Battery voltage: to boost or to connect in series.
Background.
I have a legacy battery system that is comprised of twenty-four 12V Li-ion battery modules (self-protected), enclosed in a shell with the charging connection provided as a common ground and 24 individual non-isolated charger connections. The 24 Li-ion modules are connected in parallel. Part of the circuit is the high current draw (4A) electronics running from 12V directly from batteries, with additional electronics running at low current (300mA) 5V and (200mA) 3.3V. Because of the cost of battery certification and testing, I must use the existing 12V modules. The enclosure shell imposes a height constraint on PCB+component of less than 0.5".
Current Challenge.
The high power consumption electronics is being upgraded, and can now run at up to 48V. The existing 12V modules were tested in a series connection, and they can output 4A without issue still. However, connecting the batteries in series would not allow the batteries to be charged with the existing system (due to common ground chargers), and we cannot have a "permanent" 48V Li-ion system because of the previously mentioned certification process.
Proposed approaches.
-
Create a re-configurable battery system where a 12V battery module connections can be managed with back-to-back MOSFET's when the charging system is connected. When the charging system is not connected, the battery modules would be disconnected from the charging ports and reconnected back into the predetermined configuration (4S6P). The losses from the MOSFET's operating in fully enhanced mode would be fairly small, with low Rds(ON) in the mOhm range. The component count added with this approach would be at least 24x(4xMOSFETs+2xGate drivers+4passives) = 240 SMD components or more.
-
Leave the battery modules connected in parallel with the same charging circuit and add a boost circuit to increase the voltage output to 48V and 4A. From most online resources, 90% efficiency or slightly less would be achievable. The issue here is lack of SMPS boost circuit design experience on our part, and most of the advice online is to stack batteries in series to achieve the high voltage, and buck that as necessary. There is also concern of fitting large boost circuit components into the height restriction.
I'd like to ask for advice on choosing between the two proposed approaches or introducing an alternative approach.
1 answer
To summarize the problem, you have 12 V battery packs that you don't want to change due to certifications, but you need to power a 48 V 4 A load. The battery system also needs to be compatible with an existing charger that is meant to charge each of the 12 V packs with the negative ends tied together. Components on any new board are limited to ½ inch height.
Apparently the device does not need to operate while the batteries are being charged. I would therefore consider using relays to switch the battery configuration between 48 V series connection for operating, and common ground 12 V connections while charging. The relays should be arranged so that their unpowered state is suitable for operating. When charging power is applied, the relays are energized to switch to the charging configuration, which also automatically disconnects the 48 V load.
Normally I'd want to switch things with transistors. However, by the time you deal with problems due to the inherent body diode of MOSFETs, the break-before-make issues, inconvenient gate drive levels, and voltage drops, relays may start to look simple.
I'm envisioning one SPDT relay at each end of each 12 V pack. Of course the two poles don't need to be in physically separate relays. A DPDT relay per pack would work, although each half would be used separately.
Could I use solid state relays or would that mean having to address the same range of issues that using MOSFETs would necessitate?
You could possibly use solid state relays. Their outputs are isolated from their inputs, which is certainly useful in this case. However, they usually have fairly large voltage drops relative to 12 V. SSRs are more appropriate for switching 120 V or 240 V AC where the 1 V or so drop is a small fraction of the whole. Then there is the extra complication that you really want SPDT switches with break before make.
Mechanical relays seem like a good choice here because they have very little on-state voltage drop, can guarantee break before make in SPDT configurations, and are fully on/off in one of the states without any power applied.
The usual downsides of slow operation, power to maintain the on state, and finite lifetime switching transitions, aren't much of an issue in this case. You're only switching when the charger is connected and disconnected. It will be connected for an hour or more at a time, so a 100 ms delay at each transition is inconsequential. When the charger is connected, substantial external power is available. The extra to maintain the relays state is a tiny amount more. You're only going to connect and disconnect the charger a few 1000 times over the lifetime of the product. That's a small fraction of what any reasonable relay can do.
Could the "maintaining the on state" issue be handled using latching type relays?
Yes, but that is probably not the best tradeoff. As I said earlier, I'd wire the relay outputs so that the unpowered state is operating the device from the 48 V configuration. That means power to the relays is only required during charging.
When charging, "wall power" is presumably available. We don't know what the power requirements are, but I'm assuming multiple amps at 12 V. Suitable relays to switch each 12 V battery might take ½ W or so. That's about 1% of the power of 4 A at 12 V, for example. The total power for a few relays should be less than a small light bulb in a typical room. I just don't see the power to run the relays when charging the batteries to be of any significance.
On the flip side, latching relays are usually much more expensive. Since they are much less common, the selection is also more limited.
1 comment thread