[HBR] HBR2K -- Chapter 14 -- Large Signal Performance, Part 4

[hbr@mailman.qth.net]

Mike Feher said:

> it sure seems to me that a 13 dBm IP3 is pretty darn good. I have yet
> to think about the adequacy, one way or another about the bottom end.
> Regardless, 13 dBm as you know is 20 milliwatts, measured in a
> wideband sense utilizing your techniques, and is essentially
> equivalent to the same power integrated over the bandwidth outside of
> your IF, and therefore is a heck of a lot of power. 

At the filter driver it's around 15 dbm and 82 db IFDR -- both good 
enough for this type of receiver.   The corresponding noise floor is 
close to -110 dbm -- something around a microvolt at 50 ohms.   So 
in an ideal world I'd get back to the antenna with noise-free and 
perfectly linear stages giving a gain of maybe 10 db, yielding an IP3 
of 5 dbm at the antenna and a noise floor of -120 dbm -- about a 
quarter of a microvolt.   That would be an acceptable performance, on 
a par with the usual benchmark hollow state receivers.

But when I do measurements at earlier stages it turns out that they 
are not noise free and perfectly linear:  IP3 at the 2nd mixer is still 13 
dbm but a noise floor of -84 dbm.   (IFDR 65db).   Going all the way 
back to the antenna, -7.5 dbm IP3 (no longer +) and an IFDR of 69 
db.   The corresponding NF is -111 dbm which is not all that great 
either -- Drake R-4C, -138, R-390, -137. 

The second mixer definitely contributes too much noise.   It may 
have other problems but until the noise is dealt with, they don't 
matter very much.   Probably not the mixer's fault -- it's probably 
getting 'garbage in' as noted in my earlier post.

The sharp eyed reader will notice that my numbers are slightly 
inconsistent -- the IFDR can never get better as you go backwards 
through a receiver, but I claim to have measured 65 db at the 2nd 
mixer and 69 db at the antenna.   This is at the outer limit of the 
usual error in my measurements (due to reading the S-meter, ripple 
in the filter passband ...) but might also represent some cancellation 
of distortion.

Basically the filter driver is fine and the filter itself probably okay -- it's 
handling a good fraction of a milliwatt without problems.  (I don't know 
what the max rated power level for those things is, but that certainly 
seems reasonable.)  Improvements in this area are only possible by 
improving the 1st IF stage noise figure and I have decided not to do 
that right now because the mixer stages are so bad.   

> An often made oversight is to look at the 1 dB compression point and
> equate it to the IP3 number. 

I don't think I've made that error yet, but I have made several similar 
ones.   I believe I now have the calculation right and the numbers do 
track and behave correctly, for example, the calculated IP3 is 
consistent across a range of measurement power levels and the 
distortion power does behave in a third-order fashion.

More to follow!

Walt Hutchens 
KJ4KV