[sdiy] BODE Frequency Shifter

David G Dixon dixon at mail.ubc.ca
Thu Nov 1 00:19:10 CET 2018


I'm very happy to help, Atsushi Maeda!

(Even if you are going to be selling your build as a "commercial product.")

> -----Original Message-----
> From: Synth-diy [mailto:synth-diy-bounces at synth-diy.org] On 
> Behalf Of atsushi maeda
> Sent: Wednesday, October 31, 2018 3:19 PM
> To: synth-diy at synth-diy.org
> Subject: Re: [sdiy] BODE Frequency Shifter
> 
> Thanks so much, David G Dixon
> 
> 
> I am very grateful for the very detailed explanation.
> I checked the phase difference one by one and went.
> In the process, I confirmed a very suspicious fact.
> It is a mistake of mounting parts.
> It was due to misreading of the numerical value of CAP.
> As a result, the circuit operated normally as a result of 
> replacing the part.
> We succeeded in obtaining a phase difference of 90 degrees at 
> all frequency points.
> Thank you very much.
> 
> 
> Atsushi Maeda
> 
> On 2018/11/01 3:43, David G Dixon wrote:
> > Here's my two cents on Dome filters, which I call 90-degree phase 
> > displacement networks...
> >
> > If you have built a 90D PDN with 10 or 12 allpass stages, all 
> > individually tuned with trimmers, then you should have no problem 
> > tuning each one.  Just achieve a perfectly circular 
> Lissajous figure 
> > between the fed sine and the stage output on each one at 
> the desired 
> > frequency, and then the whole thing should give you a 
> 90-degree (+/- 1 
> > or 2 degrees) difference between the two cascades over the designed 
> > range of frequencies.  You should be able to feed a slow 
> > frequency-swept sine wave to the filter and see a more or less 
> > perfectly circular Lissajous figure between the two 
> cascades over the 
> > entire range.  However, the phase shift will be 
> dramatically different once you exceed the frequency range, 
> on either end.
> >
> > At least, that's my experience.  I built a 12-stage (6 x 6) 
> filter for 
> > a range of 15 to 15000 Hz using QuadNet, and it worked 
> perfectly.  I 
> > did it two ways:
> >
> > 1) The first one was with all different monolithic ceramic 
> cap values 
> > selected to give the proper frequencies with a resistance of about 
> > 32k, and I used 27k fixed resistors and 10k trimpots and 
> tuned each one individually.
> >
> > 2) The second one was with 102, 103 and 104 film caps (4 of 
> each) and 
> > I used pairs of 1% resistors to get as close to the required time 
> > constants as possible.  For example, for the frequency of 1340Hz, 
> > given a 103 (10nF) cap, the required R is 11875 ohms (the 
> cap values 
> > were selected to keep the resistances mostly between 10k 
> and 100k).  
> > This can be obtained with an 11.8k resistor and a 75R 
> resistor, both 
> > standard 1% values.  (My biggest resistance error was 0.013%.)  I 
> > bought a small pile of 102, 103 and 104 film caps, and I 
> made a simple 
> > triple relaxation oscillator with 10k, 100k and 1M resistors that 
> > should generate outputs of 455Hz for 104, 103 and 102 caps, 
> > respectively.  Then I tested all the caps in my collection and 
> > separated them into bins depending based on 4-Hz ranges.  
> Hence, all 
> > the caps that gave between 453 and 457Hz went into one bin, between 
> > 458 and 462 in another, and between 448 and 452 into a third.  Any 
> > that were outside of those ranges were excluded (there weren't that 
> > many -- I find that caps are generally closer to their 
> stated values 
> > than the spec would lead you to believe).  I didn't select the 
> > resistors, since my design suggested that 1 or 2% tolerance 
> on the RC 
> > value wouldn't make a significant difference to the 
> performance of the 
> > PDN.  It worked well (in that it gave a more or less perfectly 
> > circular Lissajous figure between 15 and 15000 Hz) and required no 
> > trimming whatsoever.  No trimmers means a much simpler layout and 
> > quicker build.  Testing the caps went quickly, since I 
> could do three 
> > at a time (one of each value), and so within a half hour I could 
> > generate enough caps for half a dozen 90D-PDN circuits.  My 
> > spreadsheet analysis suggested that the PDN would work 
> equally well if all of the caps had the same tolerance.  In 
> other words, if all the caps were 1% low or 1% high, the PDN 
> would work just as well as if they were all perfect.
> >
> > Not sure if any of this helps, but that's my experience.  I've done 
> > similar thinking on all parts of my frequency shifter so that there 
> > are no trimmers at all and it works very well, with little if any 
> > carrier bleed.  The key thing is to AC-couple all of the 
> inputs to the 
> > multipliers, since the majority of carrier bleed is from DC 
> levels on the various signals.
> >
> >
> >> -----Original Message-----
> >> From: Synth-diy [mailto:synth-diy-bounces at synth-diy.org] 
> On Behalf Of 
> >> Dave
> >> Sent: Wednesday, October 31, 2018 7:02 AM
> >> To: synth-diy at synth-diy.org; atsushi maeda
> >> Subject: Re: [sdiy] BODE Frequency Shifter
> >>
> >>
> >>
> >> On October 30, 2018 3:45:25 AM PDT, atsushi maeda 
> >> <shine32 at kr.tcp-ip.or.jp> wrote:
> >>
> >>> The current problem occurred when adjusting DOME Filter.
> >>> It may be adjusted so as to have a phase difference of 90
> >> degrees at 40
> >>> Hz, 170 Hz, 725 Hz, 3080 Hz, 13090 Hz, but if I adjusted it to 90 
> >>> degrees phase difference at 13090 Hz, the phase difference
> >> will be 90
> >>> degrees at other frequencies It will not be.
> >>> No matter how many times you try to adjust it, it will not change.
> >>> Is my understanding wrong?
> >> Please post your measurements.  How large are the errors?
> >>
> >> -Dave
> >>
> >> _______________________________________________
> >> Synth-diy mailing list
> >> Synth-diy at synth-diy.org
> >> http://synth-diy.org/mailman/listinfo/synth-diy
> >
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy




More information about the Synth-diy mailing list