I know the FIR approach, I have seen IIR, to, but I'd like to know if it's possible to implement a Hilbert transformer in analog domain, i.e. with integrators instead of delays. Is it possible? If yes, how?

[edit] After some searching on my own, I stumbled upon an answer in sci.electronics.design - "Analog delay??" (I'm not posting the link as it seems it involves some character string and I don't know what influence will that have). A simple search will do.

Still, this involves a translation of a digital delay into its analog counterpart, clock, etc, no continuous-time values. Could this mean that there is no other way to do it (analog)?


Yes, it is possible to implement a Hilbert transform in the analog domain, although it is difficult to get the desired phase shift over a significant bandwidth when you take component uncertainties and imperfections into account. There is some discussion of the topic in this document.

I'm not sure why you think you would use an integrator instead of a delay. The analog counterpart of a discrete-time delay is a continuous-time delay. This can be implemented in various ways, although as noted before, achieving precise delays in analog electronics is quite difficult. Some methods that I've seen before for inducing delays include introducing capacitive delays (e.g. with an RC filter) or using surface acoustic wave (SAW) devices.

  • $\begingroup$ Nice document, thank you. I said integrators thinking of the general form realization of an analog filter, in building blocks. For example, for only one frequency, a 2nd order all-pass would do. I would have thought the analog HT would have been some parallel branches all summed up, FIR style. Anyway, this was just something I wanted to try in LTspice, without having to involve an (filter-)A/D-FIR-D/A(-filter) chain (computationally expensive). If I use a transmission line, I'm emulating a FIR, already, so it seems I'm back in the digital world. But I still have a trick up my sleeve :-) $\endgroup$
    – user164048
    Aug 1 '14 at 15:53
  • $\begingroup$ I marked it as the answer, but I can't vote up, not enough rep... $\endgroup$
    – user164048
    Aug 1 '14 at 15:54
  • $\begingroup$ The link to "this document" is broken, and a simple Google search didn't find an obvious equivalent pdf. $\endgroup$
    – DarenW
    Feb 8 '18 at 20:08
  • $\begingroup$ The Wayback Machine had the PDF: web.archive.org/web/20151001084040/http://home.comcast.net/… The PDF header text: 453.701 Linear Systems, S.M. Tan, The University of Auckland $\endgroup$
    – nitrogen
    Jan 15 at 3:41

I designed a so-called AcqHilbert A/D converter. It is the cascade of an hybrid Analog Hilbert transformer and a dual matched SAR A/D Converter from either Analog Devices or Maxim. The 8th order Hilbert Xformer is based on dual 12 bits CMOS Multiplying DAC's ( AD 54x5 ) and DigiPots. It is fully programmable for the positioning of the All-Pass Poles-&-Zeros transitions, over a Band-of-Interest of 1000/1, according to either the Bedrosian or Lloyd tables. The position of the B-o-I can also be positioned over a range of 4000/1, from ( 400Hz-400kHz ) to ( 0.1Hz-100Hz ). The whole system is controlled from an SPI bus.


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.