Anyone ever build a Linkwitz Transform Circuit for a subwoofer? I know they are kind of redundant/old school with todays PEQ amps and DSPs, but I was thinking of building one for fun.
I built one using opamps for a tiny two-way bookshelf. It does work, though boosting more than 10 dB might not a good idea as the low frequency driver quickly overloads. If I remember correctly, I tried to get a 3" driver to play down to 50 Hz. It sounded impressive at very low sound pressure levels but quickly started to distort when asked to play at normal levels. There is a reason why driver manufacturers do not make fs lower than they do.
I Built some back in the days and used them in homecinma applications with several drivers (8 x old radiotechnica 10", 15" Pa Woofers, a lot of cheap car-woofers, etc..). Worked fine. There was a schematic somewhere around on a german website which I used as a starting point, I can´t find it anymore.. But looking at Linwkitz-Lab, you should get all you need and propably many people put some variants online in the meantime.
You have to keep in mind that most of these designs use some sort of "under resonance principles", where the driver is put in a very small enclosure, so resonance frequency in the cab goes up above the desired frequency band. Therefore, SPL goes down a lot (about 12dB/Oct) and to achieve a certain SPL at low frequencies, you need a lot of Watts/excursion/radiation area compared to a simple ported box.
The sound was unique with these - very detailed, very clean, very punchy... Details in the lower departmend were revealed which I didn´t knew existed.. If you prefer boomy, wobbling bass, this definitely isn´t the solution for you.
Comparing to DSPs back then, the analog version was much better sounding, Headroom in the digital domain always was an issue... Might be less of a problem nowadays when dealing with more Bits internally in the DSP, haven´t tried for a while..
You have to keep in mind that most of these designs use some sort of "under resonance principles", where the driver is put in a very small enclosure, so resonance frequency in the cab goes up above the desired frequency band. Therefore, SPL goes down a lot (about 12dB/Oct) and to achieve a certain SPL at low frequencies, you need a lot of Watts/excursion/radiation area compared to a simple ported box.
The sound was unique with these - very detailed, very clean, very punchy... Details in the lower departmend were revealed which I didn´t knew existed.. If you prefer boomy, wobbling bass, this definitely isn´t the solution for you.
Comparing to DSPs back then, the analog version was much better sounding, Headroom in the digital domain always was an issue... Might be less of a problem nowadays when dealing with more Bits internally in the DSP, haven´t tried for a while..
TA Speaker Topics - Designing a Linkwitz Transform has an xls sheet to calculate for you. I am looking at doing one soon for a TPA3116 $4 amp that sounds great on some 30 year old desktop speakers to get some more low end. I only use them for ambient music at low volumes while working, so going loud is not too concerning, but getting some use out of the extra 40w or so per channel for the class D amp and getting anything under 100hz ( guessing, but it is bad) would be nice. Op amps and passive components for signal level are so cheap you can knock them out on a proto board in 30 min and spend less than the price of the beer you are drinking while puttering around.
I put together a LT circuit designer, including a sensitivity analysis for components, as an Excel spreadsheet. You might find it useful, and it helps to choose a set of components that will increase the probability that the circuit will actually perform as intended when built with real world components.
You can read about and download it here:
CFL Linkwitz Transform Designer with Monte Carlo Sensitivity Analysis
You can read about and download it here:
CFL Linkwitz Transform Designer with Monte Carlo Sensitivity Analysis
For all domestic environments, the LT corrects just one known factor. If you did no aposteriori checking, you might be a modest bit closer to good sound than if you just twiddled with any tone adjustment otherwise available to you or maybe not.
But running acoustic tests tells you where you are and where you need to EQ further. So why bother with LT?
B.
But running acoustic tests tells you where you are and where you need to EQ further. So why bother with LT?
B.
I wouldn't use a bog-standard boost circuit. I'd power the sub with an iNuke DSP amp and use the dynamic EQ feature to apply the required boost low playback levels and reduce it at higher playback levels as the peak excursion ability of the subwoofer is being approached.
Seems rather a too-straight-forward a mod to satisfy a DIYer's intellectual interests.... compared to building an ABX box or exploring how speaker sound changes a whole lot as you move them around (or move your seat around) even just 30.48 cm* sometimes.
A matter of taste in what is fun.
B.
*that's a kind of Canadian joke
Just revisited the LT calculator.
I saw that after you have entered the parameters, selected and measured the components, you can enter them (replace the calculated ideal ones in C23-C27) and simulate the response with the actual components. There is a lot of latitude.
Nice to do and gain confidence in the result!
I saw that after you have entered the parameters, selected and measured the components, you can enter them (replace the calculated ideal ones in C23-C27) and simulate the response with the actual components. There is a lot of latitude.
Nice to do and gain confidence in the result!
Found an even nicer sim tool:
CFL Linkwitz Transform Designer with Monte Carlo Sensitivity Analysis.
The Monte Carlo part is that it takes a part from the E12 or E24 series and starts to optimise from that, picking values itself for the next component, all so it fits in a certain band of error. And it shows the outcome: the error in dB you make from the ideal. Playing around here too you can see that you can get great results with your available components.
CFL Linkwitz Transform Designer with Monte Carlo Sensitivity Analysis.
The Monte Carlo part is that it takes a part from the E12 or E24 series and starts to optimise from that, picking values itself for the next component, all so it fits in a certain band of error. And it shows the outcome: the error in dB you make from the ideal. Playing around here too you can see that you can get great results with your available components.
LT-Transform, a powerful tool
I have built several LT-transforms for subs - thanks to Siegfried for his great work for diy!!!!
The problem is, and thats why it is not much used, you have to measure and calculate every unit individually.
First calculate any driver and its closed enclosure individually - there are a lot of programms ( unibox etc ) online - what sound pressure I need? what is the
closed enclosure volume etc. For this calculation the TSP-Parameters of the manufactors are enough!!.
Then build the enclosure!
After the driver has worked at 20-30Hz ( open air ) at 10W for a day or longer
, put it into the enclosure.
Then measure its DC-resistance - not with a multimeter, but precisely - I used a precision current source with 10,03mA.
The next thing is to measure the impedance curve of the speaker.I used LIMP of the ARTA Softwarefamily - be sure, that the comparing resistance is exactly.
Than go to SB-Acoustics website,they have there the mathematics, that you can determine Fs and Qtc from your measurements.
After this you should use one of the different excel-spread sheets to determine the values of the needed components ( capacitors and resistors ).
In my design I used one capacitor exactly ( 100nF - buy 1% or measure ) -
the other capacitors are 3 in parallel - just calculate and measure - you can acieve less tan 1% tolerance.
The resistors are 2 in Series - just use 1% E96 or E192!!
The Speakers in my Box are Scan Speak Revelator - 25cm - very good and
expensive Speakers.
Instead of a volume of 200l they have 47l and a fs of 15 and Qt of 0,5 ( by LT-Transform ).Before the LT-transform I used a HP-Filter with 15Hz and low ripple, because of subsonic frequencies.The Subs are powered with ICEPOWER 125ASX Modules with an upper limit of 300Hz ( minidsp 2x4HD ).
Acoustically they are overwhelming!!!!
Just hear " hell freezing over" from EAGLES " hotel california" and you know , what I mean.
It is a closed System with very low ringing - just simulate in opposition a bassreflex with this drivers!!!!
So: LT-transform is not for commercial use because of much individual measure and calculation.
For DIY-use it is an tool - giving a extremly good bass from moderate volume of sealed enclosure ( but with a lot of work!).
I have built several LT-transforms for subs - thanks to Siegfried for his great work for diy!!!!
The problem is, and thats why it is not much used, you have to measure and calculate every unit individually.
First calculate any driver and its closed enclosure individually - there are a lot of programms ( unibox etc ) online - what sound pressure I need? what is the
closed enclosure volume etc. For this calculation the TSP-Parameters of the manufactors are enough!!.
Then build the enclosure!
After the driver has worked at 20-30Hz ( open air ) at 10W for a day or longer
, put it into the enclosure.
Then measure its DC-resistance - not with a multimeter, but precisely - I used a precision current source with 10,03mA.
The next thing is to measure the impedance curve of the speaker.I used LIMP of the ARTA Softwarefamily - be sure, that the comparing resistance is exactly.
Than go to SB-Acoustics website,they have there the mathematics, that you can determine Fs and Qtc from your measurements.
After this you should use one of the different excel-spread sheets to determine the values of the needed components ( capacitors and resistors ).
In my design I used one capacitor exactly ( 100nF - buy 1% or measure ) -
the other capacitors are 3 in parallel - just calculate and measure - you can acieve less tan 1% tolerance.
The resistors are 2 in Series - just use 1% E96 or E192!!
The Speakers in my Box are Scan Speak Revelator - 25cm - very good and
expensive Speakers.
Instead of a volume of 200l they have 47l and a fs of 15 and Qt of 0,5 ( by LT-Transform ).Before the LT-transform I used a HP-Filter with 15Hz and low ripple, because of subsonic frequencies.The Subs are powered with ICEPOWER 125ASX Modules with an upper limit of 300Hz ( minidsp 2x4HD ).
Acoustically they are overwhelming!!!!
Just hear " hell freezing over" from EAGLES " hotel california" and you know , what I mean.
It is a closed System with very low ringing - just simulate in opposition a bassreflex with this drivers!!!!
So: LT-transform is not for commercial use because of much individual measure and calculation.
For DIY-use it is an tool - giving a extremly good bass from moderate volume of sealed enclosure ( but with a lot of work!).
Attachments
=1, great post.
Right, if you go to enormous effort, you can add a LT-like bass boost and then buy a DSP and mic to finish the job.
Or you can just buy a DSP and a mic and finish the job.
If your purpose is to have intellectual stimulation, certainly there must be much more stimulating activities than making an EQ.
B.
Last edited:
effort is worth
It is not enormous effort,only some hours.A stimulating activity is to think in advance. Just a dsp and a mic is not enough. Just see the minidsp 2x4Hd used in my example.
The output impedance of the 2x4Hd is quite high and the input impedance of the icepower is quite low. So my LT module acts like an buffer with high input impedance and the output impedance of a LME49710.
And think a lot of gain. Many woofers have a sensitivity less than mid/high. So may you have to add +6 to +12dB of gain to the woofer. At my pcbs you see potis. They allow adjustment of gain by +- 10dB. The minidsp has a maximum output of +8dB ( I have not tested, but I think the quality at +8dB output is not so good ) and beyond that you get heavy distortions.
But it is even worse. The LT-transform ( it is not a simple bass boost ) adds at certain frequencies, depending on design ,
another +6 to + 15 dB of gain. So may be, you have to lower the mid/high gain in the dsp by -20dB or more with consequnces for THD/N and the following amplifiers.The "discrete" solution is powered by +-15Volt and has more than 20dB headroom. It would have been possible to programm LT-transform (minidsp can use in advance biquads ), but because of input/output impedance and gain the external solution is much better.
I have used also programmed LT biquads, but in different solultions. For examle a markaudio alpair7 in 1l sealed Volume. With the Linkwitztransform you linearize the frequency down to 50Hz. Then the highpassfilter off100Hz (crossover to subwoofer ) will work perfectly.
It is not enormous effort,only some hours.A stimulating activity is to think in advance. Just a dsp and a mic is not enough. Just see the minidsp 2x4Hd used in my example.
The output impedance of the 2x4Hd is quite high and the input impedance of the icepower is quite low. So my LT module acts like an buffer with high input impedance and the output impedance of a LME49710.
And think a lot of gain. Many woofers have a sensitivity less than mid/high. So may you have to add +6 to +12dB of gain to the woofer. At my pcbs you see potis. They allow adjustment of gain by +- 10dB. The minidsp has a maximum output of +8dB ( I have not tested, but I think the quality at +8dB output is not so good ) and beyond that you get heavy distortions.
But it is even worse. The LT-transform ( it is not a simple bass boost ) adds at certain frequencies, depending on design ,
another +6 to + 15 dB of gain. So may be, you have to lower the mid/high gain in the dsp by -20dB or more with consequnces for THD/N and the following amplifiers.The "discrete" solution is powered by +-15Volt and has more than 20dB headroom. It would have been possible to programm LT-transform (minidsp can use in advance biquads ), but because of input/output impedance and gain the external solution is much better.
I have used also programmed LT biquads, but in different solultions. For examle a markaudio alpair7 in 1l sealed Volume. With the Linkwitztransform you linearize the frequency down to 50Hz. Then the highpassfilter off100Hz (crossover to subwoofer ) will work perfectly.
Linkwitz via minidsp
A minidsp will give you the LT as well as lots of digital sound processing. Even a crossover. Around $100.
A minidsp will give you the LT as well as lots of digital sound processing. Even a crossover. Around $100.
Actually no, the LT circuit transforms the natural 2nd order high pass response of a driver or system to a different high pass response. For example, the system I'm building now has an eight inch sub, a five inch mid-bass, and a one inch dome tweeter that have the following characteristics: sub in its enclosure = Fs - 61.24 Hz / Qts - .7145, mid in enclosure = Fs - 103.6 Hz / Qts - .7531, and the tweeter = Fs - 1544Hz / Qts - 1.277.
For my design I'll be using the LT to change the sub to Fs - 35 Hz / Qts - .6, the mid to Fs - 125 Hz / Qts - .707, and the tweeter to Fs - 3000 Hz / Qts - .707. Doing that along with the usual high pass and low pass electrical filters gives a true acoustical 4th order Linkwitz-Riley response.
Mike
The detailed minutiae of Mike's explication of the LT transform illustrates perfectly how someone can see the forest yet miss the trees.
As I've been saying, the LT EQ in a rather abstract math manner takes some speaker factors (some of which frankly are guesses and others are the manufacturer's wishful thinking) and precisely recalculates their output ("guesses in, guesses out").
But it misses entirely the gigantic influence of the room and everything else that influences your preference of bass sound.
Yes, some kind of bass boost resembling the LT-EQ (or Fletcher-Munson) is likely needed in every home system. Yes, LT is a good guess which might or might not be a bit better than you taking a shot at playing with your DSP controls without the wasted effort of devising a LT correction.
B.
Yet somehow my speakers manage to both measure and sound good while performing as I designed them to. And yeah, I take the acoustical environment into consideration when I build speakers.
Mike
As MB says, I use LTs in exactly the same way. I've used them both created from opamps and within DSP software. Both work well.
And as to Ben's comments, using an LT to extend a subs natural extension, or to mould a higher frequency drivers response towards a place for correct acoustical integration is a completely different thing than using EQ to adjust for room effects.
I use LTs to extend my multiple subs extension down to 20Hz and then use notch filters on each sub to integrate the subs into the room. They booth solve different problems.
And as to Ben's comments, using an LT to extend a subs natural extension, or to mould a higher frequency drivers response towards a place for correct acoustical integration is a completely different thing than using EQ to adjust for room effects.
I use LTs to extend my multiple subs extension down to 20Hz and then use notch filters on each sub to integrate the subs into the room. They booth solve different problems.
Last edited: