current sensing

I just thought of something very simple. With a pnp transistor's b/e junction set across a sense resistor placed in the current path into the battery's positive terminal, I can bring the pnp transistor's base down a few tenths of a volt by connecting it to a voltage divider across the battery, effectively reducing the voltage needed across Rs to turn the transistor on. Anybody see any problems with this?

See

The LM105/205/305 are all obsolete, but they used this technique.The problems are the tolerance on the base-emitter voltage (around 200mV) and the temperature dependence (-2mV/C).

use an opamp, or comparator.

A linear discrete amp is another possibility.

NT

Go to your local auto recycler (you may have to shop around a bit) and try to find a voltage regulator from the same vintage and model of car, clean its contacts, and slap it in. :-)

Good Luck! Rich

Have a look at:

A comparator or opamp that works at the +Ve rail is good, but it's possible to work with common discrete transistors and low-value sense resistors, if two transistors are used...

. + -----+--+-- Rsense --+----> To Batt (+) . | '-----, | . Rc | | . | c\\| | . \\e |----' . PNP |---+--e/| NPN . /c | . | R . Rload | . | | . - -----+-----+---------------> To Batt (-)

Ed, clip that conductor that shorts the upper resistor and you'll see what I was talking about. Anyway, I just thought of something a little better and just as simple.

it's possible to work with common discrete transistors

I guess I'll have to get your book, look up the circuit you just posted and read about it so I can get my naches (Yiddish for brains) around it.

It doesn't reduce the rsense wattage, and the divider isn't a divider - the top resistor is shorted.

Draw it - you'll see the problems:

• -----+---[Rsense]---+---+---> To Batt (+) | | | | | [Rdivider] \e | | PNP |------------+---+ /c | | [Rdivider] [Rload] | | |

- -----+------------------+---> To Batt (-)

You could use a .01 ohm 3W as the sense resistor, with a comparator circuit if you must avoid a high wattage resistor.

Ed

Look into antique generators and you'll find out why. It's not an alternator. It's an ANTIQUE GENERATOR. ON AN ANTIQUE BIKE.

Sorry I yelled Rick.

Win: It looks like the tempcos of the transistors cancel!

I'm wondering why you want to limit the generator output to 10 amps....

doesn't seem substantially differnt to this:

. + -----+----- Rsense --+----> To Batt (+) . | | . Rc | . | | . \\e ' . PNP |---+---|

correct

exactly what I'm doing See this link about how old GM generators work. My bike works the same. It's a two-brush generator, twelve volts

I've already built two prototype elctronic voltage regulators that work just fine. I'm not making all the ignorant mistakes people are ascribing to me. My original post was about how to implement current limiting, because the voltage regulators I built don't have it.

Read the book, see the movie. :-)

Assuming matched transistors and operating currents (not comletely possible, but anyway...), the output voltage on Rload is Vo = I Rsense * Rload/Rc.

If an opamp is substituted, better accuracy may be enjoyed.

. + -----+--- Rsense ----+----> To Batt (+) . | | . Rc | . +------------, | . | __ | | . c\\| / +|--' | . NPN |---< | | . e/| \\_-|-----' . | . Rload . | . - -----+--------------------> To Batt (-)

Note the transistor is now an NPN type, and note the opamp's inputs must be able to operate near the positive rail if the opamp is powered from the + input line.

I vaguely recall a low-drop high-side current sensing circuit along the lines of the ascii sketch below.

Rs V+ -------+-----/\\/\\----+---> Iout | | R1 R2 I1| |I2 e\\| Vbase |/e pnp|----+----|pnp /| |\\ | | +------+ +----> Vswitched | | | \\| | |/ npn|----+----|npn e/| |\\e | | 0v-------+-------------+----

Iout*Rs = I1*R1 - I2*R2

The relative values of R1 and R2 were adjusted so that Vswitched changed logic state at a defined Iout.

I haven't done any sums yet to find out the optimum value for Vbase (probably with respect to V+), etc.

Stability. The threshold depends on VBE which has a temperature coefficient of -2mV/C. Also the bias voltage is derived from the 12V which is probably not too stable. A better solution is to use a PNP long-tailed pair. This gives a good CMRR without needing presicion resistors round an op-amp