kell wrote...
Genome wrote
Fred Bloggs wrote in message news:
[email protected]...
[email protected]-spam.invalid wrote:
I have seen alot of circuits with a 7812 (12V regulator) with
a PNP transistor for current bypass (or how you explain it).
I would like MOSFET transistors instead of bipolar transistors.
. +-----------------------+ +-+-------+
. | | ^ | |
. | ===== |
. | | |
. +-----e c----+---------+---[R2]--+
. | \ / | |
. | --+-- C1=== |
. | | | |
. | | [R3] +--7812--+ |
. | | | | | |
. Vin>-+--[R1]-+-------+----+ +-----+-->Vout
. | |
. +---+----+
. |
. gnd
That looks like it might be positive feedback.
Compare Fred's circuit to:
http://www.rason.org/Projects/vreg/vreg.htm
Of course, if you replace Q2 with a mosfet you still have the dropout
voltage problem...
That needn't be so bad, at least on account of the MOSFET. A 7812 and
PNP transistor want at least 2.2 volts to operate (at 0.25A), and all
of this voltage is available as gate-source voltage for the FET. Some
power MOSFETs are happy with such a modest gate voltage, when used as
linear pass elements in their subthreshold region. This is especially
true for logic-level MOSFETs. But if the 7812 was replaced with an LDO
regulator, for reduced dropout voltage, that'd be another matter! Then
MOSFET (even a logic-level type) would become the limiting element.
One other issue for this circuit. Stabilizing the feedback loop with
all its multiple gain elements will be a bit tricky. Many folks would
be tempted to use Spice to evaluate the circuit. Let's assume you find
a good 7812 model. There's still one nasty problem, which is to create
an accurate subthreshold FET model. The sad truth is the manufacturer's
models are disasters in this region, and completely fail to work properly.
There is a solution, which I've written about here before in detail. A
Google usenet search should turn up several useful threads.