Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!cs.utexas.edu!rice!uw-beaver!sumax!amc-gw!richm
From: richm@amc-gw.amc.com (Rich Moran)
Newsgroups: sci.electronics
Subject: Re: JFET Confusion
Keywords: devices fet op-amp
Message-ID: <1165@amc-gw.amc.com>
Date: 16 Mar 90 22:18:50 GMT
References: <7310@rice-chex.ai.mit.edu>
Reply-To: richm@the-end.amc.com (Rich Moran)
Organization: Applied Microsystems, Redmond, WA
Lines: 94

In article <7310@rice-chex.ai.mit.edu> mikec@ai.mit.edu (Mike E. Ciholas) writes:
}I must not understand JFETs yet.
}
}Background...
}
}I am trying to build a circuit that measures current flow very accurately
}without inducing a significant voltage drop.  The following circuit is shown
}on page 2-25 of the Motorola Linear and Interface Integrated Circuits
}databook. ("+" indicates connection of crossing wires)
}
}                   +-----------------------------+
}                   |                             |
}            +------|------+                      |
}            |      |      |                      |
}            |      |   |-----\                   |
}            |      |   |  V+  \               |------|
}  INPUT-----+--R1--+---|-      \              |      |
}            |          |        \             | LOAD |
}            R2         | LF355   )--+         |      |
}            |          |        /   |         |------|
}            +----------|+      /    |            |
}            |          |  V-  /     |            |
}            |          |-----/      |           GND
}     2N4092 |--|          |         |
}      JFET     |<---------|---------+
}            |--|          |
}OUTPUT------+            GND
}            |
}            R3
}            |
}           GND
}
}OUTPUT = R3*(R1/R2)*(current in load)
}
}Typically, R1 is quite small, about .1 ohms.  R2 is about 100 ohms and
}R3 is about 5K.  This gives 5V per amp in load as output.
}
}Now the question...
}
}The way I understand JFETs, N channel types in particular, is that you must
}supply a negative gate voltage on the order of -5V to turn them off
}(the 2N4092 is a depletion mode JFET).
}
}Since the LF355 is supplied from the INPUT voltage to ground, its output
}range must be within this voltage, therefore it cannot turn off the
}JFET.
}
}Is this true?  Does this circuit work?
}

It does work.

This circuit is balanced whenever the voltages at the + and - inputs of
the LF355 are equal.  The voltage at the - input is fixed by the input
voltage minus the drop across R1.  The voltage at the + input is fixed
by a voltage divider consisting of R2, R3 and the JFET.  The circuit
balances itself by feeding back on the JFET, thereby modifying the
voltage divider.

With the circuit balanced, the current through R2 will be .001 (.1/100)
times the current through R1.  Since the current through R2 also passes
through R3, the voltage developed across R3 is a direct representation
of the current through R2 (and R1).

Assume a 1A current through R1.  This yields a 1mA current through R2 and
R3, and +5V at the output (and the source terminal of the JFET).  Note
that the JFET will never be completely turned off, it is used as a variable
resistance to balance the circuit.  Its gate to source junction can be
biased as a result of voltage drop across R3.  If the JFET IS ever turned
completely off, or not biased at all (completely ON), then the circuit
is operating outside of its range.

There are two limiting factors to the operation of the circuit :

First, the input voltage must be at least 5.1 times the LOAD current to allow 
the circuit to balance.  This 5.1/1 ratio can be modified by changing the value 
of R3, and thereby changing the ratio of volts/amp on the output.
If the LOAD current is too high to be measured by the circuit at the given
input voltage, the JFET will be biased completely off.
For higher currents at lower voltages, R3 should be made smaller.

The second limiting factor is low-current operation.   If a very small 
current is drawn by the LOAD, the drop across R3 will be too small to 
allow the LF355 to bias the JFET.  In this case, the JFET will be unbiased,
completely on.
For low current operation, R3 should be made larger (again, changing the 
volts/amp ratio on the output).


-- 
=============================================
Rich Moran
richm@amc.com
=============================================