[HBR] HBR2K -- Chapter 14 -- Large Signal Performance, Part 6

[hbr@mailman.qth.net]

Jim, N2EY:
> I'm beginning to wonder if the performance of the HBR2K is actually much
> better than the test setup permits to be shown?

I think I have a pretty good handle on this now, although I have had a 
remarkably hard time learning that I have to deal with it each and 
every time I make a measurement.

The short answer (which won't, of course, satisfy me) is that some of 
HBR2K's performance is indeed better than I can measure.   (The 
performance at the filter driver may be significantly better than 84.5 
db IFDR the measurements give.)   However, there are work-arounds 
that will let progress continue, I may be able to lift the measurement 
limits, and in any case I expect to be able to correctly measure the 
end result.

And the longer answer ...

After some study I have concluded that the URM-25's are pretty 
decent signal generators for this job, because:

1.  They're *powerful* -- the oscillator (tube type varies from one 
model to another) drives a 6AH6 buffer which drives a 6AG7 output 
stage.   Yeah, that's not a typo -- the output tube would work fine in a 
10 watt class transmitter.   The 'final' stage output is around 2 volts 
RMS (they're not exactly pushing it hard); at the built-in attenuator 
there's about 0.1 volts RMS.  

2.  They have pretty good isolation -- there's a 20 db fixed pad 
between the output tube and the attenuator and there's nothing that 
could be non-linear after that pad.

3.  They have no ALC to cause crossmodulation when power from 
the other generator comes in at the output port.

Now just like the QST author, I'm limited by the maximum output, but 
either his receiver was considerably better than mine -- okay, so I 
hereby commit to finding that article -- or he was testing by pushing 
his into compression rather than measuring IP3.  The measurements 
are equivalent for comparing receivers (a good compression limit or a 
good IP3 is equally descriptive of a good strong signal receiver) and 
IP3 can be measured at lower power levels -- I am now doing it at 13 
db below the 100,000 uV maximum output, or something like 20,000 
uV.

And I am also doing two measurements at different power levels 
whenever I want a serious number.   If the result at the higher level is 
better than the one at the lower level then generator (or hybrid 
combiner) distortion is limiting and the high power number is a lower 
limit for the actual performance.   

FWIW, I'm finding that just as was the case at the filter driver, the 
test setup is limiting at the 2nd mixer.  That's not a serious problem 
for same-power level 'is this modification an improvement' tests but 
does mean that absolute numbers are not meaningful.   

And the numbers at the antenna *are* trustworthy -- that is, at 
various test power levels I get the same IP3, as the theory says I 
should.   Combining that with the noise floor measured at that point, I 
get a trustworthy but less-than-satisfactory 70-72 IFDR.   And my 
guess is that when the mixer optimization is done (not necessarily 
just tweaking resistors -- I may have to redesign the 1st oscillator!) 
and (after that) when the RF stage is tweaked,  I'll be also able to 
measure the final number at the antenna in a trustworthy fashion.

The reason I say that is that the adequacy of the test setup is a 
function of how much signal it has to furnish -- smaller test demands 
are more easily supplied.  As I get closer to the antenna, that 
number goes down.

In the next few days I'll test my combiner, which is the *handbook* 
device.   (Should be easy  -- just compare third harmonic output from 
the combiner vs. a 6 db attenuator, using a general coverage 
receiver.)    If it's not good I'll either improve it or look for a 
commercial combiner.  

Also, it ought to be possible to calculate the correct value of IP3, 
given two or more of the incorrect values.    

Again, I commend the comments by John Thorpe, designer of the 
AOR 7030, at:

http://www.aoruk.com/comments.htm

This is not a mathematical treatment nor in any way unreadable for a 
ham who works on receivers and has thought some about theory; 
heck, it's even fun reading -- he has a great sense of humor.   But it's 
the soundest and clearest exposition of methods and issues in 
strong-signal testing of high performance receivers that I've seen -- 
far, far more useful than the stuff in our *Handbook*.

And a final word (again) about why this long strong-signal exercise is 
worthwhile.   Way back at the start I picked the target of showing 
that far better receivers *could* have been built at the end of the 
vacuum tube era than ever *were* produced -- basically, I believe that 
for marketing reasons manufacturers jumped for the solid state 
bandwagon while square wheels were still the solid state of the art.   
Beating the usual 1970 receiver's stability and audio quality has been 
easy to do.  But the real test of a receiver is 'what can you hear when 
the going gets rough' and that is largely determined by internal noise 
and intermodulation distortion.  

This phase has been slow going because I've been able to copy very 
little circuitry intact and have had to learn most of the 'how to' from 
near the beginning.   I believe it would have been a straightforward 
exercise for a good professional in 1970 -- say as a development 
contract for an R-390B or an Art Collins demand for a 75S4.   And 
the commercial/military and/or ham world would have then had the 
choice of a receiver that would not have been bested by the solid 
state sets for 10-15 years -- I don't know exactly. 

Thanks for the comments Jim, Mike, others -- they are appreciated.  
I often clarify my thinking or get ideas while reading or trying to 
answer.  

Back to the bench -- actually, the kitchen counter.

Walt Hutchens
KJ4KV