IIR Filter Calculator for Audio DSP Stability

IIR Filter Calculator

Design audio biquad IIR filters, normalize coefficients, inspect pole radius, and check stability before exporting DSP settings.

🎛Filter Presets

📐IIR Filter Inputs

Sample rate Used for digital frequency warping and Nyquist checks.

Filter type RBJ cookbook equations are used for coefficient design.

Cutoff or center frequency Hz. Must stay below Nyquist for a stable practical design.

Q factor 0.707 is Butterworth; high Q gives narrower bandwidth and more ringing.

EQ gain dB. Used by peaking and shelf filters; ignored by plain pass filters.

Cascaded sections Each section is one second-order IIR biquad stage.

Coefficient precision Quantized coefficient preview helps reveal narrow-filter risk.

Pole radius warning limit Designs above this radius are stable but sensitive to rounding.

Available signal headroom dB. Used to flag boost filters that can clip internally.

Coefficient format SOS or transposed forms are usually safer for audio plug-ins.

Stability status

Stable

Pole radius under 1.000

Max pole radius

0.0000

Lower radius decays faster

Approx bandwidth

0 Hz

Center frequency divided by Q

Ringing decay

0 ms

60 dB decay estimate

Normalized biquad coefficients

🧮Formula Reference

Coefficients use w0 = 2*pi*f0/Fs, alpha = sin(w0)/(2Q), and normalized denominator a0 = 1. Stability is tested from the denominator roots of z^2 + a1*z + a2 = 0; a biquad is stable when every pole radius is below 1.

📊Filter Spec Grid

0.707

Butterworth Q

1.000

Pole stability edge

Fs/2

Nyquist limit

SOS

Preferred cascade

📝IIR Type Comparison

Filter type	Main control	Typical audio use	Stability note
Lowpass	Cutoff and Q	Synth tone, sub routing, anti-harshness	High Q can ring near cutoff
Highpass	Cutoff and Q	Rumble removal, vocal cleanup	Very low cutoff needs more precision
Bandpass	Center and Q	Telephone tone, resonant emphasis	Bandwidth shrinks as Q rises
Notch	Center and Q	Hum, whistle, room mode removal	Narrow notches are quantization sensitive
Peaking EQ	Gain, center, Q	Corrective and creative equalization	Boost needs headroom
Shelf EQ	Gain and corner	Bass lift, air lift, tilt shaping	Large boosts can overload cascades

🎯Q and Bandwidth Table

Q value	Octave width	Behavior	Common setting
0.500	About 2.54 oct	Wide and gentle	Tone shaping
0.707	About 2.00 oct	Butterworth response	Crossovers and cleanup
1.414	About 1.00 oct	Focused but musical	Presence shaping
4.000	About 0.36 oct	Narrow correction	Room or resonance trim
12.000	About 0.12 oct	Very narrow	Hum or whistle notch

🔍Precision and Stability Table

Coefficient precision	Step size	Best use	Risk area
16-bit fixed	1 / 65536	Simple tone filters	Low frequency or high Q
20-bit fixed	1 / 1048576	Embedded audio control	Deep notches
24-bit fixed	1 / 16777216	Most audio DSP export	Extreme resonances
32-bit fixed	1 / 4294967296	High precision processors	Internal headroom
Float	Runtime mantissa	Plug-ins and DAWs	Denormals near silence

📈Preset Design Table

Preset	Type	Frequency	Design intent
Vocal Highpass	Highpass	80 Hz	Remove plosives and rumble
Bass Lowpass	Lowpass	5 kHz	Keep bass cabinet tone smooth
Room Notch	Notch	60 Hz	Reduce mains hum or room mode
Presence Peak	Peaking EQ	3 kHz	Add focused vocal intelligibility
Telephone Bandpass	Bandpass	1.2 kHz	Approximate narrow voice bandwidth

Stability tip: A pole radius below 1 is mathematically stable, but audio filters above about 0.98 can become touchy when coefficients are rounded for fixed-point DSP.

Workflow tip: Export second-order sections for cascaded filters so each biquad can be scaled, checked, and limited before the full IIR chain is loaded.

An Infinite Impulse Response (IIR) filter use feedback because teh output of the IIR filter is fed back into the input of an IIR filter. Because of this feedback mechanism, it is possible for the IIR filter to become unstable. An unstable IIR filter will produce digital noises instead of the mathematical signal that the filter are suppose to create.

Furthermore, the instability of the IIR filter can lead to damaged speaker or ruined audio mixes. The stability of an IIR filter can be determined by the location of the poles of the IIR filter on a complex plane. The stability of the IIR filter depend upon whether the poles of the IIR filter lie within the unit circle.

Why IIR Filters Can Become Unstable

If the radius of a pole is less than one, the pole is stable. If the radius of a pole is greater than one, the pole is unstable. In the real world, all processors has some form of rounding in their operations.

If the radius of a pole of an IIR filter is very close to one, such as 0.999, the rounding that occurs with all real world processors can lead to the pole move outside of the unit circle. If the pole moves outside of the unit circle, the IIR filter will be unstable. The Q factor of a filter determine how much the filter will “ring” at the cutoff frequency.

High values of the Q factor will lead to a resonant peak at the cutoff frequency. A Butterworth response of a filter will have a flat response in the passband, but high values of the Q factor will move the poles of the IIR filter more closer to the edge of the unit circle. As the poles of an IIR filter come closer to the edge of the unit circle, the ringing decay time of the IIR filter increase.

Thus, an increased ringing decay time will cause the IIR filter to take longer to settle after a transient. The sample rate at which an IIR filter operates will impact the behavior of the IIR filter. The sample rate impacts the digital mapping of the IIR filter.

For instance, if the digital filter has a cutoff frequency that is close to the Nyquist limit, which is half the sample rate, the digital mapping of the IIR filter begins to warp. Thus, the behavior of an IIR filter at 96 kHz will be more different than the behavior of an IIR filter at 44.1kHz. One of the factors that the designer must take into consideration for the stability of an IIR filter is the precision of the calculations of the IIR filter.

The precision of the calculations is dependent upon whether floating point or fixed point math are used to calculate the IIR filter. The majority of Digital Audio Workstations use floating point math, so most users of DAWs wont experience issues related to quantization. However, many embedded DSPs and audio effect pedal will use fixed point math.

For instance, many use 16-bit or 24-bit fixed point math. Using a 16-bit fixed point processor to implement a high Q notch filter may lead to problems because the 16-bit representation of the coefficients dont have enough resolution to maintain the position of the pole of the filter. Thus, the pole may move outside of the unit circle.

An additional factor that will impact the signal that is created by an IIR filter is the headroom for the signal. Headroom is a factor to consider if the frequency is being boost with a peaking EQ. If a peaking EQ boosts the frequency, the internal signal of the digital device will increase.

If the internal signal is too high, it may lead to internal clipping. Internal clipping will introduce digital distortion into the signal, which is a form of noise that cant be fixed with makeup gain. To ensure that the IIR filter is stable, the designer must find a balance between musicality and mathematical safety.

To ensure that the audio remains clean, the designer can check the pole radius of the IIR filter and the decay time of the IIR filter before finalizing the parameters of the IIR filter. By checking the pole radius and the decay time, the IIR filter designer can determine how close each pole of the filter is to the edge of the unit circle. By determining the distance of each pole to the edge of the unit circle, the IIR filter designer can ensure that they avoids the potential for instability caused by rounding error.