Using gnuplot, Zener diode behaviour and other tests

Software, Elektronik
Page content

Below I’ve experimented with gnuplot, in few examples:

  • Reading and displaying data from a file
  • Reading and displaying data from a GPIB-connected device (measurements from a voltmeter with open input connection)
  • Reading and displaying data from a GPIB-connected device (measurements from a voltmeter connected to a 10 volts voltage reference)

gnuplot is a nice tool for plotting data. Start gnuplot to enter plot commands. I tried this tool when analyzing a strange behaviour of a Zener diode. In the end it turned out that there was no strange behaviour.

Data file. Values separated with whitespaces:

Current [mA]    Vin     Vzener
0.100   2.420   2.396
0.200   2.710   2.652
0.300   2.894   2.862
0.400   3.038   2.911
0.500   3.158   2.997
0.600   3.260   3.068
0.700   3.359   3.130
0.800   3.447   3.183
0.900   3.528   3.231
1.000   3.605   3.274
1.100   3.678   3.313
1.200   3.748   3.349
1.300   3.816   3.382
1.400   3.880   3.413
1.600   4.005   3.469
1.800   4.123   3.519
2.000   4.236   3.564
2.200   4.344   3.604
2.400   4.450   3.641
2.600   4.552   3.675
2.800   4.652   3.707
3.000   4.748   3.735
3.200   4.843   3.763
3.400   4.938   3.788
3.600   5.030   3.813
3.800   5.120   3.835
4.000   5.210   3.857
4.200   5.300   3.877
4.400   5.386   3.896
4.600   5.472   3.915
4.800   5.558   3.933
5.000   5.643   3.949
6.000   6.057   4.023
7.000   6.459   4.084
8.000   6.849   4.135
9.000   7.234   4.180
10.000  7.614   4.220
15.000  9.462   4.365

Plotting the data

I want to output two curves (Vin, Vzener), x-Axis is first column of data file (here: Current [mA]).

Background for this graph: This graph shows regulated voltage at a Zener diode, as a function of current through the diode. Vin is Input voltage coming in at a shunt resistor of 330 ohms.

For an ideal Zener diode, on increasing Vin, the Vzener should also increase, up to the hard limit of the Zener voltage, here 3.3 volts. When Vin is further increased, output voltage (Zener voltage) should stay fixed on zener voltage 3.3 volts, so we should see a horizontal line for that part of the graph.

A real world Zener looks different.

Next line shows my plot command:

gnuplot

gnuplot> plot "zener.dat" using 1:2 title 'Vin' smooth bezier lw 2, \
  "zener.dat" using 1:3 title 'Vzener' smooth bezier lw 2, 3.3 title "3v3", 3.9 title "3v9

lw 2 means line width. using 1:2 means ‘use column 1 as x-axis and column 2 as y-Axis for curve/plotline named “Vin”’. smooth bezier means connect the graph points with some interpolating function (e.g. bezier).

All curves are speparated by a comma. There are two horizontal lines, for special values (3,3 and 3,9 volts.).

Grid:

gnuplot> set grid

Vertical line:

gnuplot> set arrow from 5,0 to 5,10

Result:

Background regarding the plot: This is a data curve from a Zener diode. I bought it as a 3,3 volts zener. Seller named it “ZF 3.3”, I found out that it should be a BZX79C 3V3.

This diode, for the 3V3 type, should regulate between 3.1 and 3.5 volts, when a current of 5mA runs through the diode.

I measured zener voltages at current values between 0.1 and 15 mA. Data set consists of 39 samples, each consisting of a current value (used for X axis), an input voltage at 330 ohms shunt resistor called Vin, and zener output voltage, measured between Kathode and Anode of diode, called Vzener.

At target current value of 5mA this diode regulates to 3.949 volts. So, from the results, it looks that this is not a 3V3 type, but a 3V9 type.

Under microscope, I finally found that this is in fact a 3V9 diode. It was a BZX55C 3V9 that somehow found its way into the 3V3 box.

BZX79C Datasheet

Look at a real 3V3 diode

After that I took a real 3V3 diode (again BZX79C 3V3, but now a correct one).

Current [mA]    Vin     Vzener
0.1 1.942      1.918
0.2 2.182   2.123
0.3 2.337   2.243
0.4 2.461   2.331
0.5 2.567   2.402
0.6 2.661   2.461
0.7 2.748   2.513
0.8 2.828   2.557
0.9 2.903   2.597
1   2.976   2.634
1.1 3.044   2.667
1.2 3.110   2.697
1.3 3.175   2.727
1.4 3.237   2.753
1.5 3.298   2.779
1.6 3.358   2.803
1.7 3.415   2.825
1.8 3.471   2.846
1.9 3.530   2.869
2   3.585   2.888
2.2 3.694   2.923
2.3 3.728   2.940
2.4 2.776   2.956
2.6 3.879   2.988
2.8 3.961   3.010
3   4.053   3.037
3.2 4.146   3.062
3.4 4.243   3.087
3.6 4.34    3.110
3.8 4.43    3.131
4   4.515   3.151
4.2 4.615   3.174
4.4 4.705   3.193
4.6 4.873   3.211
4.8 4.843   3.21
5   4.925   3.226
6   5.348   3.301
7   5.759   3.362
8   6.168   3.419
9   6.568   3.465
10  6.966   3.51
15  8.990   3.677

And the graph:

It can be seen that for 5mA, a voltage of 3.226 is regulated, well inside datasheet margins for a 3V3 type.

Dynamic plotting examples

Below I show how gnuplot can display datasets that contain more and more data points over time.

Measurements from a voltmeter, open inputs

Lets say we get measurement values from a voltmeter, via GPIB, into a file. For this simple test, the voltmeter is connected to nothing, just floating.

Data comes from a tool called HP3455A_ctl, creating a single voltage measurement output per call, which is converted to a line containing a timestamp number and the raw voltage value. This line is appended to an output file called out.dat. Skript looks like this:

#!/bin/bash
outfile="out.dat"
num=0

rm -f $outfile
while [ true ]; do
  v=$(HP3455A_ctl GET_MEAS)
  # [0] Value: -1.364690e+0
  v=$(echo $v|sed -r 's/Value: //'|sed -r 's/\[.\]//')
  echo "$num $v" >>$outfile
  echo "$num $v" 
  num=$((num+1))
  sleep 2
done

This creates a file, getting more and more lines each 2 seconds, looking like this:

0  -1.367780e+0
1  -1.385770e+0
2  -1.386650e+0
3  -1.387160e+0
4  -1.386740e+0
5  -1.404630e+0
6  -1.387100e+0
7  -1.374780e+0
8  -1.404040e+0
9  -1.387100e+0
10  -1.375040e+0
11  -1.404310e+0
...

gnuplot has a while() loop, and a command replot that will read in a data file again and updates the plot. This can be done like this:

% gnuplot
gnuplot> plot  "out.dat" using 1:2 title 'Uout' smooth bezier lw 2
gnuplot> while (1) {
more> replot
more> pause 2
more> }

This can also be all put to a file:

% cat dynamic-plot.gnuplot
plot  "out.dat" using 1:2 title 'Uout' smooth bezier lw 2
while (1) {
 replot
 pause 2
}

And gnuplot can be called then like this:

% gnuplot dynamic-plot.gnuplot

Both approaches (interactive or from the file) give same result.

Resulting plot, with roughly 300 values collected so far:

hp3455, first 700 seconds

and after 5600 seconds, we have about 2800 data points now:

hp3455, first 3000 seconds (i.e. 50 minutes)

Value goes slowly down from -1.38 volts to -1.42 volts. There is some interesting “oscillation” at beginning and end of graph.

10.422084 voltage reference over time variations

Voltage taken from a high precision voltage reference. This reference emits a voltage of 10.422084 volts.

My device measured around 10.42239 volts, so ~ +310µV deviation, maybe the HP3455A is slightly out of calibration. I cannot verify this, because the HP3455A is my most precise voltmeter.

voltage reference with hp3455a, first 600 seconds

And after around 4800 seconds:

oltage reference with hp3455a, first 4800 second

After roughly 1200 seconds (20 minutes), the volatage is stable between 10.222380 and 10.422395 volts. So, variation over time around is 15µV.

Line voltage over time variations

Final graph: 230 volts line voltage. I took the voltage from a 10 transformer that gives roughly 10 volts AC. Duration of the graph is more than 3 hours (12000 seconds)

gnuplot-netzschwankungen.png

This looks wild and random. There are many devices, some close to my power plug, some other far away, even some out of house, which switches on and off, which results in a noisy signal. To get the real voltage changes in 230 volt range, we would have to multiply all values by factor 23.