for peripherals. (That doesn't help at all for +5V sensors and peripherals, of course.)

[Peabody]

You might be able to run the AVR directly off the LiPo and use the micro to enable/disable an LDO that drives a stable 3.3 for peripherals. (That doesn't help at all for +5V sensors and peripherals, of course.)

You can provide 3.3V that way to power the peripheral, but if you power the AVR at 4.2V, its GPIO pins will be trying to communicate with the peripheral at 4.2V, which is probably half a volt above the peripheral's absolute maximum voltage on any pin - based on Vcc + 0.3V, or whatever.  So I think once you get beyond things like LEDs, you start running into these kinds of problems, and a regulator is going to be needed.  Of course some things will work even if they aren't supposed to, but you don't want to be pressing your luck too much.

[elektrolitr]

This TP4056 module is probably based on MT3608 boost IC or something like that. Those chips have enable pin, but they are hardwired to enable position. With this one is one possibility of overcharging.

Nope
TP4056 is a chip made specifically for Li charging. It should have all bells and whistles needed including the current cutout at the end of charge. See the attached picture.

I can check the behavior of the module a bit later, though.

[elektrolitr]

I've made a quick check of charging current cutoff



The chip is working good. The cutout was at 100mA which is 1/10 of programmed charging current just like in datasheet.

The only downside of these small modules is that all of them are programmed to 1A charging current which can be on the high side for small batteries and you have to change a resistor (usually of 0805 size) to bring the current down.

[kjr18]

This TP4056 module is probably based on MT3608 boost IC or something like that. Those chips have enable pin, but they are hardwired to enable position. With this one is one possibility of overcharging.

Nope
TP4056 is a chip made specifically for Li charging. It should have all bells and whistles needed including the current cutout at the end of charge. See the attached picture.

I can check the behavior of the module a bit later, though.

But of course, I know that. But you showed a module with boost added, not only TP4056 but also chip with marking B6288, which is MT3608 by Aerosemi. Try measuring cut-out while drawing larger current from boost converter.

[elektrolitr]

Thanks, I probably got the point.

Load current still goes through the 4056 and prevents it from cutting down the charging current, right?

I checked and it really doesn't cut out if there is a load connected.