Prev: TMS320LF2407
Next: Wrong question may give WRONG answer -- was[Re: True RMS and averaging.]
From: Randy Yates on 19 Nov 2009 07:27
Randy Yates <yates(a)ieee.org> writes:
> [...]

A couple of corrections...

> That's because, when the non-quantization noise is much less than the
> quantization noise, you're only going to get a repeating sequence of
> four values, and that looks like a pure sine wave at Fs/4 without ANY
> (wideband) quantization noise.

Well, it looks like a pure periodic signal with no (wideband)
quantization noise.

> Additionally, I found, using Octave, that because of the phenomenom of
> the previous paragraph, the non-Fs/2 bins

....the non-Fs/4 bins...
--
Randy Yates % "Bird, on the wing,
Digital Signal Labs % goes floating by
mailto://yates(a)ieee.org % but there's a teardrop in his eye..."
http://www.digitalsignallabs.com % 'One Summer Dream', *Face The Music*, ELO
From: Eric Jacobsen on 19 Nov 2009 12:06
On 11/19/2009 5:22 AM, Randy Yates wrote:
> "guru_kaliraj"<guru_kaliraj(a)yahoo.com> writes:
>
>> Hi Eric,
>> This explains me clearly why I'm getting less SNR. Yes!. Both sample
>> rate& signal is not synchronized& hence as you mentioned my signal phase
>> is playing a role of affecting SNR. Thanks for the reply.
>> Probably, writing some algorithm so that my sample start& end is
>> controlled (I'm planning to select samples - to get same value of digital
>> code at start& end -- and this will make sure that phase drift is
>> adjusted) will help me to increase the SNR.
>>
>> Thanks again!
>
> Wow! I'm glad the two of you know what you're talking about. I really
> don't have a clue.
>
> In fact, I ran a Matlab simulation of this and found just the opposite,
> i.e., that when Fs = 4*f, the measured SNR is way, way bigger than the
> actual.
>
> That's because, when the non-quantization noise is much less than the
> quantization noise, you're only going to get a repeating sequence of
> four values, and that looks like a pure sine wave at Fs/4 without ANY
> (wideband) quantization noise. The quantization noise translates into
> phase and/or magnitude error of the sine wave.

Isn't phase and/or magnitude error noise?

> Additionally, I found, using Octave, that because of the phenomenom of
> the previous paragraph, the non-Fs/2 bins of the FFT are identically
> zero, and therefore their magnitude is zero, and therefore the log10()
> of their magnitude is one of those special double exception values
> -Inf, making the plot unusable.
>
> --Randy
>
>
>> Regards,
>> Guru
>>
>>> On 11/18/2009 4:28 AM, guru_kaliraj wrote:
>>>> Hi,
>>>> In one of my experiment, I need to measure SNR of an ADC which has
>> fixed
>>>> sample rate of 2.4MHz& 14 bit. My input sinusoidal frequency is also
>> fixed
>>>> (decided by previous stages) and it is 600KHz. I understand that I'm
>> just
>>>> taking 4 sample in a cycle (ie. My sample rate is just 4 times my
>> signal
>>>> rate).
>>>> Will this affect SNR of the ADC? When I calculate SNR for the
>> same --
>>>> I got around 36dB only..whereas my theory says it need to be 6N -1 =
>> 83dB.
>>>> I'm losing a lot.
>>>> I'm quite sure about the algoirthm i'm using while computing
>> SNR (
>>>> I'm using FFT method, ignoring harmonic power, summing noise power in
>> the
>>>> bandwidth). I verified it with other ADCs with higher sample rate& it
>> got
>>>> related to 6N-1. But not in above case.
>>>> I'm just wondering whether this sample rate (4 times the
>> signal
>>>> freq) will affect SNR or not?.
>>>>
>>>> Regards,
>>>> Guru
>>> Just IMHO: If the test signal is synchronous to the sampling, then you
>>> likely do have a problem in that you may be limited in how much
>>> resolution you can discern. If not, and the test signal phase drifts
>>> with respect to the sampling interval, then you just need a long enough
>>> FFT to make use of the quantization steps revealed by the phase drift.
>>>
>>> I hope that makes sense. I was trying to think of a theoretical
>>> explanation for this, but I hope the intuitive approach conveys the
>> idea.
>>> --
>>> Eric Jacobsen
>>> Minister of Algorithms
>>> Abineau Communications
>>> http://www.abineau.com
>>>
>


--
Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.abineau.com
From: Randy Yates on 19 Nov 2009 13:02
Eric Jacobsen <eric.jacobsen(a)ieee.org> writes:

> On 11/19/2009 5:22 AM, Randy Yates wrote:
>> "guru_kaliraj"<guru_kaliraj(a)yahoo.com> writes:
>>
>>> Hi Eric,
>>> This explains me clearly why I'm getting less SNR. Yes!. Both sample
>>> rate& signal is not synchronized& hence as you mentioned my signal phase
>>> is playing a role of affecting SNR. Thanks for the reply.
>>> Probably, writing some algorithm so that my sample start& end is
>>> controlled (I'm planning to select samples - to get same value of digital
>>> code at start& end -- and this will make sure that phase drift is
>>> adjusted) will help me to increase the SNR.
>>>
>>> Thanks again!
>>
>> Wow! I'm glad the two of you know what you're talking about. I really
>> don't have a clue.
>>
>> In fact, I ran a Matlab simulation of this and found just the opposite,
>> i.e., that when Fs = 4*f, the measured SNR is way, way bigger than the
>> actual.
>>
>> That's because, when the non-quantization noise is much less than the
>> quantization noise, you're only going to get a repeating sequence of
>> four values, and that looks like a pure sine wave at Fs/4 without ANY
>> (wideband) quantization noise. The quantization noise translates into
>> phase and/or magnitude error of the sine wave.
>
> Isn't phase and/or magnitude error noise?

Not when you're expecting to be able to measure noise in the non-Fs/4
bins of the FFT.
--
Randy Yates % "The dreamer, the unwoken fool -
Digital Signal Labs % in dreams, no pain will kiss the brow..."
mailto://yates(a)ieee.org %
http://www.digitalsignallabs.com % 'Eldorado Overture', *Eldorado*, ELO
From: Mark on 19 Nov 2009 13:05
On Nov 19, 12:06 pm, Eric Jacobsen <eric.jacob...(a)ieee.org> wrote:
> On 11/19/2009 5:22 AM, Randy Yates wrote:
>
>
>
>
>
> > "guru_kaliraj"<guru_kali...(a)yahoo.com>  writes:
>
> >> Hi Eric,
> >>        This explains me clearly why I'm getting less SNR. Yes!.. Both sample
> >> rate&  signal is not synchronized&  hence as you mentioned my signal phase
> >> is playing a role of affecting SNR. Thanks for the reply.
> >>          Probably, writing some algorithm so that my sample start&  end is
> >> controlled (I'm planning to select samples - to get same value of digital
> >> code at start&  end -- and this will make sure that phase drift is
> >> adjusted) will help me to increase the SNR.
>
> >> Thanks again!
>
> > Wow! I'm glad the two of you know what you're talking about. I really
> > don't have a clue.
>
> > In fact, I ran a Matlab simulation of this and found just the opposite,
> > i.e., that when Fs = 4*f, the measured SNR is way, way bigger than the
> > actual.
>
> > That's because, when the non-quantization noise is much less than the
> > quantization noise, you're only going to get a repeating sequence of
> > four values, and that looks like a pure sine wave at Fs/4 without ANY
> > (wideband) quantization noise.  The quantization noise translates into
> > phase and/or magnitude error of the sine wave.
>
> Isn't phase and/or magnitude error noise?
>
>
>
>
>
> > Additionally, I found, using Octave, that because of the phenomenom of
> > the previous paragraph, the non-Fs/2 bins of the FFT are identically
> > zero, and therefore their magnitude is zero, and therefore the log10()
> > of their magnitude is one of those special double exception values
> > -Inf, making the plot unusable.
>
> > --Randy
>
> >> Regards,
> >> Guru
>
> >>> On 11/18/2009 4:28 AM, guru_kaliraj wrote:
> >>>> Hi,
> >>>>     In one of my experiment, I need to measure SNR of an ADC which has
> >> fixed
> >>>> sample rate of 2.4MHz&   14 bit. My input sinusoidal frequency is also
> >> fixed
> >>>> (decided by previous stages) and it is 600KHz. I understand that I'm
> >> just
> >>>> taking 4 sample in a cycle (ie. My sample rate is just 4 times my
> >> signal
> >>>> rate).
> >>>>        Will this affect SNR of the ADC? When I calculate SNR for the
> >> same --
> >>>> I got around 36dB only..whereas my theory says it need to be 6N -1 =
> >> 83dB.
> >>>> I'm losing a lot.
> >>>>          I'm quite sure about the algoirthm i'm using while computing
> >> SNR (
> >>>> I'm using FFT method, ignoring harmonic power, summing noise power in
> >> the
> >>>> bandwidth). I verified it with other ADCs with higher sample rate&   it
> >> got
> >>>> related to 6N-1. But not in above case.
> >>>>          I'm just wondering whether this sample rate (4 times the
> >> signal
> >>>> freq) will affect SNR or not?.
>
> >>>> Regards,
> >>>> Guru
> >>> Just IMHO: If the test signal is synchronous to the sampling, then you
> >>> likely do have a problem in that you may be limited in how much
> >>> resolution you can discern.   If not, and the test signal phase drifts
> >>> with respect to the sampling interval, then you just need a long enough
> >>> FFT to make use of the quantization steps revealed by the phase drift..
>
> >>> I hope that makes sense.  I was trying to think of a theoretical
> >>> explanation for this, but I hope the intuitive approach conveys the
> >> idea.
> >>> --
> >>> Eric Jacobsen
> >>> Minister of Algorithms
> >>> Abineau Communications
> >>>http://www.abineau.com
>
> --
> Eric Jacobsen
> Minister of Algorithms
> Abineau Communicationshttp://www.abineau.com- Hide quoted text -
>
> - Show quoted text -- Hide quoted text -
>
> - Show quoted text -

audio folks talk about this often...

we tend to call RANDOM phase nad magnitude erros noise...

and phase and magnitude erros that are corrolated to the signal we
call distortion...

It is unfortunate that "quantizing noise" is a misnomer and would be
more descriptive if it was called quantizing distortion.

Mark
From: glen herrmannsfeldt on 19 Nov 2009 15:41
Mark <makolber(a)yahoo.com> wrote:
(really big snip)

> audio folks talk about this often...

> we tend to call RANDOM phase nad magnitude erros noise...

> and phase and magnitude erros that are corrolated to the signal we
> call distortion...

Hmm. My first thought about distortion is that it should be
proportional to the signal, or some low power of the signal,
such that it goes to zero as the signal goes to zero.

> It is unfortunate that "quantizing noise" is a misnomer and
> would be more descriptive if it was called quantizing distortion.

Since "quantization noise" stays approximately constant as
the signal gets small, and so becomes proportionately larger,
calling it "noise" seems fine to me. It does go away when the
signal actually gets to zero, but for very small but still audible
signals it is important.

-- glen
First  |  Prev  |  Next  |  Last
Pages: 1 2 3 4 5
Prev: TMS320LF2407
Next: Wrong question may give WRONG answer -- was[Re: True RMS and averaging.]