Is there anyway to get an amplitude of an ac signal averaged over time
w/o some kind of rectification?

This doesn't include the trivial, i.e., dc offsets or partial cycles.

The rectification could be analog, digital or software.


Bret Cahill

Bret Cahill wrote:

> Is there anyway to get an amplitude of an ac signal averaged over time
> w/o some kind of rectification?

Sure. Connect the AC to a heater and measure the temperature.

> This doesn't include the trivial, i.e., dc offsets or partial cycles.
> The rectification could be analog, digital or software.

What are you really trying to do?


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

On Nov 5, 4:00=A0pm, Bret Cahill <Bret_E_Cah...@yahoo.com> wrote:
> Is there anyway to get an amplitude of an ac signal averaged over time
> w/o some kind of rectification?
>
> This doesn't include the trivial, i.e., dc offsets or partial cycles.
>
> The rectification could be analog, digital or software.
>
> Bret Cahill

See "Understanding DSP" by R Lyons  p 366 of the second edition..

Use a Hilbert to get Real  and imaginary components of the signal
then, the envelope is found as the SQRT (R^2+i^2).

No rectification and no low pass filtering is needed.

I found this fascinating becasue as a kid playing with AM radios, i
began to understand the limitations of a diode detector and this is
the elegant solution...

Mark

In comp.dsp Vladimir Vassilevsky <nospam@nowhere.com> wrote:
 
> Bret Cahill wrote:
 
>> Is there anyway to get an amplitude of an ac signal averaged over time
>> w/o some kind of rectification?
 
> Sure. Connect the AC to a heater and measure the temperature.
 
>> This doesn't include the trivial, i.e., dc offsets or partial cycles.
>> The rectification could be analog, digital or software.

You need some non-linear operation, which may or may not be
considered rectification.  I believe the heater/temperature
measurement has been used for RF power measurements for many years.

-- glen

glen herrmannsfeldt wrote:

> In comp.dsp Vladimir Vassilevsky <nospam@nowhere.com> wrote:
>  
> 
>>Bret Cahill wrote:
> 
>  
> 
>>>Is there anyway to get an amplitude of an ac signal averaged over time
>>>w/o some kind of rectification?
> 
>  
> 
>>Sure. Connect the AC to a heater and measure the temperature.
> 
>  
> 
>>>This doesn't include the trivial, i.e., dc offsets or partial cycles.
>>>The rectification could be analog, digital or software.
> 
> 
> You need some non-linear operation,  which may or may not be
> considered rectification. 

No nonlinear operations. Just multiply the AC by a synchronous reference.

> I believe the heater/temperature
> measurement has been used for RF power measurements for many years.

That doesn't do any rectification for sure.

VLV

> > In comp.dsp Vladimir Vassilevsky <nos...@nowhere.com> wrote:
>
> >>Bret Cahill wrote:
>
> >>>Is there anyway to get an amplitude of an ac signal averaged over time
> >>>w/o some kind of rectification?
>
> >>Sure. Connect the AC to a heater and measure the temperature.
>
> >>>This doesn't include the trivial, i.e., dc offsets or partial cycles.
> >>>The rectification could be analog, digital or software.
>
> > You need some non-linear operation, =A0which may or may not be
> > considered rectification.
>
> No nonlinear operations. Just multiply the AC by a synchronous reference.
>
> > I believe the heater/temperature
> > measurement has been used for RF power measurements for many years.
>
> That doesn't do any rectification for sure.

Positive heat with a positive signal and positive heat with a negative
signal can only be considered a form of rectification.


Bret Cahill

> > Is there anyway to get an amplitude of an ac signal averaged over time
> > w/o some kind of rectification?
>
> > This doesn't include the trivial, i.e., dc offsets or partial cycles.
>
> > The rectification could be analog, digital or software.
>
> > Bret Cahill
>
> See "Understanding DSP" by R Lyons =A0p 366 of the second edition..
>
> Use a Hilbert to get Real =A0and imaginary components of the signal
> then, the envelope is found as the SQRT (R^2+i^2).

Sqrt of sum of squares =3D absolute value =3D rectification.

> No rectification and no low pass filtering is needed.
>
> I found this fascinating becasue as a kid playing with AM radios, i
> began to understand the limitations of a diode detector and this is
> the elegant solution...

Maybe a difference measurement of peak to peak of a known waveform
wouldn't be considered rectification.


Bret Cahill

In comp.dsp Vladimir Vassilevsky <nospam@nowhere.com> wrote:
(snip regarding rectiification, and I wrote)

>> You need some non-linear operation,  which may or may not be
>> considered rectification. 
 
> No nonlinear operations. Just multiply the AC by a synchronous reference.

Multiplying by other than a constant is a non-linear operation.
 
>> I believe the heater/temperature
>> measurement has been used for RF power measurements for many years.
 
> That doesn't do any rectification for sure.

Instantaneous power is proportional to the square of the voltage
or current, not including resistance changes due to heating,
and so is always positive.  Is that rectification?  It seems
to me that any conversion of an AC signal with the DC component
removed to one that is always positive is a form of rectification,
but others might disagree.

-- glen

Bret Cahill wrote:
> Is there anyway to get an amplitude of an ac signal averaged over time
> w/o some kind of rectification?
> 
> This doesn't include the trivial, i.e., dc offsets or partial cycles.
> 
> The rectification could be analog, digital or software.
> 
> 
> Bret Cahill
> 
You need a precision rectifier..
   couple of op-amps and maybe 1 diode ..


http://sound.westhost.com/appnotes/an001.htm

jamie..

> > Is there anyway to get an amplitude of an ac signal averaged over time
> > w/o some kind of rectification?
>
> > This doesn't include the trivial, i.e., dc offsets or partial cycles.
>
> > The rectification could be analog, digital or software.
>
> > Bret Cahill
>
> You need a precision rectifier..
> =A0 =A0couple of op-amps and maybe 1 diode ..
>
> http://sound.westhost.com/appnotes/an001.htm


Poorly written OP.  The goal is getting the amplitude _without_
rectification.  Rectification" is to be very broadly construed and
includes anything that results in the final "signal" being positive at
all times.

Maybe peak to peak wil work if you know the wave form.


Bret Cahill

Bret Cahill wrote:
>>>In comp.dsp Vladimir Vassilevsky <nos...@nowhere.com> wrote:
>>
>>>>Bret Cahill wrote:
>>
>>>>>Is there anyway to get an amplitude of an ac signal averaged over time
>>>>>w/o some kind of rectification?
>>
>>>>Sure. Connect the AC to a heater and measure the temperature.
>>
>>>>>This doesn't include the trivial, i.e., dc offsets or partial cycles.
>>>>>The rectification could be analog, digital or software.
>>
>>>You need some non-linear operation,  which may or may not be
>>>considered rectification.
>>
>>No nonlinear operations. Just multiply the AC by a synchronous reference.
>>
>>
>>>I believe the heater/temperature
>>>measurement has been used for RF power measurements for many years.
>>
>>That doesn't do any rectification for sure.
> 
> 
> Positive heat with a positive signal and positive heat with a negative
> signal can only be considered a form of rectification.

1. Take 3 consequent samples x[0],x[1],x[2]
2. Solve the system of equations x[i] = A sin(Wt + Fi)
3. Now you know A, W, and Fi

> >> You need some non-linear operation, =A0which may or may not be
> >> considered rectification.

> > No nonlinear operations. Just multiply the AC by a synchronous referenc=
e.

> Multiplying by other than a constant is a non-linear operation.

Not sure what is meant by linear but that's clearly rectification.

> >> I believe the heater/temperature
> >> measurement has been used for RF power measurements for many years.

> > That doesn't do any rectification for sure.

> Instantaneous power is proportional to the square of the voltage
> or current, not including resistance changes due to heating,
> and so is always positive. =A0Is that rectification? =A0It seems
> to me that any conversion of an AC signal with the DC component
> removed to one that is always positive is a form of rectification,
> but others might disagree.

Temperature may not be as convenient as radiation or voltage to
transmit information at high frequency but there is no reason not to
treat it as a signal.


Bret Cahill

> >>>>>Is there anyway to get an amplitude of an ac signal averaged over ti=
me
> >>>>>w/o some kind of rectification?

> >>>>Sure. Connect the AC to a heater and measure the temperature.

> >>>>>This doesn't include the trivial, i.e., dc offsets or partial cycles=
..
> >>>>>The rectification could be analog, digital or software.

> >>>You need some non-linear operation, =A0which may or may not be
> >>>considered rectification.

> >>No nonlinear operations. Just multiply the AC by a synchronous referenc=
e.

> >>>I believe the heater/temperature
> >>>measurement has been used for RF power measurements for many years.

> >>That doesn't do any rectification for sure.

> > Positive heat with a positive signal and positive heat with a negative
> > signal can only be considered a form of rectification.

> 1. Take 3 consequent samples x[0],x[1],x[2]

Over several cycles?

> 2. Solve the system of equations x[i] =3D A sin(Wt + Fi)

Supposing the wave form is unknown?

> 3. Now you know A, W, and Fi-

If the signal fluctuates aperiodically, it seems you're pretty much
stuck with some kind rectification.


Bret Cahill

On 11/5/2010 1:00 PM, Bret Cahill wrote:
> Is there anyway to get an amplitude of an ac signal averaged over time
> w/o some kind of rectification?
>
> This doesn't include the trivial, i.e., dc offsets or partial cycles.
>
> The rectification could be analog, digital or software.
>
>
> Bret Cahill
>

Bret,

I suggest you think about the stated objective and whether it admits the 
constraints you've stated or assumed.

I think the answer is buried in the question:

You say you want the *amplitude of an ac signal averaged over time*.

The "averaged over time" is a linear operation - so we can disregard 
that part.
Determining amplitude *is* or implies a nonlinear operation.  That is 
all - end of story.

The example of the heater is a good one.  It does the job.  One could 
call it a nonlinear element because it's insensitive to the direction of 
flow of current.  But, I wouldn't call it a rectifier in the normal 
sense.  It's a transducer; converting current to heat.  And, one does 
lose the polarity/direction of the forcing function in the transduction. 
  So, it likely has to be modeled as a nonlinear element.

Your stated objective, if I understand its definition, would likewise be 
insensitive to the polarity of the signal wouldn't it?

Either that, or with certain assumptions, one might ignore all signal 
components of the "wrong" polarity / direction - and that's pretty much 
a half-wave rectifier.

Fred

On 11/5/2010 2:51 PM, glen herrmannsfeldt wrote:
> In comp.dsp Vladimir Vassilevsky<nospam@nowhere.com>  wrote:
> (snip regarding rectiification, and I wrote)
>
>>> You need some non-linear operation,  which may or may not be
>>> considered rectification.

Agreed.

>
>> No nonlinear operations. Just multiply the AC by a synchronous reference.
>
> Multiplying by other than a constant is a non-linear operation.

Multiplying by other than a constant or a stable periodic waveform is a 
non-linear operation.  Multiplying by a constant or a sinusoid or a sum 
of sinusoids is a linear operation.... believe it or not.

Fred

On Nov 5, 4:00=A0pm, Bret Cahill <Bret_E_Cah...@yahoo.com> wrote:
> Is there anyway to get an amplitude of an ac signal averaged over time
> w/o some kind of rectification?
>
> This doesn't include the trivial, i.e., dc offsets or partial cycles.
>
> The rectification could be analog, digital or software.
>
> Bret Cahill

In Matlab:

amp =3D abs(hilbert(x));

Then filter amp.

John

On Fri, 05 Nov 2010 15:10:00 -0500, Vladimir Vassilevsky
<nospam@nowhere.com> wrote:

>
>
>Bret Cahill wrote:
>
>> Is there anyway to get an amplitude of an ac signal averaged over time
>> w/o some kind of rectification?
>
>Sure. Connect the AC to a heater and measure the temperature.
>
>> This doesn't include the trivial, i.e., dc offsets or partial cycles.
>> The rectification could be analog, digital or software.
>
>What are you really trying to do?

Hi Vlad,
  I agree with you.  What is Brett actually trying 
to do?  What is the meaning of the word "get."
Does it mean "acquire", or does it mean "measure?"

And what does "averaged over time" mean?
To me "averaged over time" means add N time-domain 
numbers, and then divide the sum by N.  The result 
will be a single number.  

The question is too vague for me to understand.

[-Rick-]


 

On Fri, 5 Nov 2010 13:28:23 -0700 (PDT), Mark <makolber@yahoo.com>
wrote:

>On Nov 5, 4:00�pm, Bret Cahill <Bret_E_Cah...@yahoo.com> wrote:
>> Is there anyway to get an amplitude of an ac signal averaged over time
>> w/o some kind of rectification?
>>
>> This doesn't include the trivial, i.e., dc offsets or partial cycles.
>>
>> The rectification could be analog, digital or software.
>>
>> Bret Cahill
>
>See "Understanding DSP" by R Lyons  p 366 of the second edition..
>
>Use a Hilbert to get Real  and imaginary components of the signal
>then, the envelope is found as the SQRT (R^2+i^2).
>
>No rectification and no low pass filtering is needed.
>
>I found this fascinating becasue as a kid playing with AM radios, i
>began to understand the limitations of a diode detector and this is
>the elegant solution...
>
>Mark

Hello Mark,
   
If you have an American version of the 2nd Edition of 
my "Understanding DSP" book, I can send you the appropriate 
errata if you can E-mail me the "Printing Number" of 
your copy of the book.

You can determine the "Printing Number" of the American 
version (ISBN# 0-13-108989-7) of the 2nd Edition of my 
book by looking at the page just before the "Dedication" 
page.  

On that page (before the Dedication) you'll see all 
sorts of publisher-related information, including the 
ISBN number.  Down toward the bottom of the page you 
should see lines printed something like:

       Printed in the United States of America
       First Printing

indicating the "First Printing" of the book.  However, for 
later printings the second line above may have the words 
like: "Second Printing" or "Seventh Printing".

Regards,
[-Rick-]

In comp.dsp Bret Cahill <BretCahill@peoplepc.com> wrote:
(snip, I wrote)

>> Instantaneous power is proportional to the square of the voltage
>> or current, not including resistance changes due to heating,
>> and so is always positive. �Is that rectification? �It seems
>> to me that any conversion of an AC signal with the DC component
>> removed to one that is always positive is a form of rectification,
>> but others might disagree.
 
> Temperature may not be as convenient as radiation or voltage to
> transmit information at high frequency but there is no reason not to
> treat it as a signal.

There used to be a term 'detector', and 'detection' for extracting
the modulating signal from (at least an AM modulated) signal.

You can put the signal through almost any non-linear operation,
then low-pass filter the result, to extract in some form the
modulation.  In the context of RF demodulation, I am not sure
of the exact meaning of 'rectification.'   With some filtering
and such, you can extract either the half-wave or full-wave
rectified version of a modulated carrier from the input, which
would seem to qualify as rectification.  If the non-linear
operation has even powers other than zero, that is pretty close
to rectification.  If only odd powers, then maybe not.

-- glen

In comp.dsp Fred Marshall <fmarshall_xremove_the_xs@xacm.org> wrote:
(snip)

> Multiplying by other than a constant or a stable periodic waveform is a 
> non-linear operation.  Multiplying by a constant or a sinusoid or a sum 
> of sinusoids is a linear operation.... believe it or not.

Hmmm.  As I understand it (longer ago than I remember) they used
to make rectifiers out of synchronous motor driven commutators.

That is, multiply by a square wave (sum of sines) of the appropriate
frequency.  The result doesn't seem like a linear function of
the input anymore.

-- glen

In comp.dsp Bret Cahill <BretCahill@peoplepc.com> wrote:
(snip)

> Poorly written OP.  The goal is getting the amplitude _without_
> rectification.  Rectification" is to be very broadly construed and
> includes anything that results in the final "signal" being positive at
> all times.
 
> Maybe peak to peak wil work if you know the wave form.

I still have my Heathkit VTVM, christmas present from when
I was in high school.  The manual includes an explanation of
the form of rectifier used, and why the result is peak-to-peak.
Other than that, it seems unusual to rectify as peak-to-peak.

-- glen

In comp.dsp Fred Marshall <fmarshall_xremove_the_xs@xacm.org> wrote:
(snip on heater vs. rectifier)

> The example of the heater is a good one.  It does the job.  One could 
> call it a nonlinear element because it's insensitive to the direction of 
> flow of current.  But, I wouldn't call it a rectifier in the normal 
> sense.  It's a transducer; converting current to heat.  And, one does 
> lose the polarity/direction of the forcing function in the transduction. 
>  So, it likely has to be modeled as a nonlinear element.

It might be that I am remembering vacuum gauges, but I believe
it is done by spot welding two wires together where they cross.
The result, then, is a combination of heater (two of the wires)
and thermocouple (the other two), at the point where they cross.

Power goes as voltage or current squared, heating the junction,
and the thermoelectric voltage, at least in small signal terms,
is proportional to the temperature change.  It also has a built
in low-pass filter, due to the thermal mass.
 
> Your stated objective, if I understand its definition, would likewise be 
> insensitive to the polarity of the signal wouldn't it?
 
> Either that, or with certain assumptions, one might ignore all signal 
> components of the "wrong" polarity / direction - and that's pretty much 
> a half-wave rectifier.

-- glen

glen herrmannsfeldt wrote:

> You can put the signal through almost any non-linear operation,


You can put the signal into the heater and measure the temperature. 
Where is nonlinearity?

You can put the signal into a into a solenoid and measure the force 
attracting a piece of ferrous material. Or, for that matter, the force 
between two solenoids with the same current running in them.

There is a zillion of ways to measure AC without any nonlinearity or 
rectification involved.

On Fri, 5 Nov 2010 13:00:54 -0700 (PDT), Bret Cahill
<Bret_E_Cahill@yahoo.com> wrote:

>Is there anyway to get an amplitude of an ac signal averaged over time
>w/o some kind of rectification?
>
>This doesn't include the trivial, i.e., dc offsets or partial cycles.
>
>The rectification could be analog, digital or software.
>
>
>Bret Cahill

Sample it with an ADC and do the math. You can get average, RMS,
rectified mean, anything you want.

John

In comp.dsp Vladimir Vassilevsky <nospam@nowhere.com> wrote:
 
> glen herrmannsfeldt wrote:
 
>> You can put the signal through almost any non-linear operation,

> You can put the signal into the heater and measure the temperature. 
> Where is nonlinearity?

The temperature is a non-linear function of the voltage or current.
At best it is quadratic, which is pretty non-linear.  I know
it doesn't go down when the current reverses.
 
> You can put the signal into a into a solenoid and measure the force 
> attracting a piece of ferrous material. Or, for that matter, the force 
> between two solenoids with the same current running in them.

Both abs(x) and x**2 are non-linear.
 
> There is a zillion of ways to measure AC without any 
> nonlinearity or rectification involved.

I am a little unsure of the exact meaning of rectification.

The google define:rectification includes "converting AC to DC."
That seems fine in the case of power supplies, but doesn't work
so well in the signal-processing sense.  The output of either
a half or full wave rectifier, unfiltered, has a large AC component.
The important point being that the average (low pass filter)
output is non-zero.  

So, it seems to me that in the appropriate sense, any operation
which converts the modulated carrier with no DC component to
one with a DC component, a process that can't happen with
linear operations, counts as rectification.

Now, some non-linear functions, though with odd powers, don't
result in a DC component in the result.  In that case, it is harder
to call them rectifiers.   If you add a DC offset to the input,
though, then you get rectification.  (1+x)**3 has even powers.

On the other hand, I do agree that curve fitting to points
on the V(t) curve doesn't seem like rectification, and does allow
one to compute the amplitude, though not easily the envelope for 
an AM modulated carrier.


-- glen

>
> On the other hand, I do agree that curve fitting to points
> on the V(t) curve doesn't seem like rectification, and does allow
> one to compute the amplitude, though not easily the envelope for
> an AM modulated carrier.
>
> -- glen

I think the key point is not about rectification or linear vs non
linear.

I think the key point is that the usual "rectification" methods
require some kind of explicit or implicit (in the case of a heaters)
low pass filtering to remove the carrier component and keep the
modulation component.  The Hilbert method describe by R. Lyons
requires no filtering and returns an exact result regardless of the
relationship between the carrier frequency and the modulation.

That is the beauty of it.

That is the problem I saw many years ago, if you applied a lot of
filtering to the diode detector to completely remove the 455 kHz IF
signal, you would loose some of the high frequency audio components.
Of course I was using simple RC filters, I know better now of
course.

But, to me, that is the ___fundamental__ advantage of the Hilbert
method over the "rectifier" methods.

Hello Rick, I already have the received the errata from you.   thank
you.


Mark

Vladimir Vassilevsky wrote:
> 
> Bret Cahill wrote:
> 
> > Is there anyway to get an amplitude of an ac signal averaged over time
> > w/o some kind of rectification?
> 
> Sure. Connect the AC to a heater and measure the temperature.
> 
> > This doesn't include the trivial, i.e., dc offsets or partial cycles.
> > The rectification could be analog, digital or software.
> 
> What are you really trying to do?


   Troll the newsgroup, as always.


-- 
Politicians should only get paid if the budget is balanced, and there is
enough left over to pay them.

In comp.dsp Mark <makolber@yahoo.com> wrote:
(snip, I wrote)

>> On the other hand, I do agree that curve fitting to points
>> on the V(t) curve doesn't seem like rectification, and does allow
>> one to compute the amplitude, though not easily the envelope for
>> an AM modulated carrier.
 
> I think the key point is not about rectification or linear vs non
> linear.
 
> I think the key point is that the usual "rectification" methods
> require some kind of explicit or implicit (in the case of a heaters)
> low pass filtering to remove the carrier component and keep the
> modulation component.  The Hilbert method describe by R. Lyons
> requires no filtering and returns an exact result regardless of the
> relationship between the carrier frequency and the modulation.
 
> That is the beauty of it.

Hmm, interesting and a different question.  It does remind me
of asking here about synchronous demodulation of AM signals,
such that you get the right result when the modulation index
is greater than one.  I will guess that the Hilbert method
can also do that.
 
> That is the problem I saw many years ago, if you applied a lot of
> filtering to the diode detector to completely remove the 455 kHz IF
> signal, you would loose some of the high frequency audio components.
> Of course I was using simple RC filters, I know better now of
> course.

> But, to me, that is the ___fundamental__ advantage of the Hilbert
> method over the "rectifier" methods.

-- glen

> > You can put the signal through almost any non-linear operation,

> You can put the signal into the heater and measure the temperature.
> Where is nonlinearity?

It would be hard to claim _that_ was "non linear" in even the least
sophisticated sense of the term.

That isn't the issue, however.  The issue is rectification.

If some operation results in the same + sign of the magnitude of some
+/- signal then that operation could be broadly construed as
"rectification."

> You can put the signal into a into a solenoid and measure the force
> attracting a piece of ferrous material.

Same as the heater.  The operation outputs a + magnitude for a +/-
signal.

> Or, for that matter, the force
> between two solenoids with the same current running in them.

> There is a zillion of ways to measure AC without any nonlinearity or
> rectification involved.

Maybe there are several if the wave form is known.


Bret Cahill

On Sat, 6 Nov 2010 00:29:42 +0000 (UTC), glen herrmannsfeldt
<gah@ugcs.caltech.edu> wrote:

>In comp.dsp Bret Cahill <BretCahill@peoplepc.com> wrote:
>(snip)
>
>> Poorly written OP.  The goal is getting the amplitude _without_
>> rectification.  Rectification" is to be very broadly construed and
>> includes anything that results in the final "signal" being positive at
>> all times.
> 
>> Maybe peak to peak wil work if you know the wave form.
>
>I still have my Heathkit VTVM, christmas present from when
>I was in high school.  The manual includes an explanation of
>the form of rectifier used, and why the result is peak-to-peak.
>Other than that, it seems unusual to rectify as peak-to-peak.
>
>-- glen

Hi glen,
   Good God!!  When you mention a Heathkit VTVM
you strike a nerve with me.  I had one of those, 
with its cigar-sized probe and alligator 
ground clip.

http://www.heathkit-museum.com/test/hvmv-7a.shtml

When I was in high school (yes I played on the 
baseball team and got into trouble like any other 
knuckleheaded high school student), I became 
interested in electronics.

I mostly used my VTVM to test the filaments of 
vacuum tubes from the old radios and televisions 
that my neighbors threw away and gave to me.

Back then you could buy all sorts of electronic parts 
from Radio Shack  At that time Radio Shack was about 
radios and electronics hobbyists.  Now Radio Shack 
is geared toward cell phones and digital cameras.
It's too bad, really too bad.

It's a shame but American teenage boys are now far 
more interested in playing video games than 
experimenting with electronics.

glen, are you old enough to remember the chintzy 
little crystal radios that had a simple tuning coil, 
capacitor (I think), diode, earphone, and NO BATTERY?
With those little crystal radios you could "pick 
up" local AM radio while lying in your bed at night.

See Ya',
[-Rick-]
   

Bret Cahill wrote:

> Is there anyway to get an amplitude of an ac signal averaged over time
> w/o some kind of rectification?
> 
Yeah - the average of an AC signal is zero.

Hope This Helps!
Rich

glen herrmannsfeldt Inscribed thus:

> In comp.dsp Vladimir Vassilevsky <nospam@nowhere.com> wrote:
>  
>> Bret Cahill wrote:
>  
>>> Is there anyway to get an amplitude of an ac signal averaged over
>>> time w/o some kind of rectification?
>  
>> Sure. Connect the AC to a heater and measure the temperature.
>  
>>> This doesn't include the trivial, i.e., dc offsets or partial
>>> cycles. The rectification could be analog, digital or software.
> 
> You need some non-linear operation, which may or may not be
> considered rectification.  I believe the heater/temperature
> measurement has been used for RF power measurements for many years.
> 
> -- glen

Google "Bolometer".

-- 
Best Regards:
                     Baron.

On 11/05/2010 02:35 PM, Bret Cahill wrote:
>>> In comp.dsp Vladimir Vassilevsky<nos...@nowhere.com>  wrote:
>>
>>>> Bret Cahill wrote:
>>
>>>>> Is there anyway to get an amplitude of an ac signal averaged over time
>>>>> w/o some kind of rectification?
>>
>>>> Sure. Connect the AC to a heater and measure the temperature.
>>
>>>>> This doesn't include the trivial, i.e., dc offsets or partial cycles.
>>>>> The rectification could be analog, digital or software.
>>
>>> You need some non-linear operation,  which may or may not be
>>> considered rectification.
>>
>> No nonlinear operations. Just multiply the AC by a synchronous reference.
>>
>>> I believe the heater/temperature
>>> measurement has been used for RF power measurements for many years.
>>
>> That doesn't do any rectification for sure.
>
> Positive heat with a positive signal and positive heat with a negative
> signal can only be considered a form of rectification.

Troll.

-- 

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html

On 11/05/2010 01:00 PM, Bret Cahill wrote:
> Is there anyway to get an amplitude of an ac signal averaged over time
> w/o some kind of rectification?
>
> This doesn't include the trivial, i.e., dc offsets or partial cycles.
>
> The rectification could be analog, digital or software.

Don't feed the troll

-- 

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html

On 11/5/2010 5:26 PM, glen herrmannsfeldt wrote:
> In comp.dsp Fred Marshall<fmarshall_xremove_the_xs@xacm.org>  wrote:
> (snip)
>
>> Multiplying by other than a constant or a stable periodic waveform is a
>> non-linear operation.  Multiplying by a constant or a sinusoid or a sum
>> of sinusoids is a linear operation.... believe it or not.
>
> Hmmm.  As I understand it (longer ago than I remember) they used
> to make rectifiers out of synchronous motor driven commutators.
>
> That is, multiply by a square wave (sum of sines) of the appropriate
> frequency.  The result doesn't seem like a linear function of
> the input anymore.
>
> -- glen

Well, Jerry and I argued about this at length here some time ago.  The 
key is in the tests for linearity. It rather surprised me too and it 
didn't "seem like a linear system" but it worked out.

In all these things we're talking about a 2-port situation and asking 
whether the 2-port system is linear or not.

So, if the 2-port system is a 4-quadrant multiplier with a stable 
sinusoidal input (which is not one of the "ports") then the result of a 
signal on the "input" port is a sinusoid of amplitude proportional to 
that input.  And, if a different input is applied, same thing.  And, if 
the sum of those two inputs is applied, the output is the sum of the two 
independent outputs.
etc.

Now, I must say that I was pulling from memory about the composite 
sinusoidal modulating function .. but I think it still holds with that 
being used instead of a simple sinusoid...

Fred

On 2010-11-05, Bret Cahill <Bret_E_Cahill@yahoo.com> wrote:

> Is there anyway to get an amplitude of an ac signal averaged over time
> w/o some kind of rectification?

Run it through a resistor an measure heat output. The result will be
proportional to the square of the amplitiude



-- 
ɹǝpun uʍop ɯoɹɟ sƃuıʇǝǝɹ⅁

On 2010-11-05, Bret Cahill <Bret_E_Cahill@yahoo.com> wrote:
>> > In comp.dsp Vladimir Vassilevsky <nos...@nowhere.com> wrote:
>>
>> >>Bret Cahill wrote:
>>
>> >>>Is there anyway to get an amplitude of an ac signal averaged over time
>> >>>w/o some kind of rectification?
>>
>> >>Sure. Connect the AC to a heater and measure the temperature.
>>
>> >>>This doesn't include the trivial, i.e., dc offsets or partial cycles.
>> >>>The rectification could be analog, digital or software.
>>
>> > You need some non-linear operation,  which may or may not be
>> > considered rectification.
>>
>> No nonlinear operations. Just multiply the AC by a synchronous reference.
>>
>> > I believe the heater/temperature
>> > measurement has been used for RF power measurements for many years.
>>
>> That doesn't do any rectification for sure.
>
> Positive heat with a positive signal and positive heat with a negative
> signal can only be considered a form of rectification.

In that case the answer is "no"

-- 
ɹǝpun uʍop ɯoɹɟ sƃuıʇǝǝɹ⅁

On 2010-11-05, Bret Cahill <BretCahill@peoplepc.com> wrote:

> The goal is getting the amplitude _without_
> rectification.  Rectification" is to be very broadly construed and
> includes anything that results in the final "signal" being positive at
> all times.

show me a signal with negative amplitude.

-- 
ɹǝpun uʍop ɯoɹɟ sƃuıʇǝǝɹ⅁

On Sat, 06 Nov 2010 09:52:02 -0700, Fred Marshall
<fmarshall_xremove_the_xs@xacm.org> wrote:

>On 11/5/2010 5:26 PM, glen herrmannsfeldt wrote:
>> In comp.dsp Fred Marshall<fmarshall_xremove_the_xs@xacm.org>  wrote:
>> (snip)
>>
>>> Multiplying by other than a constant or a stable periodic waveform is a
>>> non-linear operation.  Multiplying by a constant or a sinusoid or a sum
>>> of sinusoids is a linear operation.... believe it or not.
>>
>> Hmmm.  As I understand it (longer ago than I remember) they used
>> to make rectifiers out of synchronous motor driven commutators.
>>
>> That is, multiply by a square wave (sum of sines) of the appropriate
>> frequency.  The result doesn't seem like a linear function of
>> the input anymore.
>>
>> -- glen
>
>Well, Jerry and I argued about this at length here some time ago.  The 
>key is in the tests for linearity. It rather surprised me too and it 
>didn't "seem like a linear system" but it worked out.
>
>In all these things we're talking about a 2-port situation and asking 
>whether the 2-port system is linear or not.
>
>So, if the 2-port system is a 4-quadrant multiplier with a stable 
>sinusoidal input (which is not one of the "ports") then the result of a 
>signal on the "input" port is a sinusoid of amplitude proportional to 
>that input.  And, if a different input is applied, same thing.  And, if 
>the sum of those two inputs is applied, the output is the sum of the two 
>independent outputs.
>etc.
>
>Now, I must say that I was pulling from memory about the composite 
>sinusoidal modulating function .. but I think it still holds with that 
>being used instead of a simple sinusoid...
>

I think this depends on what you mean by "linear".  Most of
the time, IMHO, we mean "no new frequencies produced in the
output".  Multiplying two sinusoids violates this, because
you end up with sum and difference frequencies.  

Think about amplifiers.  We talk about "nonlinear
distortion" when we get new frequency components (THD, IMD,
slew limiting, etc), while "linear distortion" (which is a
rarely used term) means only changes in amplitudes or phases
of the original frequency components.

Best regards,


Bob Masta
 
              DAQARTA  v5.10
   Data AcQuisition And Real-Time Analysis
              www.daqarta.com
Scope, Spectrum, Spectrogram, Sound Level Meter
    Frequency Counter, FREE Signal Generator
           Pitch Track, Pitch-to-MIDI 
         DaqMusic - FREE MUSIC, Forever!
             (Some assembly required)
     Science (and fun!) with your sound card!

On Sun, 07 Nov 2010 04:22:16 +0000, Jasen Betts wrote:

> show me a signal with negative amplitude.

a sin x at x>pi<2*pi

-- 
"For a successful technology, reality must take precedence 
over public relations, for nature cannot be fooled."
                                       (Richard Feynman)

On Nov 7, 11:25=A0am, Fred Abse <excretatau...@invalid.invalid> wrote:
> On Sun, 07 Nov 2010 04:22:16 +0000, Jasen Betts wrote:
> > show me a signal with negative amplitude.
>
> a sin x at x>pi<2*pi
>
> --
> "For a successful technology, reality must take precedence
> over public relations, for nature cannot be fooled."
> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =
=A0 =A0(Richard Feynman)

Very good Fred.

I searched all of the posts here to see if anyone caught that.

Posters are using the terms "amplitude" and "magnitude" as if they are
the same.

Dirk

On 11/7/2010 4:58 AM, Bob Masta wrote:
> "no new frequencies produced in the
> output"

Yes, I know: "no new frequencies produced in the
output"
that's why I was a little surprised.

The tests for linearity:

Homogeneity:
Increase the input amplitude and the output amplitude increases accordingly.

Additivity:
input "a" and get output "A"
input "b" and get output "B"
input a+b and get output A+B

Superposition - homogeneity and additivity:

Ma + Nb > MA + NB

Nowhere does it say that A can only contain the frequencies contained in 
a, etc.

So, if one does 4-quadrant multiplication for example, and one of the 
input ports to the multiplier is a steady sinusoid, then we get 
supressed-carrier modulation relative to the other input port.

While there are "new frequencies" introduced relative to a or b, there 
are no "new frequencies" introduced relative to A or B and all the tests 
for linearity are met.

http://groups.google.com/group/comp.dsp/browse_thread/thread/6c2e8d4b41ebc6e3/4288761f23f82e07?hl=en&lnk=gst&q=marshall+avins+linearity#4288761f23f82e07

Fred

On 11/05/2010 10:51 PM, glen herrmannsfeldt wrote:
> In comp.dsp Vladimir Vassilevsky<nospam@nowhere.com>  wrote:
> (snip regarding rectiification, and I wrote)
>
>>> You need some non-linear operation,  which may or may not be
>>> considered rectification.
>
>> No nonlinear operations. Just multiply the AC by a synchronous reference.
>
> Multiplying by other than a constant is a non-linear operation.

Multiplying by any m(t) is a linear operation, since it satisfies

1)   f[x(t)+y(t)]=f[x(t)]+f[y(t)]
              [x(t)+y(t)]�m(t) = [x(t)�m(t)]+[y(t)�m(t)]

and
2)   f[a�x(t)]=a�f[x(t)]
              [a�x(t)]�m(t) = a�[x(t)�m(t)]

The operation "multiply by m(t)" is a linear, time-variant system.

Pere

On Fri, 5 Nov 2010 13:00:54 -0700 (PDT) Bret Cahill
<Bret_E_Cahill@yahoo.com> wrote in Message id:
<e6a620ac-ec32-4ae8-8f56-6f948ed1f142@x7g2000prj.googlegroups.com>:

>Is there anyway to get an amplitude of an ac signal averaged over time
>w/o some kind of rectification?

RMS or peak? How much time are we talking about?

What are you measuring?

On 11/07/2010 01:58 PM, Bob Masta wrote:
> On Sat, 06 Nov 2010 09:52:02 -0700, Fred Marshall
> <fmarshall_xremove_the_xs@xacm.org>  wrote:
>
>> On 11/5/2010 5:26 PM, glen herrmannsfeldt wrote:
>>> In comp.dsp Fred Marshall<fmarshall_xremove_the_xs@xacm.org>   wrote:
>>> (snip)
>>>
>>>> Multiplying by other than a constant or a stable periodic waveform is a
>>>> non-linear operation.  Multiplying by a constant or a sinusoid or a sum
>>>> of sinusoids is a linear operation.... believe it or not.
>>>
>>> Hmmm.  As I understand it (longer ago than I remember) they used
>>> to make rectifiers out of synchronous motor driven commutators.
>>>
>>> That is, multiply by a square wave (sum of sines) of the appropriate
>>> frequency.  The result doesn't seem like a linear function of
>>> the input anymore.
>>>
>>> -- glen
>>
>> Well, Jerry and I argued about this at length here some time ago.  The
>> key is in the tests for linearity. It rather surprised me too and it
>> didn't "seem like a linear system" but it worked out.
>>
>> In all these things we're talking about a 2-port situation and asking
>> whether the 2-port system is linear or not.
>>
>> So, if the 2-port system is a 4-quadrant multiplier with a stable
>> sinusoidal input (which is not one of the "ports") then the result of a
>> signal on the "input" port is a sinusoid of amplitude proportional to
>> that input.  And, if a different input is applied, same thing.  And, if
>> the sum of those two inputs is applied, the output is the sum of the two
>> independent outputs.
>> etc.
>>
>> Now, I must say that I was pulling from memory about the composite
>> sinusoidal modulating function .. but I think it still holds with that
>> being used instead of a simple sinusoid...
>>
>
> I think this depends on what you mean by "linear".  Most of
> the time, IMHO, we mean "no new frequencies produced in the
> output".  Multiplying two sinusoids violates this, because
> you end up with sum and difference frequencies.

That is true for time-invariant systems. Look at my previous post and, 
for instance, http://en.wikipedia.org/wiki/Linear for the mathematical 
definition of linearity.

The operation "multiplying by m(t)" is a linear but time-_variant_ 
operation and is common in communication systems, such as any frequency 
conversion process:

     ..........
 >------[x]------------>
     .   |    .
     . cos wt .
     ..........  <....:linear, time-variant "box"

Incidentally, this block is often done exploiting nonlinearity, but this 
is another story and could be done differently.

Pere


>
> Think about amplifiers.  We talk about "nonlinear
> distortion" when we get new frequency components (THD, IMD,
> slew limiting, etc), while "linear distortion" (which is a
> rarely used term) means only changes in amplitudes or phases
> of the original frequency components.
>
> Best regards,
>
>
> Bob Masta
>
>                DAQARTA  v5.10
>     Data AcQuisition And Real-Time Analysis
>                www.daqarta.com
> Scope, Spectrum, Spectrogram, Sound Level Meter
>      Frequency Counter, FREE Signal Generator
>             Pitch Track, Pitch-to-MIDI
>           DaqMusic - FREE MUSIC, Forever!
>               (Some assembly required)
>       Science (and fun!) with your sound card!

On 11/05/2010 11:10 PM, Bret Cahill wrote:
>>> Is there anyway to get an amplitude of an ac signal averaged over time
>>> w/o some kind of rectification?
>>
>>> This doesn't include the trivial, i.e., dc offsets or partial cycles.
>>
>>> The rectification could be analog, digital or software.
>>
>>> Bret Cahill
>>
>> You need a precision rectifier..
>>     couple of op-amps and maybe 1 diode ..
>>
>> http://sound.westhost.com/appnotes/an001.htm
>
>
> Poorly written OP.  The goal is getting the amplitude _without_
> rectification.  Rectification" is to be very broadly construed and
> includes anything that results in the final "signal" being positive at
> all times.

If your signal x(t) is described by x(t) = A�(cos wt + fi), then A is 
called the amplitude and is positive (at all times, since it does not 
depend on t) by definition.

 From your rewritten question it seems you want something that does not 
make sense (even) in the sinusoidal case.

Pere

>
> Maybe peak to peak wil work if you know the wave form.
>
>
> Bret Cahill
>
>
>
>
>

On 11/06/2010 04:44 PM, Tim Wescott wrote:
> On 11/05/2010 01:00 PM, Bret Cahill wrote:
>> Is there anyway to get an amplitude of an ac signal averaged over time
>> w/o some kind of rectification?
>>
>> This doesn't include the trivial, i.e., dc offsets or partial cycles.
>>
>> The rectification could be analog, digital or software.
>
> Don't feed the troll
>

Sorry!

Pere

On Mon, 08 Nov 2010 11:20:27 -0500 JW <none@dev.null> wrote in Message id:
<so8gd61ru59916j21l3lrqvu4o3g0jel5e@4ax.com>:

>On Fri, 5 Nov 2010 13:00:54 -0700 (PDT) Bret Cahill
><Bret_E_Cahill@yahoo.com> wrote in Message id:
><e6a620ac-ec32-4ae8-8f56-6f948ed1f142@x7g2000prj.googlegroups.com>:
>
>>Is there anyway to get an amplitude of an ac signal averaged over time
>>w/o some kind of rectification?
>
>RMS or peak? How much time are we talking about?
        ^^^^  
p-p IM.

>What are you measuring?

On 11/7/2010 10:01 AM, Dirk Bell wrote:
> On Nov 7, 11:25 am, Fred Abse<excretatau...@invalid.invalid>  wrote:
>> On Sun, 07 Nov 2010 04:22:16 +0000, Jasen Betts wrote:
>>> show me a signal with negative amplitude.
>>
>> a sin x at x>pi<2*pi
>>
>> --

> I searched all of the posts here to see if anyone caught that.
>
> Posters are using the terms "amplitude" and "magnitude" as if they are
> the same.
>
> Dirk

Dirk,

Good point.  I've probably done that!  So you got me thinking....

It occurs to me that there's another:
"value of a function" or "value of a signal" which, obviously can be 
negative.

This goes along with:

"magnitude of a signal or function" (which I don't think implies any 
time frame).... as in "abs(value of a function)"

"amplitude of a wave" as in "a sin x"
or, amplitude of noise perhaps...
where we most often use "a" as a positive number in physics but don't 
*have to* in mathematics, eh?  In the latter case it could be said to 
have a negative amplitude but that would be most unusual wouldn't it?
Or maybe it's just implied that 0=<a ??

I'm not sure how "amplitude" really applies otherwise:

What's the "amplitude" of a*sin(wt) + b*cos(pi*w*t) ????

Then things like "rms value" or "rms whatever" usually imply a time 
frame over which the sum or integral is taken - often a single period 
but could be just some arbitrary time frame.

Fred

In comp.dsp Fred Marshall <fmarshall_xremove_the_xs@xacm.org> wrote:
(snip, someone wrote)

>> Posters are using the terms "amplitude" and "magnitude" as if 
>> they are the same.

> Good point.  I've probably done that!  So you got me thinking....
 
> It occurs to me that there's another:
> "value of a function" or "value of a signal" which, obviously can be 
> negative.
 
> This goes along with:
 
> "magnitude of a signal or function" (which I don't think implies any 
> time frame).... as in "abs(value of a function)"
 
> "amplitude of a wave" as in "a sin x"
> or, amplitude of noise perhaps...
> where we most often use "a" as a positive number in physics but don't 
> *have to* in mathematics, eh?  In the latter case it could be said to 
> have a negative amplitude but that would be most unusual wouldn't it?
> Or maybe it's just implied that 0=<a ??

In physics, at least in optics and QM, amplitude is signed.

For example, "for coherent sources add the amplitude, 
              for incoherent sources add the intensity."

Though another way to look at it is that amplitude is unsigned, but 
that one has to include the appropriate phase difference when adding.
 
> I'm not sure how "amplitude" really applies otherwise:
 
> What's the "amplitude" of a*sin(wt) + b*cos(pi*w*t) ????

(snip)

-- glen

> >>> show me a signal with negative amplitude.
>
> >> a sin x at x>pi<2*pi
>
> >> --
> > I searched all of the posts here to see if anyone caught that.
>
> > Posters are using the terms "amplitude" and "magnitude" as if they are
> > the same.
>
> > Dirk
>
> Dirk,
>
> Good point. =A0I've probably done that! =A0So you got me thinking....
>
> It occurs to me that there's another:
> "value of a function" or "value of a signal" which, obviously can be
> negative.
>
> This goes along with:
>
> "magnitude of a signal or function" (which I don't think implies any
> time frame).... as in "abs(value of a function)"
>
> "amplitude of a wave" as in "a sin x"
> or, amplitude of noise perhaps...
> where we most often use "a" as a positive number in physics but don't
> *have to* in mathematics, eh? =A0In the latter case it could be said to
> have a negative amplitude but that would be most unusual wouldn't it?
> Or maybe it's just implied that 0=3D<a ??
>
> I'm not sure how "amplitude" really applies otherwise:
>
> What's the "amplitude" of a*sin(wt) + b*cos(pi*w*t) ????
>
> Then things like "rms value" or "rms whatever" usually imply a time
> frame over which the sum or integral is taken - often a single period
> but could be just some arbitrary time frame.

The term "time ave." in the header implies the quantity is over some
time period.

If peak to peak or rms or any other measure of amplitude increases by
some percent during that time period then the magnitude of the low
pass or integral will increase by that same percent.


Bret Cahill

> > Is there anyway to get an amplitude of an ac signal averaged over time
> > w/o some kind of rectification?
>
> Run it through a resistor an measure heat output.

Temperature is a signal.  Fourier analysis _originated_ in heat
transfer.

Your resistor is taking an AC signal and putting out a positive only
signal.

What is that if not rectification?


Bret Cahill

On 11/08/2010 06:20 PM, Fred Marshall wrote:
> On 11/7/2010 10:01 AM, Dirk Bell wrote:
>> On Nov 7, 11:25 am, Fred Abse<excretatau...@invalid.invalid> wrote:
>>> On Sun, 07 Nov 2010 04:22:16 +0000, Jasen Betts wrote:
>>>> show me a signal with negative amplitude.
>>>
>>> a sin x at x>pi<2*pi
>>>
>>> --
>
>> I searched all of the posts here to see if anyone caught that.
>>
>> Posters are using the terms "amplitude" and "magnitude" as if they are
>> the same.
>>
>> Dirk
>
> Dirk,
>
> Good point. I've probably done that! So you got me thinking....
>
> It occurs to me that there's another:
> "value of a function" or "value of a signal" which, obviously can be
> negative.
>
> This goes along with:
>
> "magnitude of a signal or function" (which I don't think implies any
> time frame).... as in "abs(value of a function)"
>
> "amplitude of a wave" as in "a sin x"
> or, amplitude of noise perhaps...
> where we most often use "a" as a positive number in physics but don't
> *have to* in mathematics, eh? In the latter case it could be said to
> have a negative amplitude but that would be most unusual wouldn't it?
> Or maybe it's just implied that 0=<a ??
>
> I'm not sure how "amplitude" really applies otherwise:
>
> What's the "amplitude" of a*sin(wt) + b*cos(pi*w*t) ????

sqrt(a�+b�)

Pere
>
> Then things like "rms value" or "rms whatever" usually imply a time
> frame over which the sum or integral is taken - often a single period
> but could be just some arbitrary time frame.
>
> Fred
>

On 2010-11-07, Fred Abse <excretatauris@invalid.invalid> wrote:
> On Sun, 07 Nov 2010 04:22:16 +0000, Jasen Betts wrote:
>
> a sin x at x>pi<2*pi
>

so brett wants a low pass filter?

-- 
ɹǝpun uʍop ɯoɹɟ sƃuıʇǝǝɹ⅁

On 2010-11-08, Bret Cahill <Bret_E_Cahill@yahoo.com> wrote:
>> > Is there anyway to get an amplitude of an ac signal averaged over time
>> > w/o some kind of rectification?
>>
>> Run it through a resistor an measure heat output.
>
> Temperature is a signal.  Fourier analysis _originated_ in heat
> transfer.
>
> Your resistor is taking an AC signal and putting out a positive only
> signal.
>
> What is that if not rectification?

In what circumstances do you want a negative result?

-- 
ɹǝpun uʍop ɯoɹɟ sƃuıʇǝǝɹ⅁

> >> > Is there anyway to get an amplitude of an ac signal averaged over ti=
me
> >> > w/o some kind of rectification?

> >> Run it through a resistor an measure heat output.

> > Temperature is a signal. =A0Fourier analysis _originated_ in heat
> > transfer.

> > Your resistor is taking an AC signal and putting out a positive only
> > signal.

> > What is that if not rectification?

To be sure the quantity representing the signal changes from voltage
to temperature but arguing that isn't rectification that would be like
trying to claim the First Amendment only covers the narrowest
definitions of "speech" and "press."

Even Justice Thomas "emanated in penumbras" on that one.

> In what circumstances do you want a negative result?

Sign isn't of interest here.  The goal is to determine if there is any
way other than rectification to determine the magnitude of a signal
over a period of time.

If you knew the wave form in the time domain you could measure peak to
peak and calculate it that way.

As pointed out above amplitude is always positive.  A Fourier
transform is always positive so that might be one way to determine the
magnitude of the signal over time.

It might be hard to argue that a FFT is rectification.  When you
rectify by diode or resistor you lose some information.  You no longer
know when the unrectified signal was positive or negative.

Not true for the FFT.


Bret Cahill

On Tue, 9 Nov 2010 07:03:09 -0800 (PST), Bret Cahill
<Bret_E_Cahill@yahoo.com> wrote:
<snip>

>As pointed out above amplitude is always positive.  A Fourier
>transform is always positive so that might be one way to determine the
>magnitude of the signal over time.

Huh?  A Fourier transform produces real and imaginary
arrays, any component of which may be negative.  It is
common to display the *magnitude* at each frequency bin, but
that involves a separate step (square root of
sum-of-squares), which I assume you would call
rectification.

>It might be hard to argue that a FFT is rectification.  When you
>rectify by diode or resistor you lose some information.  You no longer
>know when the unrectified signal was positive or negative.
>
>Not true for the FFT.

If you only take the magnitude, to get the "always positive"
result you mention, then that of course has no information
about the original polarity.  You can also compute the phase
at each frequency bin, as a separate operation on the same
real and imaginary arrays.  Is that what you are talking
about?

Why the concern over whether a method is a form of
"rectification" or not?  What are you really trying to
measure?


Best regards,



Bob Masta
 
              DAQARTA  v5.10
   Data AcQuisition And Real-Time Analysis
              www.daqarta.com
Scope, Spectrum, Spectrogram, Sound Level Meter
    Frequency Counter, FREE Signal Generator
           Pitch Track, Pitch-to-MIDI 
         DaqMusic - FREE MUSIC, Forever!
             (Some assembly required)
     Science (and fun!) with your sound card!

> >As pointed out above amplitude is always positive. =A0A Fourier
> >transform is always positive so that might be one way to determine the
> >magnitude of the signal over time.
>
> Huh? =A0A Fourier transform produces real and imaginary
> arrays, any component of which may be negative. =A0

If the only information displayed in the FFT is an amplitude and a
phase angle for each frequency, and if amplitude is always positive as
suggested above, then amplitudes are always positive in an FFT.

Even a negative dc offset could be represented as a positive 0 Hz
spike with a 180 degree phase angle.

> It is
> common to display the *magnitude* at each frequency bin, but
> that involves a separate step (square root of
> sum-of-squares), which I assume you would call
> rectification.

In that case there seems to be no way to end run rectification.

> >It might be hard to argue that a FFT is rectification. =A0When you
> >rectify by diode or resistor you lose some information. =A0You no longer
> >know when the unrectified signal was positive or negative.

> >Not true for the FFT.

> If you only take the magnitude, to get the "always positive"
> result you mention, then that of course has no information
> about the original polarity. =A0

A phase angle for each frequency would allow recovery the sign of the
signal in the time domain.

> You can also compute the phase
> at each frequency bin, as a separate operation on the same
> real and imaginary arrays. =A0Is that what you are talking
> about?

Yes.

> Why the concern over whether a method is a form of
> "rectification" or not? =A0What are you really trying to
> measure?

The magnitude of a randomly fluctuating signal over a period of time.
This doesn't include the trivial case where the time the signal is
positive is integrated along with the time the signal is negative.

Peak to peak will only work with a known waveform so it looks more and
more like it must involve taking absolute values or sqrts of sum of
squares at some point which would ordinarily be considered
"rectification" even in a relatively narrow sense of the word.


Bret Cahill


"The truth has never yet clung to the arm of an inflexible signal."

-- Nietzsche

On Wed, 10 Nov 2010 08:43:04 -0800 (PST), Bret Cahill
<Bret_E_Cahill@yahoo.com> wrote:

>> >As pointed out above amplitude is always positive. �A Fourier
>> >transform is always positive so that might be one way to determine the
>> >magnitude of the signal over time.
>>
>> Huh? �A Fourier transform produces real and imaginary
>> arrays, any component of which may be negative. �
>
>If the only information displayed in the FFT is an amplitude and a
>phase angle for each frequency, and if amplitude is always positive as
>suggested above, then amplitudes are always positive in an FFT.

Don't give up your day job, Cahill.  An angle of 180 degrees is *negative*. If
the FFT "rectified" it's obvious to even a moron that information is lost and
an RFFT couldn't be done.

On Wed, 10 Nov 2010 08:43:04 -0800 (PST), Bret Cahill
<Bret_E_Cahill@yahoo.com> wrote:


>
>> Why the concern over whether a method is a form of
>> "rectification" or not? =A0What are you really trying to
>> measure?
>
>The magnitude of a randomly fluctuating signal over a period of time.
>This doesn't include the trivial case where the time the signal is
>positive is integrated along with the time the signal is negative.
>
>Peak to peak will only work with a known waveform so it looks more and
>more like it must involve taking absolute values or sqrts of sum of
>squares at some point which would ordinarily be considered
>"rectification" even in a relatively narrow sense of the word.

You haven't explained what's wrong with "rectification", and
especially true RMS for this purpose.

Best regards,




Bob Masta
 
              DAQARTA  v5.10
   Data AcQuisition And Real-Time Analysis
              www.daqarta.com
Scope, Spectrum, Spectrogram, Sound Level Meter
    Frequency Counter, FREE Signal Generator
           Pitch Track, Pitch-to-MIDI 
         DaqMusic - FREE MUSIC, Forever!
             (Some assembly required)
     Science (and fun!) with your sound card!

Bob Masta wrote:
> 
> On Wed, 10 Nov 2010 08:43:04 -0800 (PST), Bret Cahill
> <Bret_E_Cahill@yahoo.com> wrote:
> 
> >
> >> Why the concern over whether a method is a form of
> >> "rectification" or not? =A0What are you really trying to
> >> measure?
> >
> >The magnitude of a randomly fluctuating signal over a period of time.
> >This doesn't include the trivial case where the time the signal is
> >positive is integrated along with the time the signal is negative.
> >
> >Peak to peak will only work with a known waveform so it looks more and
> >more like it must involve taking absolute values or sqrts of sum of
> >squares at some point which would ordinarily be considered
> >"rectification" even in a relatively narrow sense of the word.
> 
> You haven't explained what's wrong with "rectification", and
> especially true RMS for this purpose.



        |----||----|
        |  DO NOT  |
        | FEED THE |
        |  TROLLS! |
        |----||----|
             ||
             ||
             ||
 /|\\|/||||//|||/\???\\//\\\\/|?\/\\\\/\/\/\||||\

-- 
Politicians should only get paid if the budget is balanced, and there is
enough left over to pay them.

> >> Why the concern over whether a method is a form of
> >> "rectification" or not? =3DA0What are you really trying to
> >> measure?
>
> >The magnitude of a randomly fluctuating signal over a period of time.
> >This doesn't include the trivial case where the time the signal is
> >positive is integrated along with the time the signal is negative.
>
> >Peak to peak will only work with a known waveform so it looks more and
> >more like it must involve taking absolute values or sqrts of sum of
> >squares at some point which would ordinarily be considered
> >"rectification" even in a relatively narrow sense of the word.
>
> You haven't explained what's wrong with "rectification",

Nothing is wrong with rectification.  It's not only the best way.
It's the only way.

> and
> especially true RMS for this purpose.

> Best regards,
>
> Bob Masta
>
> =A0 =A0 =A0 =A0 =A0 =A0 =A0 DAQARTA =A0v5.10
> =A0 =A0Data AcQuisition And Real-Time Analysis
> =A0 =A0 =A0 =A0 =A0 =A0 =A0www.daqarta.com
> Scope, Spectrum, Spectrogram, Sound Level Meter
> =A0 =A0 Frequency Counter, FREE Signal Generator
> =A0 =A0 =A0 =A0 =A0 =A0Pitch Track, Pitch-to-MIDI
> =A0 =A0 =A0 =A0 =A0DaqMusic - FREE MUSIC, Forever!
> =A0 =A0 =A0 =A0 =A0 =A0 =A0(Some assembly required)
> =A0 =A0 =A0Science (and fun!) with your sound card!

Bret Cahill wrote:

>>>>Why the concern over whether a method is a form of
>>>>"rectification" or not? =A0What are you really trying to
>>>>measure?
>>
>>>The magnitude of a randomly fluctuating signal over a period of time.
>>>This doesn't include the trivial case where the time the signal is
>>>positive is integrated along with the time the signal is negative.
>>
>>>Peak to peak will only work with a known waveform so it looks more and
>>>more like it must involve taking absolute values or sqrts of sum of
>>>squares at some point which would ordinarily be considered
>>>"rectification" even in a relatively narrow sense of the word.
>>
>>You haven't explained what's wrong with "rectification",
> 
> 
> Nothing is wrong with rectification.  It's not only the best way.
> It's the only way.

  Did you look at the precision rectifier links I gave you ?

  I haven't seen any reference to what BW you are interested in?


  Jamie.

On Nov 10, 11:43=A0am, Bret Cahill <Bret_E_Cah...@yahoo.com> wrote:
> > >As pointed out above amplitude is always positive. =A0A Fourier
> > >transform is always positive so that might be one way to determine the
> > >magnitude of the signal over time.
>
> > Huh? =A0A Fourier transform produces real and imaginary
> > arrays, any component of which may be negative. =A0
>
> If the only information displayed in the FFT is an amplitude and a
> phase angle for each frequency, and if amplitude is always positive as
> suggested above, then amplitudes are always positive in an FFT.
>
> Even a negative dc offset could be represented as a positive 0 Hz
> spike with a 180 degree phase angle.
>
> > It is
> > common to display the *magnitude* at each frequency bin, but
> > that involves a separate step (square root of
> > sum-of-squares), which I assume you would call
> > rectification.
>
> In that case there seems to be no way to end run rectification.
>
> > >It might be hard to argue that a FFT is rectification. =A0When you
> > >rectify by diode or resistor you lose some information. =A0You no long=
er
> > >know when the unrectified signal was positive or negative.
> > >Not true for the FFT.
> > If you only take the magnitude, to get the "always positive"
> > result you mention, then that of course has no information
> > about the original polarity. =A0
>
> A phase angle for each frequency would allow recovery the sign of the
> signal in the time domain.
>
> > You can also compute the phase
> > at each frequency bin, as a separate operation on the same
> > real and imaginary arrays. =A0Is that what you are talking
> > about?
>
> Yes.
>
> > Why the concern over whether a method is a form of
> > "rectification" or not? =A0What are you really trying to
> > measure?
>
> The magnitude of a randomly fluctuating signal over a period of time.
> This doesn't include the trivial case where the time the signal is
> positive is integrated along with the time the signal is negative.
>
> Peak to peak will only work with a known waveform so it looks more and
> more like it must involve taking absolute values or sqrts of sum of
> squares at some point which would ordinarily be considered
> "rectification" even in a relatively narrow sense of the word.
>
> Bret Cahill

I haven't read the whole thread as it seems rather long and possibly
contentious.  But if you are saying that an FFT performs rectification
in any sense, I don't think that is true.  Maybe I misunderstand what
you are saying.

BTW, why do you want to avoid rectification when performing this
measurement?  Also, how exactly do you define "rectification"?  If you
want a magnitude off the signal, by definition that will be always
positive and like would be considered rectification.  The FFT does not
inherently calculate the magnitude, rather it results in a complex
number allowing both phase and magnitude to be calculated.

Rick

On Nov 5, 5:10=A0pm, Bret Cahill <BretCah...@peoplepc.com> wrote:
> > > Is there anyway to get an amplitude of an ac signal averaged over tim=
e
> > > w/o some kind of rectification?
>
> > > This doesn't include the trivial, i.e., dc offsets or partial cycles.
>
> > > The rectification could be analog, digital or software.
>
> > > Bret Cahill
>
> > You need a precision rectifier..
> > =A0 =A0couple of op-amps and maybe 1 diode ..
>
> >http://sound.westhost.com/appnotes/an001.htm
>
> Poorly written OP. =A0The goal is getting the amplitude _without_
> rectification. =A0Rectification" is to be very broadly construed and
> includes anything that results in the final "signal" being positive at
> all times.
>
> Maybe peak to peak wil work if you know the wave form.
>
> Bret Cahill

Ok, this is making some sense.  If you want to measure amplitude, by
definition this is always going to be a positive number, right?  Even
your peak to peak measurement should provide a positive value which is
rectification according to your definition I believe.

Rick

> > > >As pointed out above amplitude is always positive. =A0A Fourier
> > > >transform is always positive so that might be one way to determine t=
he
> > > >magnitude of the signal over time.
>
> > > Huh? =A0A Fourier transform produces real and imaginary
> > > arrays, any component of which may be negative. =A0
>
> > If the only information displayed in the FFT is an amplitude and a
> > phase angle for each frequency, and if amplitude is always positive as
> > suggested above, then amplitudes are always positive in an FFT.
>
> > Even a negative dc offset could be represented as a positive 0 Hz
> > spike with a 180 degree phase angle.
>
> > > It is
> > > common to display the *magnitude* at each frequency bin, but
> > > that involves a separate step (square root of
> > > sum-of-squares), which I assume you would call
> > > rectification.
>
> > In that case there seems to be no way to end run rectification.
>
> > > >It might be hard to argue that a FFT is rectification. =A0When you
> > > >rectify by diode or resistor you lose some information. =A0You no lo=
nger
> > > >know when the unrectified signal was positive or negative.
> > > >Not true for the FFT.
> > > If you only take the magnitude, to get the "always positive"
> > > result you mention, then that of course has no information
> > > about the original polarity. =A0
>
> > A phase angle for each frequency would allow recovery the sign of the
> > signal in the time domain.
>
> > > You can also compute the phase
> > > at each frequency bin, as a separate operation on the same
> > > real and imaginary arrays. =A0Is that what you are talking
> > > about?
>
> > Yes.
>
> > > Why the concern over whether a method is a form of
> > > "rectification" or not? =A0What are you really trying to
> > > measure?
>
> > The magnitude of a randomly fluctuating signal over a period of time.
> > This doesn't include the trivial case where the time the signal is
> > positive is integrated along with the time the signal is negative.
>
> > Peak to peak will only work with a known waveform so it looks more and
> > more like it must involve taking absolute values or sqrts of sum of
> > squares at some point which would ordinarily be considered
> > "rectification" even in a relatively narrow sense of the word.
>
> > Bret Cahill
>
> I haven't read the whole thread as it seems rather long and possibly
> contentious. =A0But if you are saying that an FFT performs rectification
> in any sense, I don't think that is true. =A0

The always positive frequency domain graph is apparently generated by
taking the square root of the sum of the square of the real and the
square of the imaginary components of each frequency.

That seems like rectification.

> Maybe I misunderstand what
> you are saying.

> BTW, why do you want to avoid rectification when performing this
> measurement? =A0

Just checking to make sure that is the only way.

> Also, how exactly do you define "rectification"? =A0

Absolute value and sqrt of sum of squares.

> If you
> want a magnitude off the signal, by definition that will be always
> positive and like would be considered rectification. =A0

Maybe by definition.

> The FFT does not
> inherently calculate the magnitude, rather it results in a complex
> number allowing both phase and magnitude to be calculated.

By taking the sqrt of the sum of the squares which is rectification.


Bret Cahill

> >>>>Why the concern over whether a method is a form of
> >>>>"rectification" or not? =3DA0What are you really trying to
> >>>>measure?
>
> >>>The magnitude of a randomly fluctuating signal over a period of time.
> >>>This doesn't include the trivial case where the time the signal is
> >>>positive is integrated along with the time the signal is negative.
>
> >>>Peak to peak will only work with a known waveform so it looks more and
> >>>more like it must involve taking absolute values or sqrts of sum of
> >>>squares at some point which would ordinarily be considered
> >>>"rectification" even in a relatively narrow sense of the word.
>
> >>You haven't explained what's wrong with "rectification",
>
> > Nothing is wrong with rectification. =A0It's not only the best way.
> > It's the only way.
>
> =A0 Did you look at the precision rectifier links I gave you ?

I've been using the ones provided by LT SPICE.

> =A0 I haven't seen any reference to what BW you are interested in?

This is just a general question.


Bret Cahill

> > > > Is there anyway to get an amplitude of an ac signal averaged over t=
ime
> > > > w/o some kind of rectification?
>
> > > > This doesn't include the trivial, i.e., dc offsets or partial cycle=
s.
>
> > > > The rectification could be analog, digital or software.
>
> > > > Bret Cahill
>
> > > You need a precision rectifier..
> > > =A0 =A0couple of op-amps and maybe 1 diode ..
>
> > >http://sound.westhost.com/appnotes/an001.htm
>
> > Poorly written OP. =A0The goal is getting the amplitude _without_
> > rectification. =A0Rectification" is to be very broadly construed and
> > includes anything that results in the final "signal" being positive at
> > all times.
>
> > Maybe peak to peak wil work if you know the wave form.
>
> > Bret Cahill
>
> Ok, this is making some sense. =A0If you want to measure amplitude, by
> definition this is always going to be a positive number, right? =A0Even
> your peak to peak measurement should provide a positive value which is
> rectification according to your definition I believe.

I was thinking that myself.


Bret Cahill

In comp.dsp Bret Cahill <Bret_E_Cahill@yahoo.com> wrote:
(big snip regarding FFT and rectification)

> The always positive frequency domain graph is apparently generated by
> taking the square root of the sum of the square of the real and the
> square of the imaginary components of each frequency.
 
> That seems like rectification.

Well, it is usual to also keep atan2(y,x), in which case no
information is lost.  You have to somehow lose that information.

Consider a DVM where the minus LED is broken.  There is no
rectification, but the information regarding the sign is lost.

I tend to think of rectification as a physical process, applied
to analog signals.  As far radio receivers, it is normally followed
by a low-pass filter.  If one does the same with a digitized
version, then I would have to agree that it is rectification.

-- glen