I need to match the phase of two signals coming from a pair of transducers over the 10Hz to 4 KHz range, the difference between the two transducers is really close, however one of the transducer lags 7° in phase over the entire band with respect to the other transducer. I need something like an all-pass filter, but instead of it having linear phase shift which varies linearly over frequency, I need a constant phase shift, in my case I need it to be a constant 7° phase shift over the desired bandwidth, I know of no analog network that can perform this.

Any ideas?

  • \$\begingroup\$ Is the signal from the transducer broadband, or at any given moment is it approximately sinusoidal? \$\endgroup\$ – The Photon Oct 3 '18 at 22:56
  • \$\begingroup\$ It is approximately sinusoidal \$\endgroup\$ – S.s. Oct 3 '18 at 22:58
  • \$\begingroup\$ So you want something that delays by (1/f) * (7/360) seconds for f from 10 to 4000? \$\endgroup\$ – crj11 Oct 4 '18 at 0:26
  • \$\begingroup\$ @TonyEErocketscientist I thought that maximally flat group delay meant linear phase, not a constant not linearly changing phase. \$\endgroup\$ – S.s. Oct 4 '18 at 2:25
  • \$\begingroup\$ @S.s. Brainfa/t. So Is there another way to correct this with PLL tracking>? \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Oct 4 '18 at 2:39

Option 1: Generate the quadrature signal of your input. Sum it with the input in the correct proportions to get 7 degrees phase shift. How to generate a quadrature signal over more than 2 decades of frequency range is left as an exercise for the reader.

Option 2: Make a 1:1 phase locked loop and put an offset in the phase detector to produce 7 degrees phase shift. How to make a good VCO that spans more than 2 decades of frequency range is left as an exercise for the reader.

Option 3: Digitize everything and let software sort it out.

  • \$\begingroup\$ Option 1 is brilliant!, I cant really see how option 2 works, option 3 is precisely what I want to avoid. \$\endgroup\$ – S.s. Oct 4 '18 at 2:39
  • \$\begingroup\$ If the signal amplitude doesn’t matter a PLL IC can lock on with any phase error introduced by a type II mixer offset of Vcc/2 * 7/180 deg if Vcc/2 is 0 error but getting 3 decades of VCO is better using a log scale voltage-controlled Schmitt Trigger Relaxation Oscillator to change feedback ratio. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Oct 4 '18 at 5:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.