Hi all and thanks for your answer I am analizing (only) very short signals for which the ADC are too expensive, then analog ones are necessary. My question may be too simple for some of you, but it is really consuming my time. What do you get at the output of an analog filter? Is the magnitude of the filtered signal? is the real part of the filtered signal? or dependes on the load I use? In some paper, I saw they assume they get the real part of the filtered signal, the in order to get imaginary one an extra filterign block [1/s] is added, then the imaginary part is extracted. Well, if the previous is true, another question is how can I get the magnitude of the filtered signal without having to use both values to calculate it in the analog domain, which seems very complicated. This thigs are easily understood in the digital domain, but as you can see they are not obvious. Thanks Greg This message was sent using the Comp.DSP web interface on www.DSPRelated.com
Is the output of an analog filter real or the magnitude?
Started by ●October 18, 2005
Reply by ●October 18, 20052005-10-18
In article <5YSdnXjMYLkgK8neRVn-rQ@giganews.com>, Gregwise says...> >Hi all and thanks for your answer > >I am analizing (only) very short signals for which the ADC are too >expensive, then analog ones are necessary. My question may be too simple >for some of you, but it is really consuming my time. > >What do you get at the output of an analog filter? Is the magnitude of the >filtered signal? is the real part of the filtered signal? or dependes on >the load I use? In some paper, I saw they assume they get the real part of >the filtered signal, the in order to get imaginary one an extra filterign >block [1/s] is added, then the imaginary part is extracted. > >Well, if the previous is true, another question is how can I get the >magnitude of the filtered signal without having to use both values to >calculate it in the analog domain, which seems very complicated.Almost all analogue filters are single-variable-in, single-variable-out, i.e. real in, real out. In communications systems you can have a pair of real filters, taking in the two signals Real in and Imaginary in (I and Q as they are called in radio systems) and outputting two signals, Real out and Imaginary out. (There are some filters taking two inputs and generating two outputs which have cross-coupling between two filters, they are called "complex" or "polyphase" analogue filters.) The answer to your question depends on what signal you have. If you have a single analogue signal, that signal is - for want of a better word - real and the magnitude is just the magnitude. If you want to detect the magnitude or power of the signal, you can have a more or less complicated analogue circuit to estimate this. An "RMS detector" circuit is one of the possible solutions. If you're looking at a comunications, e.g. radio, system, you could do worse than read "Complex Signal Processing is Not Complex," Kenneth W. Martin, IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS—I: REGULAR PAPERS, VOL. 51, NO. 9, SEPTEMBER 2004. Hope this helps Jens -- Key ID 0x09723C12, jensting@tingleff.org Analogue filtering / 5GHz RLAN / Mdk Linux / odds and ends http://www.tingleff.org/jensting/ +44 1223 211 585 "Never drive a car when you're dead!" Tom Waits
Reply by ●October 18, 20052005-10-18
I'm not sure of your intention, but your mention of the method to get the complex signal' may be related to concepts such as an 'analytic' signal (one that is in fact complex) and a 'phase splitter'.>From a single analog filter you get a measure of the (filtered) realpart of the signal. Unless you in fact have two filters, each looking at some different phase (eg I,Q). I'll talk about the simple case. A real-valued signal has both positive and negative frequencies because you have not measured the phase. This is often not realistic - real-world signals may have only positive frequencies, the negative ones are due to your measurement (unless you are emsuring reltive to some carrier but let's not get too clever). This also makes it hard to do some things mathematically. For example if I shift the frequency up by some amount, then both negative and positive frequencies get shifted upwards which means some of the negative frequencies become positive and this is wrong. What I really want to do is, to get rid of the negative frequencies and then shift the (remaining) positive ones up. I can convert the real signal to a complex one by using a 'phase splitter' - which is a low-pass filter whose coefficients have been modulated with a complex exponential. This is a (digital) filter whose frequency response is not symmetric about zero - it in fact blocks negative frequencies. Filtering the signal with this (I think this may be what you are referring to?) gives you an 'analytic' signal - one that is complex, and that has only positive frequencies. So in some sense you have 'converted' a real-valued signal that was measured from the analog world, into a complex-valued signal that has no positive frequencies and so is in some sense a more 'realistic' (pun intended) signal. Chris =============== Chris Bore www.bores.com Gregwise wrote:> Hi all and thanks for your answer > > I am analizing (only) very short signals for which the ADC are too > expensive, then analog ones are necessary. My question may be too simple > for some of you, but it is really consuming my time. > > What do you get at the output of an analog filter? Is the magnitude of the > filtered signal? is the real part of the filtered signal? or dependes on > the load I use? In some paper, I saw they assume they get the real part of > the filtered signal, the in order to get imaginary one an extra filterign > block [1/s] is added, then the imaginary part is extracted. > > Well, if the previous is true, another question is how can I get the > magnitude of the filtered signal without having to use both values to > calculate it in the analog domain, which seems very complicated. > > This thigs are easily understood in the digital domain, but as you can see > they are not obvious. > > Thanks > > Greg > > This message was sent using the Comp.DSP web interface on > www.DSPRelated.com
Reply by ●October 18, 20052005-10-18
I'm not sure of your intention, but your mention of the method to get the complex signal' may be related to concepts such as an 'analytic' signal (one that is in fact complex) and a 'phase splitter'.>From a single analog filter you get a measure of the (filtered) realpart of the signal. Unless you in fact have two filters, each looking at some different phase (eg I,Q). I'll talk about the simple case. A real-valued signal has both positive and negative frequencies because you have not measured the phase. This is often not realistic - real-world signals may have only positive frequencies, the negative ones are due to your measurement (unless you are emsuring reltive to some carrier but let's not get too clever). This also makes it hard to do some things mathematically. For example if I shift the frequency up by some amount, then both negative and positive frequencies get shifted upwards which means some of the negative frequencies become positive and this is wrong. What I really want to do is, to get rid of the negative frequencies and then shift the (remaining) positive ones up. I can convert the real signal to a complex one by using a 'phase splitter' - which is a low-pass filter whose coefficients have been modulated with a complex exponential. This is a (digital) filter whose frequency response is not symmetric about zero - it in fact blocks negative frequencies. Filtering the signal with this (I think this may be what you are referring to?) gives you an 'analytic' signal - one that is complex, and that has only positive frequencies. So in some sense you have 'converted' a real-valued signal that was measured from the analog world, into a complex-valued signal that has no positive frequencies and so is in some sense a more 'realistic' (pun intended) signal. Chris =============== Chris Bore www.bores.com Gregwise wrote:> Hi all and thanks for your answer > > I am analizing (only) very short signals for which the ADC are too > expensive, then analog ones are necessary. My question may be too simple > for some of you, but it is really consuming my time. > > What do you get at the output of an analog filter? Is the magnitude of the > filtered signal? is the real part of the filtered signal? or dependes on > the load I use? In some paper, I saw they assume they get the real part of > the filtered signal, the in order to get imaginary one an extra filterign > block [1/s] is added, then the imaginary part is extracted. > > Well, if the previous is true, another question is how can I get the > magnitude of the filtered signal without having to use both values to > calculate it in the analog domain, which seems very complicated. > > This thigs are easily understood in the digital domain, but as you can see > they are not obvious. > > Thanks > > Greg > > This message was sent using the Comp.DSP web interface on > www.DSPRelated.com
Reply by ●October 18, 20052005-10-18
Gregwise wrote:> Hi all and thanks for your answer > > I am analizing (only) very short signals for which the ADC are too > expensive, then analog ones are necessary. My question may be too simple > for some of you, but it is really consuming my time. > > What do you get at the output of an analog filter? Is the magnitude of the > filtered signal? is the real part of the filtered signal? or dependes on > the load I use? In some paper, I saw they assume they get the real part of > the filtered signal, the in order to get imaginary one an extra filterign > block [1/s] is added, then the imaginary part is extracted. > > Well, if the previous is true, another question is how can I get the > magnitude of the filtered signal without having to use both values to > calculate it in the analog domain, which seems very complicated. > > This thigs are easily understood in the digital domain, but as you can see > they are not obvious. >I think we're lacking some detail about what you're doing, and you seem to be confused about some very basic stuff. First, there are simply no "imaginary" signals in this real world. While the name "imaginary" can be misleading it does mean "certainly not real", and that sense works here. I gather that you are talking about downconverting narrowband RF signals into inphase and quadrature parts, and treating the inphase as real and the quadrature as imaginary. This is a really good technique to use, but remember that imaginary numbers only exist in mathmagic land -- here in the real world they're quadrature and inphase and that's it. What you will get at the output of an analog filter is simply what you would get at the output of any other IIR filter -- a filtered version of the input signal. If you have a single-input single-output filter and you put a sine wave at 100MHz into it you will get a 100MHz sine wave out. If you need to find the magnitude of this sine wave you'll have to use some sort of magnitude detection scheme to do so. If you have a narrowband signal and you integrate it you'll change it's phase by 90 degrees which will give you a cheap and sleazy approximation of the quadrature signal. If the signal is narrowband enough for this to work then you don't need to do it -- you can just detect the magnitude of the straight signal by the means of your choice. Three popular ones -- square, average and root, envelope detection, and rectify and average all come to mind and are all quite doable with analog signal chains. So what signal do you have, what are you trying to do with it, and why? -- Tim Wescott Wescott Design Services http://www.wescottdesign.com
Reply by ●October 18, 20052005-10-18
Gregwise wrote: -snip-> > This thigs are easily understood in the digital domain, but as you can see > they are not obvious ->(in the analog domain)<-.Oh, I need to forward this statement to all the analog-only folks I know. -- Tim Wescott Wescott Design Services http://www.wescottdesign.com
Reply by ●October 18, 20052005-10-18
Gregwise wrote:> What do you get at the output of an analog filter? Is the magnitude of the > filtered signal?Depends on what you measure. If you are feeding it steady state sinusoids and using a meter then you are just getting magnitude: M=sqrt(Re^2+Im^2). If you have the ability to measure the phase shift as well, then you can get both complex components: Re=M*cos(P) and Im=M*sin(P). Bob -- "Things should be described as simply as possible, but no simpler." A. Einstein
Reply by ●October 18, 20052005-10-18
On Tue, 18 Oct 2005 10:26:21 -0700, Bob Cain wrote:> Gregwise wrote: > >> What do you get at the output of an analog filter? Is the magnitude of the >> filtered signal? > > Depends on what you measure. If you are feeding it steady > state sinusoids and using a meter then you are just getting > magnitude: M=sqrt(Re^2+Im^2). If you have the ability to > measure the phase shift as well, then you can get both > complex components: Re=M*cos(P) and Im=M*sin(P).I'm afraid that I missed Gregwise's original post, but I'm pretty sure that the answer to the question, above, is probably "the real part", unless the "analog filter" is non-linear and contains a rectifier at the end... -- Andrew
Reply by ●October 19, 20052005-10-19
Tim Wescott wrote:> Gregwise wrote: > > -snip- > >> >> This thigs are easily understood in the digital domain, but as you can >> see >> they are not obvious ->(in the analog domain)<-. > > > Oh, I need to forward this statement to all the analog-only folks I know.To me, they even more obvious with analog signals. It all depends on one's upbringing. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●October 19, 20052005-10-19
Jerry Avins wrote:> Tim Wescott wrote: > >> Gregwise wrote: >> >> -snip- >> >>> >>> This thigs are easily understood in the digital domain, but as you >>> can see >>> they are not obvious ->(in the analog domain)<-. >> >> >> >> Oh, I need to forward this statement to all the analog-only folks I know. > > > To me, they even more obvious with analog signals. It all depends on > one's upbringing. > > JerryI find that sometimes I'm more troubled by digital and sometimes more by analog, depending on the task. I just spend so much more time around folks who can't wrap their brains around the digital that seeing the statement in reverse is quite refreshing. -- Tim Wescott Wescott Design Services http://www.wescottdesign.com
