Battery Powered Power Supply??

[Ranger14]

I was looking to make a portable power supply using 18650 cells .. The plan was to use a LM317 circuit but was hoping someone had a better idea .. Also I am still a noob in electronics ...


Thanks

[ez24]

I think you need to furnish more info if you want meaningful answers.

ie  output V and I  The only useful V of one cell with a 317 that I can see is 3.3 v

What is your plan on the 18650's  ie series and parallel configuration

triva: 20% of Panasonic production of 18650s go to the Tesla, you did know the Tesla runs on 18650 ?  So if you want the voltage and current of a Tesla then that will be a challenge

[jeroen79]

What will be the specifications?
What controls do you want?

Why a linear regulator?
How about a switch buck/boost regulator?
Or a combination?

[Brumby]

A battery powered supply with linear regulation is going to give you a really low noise supply.  Buck/boost converters use switchmode opertion ... and they are anything but.

Run time might be a concern - which is why the questions about number and configuration of batteries, voltages and currents have been raised.

[Ranger14]

What will be the specifications?
What controls do you want?

Why a linear regulator?
How about a switch buck/boost regulator?
Or a combination?

I wasn't really sure on anything but looking for ideas ..
Was looking to control voltage and possible current from 0v-20 to 0amp-1amp

I feel as though you buying a buck/boost converter, along with putting 4S 2P (with protection, of course) would be your best bet.

A battery powered supply with linear regulation is going to give you a really low noise supply.  Buck/boost converters use switchmode opertion ... and they are anything but.

Run time might be a concern - which is why the questions about number and configuration of batteries, voltages and currents have been raised.

I think you need to furnish more info if you want meaningful answers.

ie  output V and I  The only useful V of one cell with a 317 that I can see is 3.3 v

What is your plan on the 18650's  ie series and parallel configuration

triva: 20% of Panasonic production of 18650s go to the Tesla, you did know the Tesla runs on 18650 ?  So if you want the voltage and current of a Tesla then that will be a challenge

I wasn't really sure on anything but looking for ideas ..
Was looking to control voltage and possible current from 0v-20 to 0amp-1amp but spec can be changed

[Brumby]

Was looking to control voltage and possible current from 0v-20 to 0amp-1amp but spec can be changed

The next question is ... How long do you want to be able to run this supply
 - at, say, 20V, 1A
 - or, say, 10V, 750mA
 - or more particularly, at your typical voltage drawing your typical current
   ?

[Ranger14]

Was looking to control voltage and possible current from 0v-20 to 0amp-1amp but spec can be changed

The next question is ... How long do you want to be able to run this supply
 - at, say, 20V, 1A
 - or, say, 10V, 750mA
 - or more particularly, at your typical voltage drawing your typical current
   ?

I would say about an hour of run time more or less because I have alot of these cells sitting around.

[kosine]

I put together a portable battery PSU a couple of years ago, and it's been really useful on many occasions. I used a 12V lead acid (because I had several given to me), but you could go lithium if you have them.

Don't have a schematic for my design because it was mostly point-to-point wired. It's basically a simple 317 regulator, but a few things you might want to consider:

I went for a four digit panel meter rather than the more common 3-digit versions (still dirt cheap on ebay), but discovered the current measurement only seemed to work on the return connection. Maybe just this unit, but something to check if you're going to add a display.

A 10-turn pot for adjusting the LM317 is pretty much essential. Also note that there's a drop out voltage on the 317, which can be up around 2V when pulling a lot of current. You will also need a good sized heatsink on the regulator, though they're thermal and current overload protected as standard.

The binding posts are very useful, but be careful buying cheap test leads. They can have a surprisingly high resistance (Mine are about 2 ohms!), which will give a noticeable voltage drop at high currents.

In the photo, yellow is raw battery out, green is battery negative. The red and black are the regulated output, with the black being the current return path that appears on the meter. So I can go from raw battery +ve (yellow) to black and measure the current if I like. There's a toggle switch top-right which shows battery voltage, so I get dual measurement (V & I) regardless of source. (It's all just wiring, the only components are the LM317, two capacitors, a resistor and the 10-turn pot. Check a 317 datasheet for suitable values.)

The blue terminal is a nice touch you might also want to consider. It's hooked up to a LP2950 which is a three-pin precision 5V regulator (+/- 25mV but usually better). Works just like a 317, but only has 100mA current capability. Still very useful though. (If you use microcontrollers, you can feed this to the Aref pin for precision ADC measurements, or even run the whole thing from it.)

Note that you MUST always buy precision components from a reputable supplier! I bought a batch of 2950s off ebay once and none of them was even close to 5V. They're usually under $1 from a good source, so no point cheaping out.

Other than finding (or making) a suitable case, it's a fairly straightforward project. I wouldn't bother with a switched mode supply to start with, though Dave did a nice video on the 34063 a while back if you're interest.

[kosine]

Replying to myself, but I forgot to point out that a 317 or similar linear regulator won't go down to 0V. They're limited to the internal 1.25V bandgap reference.

If it's essential to go down to 0V, you'll need something a bit more advanced.

[jeroen79]

Replying to myself, but I forgot to point out that a 317 or similar linear regulator won't go down to 0V. They're limited to the internal 1.25V bandgap reference.

If it's essential to go down to 0V, you'll need something a bit more advanced.

A -1.25V reference will do.
That can be made with a small extra battery. (coin cell, AAA)

[ez24]

I think there is a market for a ps using 18650s because they are getting popular.

I would look into switching them in and out according to the voltage output.   For example if you have 6 in series to get your 20v and then power 5v you will waste a lot of heat in the 317.  So rig a switch that would use only 2 cells when at 5v.  Or build several for different voltages (this would be simpler)

Some power supplies do this by switching the windings of the transformers.

[QuantumLogic]

Earlier this year I made a portable power supply from parts and modules.  Most of the stuff was from China and purchased through eBay and Banggood.
I started with a 0-20V 0-2A buck power module with LCD display, but then I upgraded it to a 0-32V 0-5A module that had the same dimensions, though I kept the input voltage at 23.25V due to other components used in the design.
It is powered by 4 18650 cells in series (16.8V fully charged), but also has a 2.5mm DC jack in the back so it can be powered (and recharged) by a 12V power supply via AC wall power for those hours-long needs.
I designed the front and back covers in Blender and printed them at Shapeways.  The original steel front and back plates were just too difficult to cut holes into cleanly.
While not a perfect design and certainly a mess of wires (you should see the bottom of the bottom PCB - yikes!), it works very well and I am very pleased with how it came out considering I picked each piece separately.
It almost looks commercial, except for the lack of silkscreen writing on the front and back panels and obvious look of the 3D printed rough surface that they have.
I should also mention that the LCD power module is extremely accurate in terms of output voltage.  Using Dave's EEVBlog BM235, the voltage output from the module matches the measured voltage +- 0.002V which is incredible since the last digit should generally be ignored.  Not sure on the current - I didn't test it.

[ez24]

Earlier this year I made a portable power supply from parts and modules.  Most of the stuff was from China and purchased through eBay and Banggood.

Nice job.  I think this could be a commercial product.   You should start a Kickstarter on it.

Are the batteries charged in series?

[QuantumLogic]

Are the batteries charged in series?

Yes.  There is a 4S BCM circuit connected to the 4S pack with individual cell monitoring (5 leads total) to ensure each cell does not exceed 4.20V or drop below 2.5V.  The BCM circuit will disconnect the battery pack from the circuit in both cases to prevent over-discharge or over-charge.  The 18650 cells I am using do not have any protection built in, so I needed the BCM circuit to provide the necessary protection.  Unfortunately the BCM circuit does not have cell balancing during charge so it is not ideal, but at least it does individually monitor each cell's voltage during charge/discharge to prevent harm.  The cells are reasonably matched and are supposed to be made by Samsung (I peeled one sticker and saw the Samsung printing underneath).

For charging, I am using a DC-DC boost convertor with 12V input and CV/CC output set to 16.8V and 1.0A (less than 1/2C to be safe).  During charging, the current drops as the voltage reaches 16.8V and the BCM circuit will stop charging the entire pack if any single cell reaches 4.20V.   On the back of the supply there is a green LED voltage display that shows current battery pack voltage so during charge and discharge I can see what the pack voltage is with the push of a button without opening the case.

[Kleinstein]

Looks really nice 

Cell balancing is not that important. Usually the cells are pretty similar. So you don't loose so much of the capacity. Many commercial battery packs in power tools also work without balancing - many even without supervising every cell separately.

An extra Display at the back, just to check the battery voltage sounds like a bit of overkill. Just showing it, when a special key is pressed should be good enough.

[QuantumLogic]

The display on the back was maybe a bit overkill, but I have no other way to know the state of charge of the internal battery pack.  The back plate was pretty empty and boring looking without it.  But for an extra hour or two in Blender and a couple bucks in parts it was a no-brainer, especially since the LED display mounting was already designed for the front panel.

Below is a picture of the back panel.

The top-left is the green battery voltage display I mentioned and to the right of it is the on/off pushbutton switch for it.
The top-right is the 40mm 24V fan that runs at 12V-16.8V when the temperature inside the case (the front LCD module, really) reaches 35 C or so.  I used an LM339 and LM35 for temperature monitoring.  There is hysteresis so the fan will stay on for a while before shutting back off.  The use of a 24V fan allowed for a slower speed and less noise.  The front panel has vents for air to enter (impossible to see in my above pic).
The LED to the top-left of the fan is red and turns on when the fan turns on.   If the red LED is on but the fan isn't spinning I know there is a problem with the fan.
The lower-left is the 2.5mm 12V DC input jack.   I use a cheap 12V 4A unregulated power supply to feed power to it.
The lower-middle switch is a 2-pole quad position slide switch.  It has 4 modes left to right:  Off, Battery, Battery charge and External DC-in.

Battery mode means the front supply panel is powered by the internal battery and any power from the DC-in jack is ignored.  The LED above that position is green.

Battery charge mode means the DC-in jack is used to take 12V and send power to a DC-DC boost convertor that is statically configured to 16.8V and 1.0A output which is sent to the battery pack to charge it in 3-4 hours from empty.  The front panel cannot be turned on in this mode.  The LED above that position is amber/green.  I used an LM339 to detect when the battery voltage reaches around 16.6V and it switches from amber to green at that point.  In retrospect, I should have used a current monitor circuit to detect when the charge current drops below some value, like 100ma or 50ma.  I almost never see the green LED during charging now because the battery pack only reaches around 16.5-16.6V and doesn't go higher than that any longer.  It used to get to 16.8V when I first finished it.  I assume it is because the cells are slightly mismatched and one cell is reaching 4.20V before the others, but I haven't checked to verify.

External DC-in mode means the DC-in jack is used to take 12V and send power to the front panel, skipping the battery completely.  This is useful when the power supply is next to an AC outlet. 

Both the battery and external DC-in are separately fused and use a MOSFET for reverse polarity protection.  I was a bit paranoid about safety due to the use of Li-Ion cells, so I figured extra protection would be worth the overhead/losses.

Due to the extensive use of DC-to-DC convertors with constant current capability, there is a huge problem of different ground potentials all over the place.  I could not just connect all of the grounds together.  I had to use a couple of 8A relays to switch between battery mode, battery charge mode, and external DC-in mode in order to guarantee grounds would not be shorted.  One relay is only activated during charge mode and the other relay is only activated during external DC-in mode.   This means in battery mode the relays are not powered which saves precious battery power.

Sorry for taking over this thread, Ranger14.  It is not my intention.  I hope you find it interesting anyway.

[ez24]

How do you match the batteries?  Voltage or capacity ?  And what % tolerance do you go by.

I have a charger that shows capacity and my batteries are all over the place like 2050, 1940, 1820  etc
I threw the 500 away.  All cells came unmarked and from Dell laptops

thanks

[QuantumLogic]

I didn't do any matching for the cells I used.  I just bought 2 sets of 2 Samsung 2500mAh cells from a USA eBay supplier hoping they would be matched "enough" (same manufacturing run/datecode) to not have to care about balancing too much.  In testing, they are (or were) rather evenly matched and the voltages stayed within 0.05V of each other during the discharge/charge process (actually they were within 0.01V of each other if I recall correctly).  I assume it will get worse over time, however, since there is no charge balancing.  If I could find a 4S BCM circuit with charge balancing I'd use it instead of the unit I bought for that extra piece of mind.

Edit: As for matching used cells, I'm not an expert on that.  A reasonable assumption for Li-Ion is that you want all cells to be at the same state of charge initially and the voltages of the cells should be within 0.05V of each other and capacity in mAh should be less than 5% different between all cells in order to allow series charging and discharging to keep them reasonably matched.   This may be difficult when testing old laptop cells unless you have lots of them to find good matches.  Also, found this: http://batteryuniversity.com/learn/article/bu_803a_cell_mismatch_balancing