Ltspice

Using a captured waveform as LTspice input

This outlines the steps necessary to use a CSV file as input for LTspice.

Capture your waveform

I have captured the following waveform using my scope, and saved it as a CSV file.

A sample of the CSV file is shown below:

x-axis,3,D0-D7
second,Volt,
-1.0000000E-03,-48.2407212E-03,0
-999.8000E-06,-48.2407212E-03,0
-999.6000E-06,-48.2407212E-03,0
-999.4000E-06,-48.2407212E-03,0
- - - - - - - - >8 - - - - - - - -
+8.9990000E-03,-168.8437350E-03,0
+8.9992000E-03,-176.8839359E-03,0
+8.9994000E-03,-48.2407212E-03,0
+8.9996000E-03,-48.2407212E-03,0

Convert the CSV file to an 8-bit wav file

Use the script below. There are a few configuration options:

Option Name	Description
CSV_FILE	The source CSV file
CSV_COLUMN	The column of the CSV file that you'd like to extract
SAMPLE_RATE	The rate that the samples were stored at, in Hz
ROOT_FILENAME	(optional) If you want to store the output somewhere else, then change this

#!/bin/bash -e
 
CSV_FILE=loaded.csv
CSV_COLUMN=2
 
SAMPLE_RATE=5000000
 
ROOT_FILENAME=${CSV_FILE%.*}
 
# end of configuration
 
cat ${CSV_FILE} \
  | tail -n+3 \
  | cut -d , -f ${CSV_COLUMN} \
  | sed -r \
    -e 's/^\+//' \
    -e 's/^(-?[0-9]+\.[0-9]+)[eE]([+-])([0-9]+)$/\1*(10^\2\3)/' \
    -e 's/\*\(10\^\+00\)$//' \
  | bc -l \
  | sed -re 's/^(-?)\./\10./' \
  | tee .${CSV_FILE}.normal \
  | sort -n \
  | uniq \
  > .${CSV_FILE}.sorted
 
cat .${CSV_FILE}.sorted | sed -n "1p;$(wc -l .${CSV_FILE}.sorted | cut -d ' ' -f 1)p" > .${CSV_FILE}.extremes
 
LO=$(cat .${CSV_FILE}.extremes | head -n 1)
if [ $(echo "${LO} < 0" | bc) == "1" ]; then
  OP=-
else
  OP=+
fi
HI=$(echo "($(cat .${CSV_FILE}.extremes | tail -n 1))${OP}(${LO})" | bc -l)
ML=$(echo "255/${HI}" | bc -l)
 
cat .${CSV_FILE}.normal \
  | sed -r \
    -e 's/^/((/' \
    -e 's/$/)/' \
    -e "s/$/${OP}(${LO})/" \
    -e "s/$/)*${ML}/" \
  | bc -l \
  | sed -r \
    -e 's/^([0-9]*).([0-9]?[0-9]?[0-9]?).*$/\1.\2/' \
  | tee .${CSV_FILE}.adjusted \
  | awk '{printf "%c", strtonum($0)}' \
  > ${ROOT_FILENAME}.raw
 
DATA_SIZE=$(stat -c%s ${ROOT_FILENAME}.raw)
CHUNK_SIZE=$((DATA_SIZE + 36))
 
WAV=${ROOT_FILENAME}.wav
 
# writeNumAsBin
# arg1: bytes used to represent the word
# arg2: value stored in the word
function wnab() {
  WORD_SIZE=$1; shift
  WORD_VALUE=$1; shift
  printf "%0$((WORD_SIZE * 2))x" ${WORD_VALUE} \
    | fold -w2 \
    | sed '1!G;h;$!d' \
    | awk '{printf "%c", strtonum("0x" $0)}'
}
 
# ChunkID
echo -n "RIFF" > ${WAV}
# ChunkSize
wnab 4 ${CHUNK_SIZE} >> ${WAV}
# Format & Subchunk1ID
echo -n "WAVEfmt " >> ${WAV}
# Subchunk1Size
wnab 4 16 >> ${WAV}
# AudioFormat
wnab 2 1 >> ${WAV}
# NumChannels
wnab 2 1 >> ${WAV}
# SampleRate
wnab 4 ${SAMPLE_RATE} >> ${WAV}
# ByteRate
wnab 4 ${SAMPLE_RATE} >> ${WAV}
# BlockAlign
wnab 2 1 >> ${WAV}
# BitsPerSample
wnab 2 8 >> ${WAV}
# Subchunk2ID
echo -n "data" >> ${WAV}
# Subchunk2Size
wnab 4 ${DATA_SIZE} >> ${WAV}
# data!
cat ${ROOT_FILENAME}.raw >> ${WAV}

Useful resources:

• https://ccrma.stanford.edu/courses/422/projects/WaveFormat/

Feed into LTspice

You'll need a voltage source (like always), but instead of providing a DC voltage, you should enter the following:

wavefile="loaded.wav" chan=0

Here is a sample sallen-key & comparator circuit that I'm testing out for my quadcopter:

When you now run the simulation, you'll get a nice output like this (probing Vin):

Setting the initial voltage of a net

This is useful, for example when you have a passive low-pass filter, you can set the initial state of the capacitor, and that can reduce the amount that you have to simulate.

1. Name the net, for example "V1" (press F4)
2. Add a spice directive ".ic V(V1)=0.333" to set the voltage to 333mV (press s)