[Amps] A tale of two IMs What happens?

[John Wilson]

A simple way of looking at this problem is to consult an oracle and work 
out the combined third order intercept point (IP3) of the two (or more) 
stages. The equation for the ip3 of two cascaded stages is;

1/ip3_tot = 1/ip3_1 + g_1 /ip3_2

The oracle in question is; “Communications receivers – principles and 
design” by Ulrich L Rohde and T T N Bucher. Unfortunately I haven't got 
it to hand at the moment to consult so I can't tell you the caveats.

Putting some figures into this, if the IP3 of the first stage is 10 dBm 
and the second stage has a gain of 10 dB and an IP3 of +20 dBm then the 
overal IP3 is about +7 dBm. Probably about what you would expect!

73 John G3UUT


Roger wrote:
> Carl wrote:
>   
>>   
>>     
>>> Jim Thomson wrote:
>>>     
>>>       
>>>> ###  per ssb  systems and circuits, et all... the consensus was/is
>>>> that if 2 x IDENTICAL.. IMD  circuits are in series...  say a TX.. with
>>>> -40db imd3... driving a linear amp also with  -40db imd3... the
>>>> author's claim the worse case you would get is -34db imd3.
>>>> The best case you could get achieve is -37db  to -38db IMD3.
>>>>       
>>>>         
>>> There are lots of claims, but I have not seen anybody post anything to
>>> substantiate these claims.
>>>
>>> I think it is a quite complex issue. I suspect an exact analysis is 
>>> possible,
>>> which would put absolute limits on the parameters, but it would not be 
>>> trivial.
>>>
>>> The fact that the 3rd order products of one device makes higher order 
>>> products
>>> in another, means you can't simply add powers.
>>>
>>>     
>>>       
>>>> If you have to get a signal over a very long distance using cables, then 
>>>> the
>>>> losses of the cables obviously become significant, and you might need to
>>>> have many amplifiers in series. So the source is connected to the load 
>>>> like
>>>> this:
>>>>
>>>> source -> CL -> A -> CL -> A -> CL -> A -> CL -> A -> load
>>>>
>>>> where:
>>>> CL = 20 dB of cable loss
>>>> A = 20 dB gain amplifier
>>>>
>>>> ## IF I remember correctly, the gain of the CATV amps was much
>>>> greater than the cable loss.  I THINK the input of each successive
>>>> CATV amp was padded down.  Sorta like  25 db gain - 20 db CL -
>>>> then 5db padding - then CATV amp, then another 20db cable loss, etc.
>>>> The CATV amps all used equalizers too, since the cable attenuation
>>>> was not uniform... but rises on the higher freqs.  IE: pre-emphasis
>>>> employed, whereby higher freqs get amplified far mote than lower
>>>> freqs.  Also known as slope equalizer's.
>>>>       
>>>>         
>>> Well, irrespective of whether there is a pad or cable loss, or 
>>> pre-emphasis, the
>>> fact remains the situation is completely different from amateur radio, 
>>> where the
>>> aim of the amplifier is to produce a bigger signal, rather than to just 
>>> overcome
>>> losses. As such, the analysis from the CATV book seems totally 
>>> inapplicable to me.
>>>
>>>
>>> Dave
>>>     
>>>       
>> I consider the CATV amp analogy a good way to understand what is happening 
>> in a receiver. There is no need to cascade amps, simply add xx number of 
>> sources into a single amp which becomes the radios front end. The 2 signal 
>> IMD method of testing an amp or a radio is equally useless except for 
>> marketing.
>>   
>>     
> My background in this topic is a bit weak, but as I see it:
> All we can deal with are analogies as the real world is dynamic and any 
> testing such as the two tone is a bit outdated. OTOH how would we 
> simulate the equivalent real world  where the signal is voice  which is 
> almost random and contains many  frequencies at once that are constantly 
> changing along with changes in the voltages on the tube?  I'd think even 
> small changes would show up as some form of modulation, superimposed on 
> the intended signal which would be a source for  additional  IM.   CATV 
> is at least a signal containing dynamic components and the results are 
> based on real world conditions albeit a bit different than the ones we 
> are working with.
>
> 73
>
> Roger (K8RI)
>   
>> Carl
>> KM1H
>>
>> _______________________________________________
>> Amps mailing list
>> Amps at contesting.com
>> http://lists.contesting.com/mailman/listinfo/amps
>>
>>   
>>     
> _______________________________________________
> Amps mailing list
> Amps at contesting.com
> http://lists.contesting.com/mailman/listinfo/amps
>
>