# Linearity

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.

#### maxpou

Hi Guys,
how can i check the linearity of my circuits? Thank you! Maxpou

A THD analyser.

#### AKSA

A triangle waveform generator and a fast CRO for a visual indication

#### maxpou

Thank you for answers!
but what's CRO?
Thank you!
Maxpou

#### gootee

maxpou said:
Thank you for answers!
but what's CRO?
Thank you!
Maxpou

A CRO is an oscilloscope (Cathode Ray Oscilloscope).

The triangle waveform generator is a good idea. And it is easy to build a fairly good one. (Do a search at national.com for "tri-wave generator".)

To check linearity, you would want to make sure that vout/vin is constant, for all expected values of vin.

Checking linearity with a triangle waveform for vin is easy: If vout is also a triangle shape, the system is linear versus input voltage. But if the rise or fall portion of the output waveform is curved, or worse, then the system is not linear.

You would probably want to check that that is also true when vin is changing, for all expected rates of change of vin. The triangle generator would also handle that, if it had variable frequency as well as amplitude.

#### Onvinyl

Hi,
interesting topic! What is the ballpark of THD-N when the triangle-wave is visible deformend?

Rüdiger

#### jan.didden

Paid Member
Onvinyl said:
Hi,
interesting topic! What is the ballpark of THD-N when the triangle-wave is visible deformend?

Rüdiger

That's a difficult one, as a perfect triangle is a sine wave + 30% 2nd harmonic. Plus 4th, plus 6th, plus.... .

Just as a perfect square wave is a sine wave + 30% 3rd harmonic. Plus 5th, + 7th + ...

That is why a tringle is a very crude way to guesstimate linearity.

Jan Didden

#### PMA

Paid Member
Should be said that depending on repetition frequency, the triangle may be distorted in shape, as no audio circuit works from DC to infinity. HF corner results in triangle corner rounding, LF corner results in a "bow" shape of triangle. It will work only for repetition freq. in the midband.

#### Pingrs

Is the triangle trace any better than the classical XY trace of I/P - O/P sinewave. Loss of linearity is immediately apparent by curvature of the (ideally) straight line at midband, or lack of ellipticity elsewhere (and shows phase to boot!)
This is how I try to detect slew rate limiting. Does that make sense?

Brian.

#### PMA

Paid Member
A "bow" shape (red curve), due to low frequency corner. Nothing in common with non-linearity.

#### Attachments

• bow.gif
10.2 KB · Views: 514

#### PMA

Paid Member
"Rounding", due to high frequency corner. Nothing in common with non-linearity.

#### Attachments

• rounding.gif
10 KB · Views: 456

#### unclejed613

a THD analyzer is best. if distortion is high enough to become visible on a scope, it's definitely far above the audibility threshold.
depending on the sharpness of the scope trace, distortion doesn't become visible until it's about 1 or 2 percent. you can approximate a medium sensitivity THD analyzer by using a dual trace scope, using the ch1&ch2 add function, with ch2 inverted (it's now subtracting) and monitoring the input signal to ch1, and monitoring the output on ch2, and adjust for the smallest waveform.

#### Pan

PC

Low distortion audio interface

ARTA

As for the interface/soundcard, Echo audio's Audiofire measures very well.. especially for the price.

/Peter

#### gootee

What is the highest frequency for which THD can be reliably measured, with any PC and soundcard?

One of the "standard" worst-case THD measurements seems to be for 20 kHz, which I assumed could not be done with a PC and soundcard.

#### PMA

Paid Member
With Lynx L22 or ESI Juli etc. you have Fs = 192kHz, so BW = less than 96 kHz. For 20kHz, you get 4th harmonic. For 10kHz, the ninth. Both Lynx and Juli have very low distortion, well under 0.001% even at 10 and 20kHz.

#### maxpou

Thank You Guys for answers,
But I am not an EE who wants to test a lot of circuits only my designs, I concluded that the least expensive solution is to buy a analyzer THD (HP334A) on ebay or i have another choice?
Thank you! Maxpou

#### gootee

maxpou said:
Thank You Guys for answers,
But I am not an EE who wants to test a lot of circuits only my designs, I concluded that the least expensive solution is to buy a analyzer THD (HP334A) on ebay or i have another choice?
Thank you! Maxpou

That should work fine. I got a recently-calibrated and cosmetically-very-good one from ebay for about \$150, a few years ago. The minimum full-scale distortion is 0.1%, which puts the smallest scale-markings down to 0.002%, which should be good for measurements at least down to 0.001%. Note that the HP334A is "manually" operated. i.e. You have to twiddle the frequency-tuning and null knobs until you get the lowest reading. It usually takes me about one to two minutes, to make a THD measurement.

On the other hand, I had a Tektronix DA4084 plug-in for a 7603 oscilloscope mainframe, for a while. It was much nicer, since it was fully automatic, and also had a digital display.

#### maxpou

gootee said:

A CRO is an oscilloscope (Cathode Ray Oscilloscope).

The triangle waveform generator is a good idea. And it is easy to build a fairly good one. (Do a search at national.com for "tri-wave generator".)

To check linearity, you would want to make sure that vout/vin is constant, for all expected values of vin.

Checking linearity with a triangle waveform for vin is easy: If vout is also a triangle shape, the system is linear versus input voltage. But if the rise or fall portion of the output waveform is curved, or worse, then the system is not linear.

You would probably want to check that that is also true when vin is changing, for all expected rates of change of vin. The triangle generator would also handle that, if it had variable frequency as well as amplitude.

Hi Gootee,
i'm very disappointed because i have an audio generator with triangle waveform and scope! Maxpou

#### maxpou

gootee said:

Note that the HP334A is "manually" operated. i.e. You have to twiddle the frequency-tuning and null knobs until you get the lowest reading. It usually takes me about one to two minutes, to make a THD measurement.
.

Hi Gootee,
are you sure it is not automatic?
Maxpou

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.