9
$\begingroup$

A PIN diode varies its capacitance based on a DC bias, and a Varicap exhibits this effect even more clearly. What limits the use of "solid state variable capacitors" like these for dynamically filtering (low power/received) RF signals?

The question at change LC parameter digitally is related, but was asked in the context of a tunable oscillator. That is already a ± "solved problem" in the SDR world. What would be useful in conjunction with a DCO that can operate in a 24–1766 MHz range would be a correspondingly flexible bandpass filter to limit receiver overload/aliasing!

$\endgroup$
1
  • 1
    $\begingroup$ See also YIG filters which can be used to sweep the centre frequency of a filter. $\endgroup$ – tomnexus Apr 2 at 5:27
13
$\begingroup$

What limits the use of "solid state variable capacitors" like these for dynamically filtering (low power/received) RF signals?

The main issue is bandwidth.

Although there are some varicaps that claim a 10:1 variation (or more) in capacitance, not all of this is useable. One end of the range may be low Q, or have a very steep curve. Once you have a varicap on a board, with pads and bias connections, you'll often find that the useable capacitance swing at the tuned circuit is no more than 4:1, giving you only 2:1 swing in centre frequency.

Even if you did have a varactor that gave you more swing, or a construction technique that reduced the effect of strays, changing the LC ratio alters the impedance of the filter. There's only so much change you can make to the impedance of the resonators before a good filter becomes a bad filter.

You may only want to operate in an octave. However you've asked specifically about SDRs, and the main benefit of those is a very wide frequency range.

You could switch inductors into the filter, with the switching strays further reducing the capacitance swing available. Or you could build multiple filters and switch between them. Both of these add to the complexity. Once you have accepted the complexity of further switching, it often makes more sense to build a few, good, fixed frequency filters with stable components and switch between them.

$\endgroup$
3
  • 3
    $\begingroup$ Welcome to the site and thanks for tackling my question! This sounds like a good summary of the kinds of "real-world" constraints I was wondering about and appreciate the breadth of practical issues you brought up. $\endgroup$ – natevw - AF7TB Apr 2 at 18:30
  • 1
    $\begingroup$ This scenario, with the OP welcoming the author of a quality answer to the site, demonstrates everything that's magical about the stack exchange community! $\endgroup$ – jpaugh Apr 2 at 21:40
  • $\begingroup$ +1. As you noted, if you want to operate in an octave or sub-octave, this approach is an option. A product I worked on earlier in my career was a narrowband receiver with a preselection bandpass filter whose center frequency was tunable by applying differing DC voltages to a pair of PIN diodes. I don't remember the bandwidth of the preselector that we were able to achieve, but it was tunable over a range of ~130-180 MHz or so. $\endgroup$ – Jason R Apr 3 at 19:17
5
$\begingroup$

Technically, the utility of a variable capacitor in a receive filter depends on the amount of available capacitance variation and its Q - the quality factor, which is the ratio of the capacitor's reactance at the operating frequency to the series parasitic resistance. (In some applications, parasitic series inductance may also be a factor.) Commercially, price is a primary concern.

Fixed-value ceramic chip capacitors commonly have a Q value of 2000 or more and, obviously, no variability - at a very low unit price. Perusing an online distributor's catalog, I see varactor diodes with capacitance ratios of around 10:1 down to around 4:1, some with Q factors over 1000 but most with Q factors in the hundreds or less - and with prices of 15X to 150X that of a fixed cap.

When taken together, fixed-value capacitors deliver the needed cost-performance balance for all but a small fraction of all such applications.

$\endgroup$
5
$\begingroup$

Nobody mentions distortion (IM2 and IM3) of PIN diodes and varicap diodes. Tuning (center frequency) limitation range, required band switching and further requirements. Best solution is to start with a good system that can handle the dynamics without filtering. As W8II Phil Frost proposes.

** Mentioned Q-factors (Brian K1LI) of 1000 are not feasible in a normal receiver and not necessary. Filter Q_loaded of 100 is practical; then unloaded Q is not necessary to be higher than 250. **

$\endgroup$
4
$\begingroup$

Another difference between using a varicap for oscillators versus for input filters is feedback. You can feed the output of a variable oscillator into a phase comparator or frequency counter to determine if the component values are correct, and run a feedback loop correction algorithm or circuit to maintain lock. This can be done while the oscillator is performing in-circuit Rx or Tx. It's a little harder to determine if an RF input filter is set correctly during actual Rx.

Because varicaps can have (large?) temperature coefficients, they present a greater problem making sure the filter passband stays at the desired frequency than with oscillator circuits that can use feedback. You might have to take the radio off-line to sweep the filter and recalibrate whenever the temperature changes.

$\endgroup$
1
  • $\begingroup$ Ah, the temperature coefficient is another interesting "non-idealized component" consideration to keep in mind, thanks! $\endgroup$ – natevw - AF7TB Apr 2 at 18:32
1
$\begingroup$

Consider increasing the dynamic range of the ADC to obtain many of the same performance benefits. This can be done by either upgrading the ADC, or low-pass filtering the ADC input below the Nyquist frequency and oversampling. Often this is the cheaper and more flexible approach, especially given the limitations of the varicap's Q-factor and capacitance range, and explained by Brian K1LI.

$\endgroup$

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.