single cell (1sXp) charger proposal

Hi,

Since I'm at the point where I need to use batteries (as my lab supply is too weak) I want to re-use some
Li-ion cells from old notebooks. Just for initial testing. Since also later I want to make my own pack
using LiFePo cells I'm now busy with the charger / BMS.

I think what I want to build is best explained by showing my 'concept schematic'

It's basically an isolated charger which runs of a 12 - 20 V supply and is capable of charging one 3.3 to 4.0 V cell. Single
cell is maybe not a good expression, it's a 1S but many P. For multiple cells in series more than one charger 'section'
must be build, the sections (because of the isolation by the transformer) each charge one cell. Cells are connected in
series by the contacts shown on the left of the schematic to form a battery pack. On the right side of the schematic,
multiple cell-chargers use the same 12-20V supply source (so the chargers are connected in parallel here)

The idea behind the operation:
The 8 pins PIC 12F617 processor is permanently supplied by the cell. This particular device runs from 2.0 to 5.5V so can even run
on an 'empty' cell. During sleep (which will be most of the time) the chip consumes around 50nA, this
is so low that the permanent connection to the cell should be no noticable load on the cell.
The supply on the right (12-20V) is the supply from which the cell will be charged. I'm thinking car
battery here or, better yet, an obsolete 20V 4A laptop charger. This supply can be either present
(for charging) or not (for when we are using the bike). You can of course also use some sort of
20V50A if you're so inclined.
All the wake-up nodes of the chargers are connected together. A (in this case positive) pulse here (from a button) wakes
up the circuit. First order of business is to send some pulses via the opto coupler to the 12-20V side
of the system. If this supply is connected a current will be induced to flow through the 0.33 Ohm current
sensing resistor, the 12F617 will be able to measure this. If the 12-20V is not connected no current
will be induced.
The 12F617 now know whether we want to charge or not.
If we want to charge we can first control the PWM signal on the optocoupler by measuring the voltage
across the 0.33 Ohm current sensing resistor to implement constant current charging. When the
cell voltage is high enough we can switch to constant voltage charging and use the current sensing
resistor to terminate this phase when the current drops below a certain threshold. After this
the 12F617 shuts down to sleep mode. A LED can be used to indicate what's happening (like burn constantly
for constant current mode, blink for contant voltage and off for when the cell is charged).
If the 12-20V supply is not present apparently we don't want to charge but we're on the move
with the bike. The 12F617 can now measure the cell's voltage and report this via the optocoupler to
the bike's central processor using (for instance) a bit-banged RS232 protocol. This can then
be used for a state-of-charge indicator and a shutdown when the cells voltage drops too low.
You can even think about reporting the cell's voltages on a display.

So... what do you guys think of this system ? Any usefull feature I forgot ?

Oh, and of course by choosing the correct type of power transistor, type of magnet
wire used for the (air core) transformer and value of current sensing resistor
this circuit can be configured to deliver whatever charging current needed. Same for
cell voltage, the changover point from CC to CV will be determined by a contant
programmed in the 12F617. This constant can be for 3.3V or 3.7V dependent on cell type.

If there's any interest, I can post full schematic and assembler code when finished.

The general plan seems doable, altough there are a few solutions which may not work well, for example the gate drive on primary, and the air core transformer, and missing resistors on LEDs.

Is the gate drive also used for communication?

Why the diode on the discharge transistor?

bearing said:

The general plan seems doable, altough there are a few solutions which may not work well, for example the gate drive on primary, and the air core transformer, and missing resistors on LEDs.

Is the gate drive also used for communication?

Why the diode on the discharge transistor?

Click to expand...

Yeah OK, but the drawing is meant as a concept schematic to show the basic idea,
not a final one.

Yes, the gate drive is also used for communication in case the charger power
supply is not there. This gate driver should then be supplied from the bike's low
voltage supply (BEC in RC terms). But again, it's a concept schematic so don't take
it litteral.

Air core versus iron core, the difference is in the coils series resistance (iron core ->
lower amount of windings necessary for certain inductance -> lower series resistance).
Since the coils are only used during charging, the lower efficiency is not a problem. Plus
what I like about air-coils is that they don't saturate.

I forgot about this diode and transistor, sorry. It's not a discharge section, it's
meant to deliver a stable voltage reference for the PIC's AD Converter. It uses
a zener diode (that's why it's in reverse and has a wiggle in the symbol) supplied
via a resistor. The transistor is there be able to turn off this current path for
when the PIC is in sleep mode. I'm not sure yet this stable reference section
is necessary but I drew it in anyway to make sure the 8 pins of the PIC are
enough...

Lebowski said:

Yeah OK, but the drawing is meant as a concept schematic to show the basic idea,
not a final one.

Click to expand...

Oh, sorry, I didn't read the line that said "concept", I get it now.

Lebowski said:

Air core versus iron core, the difference is in the coils series resistance (iron core ->
lower amount of windings necessary for certain inductance -> lower series resistance).
Since the coils are only used during charging, the lower efficiency is not a problem. Plus
what I like about air-coils is that they don't saturate.

Click to expand...

But air core transformers doesn't have closed flux path, so they will have a bad magnetic coupling between primary and secondary, right? Only like half of the coil will transfer energy and the other half will act like an inductance in series with the transformer winding. Or is it possible to layout the coils in a clever way to prevent this?

Lebowski said:

I forgot about this diode and transistor, sorry. It's not a discharge section, it's
meant to deliver a stable voltage reference for the PIC's AD Converter. It uses
a zener diode (that's why it's in reverse and has a wiggle in the symbol) supplied
via a resistor. The transistor is there be able to turn off this current path for
when the PIC is in sleep mode. I'm not sure yet this stable reference section
is necessary but I drew it in anyway to make sure the 8 pins of the PIC are
enough...

Click to expand...

You could also use an I/O to drive the Vref, and make I/O an input while device is sleeping. But I think other idea is better, to lower part count. Some PICs have internal Vref I believe, and maybe it can be measured by A/D.

bearing said:

But air core transformers doesn't have closed flux path, so they will have a bad magnetic coupling between primary and secondary, right? Only like half of the coil will transfer energy and the other half will act like an inductance in series with the transformer winding. Or is it possible to layout the coils in a clever way to prevent this?

Click to expand...

What I typically do is lightly twist the wire for primary and secondary together and
wind them as one...

The 12F617 does have an internal .6 or 1.2 V reference but this doesn't help as it can
only be used as an ADC input, not as the reference for the ADC. The ADC reference is
either the supply or an external reference connected to a pin. As far as using the
supply as a reference, I'm kind of worried about the supply bounce from the PWM current
through the current sensing resistor during charging. This will be filtered out by the digital loopfilter
but you never know about aliasing....