|
Diode Forward Voltage versus Forward
Current
01 February 2008
In electronics, as in life, it's often not what you don't
know that gets you in trouble, but rather the things you think you know but in
fact are wrong. I ran into that yesterday when I add a 1N5711 Schottky diode to
clamp the reverse base voltage across a 2SC1945 transistor 10 watt RF power
amplifier.
When I measured the reverse voltage clamping level,
I expected it to be around 0.5 volts or less. After all, a Schottky diode has a
drop around 0.4 to 0.5 volts, compared with a standard silicon diode's 0.7
volts, right? When I measured around 1.2 volts across the 1N5711, I first
thought I installed a 1N4148 silicon diode instead. The diode was clearly (well,
clearly after using a magnifying glass) marked 1N5711.
That lead me to look at data sheets for several diodes and
to measure the forward current versus forward voltage of four signal diode
types:
- 1N5711, Schottky diode
- 1N4148 Silicon diode
- 1N270 Germanium point contact diode
- 1N914 Silicon diode
This data is taken with an automated measurement system,
employing an HP6038A system power supply, an HP3456A digital multimeter and an
Agilent 34410A digital multimeter, all controlled via a GPIB
bus with a Prologix controller card. The software is home made, running in
Liberty Basic. I've inserted a 1KΩ 2 watt series resistor to make the HP6038A
into a quasi constant current power supply. (The HP6038A will function in
constant current mode, but its resolution and minimum step size is inadequate
for our purposes.)
I've also looked at three of these diode types when used
as RF detectors as might be the case in a simple RF probe. See the page at
Diodes for RF Probes
Because the DC power is continuously applied to the diode
under test, at higher current levels the diode junction's temperature will be
elevated. Since a diode's forward voltage drops with increasing temperature,
it's reasonable to expect the measured data to show increasing error with
increasing current. The error will be to measure less forward voltage than would
be found than if measured using a brief current pulse (300 μs seems to be the
commonly used value) where the forward voltage drop can be measure before
thermal effects appreciably warm the junction. I don't have suitable equipment
to make these tests in an automated environment and it would be very time
consuming to collect data manually. Accordingly, we'll accept some error in
these measurements, recognizing that in fact this constant current measured data
is more useful for purposes where the diodes are use in an environment in which
the junction is heated by the forward current.
As a comparison point, I also simulated these four diodes
using LTspice. I don't know where the various diode models I use in LTspice came
from—some may have been provide with the original LTspice installation but I've
added many models to my component library over the years. Since the accuracy of
a SPICE run depends upon the accuracy of the model, one should not ascribe
divergence between prediction and measurement to the simulator. Rather, the
model and the measurement techniques should both be questioned and further
studied.
The plot below shows measured and LTspice predictions for
these four diodes. I do not have a SPICE model for the 1N270 diode, so I instead
used a similar 1N34A Germanium point-contact diode in the simulation. Also, the
1N4148 and 1N914 SPICE models are essentially identical, so their plots overlay
each other and only one is visible.
The solid lines are measured data and dotted lines are the
SPICE simulations.|
|
