The low-frequency response of guitar speakers, however, varies greatly between speakers of different makes and models. This low-frequency response is a sharp resonance typically in the range of 50-150 Hz. The magnitude of this resonance varies from several to 20 times the nominal impedance.

The impedance of a speaker influences the response of a tube amp since a tube power amp is essentially a transconductance amplifier. It creates a current for an applied voltage. This current in turn creates a voltage across the speaker terminals that is dependent upon the impedance of the speaker. Therefore the power amp will resonate at the resonant frequency of the speaker. This causes certain notes to become emphasized as they excite the resonant frequency. Negative feedback around the power amp will reduce the amount of resonance but not all amps use negative feedback (i.e. Vox).

The increased voltage amplitude at the resonant frequency also causes the power amp to clip sooner at the resonant frequency. Think of it this way: if the power tubes are swinging, say, 200 V at the midrange frequencies, they will swing X times more at the resonant frequency where X is the ratio of the resonant impedance to the nominal impedance. So if the resonant impedance is 10 times the nominal impedance the power tubes will want to swing 2000 volts. This is impossible so they will clip. For high-gain tones this can cause the tone to sound muddy or feel spongy. For lower gain tones this can thicken the tone and make it feel, well, more spongy.

Cabinet/speaker IR data does not contain the impedance information. The only way to obtain impedance data is to measure the current vs. voltage vs. frequency (despite what modeler advertising literature would like you to think).

The Axe-Fx uses default values of LF Resonant frequency and impedance for each amp model. For models based on combo amps these values are derived from measurements of the actual amp's speaker. For models based on amp heads the values are based on measurements of the cabinet most likely to be used with that head.

You can adjust the frequency and impedance to suit your taste. Reducing the impedance (Low Res) will reduce the bass response and can give tighter bass. Raising the impedance will increase the bass response and can give a fuller sound. Altering the frequency (Low Freq) will change the frequencies at which the power amp resonates and tuning this to the key you are playing in can be an effective strategy, e.g. set it to 82 Hz if playing in E.

Don't be afraid to try drastic settings. Try turning Low Res all the way to zero. Compensate by adding some bass with the Bass knob or the EQ section. As I mentioned earlier the LF Resonance will cause the power amp to clip earlier than it will when amplifying midrange frequencies. Turning down the Master Volume will increase the headroom in the power amp and reduce this clipping. Furthermore the Transformer Match also influences when the power amp clips. So there is a relationship between LF Res, MV and Transformer Match.

Many manufacturers publish impedance data for their speakers. Eminence and Jensen and probably others publish detailed impedance data. You can look at the impedance plots and set the resonance parameters to match (roughly). The Low Res parameter is indicated from 0 to 10 and sets the resonance in dB from 0 to 24 dB (dB is a ratio of powers so it's not really the proper units for this but that's semantics). For example, the Jensen P12N has resonant frequency of about 100 Hz so you would set Low Freq to 100 Hz. The impedance at this frequency is about 40 ohms. To get the Low Res amount use the formula (20 * log10(Zr/Rdc)) / 2.4 where Zr is the impedance at the resonant frequency and Rdc is the DC resistance. For this speaker Low Res is then (20 * log10(40/6.2)) / 2.4 = 6.7.

A power amp isn't perfect though. Winding resistance in the output transformer increases Rdc, typically by a couple ohms. Therefore our above example would become (20 * log10(40/8.2)) / 2.4 = 5.7. The exact value isn't overly critical though and all this is subtle nuances.

The resonance Q is a bit more difficult to calculate. It is derived from the bandwidth at the points where the impedance "gain" is the square root of the resonance impedance gain. IOW, if the impedance is, say, 10 times the nominal impedance then the bandwidth is given by the frequencies where the response is 3.16 times the nominal impedance. For our example the resonance gain is 5 (40 / 8 = 5). So the bandwidth is the frequencies at which the impedance equals sqrt(5) * 8 = 18. From the graph this is approximately 75 Hz and 130 Hz. Q is defined as f0 / bw so our resulting Q is 100/60 = 1.67. Most speakers have a Q of around 2.0 or so. Again the exact value isn't overly critical and don't be afraid to try extreme settings (you can't break anything).

Finally, just because real speakers behave like this doesn't mean we have to adhere to this behavior. Perhaps a better speaker has no resonance (Low and High Res are zero), or maybe the Q is a lot lower or higher. In our virtual world we can design a speaker that is impossible to construct in the physical universe.

tl;dr version: Mess with Low Freq and Res if you want, or not.