Prev: Class A error
Next: Garage Door Opener - Electric Eye
From: John Larkin on 17 May 2010 14:52
On Mon, 17 May 2010 10:35:12 -0500, Vladimir Vassilevsky
<nospam(a)nowhere.com> wrote:

>
>I need to make a test source with THD ~ 0.0001% at the frequencies below
>1kHz. Leaving many details aside, the biggest problem seems to be an opamp.
>
>The typical THD vs frequency plot for a good opamp looks like this:
>
>ADTL082
>http://www.analog.com/static/imported-files/data_sheets/ADTL082_084.pdf
>
>It saturates at ~0.001% at the frequency ~2 kHz. Why is that? According
>to the GBV = 5MHz and G = 2e5, the corner should be at 25 Hz.
>
>Perhaps the limit is set by Audio Precision; it can't measure THD below
>~0.008% in direct way. But, even if 0.001% is the artifact of the
>measurement, there is still not enough of open loop gain to get to
>0.0001% at 1kHz. Would it be possible to make a cascaded amp to boost
>the open loop gain and improve distortion?
>
>Vladimir Vassilevsky
>DSP and Mixed Signal Design Consultant
>http://www.abvolt.com


I wonder how one could ever verify low distortion. I guess you could
trust Audio Precision or Agilent or somebody, at their guaranteed THD
floor. But is there any fundamental way to demonstrate the THD of a
source or analyzer, or to make a sinewave source that's known to have
below, say, 1 PPM distortion?

How do you build a lowpass or highpass or bandpass filter that's known
to have < 1 PPM distortion? Vacuum caps and toroidal inductors wound
on quartz forms? Would there be mechanical issues that create
distortion?

An RC highpass filter could be switched into the signal chain and its
effect on distortion noted. If it were a very good filter, and the
switch/connectors were to be trusted.

John

From: Vladimir Vassilevsky on 17 May 2010 15:17


John Larkin wrote:

> I wonder how one could ever verify low distortion. I guess you could
> trust Audio Precision or Agilent or somebody, at their guaranteed THD
> floor.

The direct measurement with AP provides for the THD floor slightly below
0.001%. I don't know if it is limited by generator or analyser, but this
is only -100 dB.

> But is there any fundamental way to demonstrate the THD of a
> source or analyzer, or to make a sinewave source that's known to have
> below, say, 1 PPM distortion?

> How do you build a lowpass or highpass or bandpass filter that's known
> to have < 1 PPM distortion? Vacuum caps and toroidal inductors wound
> on quartz forms? Would there be mechanical issues that create
> distortion?

You just have to operate at small voltages and currents. Noise can be
eliminated by sufficient averaging.

> An RC highpass filter could be switched into the signal chain and its
> effect on distortion noted. If it were a very good filter, and the
> switch/connectors were to be trusted.

Unfortunately, there is very little data on distortion in the passive
components. I did some experimentation with high voltage film capacitors
and good quality resistors; it is feasible to make an active filter with
0.0001% distortion provided the opamp is good enough.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com
From: Vladimir Vassilevsky on 17 May 2010 15:38


Mike wrote:

> Vladimir Vassilevsky <nospam(a)nowhere.com> wrote:
>
>>Would it be possible to make a cascaded amp to boost
>>the open loop gain and improve distortion?
>>
>
> Yes. See AN67, Linear Technology Magazine Circuit Collection, Volume III
> An Ultrapure Oscillator By Dale Eagar.
> Http://cds.linear.com/docs/Application%20Note/an67f.pdf
> Article starts on page 62
> Schematic is Figure 96 on page 65

That's what I need. Thank you, Mike.

Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com
From: John Larkin on 17 May 2010 16:07
On Mon, 17 May 2010 14:17:09 -0500, Vladimir Vassilevsky
<nospam(a)nowhere.com> wrote:

>
>
>John Larkin wrote:
>
>> I wonder how one could ever verify low distortion. I guess you could
>> trust Audio Precision or Agilent or somebody, at their guaranteed THD
>> floor.
>
>The direct measurement with AP provides for the THD floor slightly below
>0.001%. I don't know if it is limited by generator or analyser, but this
>is only -100 dB.
>
>> But is there any fundamental way to demonstrate the THD of a
>> source or analyzer, or to make a sinewave source that's known to have
>> below, say, 1 PPM distortion?
>
>> How do you build a lowpass or highpass or bandpass filter that's known
>> to have < 1 PPM distortion? Vacuum caps and toroidal inductors wound
>> on quartz forms? Would there be mechanical issues that create
>> distortion?
>
>You just have to operate at small voltages and currents. Noise can be
>eliminated by sufficient averaging.
>
>> An RC highpass filter could be switched into the signal chain and its
>> effect on distortion noted. If it were a very good filter, and the
>> switch/connectors were to be trusted.
>
>Unfortunately, there is very little data on distortion in the passive
>components. I did some experimentation with high voltage film capacitors
>and good quality resistors; it is feasible to make an active filter with
>0.0001% distortion provided the opamp is good enough.
>
>

I was thinking of using a rigid vacuum capacitor into a TIA as a
differentiator/high pass filter. If some sort of analyzer reports the
amplitudes of the fundamental and the harmonics, then you boost the
harmonic content with a known-good highpass, that sort of crosschecks
the analyzer. Or make a really good lowpass, ditto.

We've occasionally used a passive notch filter to remove the
fundamental of a sine wave before going into a spectrum analyzer. If
the reported harmonics change with/without the notch, one suspects the
analyzer.


From: Vladimir Vassilevsky on 17 May 2010 16:27



> On May 17, 11:35 am, Vladimir Vassilevsky <nos...(a)nowhere.com> wrote:
>
>>I need to make a test source with THD ~ 0.0001% at the frequencies below
>>1kHz. Leaving many details aside, the biggest problem seems to be an opamp.

George Herold wrote:

> How about a different opamp? The opa134 seems to have enough zero's
> in the THD+noise spec. (0.00008%)

That's what I planned on using initially. Unfortunately, the
availability is an issue.

David Eather wrote:

> This might be better:
> http://www.national.com/pf/LM/LME49990.html#Overview

I thought about bipolar opamps from National also. One drawback is high
power consumption, the other problem is high noise current. I am not
sure if it would be possible to make them work right with ~100k
impedances in the circuit.


VLV
First  |  Prev  |  Next  |  Last
Pages: 1 2 3 4 5
Prev: Class A error
Next: Garage Door Opener - Electric Eye