Waveform LOG, Linear?

[tony3d]

Hi All,

Been learning my new Owon AG1022F Waveform Generator! Using it to help with amplifier circuits mostly, and filter tweaking. The only function I don't understand in Linear vs Logarithmic Sweeps. When I switch from linear to log I don't see any change in the sine wave. I think I understand the difference, but how can you see this on the scope? Also, why would I need log over linear? Never had a waveform generator before so this is kinda new.

[rstofer]

When looking at frequency and phase plots, they are almost always in log scale.  Put a filter on the end of you generator and sweep over a few decades.  Maybe have a center frequency of 10 kHz and start the sweep from 100 Hz and run up to 1 MHz (5 decades).  When you look at the graph, the roll off should be a straight line of xx dB per decade.  If you use a linear scale, the distance between 100 and 1000 Hz, 1 decade, won't be very much.  The distance between 100 kHz and 1 MHz, again, 1 decade, will be enormous.

[tggzzz]

Hi All,

Been learning my new Owon AG1022F Waveform Generator! Using it to help with amplifier circuits mostly, and filter tweaking. The only function I don't understand in Linear vs Logarithmic Sweeps. When I switch from linear to log I don't see any change in the sine wave. I think I understand the difference, but how can you see this on the scope? Also, why would I need log over linear? Never had a waveform generator before so this is kinda new.

It is basic electronics w.r.t. frequency response. Look up "bode plots", and understand why linear filters (combinations of R,L,C) look simple and recognisable in the double-log bode plots.

[MosherIV]

The only function I don't understand in Linear vs Logarithmic Sweeps. When I switch from linear to log I don't see any change in the sine wave. I think I understand the difference, but how can you see this on the scope? Also, why would I need log over linear? 

The key word is SWEEP. What this does is to change the frequency (rate) of the wave not the waveform.
It is easier to describe linear, this changes the freq at a steady rate.
Logrithmical changes the freq at a changing rate, the change is very similar to a square law.

How can this be seen on a scope?
Set the time base on the scope so that you can see 3 sine wave, then do not change the time base.
Start the sweep function and you should see the number sine wave on the scope increase on screen.
Linear should increase the number at a fixed steady rate.
Logrithmical, the number increases at a faster and faster rate.

Why is Log useful?
Because many of our sense are perceived in a logrithmical maner, eg to hear a double in volumne, the electrial level must be doubled. On a Log scale, this doubling looks like a linear straight line.
Try looking at data sheets and the freq response, the freq scale is always log.

[tony3d]

Thanks everyone. Great help!

[Brumby]

Time =

0 sec.

1 sec.

2 sec.

3 sec.

4 sec.

Linear

100Hz

200Hz

300Hz

400Hz

500Hz

ie add a figure for each subsequent point - in this case, 100Hz

Log

10Hz

100Hz

1kHz

10KHz

100KHz

ie multiply by a figure for each subsequent point - in this case, 10

These are the equivalent of plot points on a graph.  The frequency will change smoothly, according to the curve of best fit of these points.  Sweeps can go down or up - depends on what your looking for.

The logarithmic scale is how we perceive music.  Each octave is 2x the frequency of the previous (lower) octave.

[CatalinaWOW]

Tied in with the perception thing is a practical element.

A linear sweep slow enough to distinguish between 20 and 200 Hz (say sweeping that range in a five seconds) would take interminably long to sweep even the entire audio range, roughly 10 hours, even longer for broader ranges.  Because most perceptible changes in response (both from our senses and from typical filters) occur over octaves or decades of frequency the required resolution is lower as the frequency increases, and thus the practical sweep rate is higher.  This is what the log sweep does, sweeping each octave or decade in frequency in the same time.  So in the initial example a sweep from 20 to 200 Hz takes 5 seconds, from 20 to 2000 Hz takes 10 seconds, from 20 to 20000 Hz takes 15 seconds.  Much more acceptable.

[Brumby]

Yes, the practicality issue is most significant.

Another example that I was thinking of using before... Let's assume you have set up your sweep generator to cover the range of a piano ... and you are going to 'play along'.

Let's start with a logarithmic 'sweep'.  If you were to play all the keys for one second starting from the bottom and going all the way to the top, it would take you 88 seconds.  At the start, the change of frequency is around 1.7Hz per second and it finishes at about 235Hz per second. 

If you take the linear approach at, say, 2Hz per second, you will press the second key in less than a second.  When you get to Middle 'C' you will have to wait 8 seconds before pressing the next key and when you get to the last but one, it will be almost 3 minutes before you can hit that last note.  All in all, it would take you close to 35 minutes to complete the sweep - and for the last 10 minutes, you'll be wondering more and more if the frequency is actually changing....  (You might need a DFM to convince you it is.)

[tony3d]

Thanks again everyone.