Re: [EVDL] Phd in Materials Science on LiFePO4 batteries video WAS: Lithium 60ah 144v
Lee Hart wrote:
> Ed Blackmond wrote:
> > You don't need IGBT modules to bypass an empty cell. You need a pair of
> > MOSFETs (or a pair of paralled MOSFETs to handle the current) that are
> > rated to handle your cell voltage in a half bridge configuration.
> > _____________
> > | _|
> > | | |
> > | H | |
> > | ___| |__
> > ===== |________o A
> > === __|
> > | | |
> > | L | |
> > | ___| |_
> > |_____________|_________o B
> This could work, and indeed, has been done in EVs where relay contacts
> replace the MOSFETs. The challenge with MOSFETs is that there are a lot
> of them, and they are carrying full pack current. It requires some
> exceptionally good MOSFETs to keep the losses (and consequentially the
> heat) low.
Agreed, but these MOSFTEs don't have to switch the full pack voltage.
Such MOSFETs with single digit mOhm Rdson are pretty inexpensive and when
paralleled to get to the required current, reduce Rdson even more. This
doesn't make any one MOSFET cooler, but the combination is cooler than a
single MOSFET with a larger Rdson. The copper connection between them can
also double as a heat sink.
> > This circuit can survive all catastrophic failures...
> MOSFETs tend to fail shorted. If one shorts, it will short the battery
> the next time the other one turns on. You're going to need fuses in
> series with each battery to protect against this failure mode.
Again I agree. I was assuming this as part of the battery. If the
battery is organized into several blocks of M cells in parallel, with
N of these blocks in series (using cells similar to those found in
iPhones, iPads, and Droids for example), then the battery needs to deal
with a cell failing with a short. MOSFET failures causing a short of the
battery is just a special case equivalent to a cell in each parallel
configured block failing short. Yes, the battery needs to deal with this
without destroying the pack or vehicle or structure it is parked in. It
may require manual maintenance to recover from such a failure (replace the
fuses), but the vehicle can still be driven without the battery.
> When built with switches, the switch was built so it was physically
> impossible for it to short the cell.
> > For cost reduction, you probably put three or four battery cells in a
> > group with a half bridge around them.
> Look up the "rectactor" circuit for examples of this. Rather than two
> switches per battery, each rectactor block has 2 batteries and 3
> switches per module (fewer parts overall). It can wire the batteries in
> either series or parallel. Repeating it lets you select series/parallel
> combinations to get every possible voltage with fewer switches overall
> than the above scheme.
I was not able to figure out how to use this circuit to meet my original
goal of protecting the battery cells. In the case of one cell operating
out of specification (voltage, current, temperature), it needs to be
disconnected from the circuit without losing a significant portion of the
overall pack capacity or power. That led me to the half bridge. The
battery with the bad cell can simply be disconnected.
Once I had a module that could ether be zero volts or V volts, the idea of
a multilevel converter made sense. The BMS can be used to generate a darn
good sine wave. Make three strings and operate them 120 degrees out of
phase and the BMS becomes a variable frequency 3-phase drive. The
multilevel converter makes a good charger too either from regen or the
This saves a lot of money. I'm of the opinion that a BMS is necessary
with high energy battery packs. Using this technique, for the price of
the BMS, you also get both the motor drive and charger. It is pretty
efficient too. There are no high frequency switching losses (no PWM) and
the harmonics of the fundamental frequency are also extremely low. It is
also safe. There is no high voltage DC and the only time it is producing
high voltage AC is when the vehicle is moving.
| REPLYING: address your message to [email protected] only.
| Multiple-address or CCed messages may be rejected.
| UNSUBSCRIBE: http://www.evdl.org/help/index.html#usub
| OTHER HELP: http://evdl.org/help/
| OPTIONS: http://lists.sjsu.edu/mailman/listinfo/ev