[Amps] A tale of two IMs What happens?

[Dr. David Kirkby]

Carl wrote:
> 
>> Roger wrote:
>>> No one has touched the question on how reducing power without returning
>>> affects IM I now have another one.
>>>
>>> Given that most of todays transceivers have an IM or roughly -35db "so
>>> I've been told", and we put a amp behind it that also has an IM of
>>> -35db, what is the resultant IM?  What if the amp has an IM of -55db?
>>> Do they add, subtract, or go with the lowest number?  IOW it is the amps
>>> job to "faithfully" reproduce the input signal, but that really only
>>> happens when running class A, if the user is lucky.
>>>
>>> Intuitively, "I would think" that the two figures would add, but if that
>>> were the case the amp with -35db and the exciter with -35db would have a
>>> pretty ratty signal. OTOH in the case of the -55db amp behind the -35 db
>>> exciter does the amp "clean up" the exciter signal? Doesn't seem likely.
>> It is not likely to clean it up, but in theory if the phase relationships 
>> were
>> just right, you could completely cancel the products. But doing that at 
>> multiple
>> frequencies, for multiple amplitudes is just not going to be practical.
>>
>> The first thing to note is that even if the exciter only generated 3rd 
>> order
>> products, and the amplifier only generated 3rd order projects, the 
>> combination
>> would (relative to the presumed perfect input, produce both 3rd and 5th 
>> order
>> products.
>>
>> I believe an exact analysis of this would be very difficult, as the phases 
>> of
>> signals matters here - not just the amplitudes. So you can't just 
>> add/subtract
>> real numbers and expect to get exact answers. Also, the fact the exciter 
>> has
>> produced undesirable products, the amplifier will amplify those, as well 
>> as
>> generating others.
> 
> 
> The CATV industry answered those questions in published papers in the 70's 
> and early 80's using then available mainframe computers. The result showed 
> the contributions of cascaded line amps (up on the poles) and how many could 
> be cascaded and maintain FCC IMD specs. As the active devices in the amps 
> improved the distribution legs became longer. The head end equipment could 
> be considered as the "exciter" as it was all one way transmision.

You are basically saying what I thought - an exact analysis is non-trivial. You 
don't surprise me this has been done before, but it needs more than the ability 
to use logs and add/subtract a few numbers on a pocket calculator.

The use of the 'mainframe computers' suggests to me this was numerical modeling. 
  Was that so? I could imagine Monte Carlo techniques could solve this sort of 
problem.

Are any of the published papers available online free? I doubt many people are 
going to want to pay $30 or so to download an IEEE paper, but perhaps some would 
read them if available free.

> When I was designing CATV component based bi-directional data networks in 
> the mid 80's the IMD as well as phase delay even in passive components were 
> of importance. At the time only C-Cor was able to provide consistent quality 
> amps to the specs required.
> 
> Carl
> KM1H

Whilst not claiming to know much about this, I suspect for amateur purposes to 
assuming the worst of the exciter or the amplifier will not be too far from the 
truth, though I could believe that approximation will be less precise if the IMD 
of both the exciter and amplifier are similar. Does that sound reasonable from 
your understanding of it?


Dave