High capacity LIPO UPS with fast charging

[lapoltba]

In the interest of making a small long-life pi tablet, I am looking for a LIPO battery solution similar to the adafruit powerboost 1000c device.

I will be powering a PI with the 7" touchscreen as well as a few minor peripherals. I would like 10Ah of battery which I can easily get with bare (protected) lipo packs. What I cannot seem to find is a charge/buck/boost PCB capable of quickly charging these batteries. A 10Ah battery can easily be charged at 1C, however I would be happy with 0.5C or even a smaller capacity battery. I can easily find cargers and buck/boost regulators, but no high power all in one units. I can incorporate heat management with heatsinks, but I need something to start with. I do not need to use micro usb to charge, I will likely use a 2.5x5mm barrel jack.

Does anyone have any suggestions, or a good way to quickly charge batteries while also powering the pi?

[JimmyN]

You'll have trouble finding a buck/boost PCB that will output enough amps to charge them quickly. If you want a high charge rate you need to look at radio control chargers, such as used for cars, planes, quads. They provide the protection circuitry needed when charging (the risky part with lithiums) and can also provide the amps required for charging high capacity cells in a timely manner. Charge a 10000Mah pack to ~85% in 15 minutes rather than ~6 hours or more. RC'ers are not going to wait around all day for a pack to charge, when the one they're running is low the second pack on the charger needs to be ready to go so they can swap them out. You can get RC chargers up to 1200W output or more.

You can not charge a Li-Ion/Li-Poly cell while it's under load (in use). It has to be disconnected from the load to charge it due to the strict voltage limitations for lithium cells. Attempting to do so you may experience a "vent with flame" event, and it's not pretty. The smoke is very toxic and corrosive to any metal it contacts, plus it will set fire to anything around it.

I pulled these two out of my refrigerated storage to make up a pack for my RPi with the 7" display. I buy them when on sale, charge to ~3.80V storage voltage if they're not there already, and stick them in the garage fridge. They'll keep nicely there for years until I need them.

These are 40C rated cells (200A burst) and have a manufacturers claimed 5C charge rate, though I will only use 4C as a max. I can charge these at 20A (40A with two in parallel) up to about 85% when it has to switch to constant voltage charge. Then it's just a matter of how long I want to leave them on there. The undeniable fact is that it takes at least an hour and a half to fully charge a cell to 4.2V no matter what you do. Since the charger can never exceed 4.2V when the cells get to 4.1V there is only a tenth of a volt potential difference, so charge current drops way off. That's why many RC'ers, and the supposed "quick lithium chargers" for phones/tablets only do a partial charge rather than top if off all the way.

So I'll charge these with an RC charger, probably about 20 minutes or so, then I'll have enough to run the RPi with display for about 6~7 hours, maybe a little more. Then another 20 minutes to juice it back up and go again.

5000_cells.jpg (59.34 KiB) Viewed 4303 times

[lapoltba]

That was the way I was going to go, but that requires removing the battery to charge it. To be clear I didn't mean a buck AND boost converter, it was one or the other depending on batteries being series or parallel. I wouldn't mind going to a 2s configuration as buck would be more efficient than boost, but I have not found a UPS type device for 2s configurations. I am quite familiar with RC lipo packs and I do have a LIPO charger for 1-6s batteries that can supply 5A. It still doesn't make a nice solution where the RPI can remain powered while charging.

There must be a way to do this, as laptops and all other portable electronics do it all the time. The Adafruit powerboost 1000c also does exactly this as it can charge and supply power at the same time. That is my current solution, although still not ideal. I'll use 2x 2500 mAh packs in parallel (for packing efficiency). Charging a 1s2p 5Ah-10Ah pack at 1A is painfully slow.

I think Adafruit would have a good market if they could produce a 2s balance charger capable of 2A+ of buck current @5V and 2-4A charge current.

[JimmyN]

I'm using a 2.1mm barrel connector to power the whole case, and that powers both the RPi and the display. It's a switched type barrel jack, so when the power connector is plugged in it uses the external power and disconnects the battery. Remove the power plug and it switches back to battery power.

So plugging in the power disconnects the battery, and powers the RPI, and with some additional circuitry (you can buy Li-Ion charge controller IC's) it could also charge the battery at the same time since it's disconnected. The problem is the amps needed to quickly charge a high capacity cell. 2A or so is feasible using a charge controller IC, and your power adapter that is powering the whole setup, but that's not going to give you a "fast" charge (minutes rather than hours).

What I'll do is have a charge connector and a balance connector (I'm going to use 2S as well, with a buck convertor) on the battery pack. When I apply power to the RPi with the barrel connector it will disconnect the battery and I can then plug the charger connections into the battery pack and charge it up. That will provide a higher rated power plug and larger wiring just for charging purposes, rather than charging using the RPi's input power from an adapter.

You get about the same capacity whether using two cells in parallel or series. 10000mAh at 3.7V gives you about 7400mAh at 5V, just as 5000mAh at 7.4V gives you about 7400mAh at 5V. And I too prefer a buck convertor over a boost convertor. I'm going to use 2S as I can have a max charge rate at 20A, versus 1S where I would need 40A to get the max charge rate, thus smaller wire, smaller charge connector, easier to handle.

[lapoltba]

Good info thank you.

Can you link me to the switched barrel jack you used? Is it a mechanical switch or solid state? I fear that a mechanical switch would cause noise and reset the pi. I have already done this accidentally by hitting the stupid micro-USB cable several times.

[JimmyN]

It's mechanical, but I haven't had any trouble with them in the 2.1mm size, they're pretty sturdy and have good contact area. Any barrel jack with three pins is "switched", most are.

A quick Google turned this up and you can see a diagram of how it switches. Positive to battery and RPi would be connected to pin #1 (center terminal). Negative from the battery would go to pin #2, and negative going to the RPi would be pin #3 (outer terminal or shell of the plug).
http://electronics.stackexchange.com/qu ... t-contacts

With no plug inserted negative from the battery goes in pin #2 and out to the RPi thru pin #3. When the plug is inserted the negative connection from the battery is broken, and the plugs outer shell provides negative to the RPi thru pin #2

[lapoltba]

Yeah, I suppose that would work, but still not the fully integrated solution I was hoping for.

I found the PIJuice kickstarter, but there are many technical details lacking. I'm still looking for more info.

[JimmyN]

In your original post you were looking for a fast charge method for a 10Ah battery, so I was pointing out some ways to do that. I don't think you're going to get the output needed for a fast charge on a 10Ah battery using PiJuice or any of the other RPi solutions, they're just not going to provide that kind of amperage.

Or maybe the difference is that you consider 3~4 hours a fast charge, and I consider 20 minutes to be a fast charge for a battery of that size because I do that for RC usage?

[lapoltba]

I don't think the PIjuice will do it either, but they do mention a 3Ah option, which was at the low end of what I want. I could suffice with that, but if they are only able to charge at 1A max, it's basically useless.

I found a couple things for 3s batteries available off the shelf, but they are direct from china, and getting specific technical details is nearly impossible. Something like this: http://www.batteryspace.com/CMB-for-11. ... -char.aspx Charge current is only 2.7A, but a 3s battery would only need to be maybe 3Ah and get charged in roughly 1 hour. I just don't like the fact that the board is so large (relatively) and it has other stuff like the fuel gauge on it that I don't need.

If I had the time and patience I would just take a TI or Linear chip and build one myself. The datasheets are fairly comprehensive and the circuits straight forward. Anyone with the proper tools and time could design up a board with little effort. Most of the datasheets have demo circuits that would suffice. Something like the LTC4006 which also has power path functions handles the switching between DC input and battery power and has 4A charge current. This would require an external DC-DC buck converter, but there are many options for that.



I don't need to charge in 20 minutes, but a 2-5A charge rate would be nice. Obviously any USB battery or charging system is out as they are pretty much limited to 1A charge.

[JimmyN]

2S or 3S does work better, as you have lower current both in and out, but then you have to monitor individual cell voltages while charging. That's why the RPi solutions and USB power packs, as well as cell phones/tablets, typically only use one cell (or cells in parallel) as it makes things a lot simpler, and cheaper.

Way back when li-Ion was first gaining popularity RC'ers were ripping the cells and protection circuit boards (PCB) out of Qualcomm cell phone batteries and laptop computer battery packs. We'd use the protection circuits to charge them and the cells were half the weight of NiCad for the same energy, so they were perfect for RC planes. Later on manufacturers would start to produce balancing chargers, and balance connectors on the battery packs, negating the need for a PCB as the charger took care of that. I'd say you could salvage a PCB from a laptop battery pack, which would work great, but they are the width of the laptop battery so a bit unwieldy due to the size. At the time I did get some Lithium charging IC's from Maxim, rather than use salvaged PCB's. I don't know if they still do it, but Maxim would give away single samples for many items, so I got them for free. You may want to look there and see if they have anything suitable for 2S/3S with adequate current ability, and you may be able to try it at no cost, they even paid postage.

I have thought about designing an actual UPS setup for the RPi, but you run into the other problem with Li-Ion/Li-Poly cells, if you keep them fully charged they deteriorate rather rapidly. Manufacturers ship the new cells at about 3.85V, and that's about the same voltage they'll be if you buy a product that has Li cells inside. At that voltage they can sit on shelves for many months with no harm and very little self discharge. Charge them to 4.2V and let them sit for a week and you'll have lost about 25% of capacity and things will continue downhill from there.

For RC use I may have $50~$100 in a battery pack and if I had it charged, but didn't use it, the first thing I do is put it on the charger and hit the "storage" function, which will discharge them down to storage voltage. If I just put them on the shelf fully charged when I get home, they'll never be the same, and that gets expensive. That's another reason for the "fast charge". Since they are never stored fully charged I need to boost them up before use and I don't want to wait hours before I can play, so with a "fast charge" they're up and ready in about 15 minutes.

So with a UPS you have the dilemma of not being able to keep them fully charged and on standby. You could double the capacity above what you actually need and only charge them to 3.9V~4.0V which would probably work, but the pack itself would be physically larger.

[lapoltba]

I'm not too concerned about balancing and protection since there are many off the shelf solutions for 18650 and lipo pouches that both protect and balance. Charging with an integrated solution is still the limiting factor.

As far as I know the charging circuitry for laptops is still on the motherboard itself not in the battery, but I could be wrong. It would be nearly impossible to get much info I imagine due to IP and the proprietary nature of the systems used.

Yes there are plenty of single chip solutions available, but like I said, I don't have the time or resources to get one working.

[JimmyN]

As far as I know the charging circuitry for laptops is still on the motherboard itself not in the battery.

The charging circuitry is in the laptop, and the ability to use it while charging the battery. But individual cell protection is in the battery pack. That way the charge circuit only has to output one constant voltage to the pack, the PCB monitors individual cell voltages inside the pack and will shut down if any cell goes out of range, and it also stops input current when fully charged. It also protects the pack from short circuit when it's not plugged into the laptop, and over discharge. If the battery pack is in a discharged state and sits for a long time the cells could drop below the minimum safe voltage, so if any cell ever goes below ~2.5V the PCB will turn the pack off permanently and it can't even be recharged because of the fire hazard.

[lapoltba]

The pack only has balancing and gross protection against a fault condition, High charge/discharge current, high/low cell voltage and over temperature. The external circuitry in the computer/tablet phone handles the multi stage charging and communicates with the battery, usually through smbus.

It seems these integrated chargers are tailor built by the OEMs and off the shelf solutions aren't readily available.