[R-390] Filter foolishness

[Bob Camp]

Hi

The loaded Q of the driver coils themselves is far from the whole story. The Q on each end of the test circuit in the filter spec is 1/2 of the Q of the drivers alone. It's still significant if you want to hit your filter passband numbers. Adobe is going a bit nuts at the moment so I can't open my copy of the Y2K manual. Wonderfully buggy software… off to paper copy land. 

There's no obvious loading on either end of the filters. The in circuit Q's of the drivers could be anywhere between 33 and 16. Vacuum tube grids rarely are low impedance, so I 'd bet that at least one end is lightly loaded. 

The K/Q values for fancy IF transformer sets were a design variable. They generally did a pretty good job with graphical tools to get the design done. Of course most IF strips are pretty simple. 

Mechanical or crystal filers were a very different thing. There big rooms full of mechanical four function calculators were the norm. Parallel rooms doing the calc's and cross checks every few hours to catch the inevitable errors. Needless to say they spent a lot of time on any one design. 

The best way to look at most IF strips is as a "synchronously tuned filter". "Handbook of Filter Synthesis" by Anatol I Zverev ($93 on Amazon) is the classic reference on a lot of this good old stuff. Zverev was working for Westinghouse at the same time Collins was designing some of this stuff. One way to look at a synchronous filter is as a K/Q filter where none of the sections are over coupled. In some over coupled filters, you de-q the sections to it into a synchronous filter for alignment. 

If you want to do a full computer simulation of the IF strip, Linear Tech's LT-SPICE program is free. It's quite capable of doing a full run up of the 455 KHz strip. You can indeed do the tubes. I'd probably keep it simpler than that.

Bob


On Feb 18, 2013, at 10:45 AM, Tisha Hayes <tisha.hayes at gmail.com> wrote:

> Quote;
> Bob Wrote:
> "The loaded Q of the driver coils is indeed in the vicinity of 33 (input
> resistance 100K ohms, resonating cap 120 pf, 120 pf = 2916 j ohms at 455
> KHz). Since the 3 db bandwidth is Fc / Q you get 455 / 33 = 13.8 KHz.
> That's just for the driver coil on one end. The Q of the coil it's self,
> unloaded by the filter is significantly higher.
> 
> In order to be "much wider" than 16 KHz, you might want 160 KHz on both
> ends. 455/160 gives you a Q of 2.8. For an input resistance of 100K ohms,
> that would be a resonating capacitance of ~ 10 pf. In order to hit
> something like that you would need to tune the input to each filter. That's
> not something the Collins guys wanted to do. Eight more trimmers cost
> money, and take up space.
> 
> Bottom line - yes the coils do get into the bandwidth of the IF filters to
> a greater or lesser degree. The response is a composite of the driver coils
> and the disks. You don't want to do anything dramatic to those coils. You
> will certainly mess up the passband of the wider filters if you do. The
> whole filter was calculated / designed with the impact of the driver coil
> Q's taken into account."
> -----------------------------------------------------------------------------
> 
> Yes, that is pretty close to what I was seeing as well, it was a fun
> exercise to work through. If you played around with significant coil
> variations for the different filters you are also messing with the
> insertion loss and might be blasting the headphones off of your head
> whenever you turned the bandwidth switch. That would be if you were looking
> at a different set of specifications for each individual filter but would
> make no sense for a family of filters that should be more or less matched
> to each other.
> 
> I did not believe the Q values at first and had not considered that when
> first looking at things. The software tool I was using gave me the value
> but I decided to not emphasize that in what I wrote. Also my values for "k"
> on calculating mutually opposed impedance of the coils was extrapolated
> (wing and a prayer) and I was unsure how to consider the magneto-mechanical
> impact upon the magic wire for the center cheekpiece spacing and coil
> length (the coils could be compressed into a smaller dimension for a
> greater flux density).
> 
> I was intrigued by their choice of wire gauge, turns ratio and coil
> spacing. With modern computing methods we could run through in an afternoon
> what would of taken a room full of people several weeks to calculate by
> hand. It just increased my admiration of what they did on what is such a
> small device inside of the receiver. I had to go crack open a few books on
> acoustics and speaker design to get a better understanding of the magic
> wire and the dancing disks.
> 
> Tisha
> ______________________________________________________________
> R-390 mailing list
> Home: http://mailman.qth.net/mailman/listinfo/r-390
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:R-390 at mailman.qth.net
> 
> This list hosted by: http://www.qsl.net
> Please help support this email list: http://www.qsl.net/donate.html