On 02/16/11 04:56 pm, Dave Platt wrote:

N5BIA offers a kit, and I was wondering whether this could be beefed up
to handle 25A by adding 12ga wire to all the charging-current traces and
substituting higher-current pass transistor(s).

... and huge heatsinks.

Of course. His kit includes only the PC board and the PCB-mounted
components anyway, so I would have to provide my own enclosure and heat
sink (well, maybe it does include a clip-on heat sink; I don't recall).

I'd have to recalculate
the resistor values to suit a flooded battery, of course. And what about
using a P-channel MOSFET device, such as the STP80PF55 that the "Micro
M+" uses?

The thing about a low-Rds-on MOSFET, or a low-Vce-sat PNP, is that it
really only gains you a benefit under one circumstance: when it's
"hard on", acting as much as possible like a short-circuit. This will
happen only during the "bulk" fast-charge stage... and only if the
charge controller "sees" that the raw (unregulated) power supply
circuit isn't capable of pushing more amps into the battery than the
design allows.

If the charge control circuit finds it necessary to reduce _either_
the charge amperage, or the voltage being delivered to the battery, in
order to charge the battery safely, then the pass transistor will be
"partially off". There will be a significant voltage across it
(roughly speaking, Vsupply - Vbattery) and lots of amperage, and so it
will be dissipating a lot of energy as heat.

At that point, the actual Rds-on of a MOSFET, or the Vce-sat of a PNP,
will matter not at all. You'll have to dissipate (Vsupply-Vbat)*Icharge
watts of heat in the transistor.

My idea of using a MOSFET was to avoid the voltage drop of a junction
transistor so that it could be fed from a regular P/S that has been
cranked up only a little -- as with the "Super PwrGate," which has a
voltage drop of no more than 0.5V; any idea what West Mountain Radio
uses to accomplish that? So Vsupply - Vbattery would be low, and also
the power dissipation.

Now, if you happen to have been careful (or lucky) enough in the
design of your "raw" power supply, things will look good. By "careful
or lucky", I mean that you've put together a raw supply which just
happens to run out of "oomph" at exactly the right moment... the
effort of delivering 25A into the battery just happens to cause the
supply to sag down to the right voltage (equal to a voltage in the
range you want to be charging at). Under those conditions, the
whole system will be running "flat out", the pass transistor will be
turned on as hard as it can be, and heat dissipation in the transistor
will be minimized by using a low-voltage-drop transistor of some sort.

However, this approach has pitfalls... it will be finicky to get right
(component selection will be difficult) and it will probably be very
sensitive to variations in the AC power-line voltage. In real life,
you'd find that much of the time, either you aren't getting the full
25A of charge current (line voltage too low), or the voltage and/or
current are potentially too high and the charger is having to back off
turn down the pass transistor (at which point there's no longer an
advantage to a low-voltage-drop transistor).

If you want to deliver a high charging current, with good control and
low losses, under a fairly wide range of conditions, I think you'd
probably want to use a different approach... use a buck-mode switching
regulator rather than a linear pass-transistor system. With that
approach, the pass element would almost always be fully-on or
fully-off, and thus there'd be a real benefit to a low-resistance
MOSFET or a low-saturation power bipolar part.

I'm trying to avoid switching-type circuitry, since I already have
RFI-quiet regulated power supplies that are capable of supplying the
desired maximum voltage and current. It's just a matter of reducing the
voltage at the appropriate stages of the charging/maintenance process.

I was looking at the Super PwrGate until WMR's tech guy pointed out that
the absorption voltage is too low and the float voltage possibly too
high for flooded batteries. If I knew that it used a UC3906 and did not
use SMT components I might be willing to try modifying one, but I don't
have that information.

"Perce"