Reply by Fred Marshall●February 24, 20052005-02-24
"Fred Marshall" <fmarshallx@remove_the_x.acm.org> wrote in message
news:0fednQVKW-v_DoPfRVn-og@centurytel.net...
>
> "Max Hauser" <maxREMOVE@THIStdl.com> wrote in message
> news:111oecet5h9b407@corp.supernews.com...
>> "Allan Herriman"
>>>
>>
>> Neglecting novelty items like recursive FIR implementations, there are
>> other famous nonrecursive structures that also realize Finite Impulse
>> Responses but lack the transversal (tapped-delay) structure. A parallel
>> set of delay lines of different lengths or delays, each weighted at their
>> input or output, and finally summed, is such a structure. (It has been
>> useful in some delay-line technologies that had constraints, for example
>> certain CCD technologies circa the 1970s.)
>>
>> -- Max Hauser
>
> In the olden days when delay lines of any appreciable length were
> expensive we used a relatively wideband torsional wire acoustic delay line
> over the range 100kHz to 1MHz. The line wasn't tapped. We wanted to
> delay signals up to 100kHz or so and create a FIR filter sort of response.
> So, we used frequency domain multiplexing on the line and used the line
> multiple times:
>
> output at baseband at t
> first pass at 150kHz downtranslate a tap out at t+T
> uptranslate for:
> second pass at 300kHz, downtranslate a tap out at t+2T
> upstranslate for:
> third pass at 450kHz, downtranslate a tap out at t+3T
>
> End result, one delay element of length T for a FIR filter of length 4.
> So, in a way, these were different length delay lines in parallel even
> though the physical line was the same one.
> Note that this isn't recursive because all the signal paths are
> independent / not added - well they are added in the delay line but
> separated in frequency so the end result is not recursive.
>
> Fred
Well, on second thought, it seems that this implementation was more serial
than parallel.
Fred
Reply by Fred Marshall●February 24, 20052005-02-24
"Max Hauser" <maxREMOVE@THIStdl.com> wrote in message
news:111oecet5h9b407@corp.supernews.com...
> "Allan Herriman"
>>
>
> Neglecting novelty items like recursive FIR implementations, there are
> other famous nonrecursive structures that also realize Finite Impulse
> Responses but lack the transversal (tapped-delay) structure. A parallel
> set of delay lines of different lengths or delays, each weighted at their
> input or output, and finally summed, is such a structure. (It has been
> useful in some delay-line technologies that had constraints, for example
> certain CCD technologies circa the 1970s.)
>
> -- Max Hauser
In the olden days when delay lines of any appreciable length were expensive
we used a relatively wideband torsional wire acoustic delay line over the
range 100kHz to 1MHz. The line wasn't tapped. We wanted to delay signals
up to 100kHz or so and create a FIR filter sort of response. So, we used
frequency domain multiplexing on the line and used the line multiple times:
output at baseband at t
first pass at 150kHz downtranslate a tap out at t+T
uptranslate for:
second pass at 300kHz, downtranslate a tap out at t+2T
upstranslate for:
third pass at 450kHz, downtranslate a tap out at t+3T
End result, one delay element of length T for a FIR filter of length 4.
So, in a way, these were different length delay lines in parallel even
though the physical line was the same one.
Note that this isn't recursive because all the signal paths are independent
/ not added - well they are added in the delay line but separated in
frequency so the end result is not recursive.
Fred
Reply by Fred Marshall●February 24, 20052005-02-24
"Max Hauser" <maxREMOVE@THIStdl.com> wrote in message
news:111oecet5h9b407@corp.supernews.com...
> "Allan Herriman"
>>
>.......................A parallel set of delay lines of different lengths
>or delays, each weighted at their input or output, and finally summed, is
>such a structure. (It has been useful in some delay-line technologies that
>had constraints, for example certain CCD technologies circa the 1970s.)
>
> -- Max Hauser
Max,
A parallel set of delay lines *is* transversal.....
Transversal doesn't say anything about "serial" only about "across" which
the delay lines of different length accomplish.
Fred
Reply by hugo●February 24, 20052005-02-24
Hello Steve, Max, Allan, Fred, Randy, Jerry,
Thank you for all the feedback. The answer could not be clearer!
Regards,
Hugo
Reply by Steve Underwood●February 24, 20052005-02-24
Jerry Avins wrote:
> Allan Herriman wrote:
>
>> On Wed, 23 Feb 2005 00:52:57 -0800, "Max Hauser"
>> <maxREMOVE@THIStdl.com> wrote:
>>
>>
>>> "Fred Marshall" in news:a9ednW1bu_E7jYTfRVn-iA@centurytel.net...
>>>
>>>> ...
>>>> Before there were truly discrete, digital filters, there were
>>>> continuous time or analog filters (such as delay lines) that were
>>>> "tapped" at discrete points.
>>>
>>>
>>> Not just _before_ digital filers by the way, but also after. Here
>>> are some modern, monolithic continuous-time FIR examples. (Click on
>>> "Images" for the figures.)
>>>
>>> http://tinyurl.com/548jv
>>>
>>> (Remember: Not all FIR filters are transversal. Not all FIR
>>> filters are digital. Not all non-digital FIR filters are even
>>> discrete-time.)
>>
>>
>>
>>
>> A SAW (Surface Acoustic Wave) filter is another good example of a
>> continuous time transversal filter.
>>
>> The surface acoustic waves are launched by "interdigital transducers"
>> which are just interleaved electrodes deposited on the surface of the
>> quartz. The gain (& hence the weight of that 'tap') is proportional
>> to the overlap between the electrodes.
>>
>> http://koigakubo.hitachi.co.jp/~cs/cd/eng/technical/device/saw/
>>
>> Regards,
>> Allan
>
>
> SAWs are frequency selective depending on the spacing of the
> electrodes relative to the speed of the wave. If a pulse is launched
> down the substrate and under a set of electrodes with progressively
> decreasing spacing, a chirp is produced. When that chirp is applied to
> to the variable-pitch electrode of an identical device, pulses emerge
> from its other electrode. Even if the chirps overlap in time, the
> pulses can be distinct. There are several very useful applications.
>
> Jerry
Many pulse radars used to use that technique for generating and
recompressing their chirps. It offers a somewhat limited dynamic range,
though, as all SAW devices do. Bulk waves and other side effects of the
generation of the surface wave are a key cause. Another is the fact that
things don't stop abruptly at the edges of the inter-digital
transducers. SAW devices can generally be characterised as reproducable,
predictable, stable, but never really great performers.
This is getting away from the original issue. The type of SAW device
Allan referred to is an exact analogue of a conventional DSP FIR filter,
and the same design methods apply to both. The only difference is, the
DSP version doesn't require fiddle factors to compensate for the effects
around the edges of the transducers.
Regards,
Steve
Reply by Jerry Avins●February 23, 20052005-02-23
Allan Herriman wrote:
> On Wed, 23 Feb 2005 00:52:57 -0800, "Max Hauser"
> <maxREMOVE@THIStdl.com> wrote:
>
>
>>"Fred Marshall" in news:a9ednW1bu_E7jYTfRVn-iA@centurytel.net...
>>
>>>...
>>>Before there were truly discrete, digital filters, there were continuous
>>>time or analog filters (such as delay lines) that were "tapped" at
>>>discrete points.
>>
>>Not just _before_ digital filers by the way, but also after. Here are some
>>modern, monolithic continuous-time FIR examples. (Click on "Images" for the
>>figures.)
>>
>>http://tinyurl.com/548jv
>>
>>(Remember: Not all FIR filters are transversal. Not all FIR filters are
>>digital. Not all non-digital FIR filters are even discrete-time.)
>
>
>
> A SAW (Surface Acoustic Wave) filter is another good example of a
> continuous time transversal filter.
>
> The surface acoustic waves are launched by "interdigital transducers"
> which are just interleaved electrodes deposited on the surface of the
> quartz. The gain (& hence the weight of that 'tap') is proportional
> to the overlap between the electrodes.
>
> http://koigakubo.hitachi.co.jp/~cs/cd/eng/technical/device/saw/
>
> Regards,
> Allan
SAWs are frequency selective depending on the spacing of the electrodes
relative to the speed of the wave. If a pulse is launched down the
substrate and under a set of electrodes with progressively decreasing
spacing, a chirp is produced. When that chirp is applied to to the
variable-pitch electrode of an identical device, pulses emerge from its
other electrode. Even if the chirps overlap in time, the pulses can be
distinct. There are several very useful applications.
Jerry
--
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������
Reply by Allan Herriman●February 23, 20052005-02-23
On Wed, 23 Feb 2005 00:52:57 -0800, "Max Hauser"
<maxREMOVE@THIStdl.com> wrote:
>"Fred Marshall" in news:a9ednW1bu_E7jYTfRVn-iA@centurytel.net...
>> ...
>> Before there were truly discrete, digital filters, there were continuous
>> time or analog filters (such as delay lines) that were "tapped" at
>> discrete points.
>
>Not just _before_ digital filers by the way, but also after. Here are some
>modern, monolithic continuous-time FIR examples. (Click on "Images" for the
>figures.)
>
>http://tinyurl.com/548jv
>
>(Remember: Not all FIR filters are transversal. Not all FIR filters are
>digital. Not all non-digital FIR filters are even discrete-time.)
A SAW (Surface Acoustic Wave) filter is another good example of a
continuous time transversal filter.
The surface acoustic waves are launched by "interdigital transducers"
which are just interleaved electrodes deposited on the surface of the
quartz. The gain (& hence the weight of that 'tap') is proportional
to the overlap between the electrodes.
http://koigakubo.hitachi.co.jp/~cs/cd/eng/technical/device/saw/
Regards,
Allan
Reply by Max Hauser●February 23, 20052005-02-23
"Fred Marshall" in news:a9ednW1bu_E7jYTfRVn-iA@centurytel.net...
> ...
> Before there were truly discrete, digital filters, there were continuous
> time or analog filters (such as delay lines) that were "tapped" at
> discrete points.
Not just _before_ digital filers by the way, but also after. Here are some
modern, monolithic continuous-time FIR examples. (Click on "Images" for the
figures.)
http://tinyurl.com/548jv
(Remember: Not all FIR filters are transversal. Not all FIR filters are
digital. Not all non-digital FIR filters are even discrete-time.)
--------
"To converse at the distance of the Indes by means of sympathetic
contrivances may be as natural to future times as to us is a literary
correspondance." -- Joseph Glanvill, 1661 [as quoted by Scott Dorsey,
news:4723@pyr.gatech.edu, rec.audio, 1988]
Reply by Max Hauser●February 23, 20052005-02-23
"Allan Herriman"
>
> I assume that 'transversal' is a property of a filter implementation,
> whereas 'FIR' is a property of the impulse response of a filter.
>
I thought that a good summary, and in accord with what much of the
literature has used in the past 40 years or so.
Neglecting novelty items like recursive FIR implementations, there are other
famous nonrecursive structures that also realize Finite Impulse Responses
but lack the transversal (tapped-delay) structure. A parallel set of delay
lines of different lengths or delays, each weighted at their input or
output, and finally summed, is such a structure. (It has been useful in
some delay-line technologies that had constraints, for example certain CCD
technologies circa the 1970s.)
-- Max Hauser
Reply by Fred Marshall●February 22, 20052005-02-22
Fred said:
>>OK - if we include recursively implemented FIRs then they aren't
>>transversal.
>>But, that's a special case that only applies to a small subset of FIRs.
>>Correct but limited. I like to think of the recursive implementation of
>>FIRs as sort of a curiosity with limited application - no matter how cool
>>and even valuable some of those implementations may be. Let's not have
>>the
>>tail wag the dog.
>>
>>What if we turn it around and say:
>>"In some cases a FIR can be implemented recursively and then isn't a
>>transversal filter. However, a FIR filter can *always* be implemented as
>>a
>>sum of delayed inputs and thus, can always be transversal" ??
>>In that sense, FIR can very reasonably imply transversal.
>
Allan said:
> They never should have stopped teaching logic in schools.
>
> Allan
Oh geez, now I have to construct a syllogism for testing.... ??
I agree with what you said. Thanks for the clarification - it took me a
while.
Fred