remove distortion using Total Harmonic Distortion (THD) Test Circuit

Thread Starter

artmaster547

Joined Jan 6, 2016
409
Hi
Would like any guidance about removing harmonic distortion from a sine wave created from a phase shift oscillator, I have come across the following two circuits but not sure if I can use them to remove any unwanted harmonics I would like a fourier transform with a peak only at 1kHz and other frequencies to be attenuated.
Thanks Art
I have attached documents with two kinds of circuits, one called
-Total Harmonic Distortion (THD) Test Circuit
and the other:
-Precision Absolute Value Circuits
 

Attachments

Lestraveled

Joined May 19, 2014
1,946
How do you wish to remove the distortion, by adjustment or filtering?

The AN5420 paper applies to amplifiers, mainly op-amps. You need a different approach for measuring distortion in a oscillator. What are your goals?
 

Thread Starter

artmaster547

Joined Jan 6, 2016
409
How do you wish to remove the distortion, by adjustment or filtering?

The AN5420 paper applies to amplifiers, mainly op-amps. You need a different approach for measuring distortion in a oscillator. What are your goals?
hi thanks for replying, I have a feeling using filtering, i believe the active components will effect the harmonics further or am i incorrect in my thinking and I should just use a bandpass filter so the sine wave oscillates at the desired frequency. My goal is to produce a sine wave with as minimal distortion as possible which can be viewed when carrying out an FFT.
 

Thread Starter

artmaster547

Joined Jan 6, 2016
409
I have created a sine wave which I want to create using a phase shift oscillator I know its easier with Wein bridge, and I have used a JFet to stabilise the amplitude to a certain degree but would like it to be even better
 

Lestraveled

Joined May 19, 2014
1,946
................... but would like it to be even better
That is my question, how do you want to make it better, by adjusting the oscillator or adding a filter?

Also, if you do this, how will you measure the improvement to the distortion? This will be far more expensive than the oscillator.
 

Thread Starter

artmaster547

Joined Jan 6, 2016
409
I was thinking a filter since I've already added a JFet in the circuit but if you think the oscillator could be adjusted with some topology I would be open to suggestions. In terms of improvement, this is mainly just me creating simulations using LTSpice so cost is not a consideration for this at this stage anyway, I think for measurements I would have feeling it would involve some feedback component
 

MrAl

Joined Jun 17, 2014
11,389
Hi,

What frequency do you want to operate at?
I bandpass filter will work in removing both higher and lower harmonics, but a low pass filter might work too if you dont mind some attenuation. Depends on how fancy you want to get. A first order filter does some filtering, a second order does more, a third order more yet, etc., but at some point it gets harder to adjust the components if you need an exact setting, so usually a second order filter is good enough. I used a 6th order passive high pass filter one time to remove effects of room lighting 60 or 120Hz oscillations from interfering with the signal. Just 6 resistors and 6 capacitors. Quite a bit of attenuation, but an op amp boosted the signal again.
 

Nykolas

Joined Aug 27, 2013
115
The OP did not state how much "little as possible" distortion is. A well designed Wien-Bridge oscillator easily achieves distortion levels below 0.01%, a phase-shift oscillator not so much. Unless you are forced to use a phase-shift, I would suggest using a low distortion oscillator to start with. E
 

Thread Starter

artmaster547

Joined Jan 6, 2016
409
Hi,

What frequency do you want to operate at?
I bandpass filter will work in removing both higher and lower harmonics, but a low pass filter might work too if you dont mind some attenuation. Depends on how fancy you want to get. A first order filter does some filtering, a second order does more, a third order more yet, etc., but at some point it gets harder to adjust the components if you need an exact setting, so usually a second order filter is good enough. I used a 6th order passive high pass filter one time to remove effects of room lighting 60 or 120Hz oscillations from interfering with the signal. Just 6 resistors and 6 capacitors. Quite a bit of attenuation, but an op amp boosted the signal again.
thank you so much for confirmation and for replying :)
 

Veracohr

Joined Jan 3, 2011
772
I've been working on exactly this for myself. My function generator's output isn't the cleanest so I'm making a filter for 1kHz and 100Hz distortion tests. The attached circuit is a modified elliptical filter with a Sallen-Key low pass on the output. The intention was to attenuate the second and third harmonics a lot, then standard low pass the rest.

Simulation shows -3mdB at 1kHz, -51dB at 2kHz, -66dB at 3kHz, and greater than 43dB attenuation at all frequencies above 2kHz. In the simulation I have NE5532 opamps but I'm actually going to use LME49740 which boasts 0.00003% distortion.

I haven't built it up yet so I can't verify how it measures up against the simulation.
 

Attachments

Thread Starter

artmaster547

Joined Jan 6, 2016
409
I've been working on exactly this for myself. My function generator's output isn't the cleanest so I'm making a filter for 1kHz and 100Hz distortion tests. The attached circuit is a modified elliptical filter with a Sallen-Key low pass on the output. The intention was to attenuate the second and third harmonics a lot, then standard low pass the rest.

Simulation shows -3mdB at 1kHz, -51dB at 2kHz, -66dB at 3kHz, and greater than 43dB attenuation at all frequencies above 2kHz. In the simulation I have NE5532 opamps but I'm actually going to use LME49740 which boasts 0.00003% distortion.

I haven't built it up yet so I can't verify how it measures up against the simulation.
that's super useful thank you so much I'm having problems openning this simulation however because of NE5532 opamps, is there an alternative it doesn't matter if its worse, but preferrably readily available in LTSpice
 

Veracohr

Joined Jan 3, 2011
772
Here's the NE5532 model. Change the extension back to .sub (the forum software doesn't allow attaching .sub files) and add it to the sub folder.
 

Attachments

Top