[HBR] HBR2K -- Chapter 14 -- Large Signal Performance, Part 3

[hbr@mailman.qth.net]

(In previous parts I focused on trying to understand and improve the 
IMD dynamic range at the input to the crystal filters leading to a need 
to increase the IF gain.)

Measurements on 1st 3180 kcs IF stage showed a noise floor (10 db 
s+n/n) of 0.3 uV and an AGC threshold of 4 uV.   Certainly a 
respectable noise performance however, we need to do better than 
'respectable.'   Since the stage was driven directly from the 500 ohm 
filter, the first thing to try was impedence matching.   

After a week's work and several dead ends I had a 1:8 toroidal 
transformer with a tuned secondary.  Now *that* improved sensitivity -
- or was it just pickup of things I didn't want to hear in the new high 
impedence grid circuit?   It took a day to shield the carrier oscillators 
(on the other side of the chassis but still close enough to radiate into 
that tuned grid circuit), re-route all non-essential wiring away from the 
stage, and put a tin sheld across the 1st IF socket.   That fixed the 
spurious pickup except for one very strong signal that comes in in 
the evening; a tin shield over the grid components takes care of that.

The noise floor improved to 0.05 uV  (not very accurate!) with an AGC 
threshold of 1 uV.   That's more like it -- should be plenty of gain.  
But the IMD dynamic range at the filter input was now 61 db -- 
certainly no better than the 62 db or so before the sensitivity 
improvement.   Lowering the amplitude of the signals going through 
the filter should allow a *higher* ratio because the 3rd order IMD 
spurious signal rises or falls as the third power of the input.   How 
could there be no improvement?

I had been using two URM-25D's to do the tests.   I *thought* they 
were okay, but ... Thinking some about the problem of 
crossmodulation between the generators (which would mimic IMD 
occuring in the receiver) I came up with an easy test.  I set the 
carrier level to the maximum available without visible distortion; this 
turned out to be about 0.6 volts p-p.   Then I turned the two output 
attenuators down one step, thereby adding 40 db more generator-to-
generator isolation to the 30-40 db provided by the hybrid combiner.   
The result was still enough signal to do a dynamic range test and -- 
voila!  Doing the same test that had produced the 61 db result, I got 
74 db.   

So (a) the filter and driver section is good enough for now, and (b) the 
signal generators are a problem.   They do not limit measurements 
significantly below 74 db so I can use them to try to improve the 55 
db ratio set by the second mixer but to actually *measure* dynamic 
ranges 74 db and above (when the second mixer problems are 
solved) I'll have to come up with better instruments.  

Basically I need two oscillators capable of a sizeable fraction of a volt 
output, pure sine wave, and immune to comparable signals being 
injected at the output.   Fancier signal generators aren't necessarily 
better for this job; the HP8640 is considered poor.  John Thorpe (in 
the reference cited last time) discusses the problems.   

There's no obvious reason the URM-25's shouldn't work.   There's 
some waveform distortion at extremely high (above spec) signal 
levels due to the cathode follower output but since they lack ALC 
(you set the level by hand at each frequency) the problem found in 
fancier units simply can't happen.   But since they don't work well 
more analysis is needed ... maybe the diode circuit used to drive the 
output meter?  Maybe the output stage itself?    I am trying to 
download and print a BAMA manual but the software needed for the 
new space-saving format seems not to support printing if you run 
WIN 95.

The 3180kcs IF gain increase was still a good idea.   The road ahead 
includes reducing distortion in the second mixer as much as 
possible; some gain may be lost there.   Then substantially reducing 
the max gain of the RF stage -- maybe to the 6 db range or so, from 
the current 20 db.   Definitely couldn't do the latter without the higher 
IF gain.

Then another light went on again ... I remembered I had a specialized 
military test set from some long-forgotten hamfest.   I had never done 
anything with it but I remembered the manual saying something 
about testing of SSB receivers and transmitters.   Retrieved the unit 
from the archives and, wow, it is a URM-144, designed specifically to 
do distortion testing with two RF signals.   Crystal controlled on 2, 
3.6, 4, 7.2, 8 ... 28.8 Mcs.    It consists of two solid state Butler 
variable-freq. crystal oscillators with buffer amps and/or three 
doublers switched in as necessary to reach the output freq.   The 
combiner is built in, as is a 120 db step attenuator.   Unfortunately 
the frequency separation is only a couple of kcs at the lowest 
frequency so only the highest bands would allow tests at 20 kcs 
spacing.   There is some frequency adjustment available but not 
enough to make it work.

However, substituting a 3.579 Mcs clock crystal for one of the 3.6 
Mcs crystals gave a spacing of about 25 kcs and I was able to 
measure an IMD dynamic range at the antenna jack of the HBR2K of 
60 db at that frequency.   That's consistant with the 50's numbers 
from before, considering the greater spacing (which allows pre-crystal 
filter selectivity to do more) and testing on 80 rather than 40 meters.

This set is intended for pass-fail testing of in-service military gear; 
closer frequency spacing tests a radio all the way to the detector.   
However, for receiver development work, 20 kcs is more useful 
because it lets us focus on designed-in distortion ahead of the 
narrow filter.  The set could be modified so the bandswitch selects 
offset crystals to give a frequency difference of 20 kcs.    All 
frequencies start with 2 Mcs or 3.6 Mcs so if you used 2.020, 3.620, 
(two fundamental frequencies), 2.010, 3.610 (for 4 Mcs and 7.2 Mcs), 
and so on, you'd have it.  Expen$ive, but just doing the 3.6 Mcs 
crystals would take care of all the ham bands.

Distortion is spec'ed at 60 db below output rather than in terms of 
IMD range.   If I knew what gear it was intended to be used with I 
probably could find out what it's intended to measure -- hafta look 
into that. 

Anyway I have what I think is a reliable way to measure IMD for the 
whole receiver and a technique that will allow using the URM-25's to 
test later stages on other frequencies.   

This ought to be a lot more than one would want to learn about 
receiver design, but I'm actually finding it pretty interesting.

Walt Hutchens
KJ4KV