Influence of fs (driver resonance frequency) on active/passive XO development?

[Shefffield]

Hello forum,

I hope this is the right section of the forum to discuss this topic. I'be been wondering about which or how much influence the resonance frequency of a driver/driver in cabinet has - especially since I am more interested in active crossovers.

Sure, I am aware that for passive crossovers the impedance matters a lot. It interacts heavily with the passive XO. However, an active XO before the amp doesn't care about impedance at all.

If I look at woofers, it is common and accepted to use them well over their fb/fs. Does this have any downsides?
From what I understand, distortion rises when excursion gets higher. Is there any link between fb/fs and distortion?

How about midrange drivers? Given that they are not excursion limited, would there be any downside in using them well into their fb/fs region in an active setup?

In my current digital active system I am using Ciare 38 mm dome tweeters with fs 1300 Hz. (https://www.ciare.com/en/product/?codice=PT383#mount-ship10 - Yes, the data sheet claims 1000 Hz, but I've seen higher numbers in older versions of the data sheet.) So far I have only tried it down to 1600 Hz crossover frequency (200 dB/oct digital filtering, so excursion really shouldn't be a limiting factor here). Can't make new trials right now since I am travelling, but I am interested in a theoretical understanding of the matter anyway.

What's your take on this?

 

[KSTR]

Basically you can use a driver at any frequency, but distortion/error behavior is depending on a lot of things, simplified:

• Motor force linearity BL(x) : Obviously will be dominating the error at large excursions, no matter what frequency.
• Spring linearity Cms(x) : Dominates distortion below fs, but not necessarily below fb. Below fs the driver is more and more spring-controlled *1) and when the spring is nonlinear you'll get errors, regardless of motor force linearity. Air spring is way more linear than driver spring, that's why fb can be high and the driver is mainly spring-controlled but still low error.
• System Q plays a major role. The driver will ring governed by the system Qb at fb for any error signals, including it's own errors, so Qb should be kept low (< 1). In an active system you can tailor Qes which determines Qb by altering drive impedance (including negative values up to -Re, theoretically establishing full velocity feedback in the driver)

*1) Imagine a driver near DC, way below fs: There is almost no internal velocity feedback in the driver because the ratio of required drive voltage (for flat SPL) vs microphonic voltage is large. In other words, the voltage developed across the static VC impedance (Re) is large vs. back-EMF, reducing effective feedback. Therefore, spring nonlinearity dominates even with a perfect BL(x). Current produces force via BL, Newtons per Amperes, and those Newtons work on the spring compliance Cms, Meters per Newton, to finally form the momentary excursion value. Velocity feedback is highest at fs (least amount of current required) and with a flat BL(x) you get lowest distortion here.

200 dB/oct? I would never use that, at least not when there is a sharp kink in the transition. Acoustic target any steeper than 4th order (reducing excursion to second order -- 12dB/oct below XO) does't have much benefits. The transition region is what counts in the end (for seamless integration and for effective excursion control efficiency at and below XO), and I prefer no kinkier than the approximate shape 2nd order. In analog XO's I often use 3rd order final slopes (reducing excursion 6dB/oct below XO) with a almost 2nd order transition to get better phase matching (60deg phase offset rather than the 90deg from a Butterworth 3rd order) and better excursion efficiency.

That Ciare is very high and fixed Qts (closed back)... hhm. Underhung, so BL(x) is quite good but will go berserk when overdriven into large excursion.

 

[DanielT]

Thanks, KSTR. Interesting. Well explained. There I learned something new!

Most things start to distort (if you see distortion as deviations from how gadget x is meant to perform).

Take any mechanical thing. Hm a bicycle, car engine (where you turn up the turbo pressure) and so on. They work within certain limits, then well they function/work worse and in the end, in the event of heavy power / heavy use (although you probably have to really step on the pedals with the bicycle) they break.

 

[bigjacko]

@KSTR Thanks for the information. Regarding Q of driver, does driver start to ring when going past 0.707, and rings for longer when Q is higher? When we cannot use feedback, what else can we use to reduce the Q without changing driver and enclosure?

When we use normal drivers, do drivers normally only usable above fs? I know tweeters generally need to be crossed 2x fs, but subs can be used below fs. What about cone mid and dome mid? Do manufacturers normally optimize bandwidth above fs rather than everything?

Thank you very much!

 

[KSTR]

@KSTR Thanks for the information. Regarding Q of driver, does driver start to ring when going past 0.707, and rings for longer when Q is higher? When we cannot use feedback, what else can we use to reduce the Q without changing driver and enclosure?

When we use normal drivers, do drivers normally only usable above fs? I know tweeters generally need to be crossed 2x fs, but subs can be used below fs. What about cone mid and dome mid? Do manufacturers normally optimize bandwidth above fs rather than everything?

Thank you very much!

At Q>2 things start to fall apart. Q=1 is no big deal.

There is nothing you can do reduce Q, other than using amps with negative output impedance. Or pack a huge extra neodymium magnet on the driver ;-)

Most VC-based drivers are designed to be operated at or above the system Q in the cabinet types they've been optimized for, othewise extra EQ is needed. Flat SPL at reasonable efficiency is the target, ususally (for specialized PA drivers not necessarily).

 

[hex168]

In passive crossovers, it is not unusual to use either 1) a resonant tank circuit to reduce impedance at driver resonance, or 2) a resistor in parallel with the driver for a similar but less frequency-specific effect with fewer and cheaper parts. Also, even-order high-pass crossovers have an inductor in parallel with the driver that can have a similar effect. Simulations are better than formulas, but this should be a start:
1) http://www.mh-audio.nl/Calculators/NFC.html
2) This thread is worth reading ("Wolf" posts here at ASR, by the way):

Tweeter crossover - when resonance gets in the way - NE25 - Techtalk Speaker Building, Audio, Video Discussion Forum

Want a second or third opinion about your speaker cabinet design or other audio related problem? Post your question or comment on the Technical Discussion Board. Hundreds of technicians, engineers, and hobbyists, nationwide read and discuss electronics related questions each week. We welcome...

techtalk.parts-express.com

 

[Wolf]

Thanks for mentioning me, even if that was an 11 year old thread-

No, active xovers before amplifier do not care about the impedance.

The magnitude of the Fs is also very critical. If the peak is only 8-15 ohms, you may not need to do anything. Above there it gets more easily excited. This goes for all drivers. Higher the magnitude, the less voltage it takes to excite the frequency.

Since most of this above converses about tweeters, then when it comes to mids and woofers...

If you have a 3-way with a mid and nothing to counteract the Fs of the mid, it can make the highpass less effective passively, and/or actually cause a bump in the response. 4 options- xover considerably above the Fs, series LCR (parallel to driver in series string) to remove it, parallel R to minimize it, or make a dual-chamber aperiodic box to damp it acoustically (internally a multi-chamber box with rear lined and front stuffed, and adding an open-cell-foam covered channel from front to rear chambers).

With a woofer, this can cause a HUGE bump in the response with a high-magnitude Fs for passive xovers. On a woofer, there are 4-ways to take care of it if need arises- full series LCR across driver (large parts values), higher wattage resistor across woofer (Try to keep the current low by upping the value, recommended 40-50 ohms), aperiodic box types, or a higher-DCR coil (smaller gauge/diameter wire) in the lowpass circuit. Do not place a resistor in series with a woofer as it will likely get warm.

With the tuning of woofers in vented boxes (2 peaks in impedance), the best at controlling overshoot is by tuning to Fs = Fb. Peaks are equal in magnitude.
If Fs > Fb, the design is likely looking to extend a little below Fs and is not unheard of. This also can control Xmax to a lower freq range if needed where overshoot or unloading starts further down. This can allow overshoot above tuning if not careful. Visually in impedance, the lower freq peak is smaller in magnitude than the higher peak with tuning at the low point of the valley between.
I rarely tune where Fs < Fb, as it protects the driver less in most cases. There are cases however, where the response of the driver droops without doing it this way, and it can level the response off before rolloff, or smooth out the curve for a non-standard vented rolloff.

 

[Tangband]

Basically you can use a driver at any frequency, but distortion/error behavior is depending on a lot of things, simplified:

• Motor force linearity BL(x) : Obviously will be dominating the error at large excursions, no matter what frequency.
• Spring linearity Cms(x) : Dominates distortion below fs, but not necessarily below fb. Below fs the driver is more and more spring-controlled *1) and when the spring is nonlinear you'll get errors, regardless of motor force linearity. Air spring is way more linear than driver spring, that's why fb can be high and the driver is mainly spring-controlled but still low error.
• System Q plays a major role. The driver will ring governed by the system Qb at fb for any error signals, including it's own errors, so Qb should be kept low (< 1). In an active system you can tailor Qes which determines Qb by altering drive impedance (including negative values up to -Re, theoretically establishing full velocity feedback in the driver)

*1) Imagine a driver near DC, way below fs: There is almost no internal velocity feedback in the driver because the ratio of required drive voltage (for flat SPL) vs microphonic voltage is large. In other words, the voltage developed across the static VC impedance (Re) is large vs. back-EMF, reducing effective feedback. Therefore, spring nonlinearity dominates even with a perfect BL(x). Current produces force via BL, Newtons per Amperes, and those Newtons work on the spring compliance Cms, Meters per Newton, to finally form the momentary excursion value. Velocity feedback is highest at fs (least amount of current required) and with a flat BL(x) you get lowest distortion here.

200 dB/oct? I would never use that, at least not when there is a sharp kink in the transition. Acoustic target any steeper than 4th order (reducing excursion to second order -- 12dB/oct below XO) does't have much benefits. The transition region is what counts in the end (for seamless integration and for effective excursion control efficiency at and below XO), and I prefer no kinkier than the approximate shape 2nd order. In analog XO's I often use 3rd order final slopes (reducing excursion 6dB/oct below XO) with a almost 2nd order transition to get better phase matching (60deg phase offset rather than the 90deg from a Butterworth 3rd order) and better excursion efficiency.

That Ciare is very high and fixed Qts (closed back)... hhm. Underhung, so BL(x) is quite good but will go berserk when overdriven into large excursion.

For an amateur DIY:er , would it be a reasonable advice to never tune the basstube lower than the drivers fs , because this advice would be good sounding in most cases, where tuning below fs depends very much of how the driver behaves below its fs ?

 

[Wolf]

Best advice for newbs is to tune to Fs, but no, I don't think your statement is without problems. The tuning Fb, not the Fs, is where the problems start below most of the time.

 

[KSTR]

For an amateur DIY:er , would it be a reasonable advice to never tune the basstube lower than the drivers fs , because this advice would be good sounding in most cases, where tuning below fs depends very much of how the driver behaves below its fs ?

As a general rule I would certainly agree. Don't operate a driver in a cabinet way below its natural Fs. There are, of course, exceptions. Very strongs motors (extremely low Qes) allow for more intrinisic corrective feedback below resonance that weak drivers. But they also need to be very linear for this, flat BL(x) curve on the Klippel plots.