[R-390] OT: Problem with HP-606A linearity

[Barry]

> I've been working on an HP-606A for a while and have come up on something 
> I
> don't know how to correct.  The linearity is not good.  The specs say this
> thing should be within 1%, but I'm not getting that.
>
> I could get the endpoints on target, but as the frequency went from lowest
> to highest, there was a gradual increase in the output frequency, peaking 
> at
> a little over 4% at about 3/4 of the way up the dial and then dropping off
> rather sharply as it approached the upper endpoiint.
>
> Someone bent the fins outward on the tuning capaitor in an apparent 
> attempt
> to correct the linearity (or, perhaps more accurately, alignment with the
> dial markings); however, from what I could tell, this was the wrong thing 
> to
> do.  Since the frequency was too high at the "bad" points, then according 
> to
> the LC formula, there isn't enough C (or perhaps not enough L?) at that
> position.  At any rate, decreasing C didn't seem right and I straightened
> the fins.
>
> The linearity is better now (about 3% max.), but it's still on the high
> side.  Again, I assume this indicates eiher not enough C or not enough L 
> at
> these positions.
>
> Looking at the capacitor, it appears it is a custom-made job that was
> designed for a particular capacitance "curve".  The edges of the rotors 
> are
> not round but have flats along the way.  I presume this is to "shape" the
> capacitor to agree with "curve" so as to align with the dial markings (I
> hope that made sense).
>
> At any rate, it appears there isn't enough C at or near the midpoint of 
> the
> rotation so the output frequency doesn't match the dial all along its
> travel.
>
> Is there a common-knowledge way to get around this?  While I can always
> ignore the dial markings and use the frequency counter, I'd like it to be
> within spec if possible.
>
> The trimmer cap for the band I'm working on (530kc to 1800kc) is close to
> mid point (about halfway meshed) so that's a good thing.  I just don't 
> know
> what to do to make this thing "linear" again.
>
> I tried changing the position of the dial for this band by resetting the
> endpoints up or down 50kc.  In other words, 530kc to 1800kc became 500kc 
> to
> 1730kc.  This way, 1000kc was really 930kc, etc., but the problem didn't
> improve.  I tried going the other way (580kc to 1850kc) with similar
> results.  I was told there is a "sweet spot" where the dial plate could be
> repositioned and it might correct this problem but rather than move the 
> dial
> physically, moving the endpoints like this is effectively the same thing.
>
> I had thought about bending the fins on both ends of the capacitors, thus
> reducing the capacitance at both end points and effectively increasing the
> capacitance near the middle but this seems a bit of a far-out method.  Are
> there other things that might effect this?  Maybe someone on the list has
> had a similar problem with a 606A and could give me some Pearls of Wisdom?
>
> Thanks guys,
>
> Barry - N4BUQ

Just thought I'd post a follow-up to this.  I found a very badly botched job 
someone did of replacing a resistor in the plate circuit of the oscillator 
and finally got a replacement for it this week.  I was thinking the resistor 
someone had used might have some inductance that was upsetting the circuit 
so I replaced it with a carbon composition.  After replacing the resistor, I 
rechecked the linearity and it hadn't changed - at least not significantly.

Someone suggested that reorienting the dial in relation to the tuning cap is 
the correct way to do this.  I was skeptical since I thought that changing 
the endpoint frequencies was equivalent to "slipping the dial"; however, I 
gave it a try.  I was pleasantly surprised to find this had a very signicant 
effect on the linearity.  The first test got me very close and the second 
attemp (moving the dial a little more in the same direction) got me pretty 
much within spec.  Worst case is 1.1% (spec is +/- 1%).

I created an Excel spreadsheet with some charts that show the baseline 
deltas (the "before" condition), the changes after straightening one set of 
fins, then the other set of fins, and finally the curve after moving the 
dial.  Pretty interesting.  I think if I move the "center" point (the 
linearity goes a bit negative and then a little bit positive), I think I can 
get the entire dial within 1%.

Just thought I'd report back in.  Maybe someone else will have the same kind 
of issues in the future so this might serve as reference material (as it 
doesn't exactly apply to an R390).

Regards,
Barry - N4BUQ