opamp sine wave oscillator

[John Popelish]

Our 2ppm may well be coming from the caps.

Down at 2ppm, I would start worrying about resistor
linearity, also.

 

[Phil Allison]

"Vladimir Vassilevsky"

It largely depends on the excursion of the cone, which is higher at low
frequencies. Can be anywhere from ~10% at low to 0.1% or so at the middle
to high frequencies. For the speakers, the IMD is much more
representative parameter then the THD.

** No point in testing a multi-way speaker ( ie the vast majority ever
sold) for IM distortion.

The famous Quad ESL63 electrostatic speaker has THD in the order of 0.03% at
97dB SPL.


....... Phil

 

[Phil Allison]

"Arie de Muynck"

I still wonder if sending e.g. 100 Hz and 5 kHz at the same time to a
speaker wouldn't give a 100Hz FM modulation of the 5 kHz tone (doppler
effect caused by the cone movement).

** Cone displacement modulates the *phase* of the high frequency tone - but
it ain't audible.

Any resulting Doppler effect is undetectable since the sidebands are
submerged beneath similar ones produced by the IM distortion of the driver
when operated with large cone excursions.

But maybe the microphone has an inverse effect?

** Nope.


...... Phil

 

[MooseFET]

MooseFET wrote: [...]
!
=== C1
!
!------+
Control-->! !
!-- === C2
! !
GND GND
helps.
It divides down the amplitude the JFET sees and C2 also tends to short
out the harmonics.
Cute--that's like tapping a varactor down on a tank circuit. With an op
amp it isn't hard to know the total range of gains you can have, so you
can use a vernier control and get better nonlinearity.

The down side is that the frequency of operation is more sensitve to
component values. BTW

That's because the phase shift depends on the control voltage. If you
used two resistors instead, you could get the same effect, minus the
couple of dB worth of harmonic rolloff, and no significant phase shift.

Yes, I knew the cause. The few dB were worth the frequency error.

 

[Glen Walpert]

Down at 2ppm, I would start worrying about resistor
linearity, also.

How about ppb instead of ppm:

Linear Technology Magazine, Feb 94 pp 26-28 - LTMag_V04N1_Feb94.pdf
at http://www.linear.com/lt_magazines.jsp

"Using Super Op Amps to Push Technological Frontiers: an Ultra-Pure
Oscillator

An Ultra-Low-Distortion, 10kHz Sine-Wave Source for Calibration of
16-Bit or Higher Analog-to-Digital Converters

.... The measurement of the harmonic distortion of this oscillator
defies all of our resources, but appears to be well into the
parts-per-billion range."

And only 11 Op amps .

ISTR another LT article on measuring the distortion of this oscillator
using a 120? dB notch filter also designed with the "Super Op Amp"
idea, but can't find it now.

 

[Vladimir Vassilevsky]

...


I still wonder if sending e.g. 100 Hz and 5 kHz at the same time to a
speaker wouldn't give a 100Hz FM modulation of the 5 kHz tone (doppler
effect caused by the cone movement).

It will certainly produce the FM however the Doppler effect is not the
biggest of the problems. The velocity of the cone is only at the order
of meters per second. Other nonlinearities are much higher.

But maybe the microphone has an inverse effect?

If the membrane moves with the same velocity as the cone. This is not
realistic.

But then, what about a pressure sensor as microphone (assuming no diafragm
movement)?

The DSP can be successfully applied to compensate all sorts of
distortion in a speaker. However it is much cheaper to make a reasonably
good speaker then to make a good speaker from a bad one by means of the DSP.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

 

[John Larkin]

"Vladimir Vassilevsky"


** No point in testing a multi-way speaker ( ie the vast majority ever
sold) for IM distortion.

The famous Quad ESL63 electrostatic speaker has THD in the order of 0.03% at
97dB SPL.

How can that be measured? Do you believe the number?


John

 

[Phil Allison]

"John Larkin"

"Phil Allison"

How can that be measured?

** Huh ?

No idea how THD can be measured ?

Do you believe the number?

** No faith required, it ain't religion.



........ Phil

 

[John Larkin]

"John Larkin"


** Huh ?

No idea how THD can be measured ?

No idea of how to measure THD in a sound wave to this sort of
resolution. Are microphones this good? If so, how do people measure
*them*?

** No faith required, it ain't religion.

A lot of audio stuff seems to be. It sure ain't science. Do you
believe that you can buy cables that are six-nines pure copper?


John

 

[Phil Allison]

"John Larkin"
"Phil Allison"

No idea of how to measure THD in a sound wave to this sort of
resolution. Are microphones this good?

** True condenser ( not electret ) mics are easily good enough.

If so, how do people measure *them*?

** Tested with SPLs way above 97dB, THD from the mic is still below 0.1 %.

It is easy to distinguish harmonics produced by the mic from those generated
by a source by simply moving the mic further away - if the THD percentage
falls, then the mic is responsible for some or all of it.

A lot of audio stuff seems to be.

** As usual, you confuse audiophool marketing drivel with audio
engineering.



........ Phil

 

[Joseph2k]

Joerg wrote:
Rich Grise wrote:
How does booze "[go] bad"?
No idea but...
Actually I think Rich was asking how the booze doesn't get
consumed immediately it's in your possession, instead being
kept for long enough for *anything* else to happen to it

Well, visitors mostly drank it. Before leaving Europe for good we did
throw one heck of a good-bye party for neighbors and friends but they
all wanted beer and wine.

Vodka, brown sugar, and seltzer tastes just like root beer. ;-)

Cheers!
Rich

No. You also need to ad some Angostura bitters.

 

[Phil Hobbs]

No idea of how to measure THD in a sound wave to this sort of
resolution. Are microphones this good? If so, how do people measure
*them*?

One method is to assume that the microphone is time-invariant at small
amplitudes, pick a repetitive signal, and do a lot of signal averaging.
By attenuating the sound with distance, the distortion polynomial
(i.e. the Nth harmonic goes as amplitude**N for small amplitudes) allows
you to distinguish between microphone and speaker distortion.

Not too much fun, but conceptually not too bad.

A lot of audio stuff seems to be. It sure ain't science. Do you
believe that you can buy cables that are six-nines pure copper?

I believe that some are _advertised_ to be six nines...of course that's
before they went through the drawing die, not to mention the oxidation
due to using thin strands...even assuming that people who would stoop to
fleecing audiophools have the elementary honesty to _try_ to deliver six
nines wire.

Cheers,

Phil Hobbs

 

[MooseFET]

One method is to assume that the microphone is time-invariant at small
amplitudes, pick a repetitive signal, and do a lot of signal averaging.
By attenuating the sound with distance, the distortion polynomial
(i.e. the Nth harmonic goes as amplitude**N for small amplitudes) allows
you to distinguish between microphone and speaker distortion.

Not too much fun, but conceptually not too bad.





I believe that some are _advertised_ to be six nines...of course that's
before they went through the drawing die, not to mention the oxidation
due to using thin strands...even assuming that people who would stoop to
fleecing audiophools have the elementary honesty to _try_ to deliver six
nines wire.

There really was at least one wire maker that really did deliver on
the 6 nines pure claim.

I think they used ceramic or copper in all of the wire pulling
hardware. Unlike the Phelpsdodge wire it didn't have to be pickled
before coating to remove the outer layer.

 

[John Larkin]

One method is to assume that the microphone is time-invariant at small
amplitudes, pick a repetitive signal, and do a lot of signal averaging.
By attenuating the sound with distance, the distortion polynomial
(i.e. the Nth harmonic goes as amplitude**N for small amplitudes) allows
you to distinguish between microphone and speaker distortion.

Not too much fun, but conceptually not too bad.

Yeah, that sounds labor-intensive.

I believe that some are _advertised_ to be six nines...of course that's
before they went through the drawing die, not to mention the oxidation
due to using thin strands...even assuming that people who would stoop to
fleecing audiophools have the elementary honesty to _try_ to deliver six
nines wire.

It's all Belden.


John

 

[Don Klipstein]

What sort of harmonic distortion does a speaker have? I've seen
numbers like 6% called "good."

This varies wildly. Not only does THD vary with signal amplitude, but
frequency complicates it. THD gets closer to a function of cone movement
and of voice coil current.

Generally, loudspeaker THD ends up mostly being second and third
harmonics, which are generally less audible as distortion products than
the higher ones. (Obviously, there is an exception for harmonics at
frequencies past the range of human hearing!)

Audibility of lower harmonics varies with frequency. I remember many
years ago reading somewhere a graph showing % THD (in second harmonic
only) required to hear the distortion as a function of frequency, and IIRC
that was about 20% or 40% at 20 or 25 Hz or so. That means a 20 Hz
sinewave with 20% THD entirely in second harmonic sounds like a pure 20 Hz
sine wave.
And a few percent of third harmonic (in lieu of a lot of second) is also
inaudible here.

As poor as most loudspeakers are at radiating acoutstic energy at
frwequencies in the lowest octave, they can easily produce very high THD
at such frequencies due to being better able to radiate the harmonics. So
it gets easy for a loudspeaker to make a 20 Hz sinewave sound distorted,
whether from a lot of harmonics 2-4 or a little bit of higher ones or
both.

Meanwhile, distortion produced by a "gentle" distortion device, perhaps
a "diode clipper" (especially one made with germanium diodes), is
generally not nearly as audible as that from hard clipping.

- Don Klipstein ([email protected])

 

[Don Klipstein]

I'd imagine that most capacitors add distortion, too.

I know some do.

The electric field applies a force that squishes the dielectric, and
that will change the capacitor's geometry to some extent. I have heard
some change shape enough to audibly vibrate when AC is applied. I exect
the capacitance to be not perfectly constant through a cycle of AC.

I expect that use of capacitors with a higher working voltage (thicker
dielectric) will mitigate that one. I expect this to be worse with more
compressible dielectrics - such as paper soaked with oil.

I also suspect that dielectric constant varies a bit with electric field
due to electron orbits in the dielectric's molecules being squished out of
shape. I expect that effect to be mitigated by use of a thicker
dielectric (higher working voltage).

Some ceramic capacitors have dielectric constant so high that I would
not trust the dielectric constant to be all that constant. This is
probably worse with ones whose dielectric constant varies too much with
temperature anyway for the capacitor to be useful to make a good
oscilltor.

Not that electrolytics are used to make a good oscillator, but their
capacitance often varies with voltage. As I understand it, discharging
slightly erodes the dielectric and charging restores it. That would make
the capacitance run high at lower voltages. Most of the few times I
checked into that many years ago, I found this to be true.

- Don Klipstein ([email protected])

 

[Tim Shoppa]

On Feb 5, 1:23 pm, John Larkin

An analog multiplier is the obvious choice.

Obvious, but not nearly as linear as a cheap photocell or light
bulb . And nowhere near as economical!

The fact that the gain of a light bult or photocell can vary by a
factor of a few depending on room temperature or recent light history
etc. is not all that important if you're searching for ultra-low
distortion.

Tim.

 

[MooseFET]

On Feb 5, 1:23 pm, John Larkin


Obvious, but not nearly as linear as a cheap photocell or light
bulb . And nowhere near as economical!

The fact that the gain of a light bult or photocell can vary by a
factor of a few depending on room temperature or recent light history
etc. is not all that important if you're searching for ultra-low
distortion.

A couple of others have been missed:

A resistively heated thermistor can be made to have a very long time
constant. It has the advantage over the light bulb that the
resistance value can be higher and the life expectancy very long.

If mechanically varied resistances weren't so microphonic, it would
have an advantage of making very long time constants easy to do.

 

[MooseFET]

On Feb 10, 2:54 pm, [email protected] (Don Klipstein) wrote:
[...]

I also suspect that dielectric constant varies a bit with electric field
due to electron orbits in the dielectric's molecules being squished out of
shape. I expect that effect to be mitigated by use of a thicker
dielectric (higher working voltage).

Some ceramic capacitors have dielectric constant so high that I would
not trust the dielectric constant to be all that constant. This is
probably worse with ones whose dielectric constant varies too much with
temperature anyway for the capacitor to be useful to make a good
oscilltor.

All of the high K ceramic materials are piezo electric and nonlinear.
The internal structure distorts under bias.

You can make a parametric amplifier using ceramic capacitors.
Unfortunately the capacitor makers always try to reduce the effect
that the parametric amplification needs so unless you get into some
sort of a relationship with the maker they may quit making ones that
work well.

 

[John Larkin]

On Feb 5, 1:23 pm, John Larkin

Obvious, but not nearly as linear as a cheap photocell or light
bulb . And nowhere near as economical!

The fact that the gain of a light bult or photocell can vary by a
factor of a few depending on room temperature or recent light history
etc. is not all that important if you're searching for ultra-low
distortion.

Tim.

Don't LDRs also have a nonlinear voltage effect? Light bulbs do at low
frequency. If you're hunting ppm distortion, may as well do it right.

Or use a 20-bit DAC with feedforward distortion tweaking.

John