Battery or Supercapasitor in portable bench power supply?

[Simenhs]

Hi!
I have been developing a small portable bench power supply for some time now and I have finally gotten it to a point where it kind of works. This is, of course, when someone points out an even better solution, and I am contemplating doing a major redesign. Should I replace my battery with superconductors? I have started on a pros and cons list, but hope some of you have additions or other insights. Thanks you
For more info on the power supply project, take a look at : https://github.com/Blue-Smoke-Electronics/Ember

I am contemplating two alternatives:
2x 100uf 2,7V supercapacitors in series
SCCV60B107SRB
Or: 
1x
3.7V 2500mAh 25C LiPO Battery
803465AX1


Pro supercap

• Lipo batteries are known to catch on fire when damaged or shorted. The battery is quite well protected, but if someone drives a nail through the power supply there may be more than smoke. The potentially  dangerous failures may also discourage some from tinkering with, modifying or repairing the device. 
• Battery lifetime. We have all had an old phone that has almost no battery life left. And if the power supply is in much use, that may happen to it asswell. Having the battery being replaceable should help, but with capacitors this should not be a problem.
• When used with USB plugged in, battery capacity doesn’t matter as much. Capacity only affects how long you can draw more than 2.5W. When high power draw is continues, not in pulses, a battery may not be enough either.
• For long sessions away from the grid, a usb power bank can be used.
• Overheating may be less of a problem when small capacitors guarantee short periods with high power usage. I may be able to turn up the max current limit without fear of overheating.
• Can be bought from Digikey, battery comes from ebay. Sourcing of batteries may be a problem in the future. Lipo batteries are banned as airplane cargo, I think that's why they are not available from serious stores. 
• Capacitors can be soldered straight on the bord, no connector. This makes production easier and faster, also chances of the connector being plugged halfway in.
• Faster recharge time, in theory, in particle recharge is limited by USB’s 500mA, but there will be no slowing down for the last 20% when the charger is in constant voltage mode for batteries. 
• Lower ESR, which should lead to smaller losses and less heat.
• Predictable capacity. The capacity can be estimated by looking at the voltage over a capacitor, output of a battery is not linear and I now have to integrate power usage over time.   

Con supercap

• Cappasity supercap is 2x0.1013 Wh, battery is 5.5 Wh. That is Around 30 times larger capacity in the battery.  In runtime this is 15 min of maxload with battery, 30 sec with supercap. When idle (display on) 30 hours with battery, 1 hour with supercap.
• Batteries are sheeper. 1 battery is around half the cost of two supercaps.
• Larger voltage range. Battery voltage ranges from 4.1V to 3.0V, supercap voltage will be from 2.5V to 5V.
  The larger range will maybe make it harder to make the boost converter stable for the whole range, and will lead to larger losses in the linear part of the regulator. 
• 1/4 of the capacity is inaccessible. When the voltage reaches 2.5V the capacitors are not empty, but the electronics need at least 2.5V to operate. 

I guess, what it really comes down to is if a power supply that can only deliver a 30 second burst of max output power is still usable? What is enough battery capacity?

Somewhere in between is also LITH HYBRID capacitors. best/worst of both worlds?

[Wolfram]

Nice project! How about an 18650 cell in a holder? This way, the battery is replaceable, with spares available anywhere, and with tons of well tested and reputable cells on the market.

[Shay]

Nothing like a PSU that works for 30 seconds.
I don't see how anyone would use a general purpose psu that has such a low usage time. Batteries are the way to go. Just make sure you design all the required protections (short circuit, tempature, over/under voltage, overcurrent) and the proper charging/balancing circuit.

[Peabody]

You could use a USB power bank to recharge the capacitor, but probably not to run the power supply.  Power banks have a minimum load current to keep them turned on.

You certainly can buy lithium batteries from Digikey.  At least 18650s.

I think it's really just a question of energy density, and batteries just have a lot more of that.  If you look at Cubesat power supply design, I think you'll see that they still use batteries.  And that's with everything going for supercaps - "nighttime" is only 40 minutes long, and in daylight there are no clouds.  But they still use batteries because they just can't stuff enough supercaps into a Cubesat to provide the needed power level overnight.

And remember that batteries have relatively flat discharge curves, where the voltage drops steeply only when the battery is almost discharged.  A capacitor has the steepest voltage drop right at the beginning, and your boost converter will be drawing ever increasing current from the supercaps as their voltage drops.

It just seems to me that a standard 18650 charger circuit with load sharing would let USB, or a USB power bank, power the bench supply and recharge the battery at the same time, and that's going to give you the most flexibility and the longest run time, and of course the lowest cost.  And I can't imagine that your customers would prefer a supercap circuit.  But that's just me.

[Simenhs]

Nice project! How about an 18650 cell in a holder? This way, the battery is replaceable, with spares available anywhere, and with tons of well tested and reputable cells on the market.

Guess I did a bad job when researching 18650 the first time. I need around 10A of max current draw, and the batteries I looked at was limited between 5A and 8A. I shouldn't have stopped there, when I look now, 18650 comes with all kind of different current ratings. And, yeah, they are the perfect size, and more trustworthy than the lipo I use now. I also think I will get almost twice the capacity with a 18650 (around 3000 mAh compared to 1500mAh). I will definitely test out a few 18650. 

Nothing like a PSU that works for 30 seconds.
I don't see how anyone would use a general purpose psu that has such a low usage time. Batteries are the way to go. Just make sure you design all the required protections (short circuit, tempature, over/under voltage, overcurrent) and the proper charging/balancing circuit.

 That is 30 seconds on max load, which is only possible with 15V, 1A (15ohm load). In almost all real applications you would the a longer runtime. I see you point though, it would almost be unusable when not plugged into a usb charger. 

You could use a USB power bank to recharge the capacitor, but probably not to run the power supply.  Power banks have a minimum load current to keep them turned on.

I had not thought of that, thanks for the reminder. Do you know how much the minimum load current usually is? It would be nice to find a way to run it from power banks, even with a battery. 

You certainly can buy lithium batteries from Digikey.  At least 18650s.

I only find one marketplace part with maximum 2.2A current draw when searching for 18650. How do you find them? 


Battery 3 votes, Supercap 0 votes.
Seems like i may not need to do a major redesign afterall  

[coppercone2]

I think we need solar powered miniature test equipment like desk calculators that you can run off fluorescent lights

[Geoff-AU]

18650s are a no-brainer.  Mechanically strong, great capacity, add more in parallel if you need higher current output or more capacity.  Add em in series if you need higher voltage, downside is you need to deal with balancing.

Supercaps are great for dumping a lot of energy fast.  I have a rainy day project for a motor that needs to run for 30 seconds at a time and can slow charge a supercap bank via small solar panel for the rest of the day.  Your PSU needs to use batteries, the energy density of supercaps is poor.

Cool little project you've got there 

[Peabody]

If you Google '18650 10A'  or 20A, you will get some hits.  I think LG make a 10A, and Samsung makes a 20A.  And you can use more normal cells in parallel.

With respect to the auto-off "feature" of powerbanks, I think the minimum current varies a good bit. I've seen posts referring to 100mA in some cases, but I think 50-60mA might be more typical.  But that should not be a problem if you use a  powerbank to recharge your battery. After all, that's what powerbanks are designed to do.

What converter are you using for the output of your supply?  Is it boost, buck/boost, or what?

[Simenhs]

If you Google '18650 10A'  or 20A, you will get some hits.  I think LG make a 10A, and Samsung makes a 20A.  And you can use more normal cells in parallel.

I have ordered some of the samsung ones, will test them when they arrive. The power supply will need to be a bit larger to fit two cells, but not very much, I can fit approximately 1.7 cells  It seems like cells with lower current rating often have larger capacity rating, so two cells in parallel may give me more than twice the capacity. 

What converter are you using for the output of your supply?  Is it boost, buck/boost, or what?

After the battery, there is a swichmode booster taking the voltage up to 2V more than the output voltage. Then there is a linear regulator taking the voltage back down to the target output. When the output voltage is lover than the battery voltage, only the linear regulator is used. 

 

[tooki]

Nice project! How about an 18650 cell in a holder? This way, the battery is replaceable, with spares available anywhere, and with tons of well tested and reputable cells on the market.

Guess I did a bad job when researching 18650 the first time. I need around 10A of max current draw, and the batteries I looked at was limited between 5A and 8A. I shouldn't have stopped there, when I look now, 18650 comes with all kind of different current ratings. And, yeah, they are the perfect size, and more trustworthy than the lipo I use now. I also think I will get almost twice the capacity with a 18650 (around 3000 mAh compared to 1500mAh). I will definitely test out a few 18650.

And bear in mind that 18650s are capable of delivering enormous currents, far above what they’re rated at. (The rated current is what you can draw without damaging the battery.)

I had an oopsiedaisy with a 1s3p battery (in all-metal holders, which shorted when one battery’s plastic sleeve was damaged with a tiny pinhole right near the positive terminal). Within seconds — and I mean literally 2 or 3 seconds — the battery holders were glowing bright red hot. I pried the batteries out within 5-10 seconds, but even so the damaged one then spewed its contents on the floor. Luckily nothing caught fire.

Anyhow, some quick research found that 18650s are capable of short-circuit currents of 50-100A EACH, so my three in parallel shoved 150-300A through a battery contact!

Those were Samsung cells rated for 8A, IIRC.

Two lessons: 1. Don’t rely on the battery’s insulating sleeve alone. 2. Make sure your protective circuits are in order.

So using 18650s, sufficient current is the least of your problems.

Don’t invent your own lithium charging, protection, or battery level monitoring circuits. There are dedicated ICs for all of those. (The latter are called “fuel gauges”, by the way.)

[tooki]

If you Google '18650 10A'  or 20A, you will get some hits.  I think LG make a 10A, and Samsung makes a 20A.  And you can use more normal cells in parallel.

I have ordered some of the samsung ones, will test them when they arrive. The power supply will need to be a bit larger to fit two cells, but not very much, I can fit approximately 1.7 cells It seems like cells with lower current rating often have larger capacity rating, so two cells in parallel may give me more than twice the capacity.

Correct. High-discharge-current cells have lower capacity and reduced cycle life. Highest capacity cells also trade capacity for cycle life. Basically, a less-ambitious cell will have the longest cycle life. 

What converter are you using for the output of your supply?  Is it boost, buck/boost, or what?

After the battery, there is a swichmode booster taking the voltage up to 2V more than the output voltage. Then there is a linear regulator taking the voltage back down to the target output. When the output voltage is lover than the battery voltage, only the linear regulator is used. 

Analog Devices has some linear regulators (like the LT3045-1) that have an output specifically for controlling an upstream switching regulator (by directly controlling its feedback pin), should you want to use an off the shelf part. 

[Martinn]

Nice project! How about an 18650 cell in a holder? This way, the battery is replaceable, with spares available anywhere, and with tons of well tested and reputable cells on the market.

Guess I did a bad job when researching 18650 the first time.

It would seem so. Check https://lygte-info.dk/info/batteryIndex.html

[Simenhs]

Analog Devices has some linear regulators (like the LT3045-1) that have an output specifically for controlling an upstream switching regulator (by directly controlling its feedback pin), should you want to use an off the shelf part.

I did use two lt3080 in an earlier revision. The reason I switched to using a npn and a opamp was firstly that they were completely out of stock, secondly, 2x7$ is a substantial part of the bom cost, thirdly, i could put the output protection diode inside the feedback loop.

 

It would seem so. Check https://lygte-info.dk/info/batteryIndex.html

Thank you so much, that page will be super useful!