Battery Operated Device Power supply

[paulca]

I just got some interesting results.

I was running a breadboard project (prototype of a binary clock) with 6 LEDs switching on and off constantly, maybe an average of 10% LED usage.  An Arduino Nano, an RTC and a 74HC595 shift register.

I connected a nearly fully charged 18650 (4.1V) directly to the breadboard and left it running to see how long it would go for.

A week later (last night) the bits for the rest of the proposed power supplied arrived.

A TP4056 + 5V boost module.

So I checked how the battery was doing after the week, 3.56V.  So it was doing fine.

I decided as 3.56V was getting awfully close to when the Nano would brown out and crash I would add the TP4056 and Boost module to see how long it would continue running.

It was dead this morning.  So it ran for a week on 4.1V - 3.56V, but lasted only 9 hours with the modules in place.

Even if you include inefficiencies of the boost converter that performance stinks!  The boost converter was claimed to work down to 2V, but the TP4056 would cut off at around 2.5V.

Of course this make the concept of the power supply rubbish and I would be better off running it direct from an 18650 which I guesstimate would run it for 2 weeks.

[danadak]

If I look at this datasheet I do not see anywhere it is guarenteed down to 2V
Vin operation ?


https://dlnmh9ip6v2uc.cloudfront.net/datasheets/Prototyping/TP4056.pdf


Regards, Dana.

[paulca]

The boost converter module claimed it operated down to 2V (1.5V actually).  Of course the TP4056 cuts out around 2.5V to protect the Lithium cell.

The annoying thing is how it managed to get the battery from 3.5V down to 2.5V in 9 hours.

I need to test the current draw and compare it to the raw 18650.  Obviously the AVR chip will brown out, probably around 3V, but if the boost converter is going to suck up that much power it would run longer with out it.

Also, it is the TP4056 V2 board which include the protection IC as well which disconnects the cell when it's voltage drops below around 2.5V, it's not the actual TP4056 which cuts it off.

[StillTrying]

"The annoying thing is how it managed to get the battery from 3.5V down to 2.5V in 9 hours."

I only use a couple of 18650s, during discharging and charging I get the impression nearly all of their capacity is between 3.6V and 4V, and very little of it between 3.6V and 2.7V.

One day, I might* do a battery powered LED clock, with a photo transistor facing forward so that it turns the LEDs off if it hasn't seen any movement for a while.

*= probably never.

[soubitos]

This is normal behavior..
The battery discharges the bulk of its capacity from 4.2 down to around 3.2V

The useful range of the battery is from 2.8-4.2V
The TP4056 module you got most probably has a DW01x protection IC which allows for deeper discharge.
I use FS312F-G chips which will shut down at 2.9V...
If you use the battery as is with any load, you risk to over-discharge it... do this a few times and your battery will wear out too soon. Also, you must take care when recharging it after such a deel discharge, 100mA for a new 18650 until it reaches 3-3.1V is normal...

Even if you dont use the booster circuit and power your load directly from the TP4056 module, it is far better than powering your load from the battery without any protection.

[Peabody]

The TP4056 datasheet says that when it's shut down (USB not connected), the BAT terminal sinks less than 2 uA.  And the DW01 datasheet says its operating current is typically 3 uA.  So it's unlikely either of them would run down the battery very fast.

I assume your boost converter's efficiency is on the order of 90%, but you could probably test that with your meter.

Then there would be the question of the power consumption of your load circuit when run at 5V versus when it's directly powered by the lipo at a lower voltage.

Maybe you could do a test with and without your charger and booster, with the lipo fully charged in both cases, and see what the difference is in how long it takes the battery voltage to drop to, say, 3.8V.

[paulca]

Maybe you could do a test with and without your charger and booster, with the lipo fully charged in both cases, and see what the difference is in how long it takes the battery voltage to drop to, say, 3.8V.

This is the plan.  I am pondering how I log the current draw over time.  I suppose an arduino with a current sensor module would do.  I can ping the amps and volts out over serial and save it on the PC.  Alternatively I could use an ESP8266 and send it over Wifi.  (Obviously I don't have a data logging DMM).

Still to test the discharge for the life of the battery could take over a week!  I think I will just do a 12 hour test or something.

[paulca]

This is normal behavior..
The battery discharges the bulk of its capacity from 4.2 down to around 3.2V

At 3.56V it should have been somewhere around 40-45% capacity remaining.  It started at at least 90% and ran for a week.  So 50% capacity ran it for a week, but 40% capacity with the boost converter ran it for 9 hours.

My suspicion is the boost converter is using too much power.  I might desolder it's LED but a single LED shouldn't be that much of an issue as the Nano has a power LED and the project itself blinks 6 LEDs.

[mikerj]

This is normal behavior..
The battery discharges the bulk of its capacity from 4.2 down to around 3.2V

At 3.56V it should have been somewhere around 40-45% capacity remaining.

Nowhere near 40%.  Maybe 5% or so at best.

[paulca]

This is normal behavior..
The battery discharges the bulk of its capacity from 4.2 down to around 3.2V

At 3.56V it should have been somewhere around 40-45% capacity remaining.

Nowhere near 40%.  Maybe 5% or so at best.

Not from the evidence I have seen.  What is your reference for that?

5% would occur around 3.0V or even 2.9V

Also it depends on the load as to what the voltage would be.  At sub 100mA, 3.5V is roughly half way down the nominal flat area on the discharge curve, so around 40/45%.

[nrxnrx]

Hi!

https://www.google.com/search?q=li-ion+discharge+curve&tbm=isch

The above link is an image search. I couldn't decide which graph to attach. 

[paulca]

It's interesting because if you change the search to 18650 discharge curve you get completely different graphs.  However both searches show a wide variety.

In my experience when my 18650s are reading 3.6V there is plenty of time left and if you consider it, the "nominal" voltage is 3.7V.  The nominal voltage is the usually the mean point on the main "flat" part of the discharge curve.

In my experience, granted mostly with LiPos...

4.20V to 3.8V is a very steep drop covering around 15-20% of capacity.  The bulk of capacity is between 3.8V and 3.3V when it starts to steepen off again.

All of this highly depends on the load, a lot of those graphs are probably wrong as they are constant current at 1A discharges which is not totally representative of actual capacity.

If you "flatten" an 18650 at 1A to 2.9V and then wait the voltage will rebound back to around 3.2V or higher.  If you then discharge it further at 100mA you will get quite a bit more out of it.

[mikerj]

It's interesting because if you change the search to 18650 discharge curve you get completely different graphs.  However both searches show a wide variety.

In my experience when my 18650s are reading 3.6V there is plenty of time left and if you consider it, the "nominal" voltage is 3.7V.  The nominal voltage is the usually the mean point on the main "flat" part of the discharge curve.

If you were getting ~3.6v under a decent amount of load I'd agree, but it sounds like you weren't drawing much current (I suspect well under 1C).

[paulca]

I'll accept the battery voltage under light load, for now.

Anyway.  So I let the 4056 charge it until it was happy it was fully charged.  The blue LED was flickering on and off for an hour.  Probing the battery showed 4.13V.  Not exactly accurately fully charged, but it was still under a small load from the prototype.

It lasted 24 hours.

So the efficiency is in the order of 20% compared to the direct 18650.

I'm not sure I can be bothered to set up a logging feature to record it's current draw but I will test instantaneous draw today to see what is what.  Without anything, with just the TP4056 and just the boost converter.

Obviously this solution is not going to work.  I would like the device to run for at least a week on 1 cell.  Basically I don't want to have a permenant USB power connection to it.  I just want to plug it in from time to time to refresh it while I'm charging my other gadgets.

A few asides.  The TP4056 seems to default to 0.5A charge.  I'm not sure if I should investigate the sense resistor, maybe the V2 boards come configured for 0.5A, the V1's I believe came configured for 1A.

The other thing that makes me wonder.... I cleaned the boards after soldering with what I was sold as "99.9% Pure Isopropanol Alcohol" off ebay (I know I should have just gone to the chemist).  However the boards are sticky.  So the alcohol has left a residue.  Now what are the odds the residue is semi conductive and drawing power.  I think it's a long shot as the boards are not getting warm to the touch, but then again I am working with a few tens of mA.

[soubitos]

is your battery good?

[paulca]

is your battery good?

Brand new, though cheap, generic laptop cell, 2600mAh.

[Kalvin]

Here is a typical Li-Ion battery discharge curve:

http://www.ibt-power.com/Battery_packs/Li_Ion/Lithium_ion_tech.html

If the battery voltage was at 3.56V it had only about usable 10% energy left, which matches with your experience.

[paulca]

I'm having one of those "Mandela effect" moments.  But I concede.

Thinking about it, I have always dealt with 3 cell packs.  The general rule of thumb there is, anything over 10V and it's fine, with 11.1V being the mid charge point.  When it gets below 10V it's going to run out soon, when it starts to fall from 10V it will fall quickly.

I have negative re-enforcement in the form of an RC LiPo alarm set at 2.8V.  If the cells get there I get a very loud beeping siren like a smoke alarm for my pleasure.  Scares the shit out of you.  PHEEPEPEPEPEPPEPEPEPEPEP!

RC people don't do low voltage cut outs.  If you are flying a plane or worse a heli, the battery is worth a lot less than the model, so a loud audible alarm you will hear over the motor is used.  Some speed controllers do have a soft cut which drops your power output around 3.3V to give you warning and slowly ramps it back to nothing below 3V.

In those spaces where size and weight are important, you are usually pulling 10C or 20C from a pack .  My binary clock is pulling around 0.01C

That all said, it went from fully charged to cut off in 24 hours with the PSU modules in place.  Without the modules it went for a week.

[Peabody]

@paulca, I hope you will run this down and find an answer.  I'd certainly like to know why there is such a big difference.  I mean, there are things that would contribute to there being some difference, but nothing comes to mind that would account for what you're seeing.  So I hope you'll do the current measurements, etc., and give us a full report.  :-)

[Kalvin]

That all said, it went from fully charged to cut off in 24 hours with the PSU modules in place.  Without the modules it went for a week.

Oops, I missed that one. Probably you could do some current measurements and figure out what is going on. You could measure the current consumption of the Nano at 3.5V and 5V, and check the module's current consumption with and without Nano while varying the module's input voltages (4V to 3V).

[paulca]

So I pulled the boost converter an reconnected the TP4056 on it's own powering the circuit.  The first thing that happened was the TP4056 instantly became convinced the battery was fully charged.  Multiple resets confirmed.  So the boost converter is pulling the voltage down.  Battery measures 4.12V, I'd prefer 4.20V, not great, but that's fine.  With the boost converter connected I could barely get a flicker out of the fully charged light.  Maybe something wrong there with the boost converter.

I'll run it without the boost converter see how long it lasts with just the 4056 and might play with the boost converter on it's own on the breadboard, see if I can come up with a way to test it's efficiency and see if it has a fault or is just rubbish.

[soubitos]

what chip is on the boost converter module?

[Rick Law]

I just got some interesting results.

I was running a breadboard project (prototype of a binary clock) with 6 LEDs switching on and off constantly, maybe an average of 10% LED usage.  An Arduino Nano, an RTC and a 74HC595 shift register.

I connected a nearly fully charged 18650 (4.1V) directly to the breadboard and left it running to see how long it would go for.

A week later (last night) the bits for the rest of the proposed power supplied arrived.

A TP4056 + 5V boost module.

So I checked how the battery was doing after the week, 3.56V.  So it was doing fine.

I decided as 3.56V was getting awfully close to when the Nano would brown out and crash I would add the TP4056 and Boost module to see how long it would continue running.

It was dead this morning.  So it ran for a week on 4.1V - 3.56V, but lasted only 9 hours with the modules in place.

Even if you include inefficiencies of the boost converter that performance stinks!  The boost converter was claimed to work down to 2V, but the TP4056 would cut off at around 2.5V.

Of course this make the concept of the power supply rubbish and I would be better off running it direct from an 18650 which I guesstimate would run it for 2 weeks.

I think before we look too deep about efficiency...  Let me first understand...

From your description, I assume you are running it like this:

battery -> booster -> TP4056 -> Nano

This is not a very good layout.  TP4056 is a charge-management IC.  It makes no sense to supply it with the battery.

Each step you are loosing power.  So, from battery (3.x to 4.2 volt) you first boost to 5V (efficiency lost), then feed it into to TP4056 which regulates it back down to charge voltage (4.2 - 3.x volt), and if you tap the battery into NANO Vin, you have yet another regulator which drops the voltage a bit further.

Each step you are loosing power.  The worst lost is at the boost stage - even if component is working at 100% efficiency, for the same amperage drawn, boosting from 4V to 5V means you lost a quarter of your power right there.  Since the battery operation is working well for you, there is no reason to first boost it.  With that in mind, you should reconsider how you stage each module.

If you want to charge (that's what the TP4056 is designed for), it would be silly to run the TP4056 via the battery.
So, the running would be:
Battery-->NANO +5V (bypass the voltage regulator in NANO at Vin)
or if you want +5V for the NANO
Battery-->Boost-->NANO +5V

(Warning, I am a hobbyist with limited experience, but this is the way I would try)

If you want to charge with the TP4056 while you run, you need an external source of 7.5V ish.
+7.5V --> TP4056-in
+7.5V --> NANO Vin (not NANO +5V) using NANO's voltage reg to bring 7.5V to 5V.
plus
TP4056-out --> battery
TP4056-out --> NANO USB Vin (which has a power selector diode)

When 7.5V external power is ON, NANO is powered from Vin, and TP4056 is powered.  Battery is charging via TP4056-out.  TP4056-out also connects to the NANO USB Vin giving it up to 4.2V max.  But the USB Vin has a "power selector diode".  So, it will draw power from the 5V regulator and the diode also prevents current to flow from NANO's 5V back to battery.
When 7.5V external power is OFF, NANO is powered via the USB's Vin which is your battery (connected to TP4056-out).

Since there are a ton of NANO variations out there, I'd experiment with it first.  There may be NANO's out there without the "power-selector diode".   Run it with TP4056 powered but without battery to make sure that the TP4056-out (USB Vin) is sitting at or below 4.2V.  Connect the battery and make sure it is <= 4.2V, and it terminates charge when full.  Otherwise, you would be flow charging the battery which would not be good.

Note that your voltage at the NANO's +5V is really 5V ONLY when the 7.5V source is ON.  When it is off, the NANO's +5V is your battery voltage after the diode drop, which would be at best close to 4.2V.  Inserting your Boost module there (between battery and NANO USB in) would mean that the TP4056-out would be connected to both the battery and boost module.  That could mess up charge management a little or a lot.

It mess up by a lot when your boosted voltage exceeds the NANO's regulator output voltage.  If (say for example) your boost output is 5.1V into the NANO's USB-Vin and your NANO's Vin regulator output is lower at just 5.0V, the "power selector diode" is not blocking and power is therefore coming from USB-Vin (ie: from battery).  So, the TP4056 is actually powering the battery charging and the NANO plus breadboard at the same time.  You will likely be drawing enough current for the TP4056 never to reach full-charge-shutdown.  So your battery would be float charging until 7.5V source is off.  To prevent that, your boost must be lower than the NANO's 5V regulator output.  I would prefer not to boost at all (clean), but if I must, I would set boost to about 4.5V ish to make sure that boost-out voltage is always less than NANO's 5V regulator-out.  Then, I am sure when the 7.5V external is ON, my boost circuitry is mere idling supplying no power to the NANO.  The TP4056 terminates at about 10% charge current, so when your boost circuitry is merely idling, the current draw may be low enough (less than 1/10 C) for the TP4056 to decide "charge full".  You need to experiment with it well.  Float charging will kill your battery for sure.

Good luck.

[paulca]

Battery -> TP4056 Bat +-
TP4056 Out +- --> Booster IN +-
Booster 5V +- --> Breadboard supply rails
Breadboard supply rails --> Nano 5V pin (+ the RTC chip + The 74HC595 shift register + the LEDs)

The TP4056 board has a 5V input via a Micro USB for charging.  The tp4056 board is designed to drop 5V to 4.2V for charging with a current limiter, so CCCV charging.

The only downsides of running the circuit off the Bat +- direct or the output of the TP4056 direct is that the LEDs are not as bright and they gradually get dimmer as the battery discharges.  The Nano will be experiencing the same thing.

The DS1302 RTC has a Vcc range of 2V to 5.5V
The 74HC595 has a Vcc range of 2V to 6V
The nano 5V Vcc power pin has very little documentation, some claim 6V max, but the AtMega328P voltage range is 1.8V to 5.5V, however it will not operate with the 16Mhz clock 1.8V, the datasheet lists 10Mhz as 2.7V and 20Mhz as 4.5V, so it will brown out, probably around the time the TP4056 cuts it off anyway.

Of course this is a prototype.  The 74HC595 will be replaced with a TLC59282 LED Driver and drive 16 LEDs in total in the real thing.  The Nano will be replaced with an actual ATMega328P.

The extra leds and the longer duty on them will make the battery last less long.  Of course, it is possible the boost converter needs a minimum load and the load I'm currently pulling might make it highly inefficient.  Like a camel it might run better with a heavier load.

I can also put the 328p to sleep, maybe even deep sleep and have the RTC wake it up every seconds.  Needs investigation.

This is the boost:
https://www.ebay.co.uk/itm/Dc-Dc-Boost-Converter-step-up-Power-Module-1-5-5v-to-5v-500mA/263493774340?hash=item3d5973ec04:g:uPYAAOSwuGxZlnwv

[soubitos]

This is the boost:
https://www.ebay.co.uk/itm/Dc-Dc-Boost-Converter-step-up-Power-Module-1-5-5v-to-5v-500mA/263493774340?hash=item3d5973ec04:g:uPYAAOSwuGxZlnwv

I havent seen this module before.. can you check the "active" parts and post what they are (codes/markings obviously)

[paulca]

It has a 3 pin + tab with marking

L5  (might be E5)
OP

Which I assume is the boost IC.

A large 2 pin device marked

SS14

Which I assume is a diode.

An inductor marked 220 which I assume is a 220nH inductor.

An LED with a 222 SMD resistor and a few caps.

[paulca]

E5OP would make it a TP8350 boost IC.

[Ian.M]

For maximum battery life, you'd probably do a lot better with a nominally 3.3V 8MHz Arduino, board, (or better a bare Atmega328P so you don't have any extra power drain)  and a low quiescent current buck converter down to 3.0V.    Even bucking to 3.3V then letting the output voltage follow the input as the 18650 cell drops below about 3.4V  would be a vast improvement.   

As you are building a clock, you *may* be relying on the CPU clock for accurate timing.  Rather than running a 8MHz crystal primary oscillator all the time, you'll save power by providing a 32.768KHz crystal for the Timer/Counter 2 Oscillator, and use the 8MHz calibrated internal osciillator as the primary clock, sleeping as much as possible with wakeup by timer 2.   See http://kineticsandelectronics.com/sleep.html

[paulca]

I'm using a DS1302 RTC with NiMH battery backup.

[soubitos]

It has a 3 pin + tab with marking

L5  (might be E5)
OP

Which I assume is the boost IC.

A large 2 pin device marked

SS14

Which I assume is a diode.

An inductor marked 220 which I assume is a 220nH inductor.

An LED with a 222 SMD resistor and a few caps.

I am no expert to calculate how the changes from manufacturer recommended values vs installed values affect efficiency in this module but....

L1 is 47uH (resistance <0.1 ohm)
a 47uF cap is in the input and a 100uF cap or a 20uF with a 0.05 ohm resistor in series are in the output

With these, they claim 80% efficiency when 3.2V<Vin<4.2V. Vout is 5V and Iout is 150mA<Iout<500mA

I think with a very small load the efficiency of this chip is going way down. Same for Vin bellow 3.2V with 5V output

[paulca]

With these, they claim 80% efficiency when 3.2V<Vin<4.2V. Vout is 5V and Iout is 150mA<Iout<500mA

I think with a very small load the efficiency of this chip is going way down. Same for Vin bellow 3.2V with 5V output

Yep.  Sounds fair.  My prototype is probably only pulling at most 20mA.

I know I promised current measurements, but I'm trying to fix a PC OS issue that came up which has broken a number of things on me.

[soubitos]

Actually, its mentioned in the datasheet
4) The customer can reduce the inductance (about 22uH) if it drives large current load (more than 150mA) and the ripple requirement is not high. Customers who drive small current loads (less than 50mA) and want low ripple output voltages can increase their inductance

If you have an inductor around 47uH with Isat over 1A try to replace the 22uH inductor with it... also i would try replacing the output capacitor with an electrolytic 100uF in parallel with a 100nf ceramic

[Kalvin]

If I have understood correctly you want to run Nano and the LEDs from the re-chargeable Li-Ion battery, and you want to keep the LED intensity constant. As the battery voltage drops the LEDs will be getting dimmer, you want to run device from +5V power supply.

About the Li-Ion battery: The battery is almost dead when it reaches 3.3 so the voltage range you effectively get is 4.2V down to 3.3V, and you have used more than 90% - 95% of the battery capacity when the battery voltage is 3.3V.

You may want to run Nano directly off the battery voltage 4.2V ... 3.3V as the Atmega328 runs fine between +1.8V ... +5.5V.

The 74HC595 is specified for 2V ... 6V operating voltages, so you can run it directly from the battery as well.

About the LEDs: The LEDs will consume few milliamps each, and a low-power linear regulator will consume around 100 uA maximum. So, one option might be to use a low-power low-drop voltage regulator to power the LEDs in order to keep the brightness constant without affecting the power consumption too much. There will be some losses due to the nature of a linear regulator, but you are using the current limiting resistors with the LEDs anyway, so there won't be any extra loss compared to running the LEDs directly from the battery with current limiting resistors.

[paulca]

Interesting, as I was going to be using the TLC59282 LED Constant current Driver that allows me to set the current for all LEDs with 1 resistor.  It sinks the current from the LEDs ... so I can run them off a 3V rail.  (I'll have to check the IC doesn't mind).

Actually, I added an I2C header on the board for an optional oLed.  When I start talking about oLeds and potentially a dozen LEDs on at a time, a battery solution that will run for a few weeks it probably going to be looking at multiple 18650s in parallel, even if I do get it to be efficient.

I'll have to test the oLED, but 10mA sounds fair. 

If I consider a ball park duty cycle of 50% of the LEDs on at a time, that's 8 LEDs (8.5 if you count the AM/PM light).

10mA oLED screen
32mA 4mA*8 LEDs
uC 10mA
neglibilbe RTC

That's 52mA, call it 60mA.  2600mAh battery, even if it gave it's full capacity range, would last 43 hours.  More likely 36, one and a half days.

To get a week I would need 4 18650s in parallel.

It's starting to sound like a USB powered device.

[Peabody]

My guess is that the efficiency of your converter is not good at the current you're drawing.  I just finished converting my scope to lipo power using an MT3608 boost converter module, and it seems to work fine going from a single cell lipo to 8.2V.  It's more expensive, but runs at 1.2 MHz, and from the datasheet it looks like you'd be at over 90% efficiency even at low current.

But you have several things working against you as far as battery life is concerned when using the boost converter versus powering directly from the battery:

Your circuit probably uses less current when run at 3.5V than at 5V.

Even with 100% efficiency, converter power in must at least equal power out.  So if you need 25ma at 5V for your circuit, you'll need 36ma at 3.5V coming out of the battery.  Powered directly from the battery, you might only need 20ma at 3.5V.

No converter is 100% efficient, and may require tweaking for the current you will be drawing.

Drawing more current out of the battery drops its instantaneous voltage a bit more, which makes it hit the protection low-voltage disconnect earlier.  So you're actually using a lower percentage of the battery's capacity.

For all these reasons, it wouldn't be at all surprising if battery life was cut in half using the converter, even with a very efficient converter.  But you're off by 6 to 1, and unless you just got a bad copy of the converter, I suspect it just isn't at its best at low current.

The current measurements should tell you a lot about what's going on.

[paulca]

Well I sent the board for manufacturer, so I'm commited now.

Of course the board does not include the PSU   It just +5V and GND headers.

I think what this will end up is a USB desktop powered device that has a battery backup that will run it for a few days.

[paulca]

On the prototype power situation...

It's still running on the same charge since my last post on removing the boost module.  Cell voltage is 3.2V, Nano and LEDs all seem happy though a little dimmer of course.

I wonder how long it will go for.  Anyone wanna place bets?  I recon it will make it until Tuesday.

[paulca]

Nope.  It just died.  Nano locked up first, LEDs dimmed quite rapidly and the TP4056 cut out.

[soubitos]

there is something wrong about this...

[soubitos]

If i remember correctly, you mentioned you use a brand new cell...
1. have you done any capacity test on it?
2. have you measured the current draw before and after the booster?
3. In ideal conditions a 2600mAh cell directly powering a 100mA load should last... 26hrs if i am not wrong?
With a booster working at 80% efficiency this would drop to near 20hrs... so, its either you expect it to keep going for much longer than it can, the battery is of much lower real capacity or the current draw is much higher than you estimate or a combination of the above... plus the fact that MT3608 at low loads is NOT efficient at all

[Peabody]

And he should look at the circuit's current draw at the booster's 5V output vs when powered directly by the lipo at, say, 3.6V.  The circuit will likely draw more current at 5V.

Also remember that even if the converter is 100% efficient, when going from 3.6V to 5V, the current drawn from the battery will be 139 ma to provide 100 ma at 5V.  Power in must equal power out.  Then for 90% efficiency, that would change to  154 ma out of the battery.

About the MT3608, the graphs in the datasheet show the efficiency to be quite good, especially at low currents, down to about 40 ma.

[viperidae]

Your boost converter could be chewing 2mA quiescent current.
Your leds will be using a lot more current at 5v too. That extra volt is going to end up across the current limiting resistor with a corresponding increase in current flow.

[paulca]

Sorry, I think I've confused everyone.

The final circuit will run 16 LEDs and an oLED constantly with an estimated average 8 LEDs on at a time.  It uses a constant current LED driver IC, no current limit resistors, I have guesstimated 100mA total for that.  So yes, a fully charged 2600mAh 18650 should get me at least 24 hours.  Making it more of a UPS style thing rather than a battery operated device.

The prototype on the breadboard only has 6 LEDs with current limiting resistors (470R).  Without the boost converter the board ran for over a week!  With the boost converter it lasts less than 24 hours.

I'll try testing the boost converter current now.  BRB.

[paulca]

So, test circuit:

4 x White LEDs --> 470R current limiters.  Common Anode.
Input voltage: 4.0V


With boost converter:  0.029A  Output voltage: 5.0V
Without boost converter: 0.006A Voltage: 4.0V
Without boost converter: 0.011A Voltage: 5.0V

An efficiency of 20% or 37% depending on how you look at it.

Also, it randomly squeals quite loudly when unloaded.

I don't have anything to hand right now to test it under a more decent load.  But for the project in question it is not going to work.

[Ian.M]

I said back in reply #27 you need to *BUCK*, not boost.   If you buck to the lowest voltage your design can tolerate, you'll save most of the power being dissipated in those resistors or constant current LED drivers.

Also, add a photosensor so you can auto-adjust the LED intensity to track ambient light levels.  The LEDs only need to be bright in full daylight, so there are considerable power savings to be made overnight.

[Peabody]

I'm not sure a load that small tells you much.  But it may well be that your boost converter just isn't that good.  That doesn't mean you couldn't find one that works much better.

[soubitos]

hold on.. the booster in this case is NOT MT3608 but TP8350.
And its not as efficient at low current draw.. not at all..
https://datasheet.lcsc.com/szlcsc/TP8350-5-0_C21998.pdf

Plus, the module is not following datasheet suggested values either in inductor or capacitors..

In the datasheet is also mentioned:
4) ? 150mA???22uH ?
Which means, you can use a 22uH inductor IF your current draw is over 150mA and you dont care much about ripple.
Its normal spec is for a 47uH coil and in the module is a 22uH yet the current draw is well bellow 150mA and the capacitors are definitely not as recommended... This speaks in Chinese "I am not a good module for your application, take an MT3608 or better" LOL

[soubitos]

HAHA Chinese didnt go through.. anyway. it would be VERY interesting to see how this little module would operate with a 47uH coil and a 100uF//100nF caps in its output

[paulca]

HAHA Chinese didnt go through.. anyway. it would be VERY interesting to see how this little module would operate with a 47uH coil and a 100uF//100nF caps in its output

Well unless I have a project for it that pulls circa 500mA it's a superfluous device.  Meaning, I now have no other purpose for it and can modify it.

When my hot air gun arrives I will try and replace a few things and see how it goes.

Can you post the links to datasheets?

[soubitos]

https://datasheet.lcsc.com/szlcsc/TP8350-5-0_C21998.pdf

[soubitos]

Its in Chinese but somehow google translate gives a clear view if you copy paste text

[paulca]

I said back in reply #27 you need to *BUCK*, not boost.   If you buck to the lowest voltage your design can tolerate, you'll save most of the power being dissipated in those resistors or constant current LED drivers.

This is why for my solar setup I charge 12V battery packs and buck to 5V for USB device charging. 

(which as an aside all of the testing in this post has been solar powered).

The problem here is the margins involved.  The circuit seem capable of running down to about 2.9V when the MCU browns out and eventually crashes.

Bucking the 4.2V to 2.7V range of the 18650 limits me to the 4.2 to 3.x voltage range.  The x dependant on the minimum input of the buck converter.

I may misunderstand correctly but a buck converter only operates well when the input is above the output.

Thus I would need to buck down to 3 volt. 

That's possible, would be intersting to see the difference, but ... I'd need to build the buck converter or find and use a cheap module.  I can see there being 3.3V modules available easily, but not 2.9/3.0V modules.

[soubitos]

I can send you a TP4056FlexAdv mate.... on the house... I am sure you will solve all your issues... it can be charged by solar directly, take USB 5V, wall adapters of any kind (imp 8-45V), will charge and protect your lithium and has an MT3608 as a booster ... i say, you pm your postal address, i will put one in the mail in a couple days max, you will receive it in a week or so and you will post here how you find it!

[paulca]

I'm willing to test this out.