Hacker News Reader

Comment by teraflop | Hacker News Reader

This looks like a pretty cool device!

However, I was immediately curious about how the "dual battery" feature works. The IP5306 power-management IC seems to be designed only for a single battery, and as far as I can tell from the schematic[1], the two battery connectors are just directly connected to each other in parallel (across VBAT and GND).

This seems really sketchy. If you plug in two batteries that are not at the same state-of-charge, then you're going to get a very large current flowing from the higher-voltage battery to the lower-voltage one, probably significantly exceeding the batteries' rated current limits. At best this wastes a lot of power and generates a lot of heat, and at worst it could be a fire hazard.

[1]: https://github.com/ZitaoTech/Hackberry-Pi_Zero/blob/main/Sch...

View on HN

I was going to launch into a lecture about how battery isolation is a Thing, very simple, etc and the author couldn't possibly be dumb enough to not isolate them with diodes, but then I looked at the schematic and yiiiiiiiikes, they're both just tied together, I think. There's a charge management IC and a MOSFET as part of the USB power input...and that's it.

I was sketched out by BSI not being connected to anything, but it seems it's a fixed resistance battery size/chemistry ID function, not a temperature sensor. That said, I don't see any over/under temperature protection except for some sort of vague temperature limit in the charge IC (which is not thermally coupled to the battery in any way, so only a general "the device is way too hot"), so hopefully that's in the battery - don't go buying any cheapo clones.

I also don't see any fusing, which is a huge no-no. A polyfuse is a twenty cent (ish) part. Again, an official battery would have an internal BMS circuit and prevent overcurrent events, but people are probably going to go for the almost-cheapest battery on Amazon/Ebay. Not to mention counterfeit problems even if you do try to get an OEM battery.

Oh, houses can burn down by not putting in 2 diodes and ptc fuses

I agree, this is a pretty bad idea. I've used batteries in parallel many times, but it's always in the context of treating it as a single "bank" that is always charged at the same level, not as individual batteries which can be swapped separately. Swapping a full battery in when the other one is low is going to result in a rapid rebalance of power until they get to the same level!

You can think of it like having two big swimming pools side by side, slowly draining out into a little stream. If the two pools are at the same level and the wall between them disappeared, they're largely going to be fine and keep on going.

Now imagine one pool is nearly empty and the other is full. As soon as you disappear that wall in the middle, there's gonna be a very powerful wave as the water from the high pool rushes in to fill the empty pool.

At best this wastes a lot of power and generates a lot of heat, and at worst it could be a fire hazard. - you just described my high spec HP laptop.

It looks like two BL-5Cs, which normally come with their own overcurrent protection (AFAIK even the no-name generic clone cells do), so what's more likely to happen is you trip the protection on one of the cells being swapped. Fortunately these are not very high capacity nor current cells, being ~1000mAh at most. They seem to be pretty robust in general, as a series of hugely popular MP3/MP4 players that used them had nothing more than a diode from USB5V for charging.

On the other hand, I'll gripe about that "schematic" being not actually a schematic but just loose parts with labeled pins.

Barely more than a netlist.

It has no solar panels , runs shorter than my cell phone on a single charge. It is thick. No ethernet socket Any small UMPC is better than this thing. I don't mercifully mention the blobs encapsulated in the raspberry pi itself.

It’s got usb-2.0 ports though, you can get an usb-ethernet adapter if you want

for power?

I understood it as it'll use only one of the batteries, but you can swap between which is used. So initially, you have two charged batteries, while using only one. Once you run out of power in the first one, you'll switch to the second, and now the first one could be swapped to a fully charged one. Blue/green deployments, but for batteries basically.

It would be convenient if it worked that way, but since the batteries are connected across each other in parallel, they will both be discharged simultaneously. And as soon as you hot-swap one of the nearly-discharged batteries for a charged one, it'll be more-or-less short circuited across its discharged counterpart.

To do what you describe, you would need additional components to "switch" one battery at a time into the power path. (This can be done with a single transistor if you're only worried about current flowing in one direction, but I believe it's trickier if you want to support both charging and discharging in the same circuit.)

I asked the following above - does this work?

>> A mid-stream capacitor is holding enough charge for 10 seconds to allow indivdual swapping out -- and also this is meant? to mean that you can swap TWO NEW batteries in without having to shutdown/sleep/power interrupt - thus the 10 seconds.... that dope

EDIT: I was imagining the capacitor holds up the whole thing for ten seconds to swap out both bats...

In theory you could do that, but it has its own costs and complications. And if you're going to add a big enough capacitor to ride out a battery swap, why bother with two batteries instead of just one?

Hotswap of two parallel batteries must replace both with a matched pair. Otherwise the one at higher state of charge starts charging others without current limit.

null

Yes, that is exactly what I said at the beginning of this thread.

I read this part:

> Replace your battery in 10 seconds without killing the power!

as a suggestion that you'd have four batteries total, and you'd have two that're fully charged, and you'd replace one battery, and within seconds you'd replace the other. Or at least that's how I'd do it.

I've recently read up about power management and battery charging, and want to make a charge controller than can connect two separate banks. I wonder how hard it'd be to change the IP5306 in the Hackberry Pi Zero to handle the two batteries separately.

And if you screw up which battery is which?

"Do things exactly right, quickly, or the device bursts into flames*" is not acceptable electronics design, even for something you intend to use yourself.

* do you really want to trust a generic battery's built-in protection IC?

Maybe I’m being too generous but I feel like 99.9% of the people using something like this would do it properly

Now that I'm aware, sure. If I wasn't told, I would assume that they were designed so that a new battery could be swapped in while one battery remained inserted to maintain power (much like an old Thinkpad I used to own).

.. until their curious child decides to have a look at it.

If there's a capacitor... what about adding a switch to allow you to select which battery you are using? Then you can replace whatever you are not using without issue.

A mid-stream capacitor is holding enough charge for 10 seconds to allow indivdual swapping out -- and also this is meant? to mean that you can swap TWO NEW batteries in without having to shutdown/sleep/power interrupt - thus the 10 seconds.... that dope

It should definitely use the plural, batteries, if that's the intention.

acktually, traditionally, 'batteries' referred to two or more cells. This keeps with the etymology of the word as an emplacement of cannon or guns. If OP meant a single cell, they should have written 'cell', not the singular 'battery'.

People often do this on their boat, but with 100Ah batteries and those rotary switches that select between "1, 2, BOTH".

They don't realize how much current can flow if they slam it to "BOTH" when two batteries with very different voltages are connected.

I'm not understanding the appeal here. Why use this instead of a regular smartphone?

If you ran it with one battery would you not have the problem ?

There's no problem if you use one battery, and there should also be no problem if you use two batteries and charge/discharge them both simultaneously (because then the voltages are matched).

The problem shows up when you try to "hot-swap" just one of the batteries and replace it with one at a different state of charge, as the README claims you can do.

What is the expectation here about functionality? They are using some kind of COTS battery system to keep the cost down, but at the consequence of this safety qustion. Should these people expect any buyer to antipate this? Its not really a consumer product after all.

I think there is no shortage of battery management ICs, but the number that can arbitrate between external power/battery A/Battery B certainly eliminates 97% on your Digikey parametrics.

IMO my expectation is that they not do that. If they want to offer hot-swappable batteries, they should either do it right (with a more functional battery management IC or a little homebrewed FET-switching circuit), or they should at least come up with a hack that doesn't threaten the battery safety. For example, have a physical switch to select between the two batteries and instruct the user to flip it when one gets low. Use a super capacitor to cover the short time while the switch is between connections.

For example, with this part:

https://www.analog.com/media/en/technical-documentation/data...

It feels to me that the fact that there are no readily available BMS chip that supports this type of circuit indicates something. But maybe I'm mistaken...

What about having 2 diodes and problem solved

Reply on news.ycombinator.com