[ARC5] Smart People: IF Curve- What Am I Doing Wrong?

[David Stinson]

I've aligned hundreds of IFs and everything worked fine.
But trying to draw a curve for the ARC-2 IF is "strange."
Now, I don't believe what my figures are telling me.

Here's the "set-up:"
With a "fixed" IF, I'd align at the IF frequency.
However, the ARC-2 has a variable IF.
The manual procedure calls for aligning it with a Fo 
input.   The procedure says to set the rig on the 
lowest band at Fo 2.1 MC, AM mode, set the 
modulated signal generator level well below 
AVC action, meter the audio output and follow 
a step-by-step adjustment list for the "Main" slugs.
Then move Fo to 2.9 MC and do the same for the 
"Trimmer" slugs.  I do all that, except I use a scope 
on the output rather than an audio level meter.

I don't have a sweep generator, network analyzer or 
Star Trek Tricorder, nor am I likely to get any of these.
To plot the curve, I set the rig Fo to 3.890 MC and
leave it there.  Using an external freq standard 
(Icom R-70) to vary the signal generator frequency,
I move the sig gen 1 KC at a time and note signal level 
on the scope, going  +/- 20 KC.

Here's the problem:  My last attempt at "tightening" 
the IF bandwidth was to replace the 5 pFd IF 
coupling caps with 3 pFd and re-align.
After doing so, the curve looked only slightly better, 
so I left the 3 pFd in place and  buttoned-up the rig.  

I turned it on this weekend to listen to some AM.
Before, AM sounded sweet when the band wasn't 
crowded.  But now, it sounded "pinched;" just as you'd
expect when using a "too-narrow" filters, yet the curve
showed no real change in bandwidth.
I put the 5 pFd caps back in service and the sweet,
broad AM audio is back.

OK, Smart People:  What's the really obvious mistake
that I'm missing here?

73 DE Dave AB5S