[R-390] A vs non A

[n4buq]

I, for one, always look forward to reading Don's posts.  They are 
intelligent, informative, and usually written so that someone like me 
(who isn't an electronics engineer by trade) can understand.  Don can 
dive deeply on technical subjects without being stuffy or pretentious.

While he took a different slant on the subject than Roger, nowhere did I 
see any name calling or arrogance in his post.

Thanks, Don, for the information.  I appreciate it.  I think it might 
keep me from destroying the alignment in the IFs in my R390 as I would 
have probably thought the process was a bit more straightforward.

I would guess the IFs could be set a number of different ways and would 
sound "okay" (or not) but it appears there's a depth to this subject 
that needs to be well understood to do it correctly (or at least the way 
the engineers had in mind when they designed it).

Regards,
Barry

:
> Oh boy, Here we go again.
>
> The *ONE* that *NEVER* did this for a living for Uncle Sam is once again
> making proclamations that Roger is wrong.
>
> Does no one else find this arrogant?
>
> N0DGN
>
>
> On 10/13/2012 3:26 PM, 2002tii wrote:
>> Roger wrote:
>>
>>> R390 IF's are no more complex than the R390/A
>>> Either can be done very well with a simple signal generator and volt meter.
>>>
>>> The Sweep generator myth is in with moster gold plated Audio cable.
>>> Can I sell you some old bridge stock? I have original paper certificates.
>> Actually, if you want to adjust 390 IFs to factory alignment (linear
>> phase, maximally-flat, constant group delay), sweep alignment really
>> is a practical necessity.  The reason is that 390 IFs are not
>> strictly "stagger tuned," as that term is conventionally used by
>> filter designers.  Stagger tuning (strictly defined) refers to
>> broadening the passband of a multi-stage filter (like an IF) by
>> peaking various stages at slightly different frequencies.  Imagine
>> drawing a single, peaked response curve at the IF center frequency,
>> then drawing two more, offset to the left and right so that the upper
>> -3 dB point of each filter coincides with the lower -3 dB point of
>> the next higher filter.  Now, imagine the overall response of the
>> chain -- it will more or less follow the skirts of the upper and
>> lower offset filters, and the top will be a wavy line that averages
>> the three filter responses.  (I just used coinciding -3 dB points as
>> an example.  In practice, the offset frequencies could be chosen so
>> that the -1 dB, or -0.5 dB, or other close-in attenuation points
>> coincide, or so the -1 dB points of the offset filters coincide with
>> the -3 dB points of the center filter, or ... on and on.)  Some
>> people stagger tune 390A IFs to make sure the overall response is
>> wider than the widest mechanical filter, and instructions have been
>> published and are available for doing that.
>>
>> This "true" stagger tuned IF can be tuned (to a reasonable
>> approximation) with a sig gen and voltmeter, because each stage is
>> peaked at some frequency.  If you know what frequency each stage is
>> supposed to be tuned to, you can change the sig gen frequency to each
>> of these frequencies in turn and peak the appropriate
>> stage.  However, this only gets you to an approximation because it
>> assumes that the Q of each stage is the same as the design value.  If
>> it isn't (and it probably won't be, at least not to high precision),
>> you won't get the -3 dB points (or whatever alignment points it was
>> designed for) to exactly coincide, so the passband ripple will not be
>> optimized.  To optimize the passband ripple, sweep tuning is a
>> practical necessity even for a "true" stagger-tuned IF.
>>
>> Note the term "passband ripple."  Therein lies the limitation of
>> "true" stagger tuning -- it results in passband ripple, not a true
>> linear-phase, constant group delay response.
>>
>> The linear-phase, constant group delay response of the 390 IFs is not
>> achieved this way.  Rather, each stage is centered on the nominal IF
>> frequency, but some stage pairs are undercoupled for a peak response,
>> and others are overcoupled for two peaks with a valley at the center
>> frequency.  Perhaps this should be called "stagger coupled."  By
>> adjusting the coupling, a maximally flat, linear-phase, constant
>> group delay response can be achieved.  Because all of the filters are
>> tuned to the center frequency, and some are not peaked, but rather
>> overcoupled with two peaks and a valley, there can be no list of
>> frequencies where each stage should be peaked.  One might think that
>> you could instead specify the depth of the valley of the overcoupled
>> stages instead, and you can -- but you can only measure this with
>> those two stages in isolation, not with the IF strip in the radio,
>> because the response of the other stage couplings makes the overall
>> measurement worthless for tuning the individual stage pairs.  (Even
>> with the stages isolated, this would be very tedious with a sig gen
>> and voltmeter -- you would need to find and measure the
>> center-frequency response, find and measure each peak, adjust the
>> coupling, repeat, repeat, repeat, ....)
>>
>> So, one may call the 390 IF "stagger tuned" using that term loosely,
>> but it is not truly stagger-tuned in the sense that each stage is
>> peaked at a certain frequency.  And there is no practicable procedure
>> for tuning "stagger coupled" filters, like the 390 IF, without
>> sweeping them.  This is presumably why the 390A manuals caution you
>> not to undertake the procedure they provide unless the IFs are AFU
>> and you are desperate -- it will not return the IF strip to its
>> proper maximally flat, linear-phase, constant group delay response,
>> but it may get the radio back on the air.
>>
>> Best regards,
>>
>> Don
>>
>>
>> Copyright (c) 2012.  Not for redistribution
>>