[Elecraft] K3 related - non-linear audio transformer behavior

[Jack Smith]

Have you considered the difference between hitting a  transformer with a 
fast rise / fast fall square wave and hitting it with one that is 
narrowly band limited?

Ignoring for the moment the difference in filter shape between Gaussian 
and the complex shape of the K3 crystal and DSP filters, the rise/fall 
time of a perfect square wave fed into a Gaussian filter is 0.34/BW, or 
for a 1 KHz bandwidth, 340 microseconds.

I have not yet measured but hope to later today, the transformers with a 
slow rise/fall waveform.

A rather different story can be found, I suspect, in other audio 
applications where the source bandwidth may easily be 20 KHz or greater, 
yielding a rise/fall of 17 us or less. The slowest ringing rate I found 
was around 125 KHz, or 8 us for a period. I can see how that might be 
excited by a 17 us rise/fall square wave. I just don't see the same 
effect occurring with 340 us rise/fall "square wave." I use quotes 
around "square wave" because that leads to the image of a waveform with 
an abrupt rise/fall which is not the case where the bandwidth is limited 
to a few hundred Hz or a KHz at most.

That's why I question the applicability of a ringing test with a square 
wave having a rise/fall time measured in nanoseconds, as the audio that 
makes it through the K3's crystal and DSP will not remotely resemble 
that waveform.


Jack K8ZOA

Jim Brown wrote:
> On Sun, 21 Sep 2008 09:11:22 -0400, Paul Christensen wrote:
>
>   
>> the square-wave impulse response 
>> of the transformer becomes less meaningful.
>>     
>
> As long as there is impulse noise, the square wave response of the 
> line out transformer is definitely relevant! As the folks at 
> Elecraft have noted, the primary function of the Line Out is to 
> drive data decoders of one sort or another. Those decoders see 
> signal plus noise. Any distortion in the transformer is additional 
> noise as far as the decoder is concerned. The square wave response 
> of the circuit is simply another way of LOOKING at those 
> distortion products. 
>
> The analysis of the K3 Line Out that I did back in June using band 
> noise as a source of excitation show the same problems as Jack's 
> analysis using very different signal sources, but my excitation, 
> being broadband, exposes the IM distortion as apparent broadening 
> of the filter skirts. Jack measures IM using traditional two-tone 
> methods. In other words, we both see the same problems using very 
> different methods to study them. 
>
> Those of us working in pro audio learned long ago that broadband 
> noise (we use pink noise) is a VERY powerful analysis tool. 
>
> 73,
>
> Jim Brown K9YC
>
> Audio Systems Group, Inc. 
>
>
> _______________________________________________
> Elecraft mailing list
> Post to: Elecraft at mailman.qth.net
> You must be a subscriber to post to the list.
> Subscriber Info (Addr. Change, sub, unsub etc.):
>  http://mailman.qth.net/mailman/listinfo/elecraft    
>
> Help: http://mailman.qth.net/subscribers.htm
> Elecraft web page: http://www.elecraft.com
>
>