[andyr]

for those ( like me) who didn't know to much how to implement a Zobel:

Thanks Kyrill - great article!   

Regards,

Andy

[andyr]

Hi Dave,

Also, I'm wondering if this type of RC Zobel is recommended then why stop there.  Why not add an LCR conjugate in shunt with the driver to address driver resonance at lower frequenices as well??

We did try that, and found that the components were very expensive (10,000uF electro and very large inductor) and the audible benefits were zip......

Hugh

Hi Hugh,

In an active setup, why wouldn't you simply use a HP filter to roll off the mid-range or tweeter (well) before the driver LF resonance point ... so there is no need to Zobel-correct the driver?

Regards,

Andy

[AKSA]

Hi Andy,

Hmmm.  Four reasons:

#1  GNFB amplifiers are badly affected by phase differences presented at their outputs by drivers.
#2  These disturbances affect the nfb loop at that very important crossover point, where much of the spatial information is held in the music.
#3  Interactions taking place at the driver/amp interface cannot be corrected at the input to the amp, only at the driver voice coil.
#4  Phase correction on active filters is centred around the crossover points where one driver hands over to another;  this is a separate issue to the phase shift introduced at the driver and its attendant effect on the amplifier.

Clear as mud?

Hugh

[andyr]

Hi Andy,

Hmmm.  Four reasons:

#1  GNFB amplifiers are badly affected by phase differences presented at their outputs by drivers.
#2  These disturbances affect the nfb loop at that very important crossover point, where much of the spatial information is held in the music.
#3  Interactions taking place at the driver/amp interface cannot be corrected at the input to the amp, only at the driver voice coil.
#4  Phase correction on active filters is centred around the crossover points where one driver hands over to another;  this is a separate issue to the phase shift introduced at the driver and its attendant effect on the amplifier.

Clear as mud?

Hugh

Sorry, Hugh ... you've answered the Q I already know the answer to (because of previous posts).

I'll re-phrase the Q that I would like the answer to:

"Instead of using a Zobel to flatten the LF resonance peak of a driver (which you said would typically involve a very big inductor and 10,000uF of capacitance), why not simply cut off the LF extension of the driver with an active high-pass filter, well above the resonant frequency?"

(Yes it is necessary to use a Zobel circuit to "tame" the HF inductance rise.)

Regards,

Andy

[Davey]

Andy,

You can't cut off the LF extension of the driver because it's the woofer (primarily) that we're talking about here.  The resonant peak is in the area of interest. 

However, it appears that Hugh's amps (or others) must be doing a decent job of handling these phase angles above/below the resonant peak because when he tried it "the audible benefits were zip." 

Also, I'm not sure where Hugh got the "very big inductor and 10,000uF capacitance" numbers.  I just flattened the resonant peak (~38Hz) of the W22 woofer in my Orion's with a 423uF capacitor and 27mH inductor.  Large component values yes, but a LONG way from very big and 10,000uF.

In any case, a person has to be very careful hanging these types of component values across an amplifier output.  Yes, the calculation says the impedance is now resistive, but there's a considerable amount of stored energy in a resonant circuit like that.  That type of circuit can toast the output stage of many amplifiers. 

What might be worth doing for those using direct drive connections is to add a Zobel at the far end of the speaker cable to terminate the line for above audio frequencies.  Say a 10 ohm resistor and 0.27uF capacitor.  That would prevent any possible antenna pickup from the speaker cables being applied to the amplifier output.

My two cents.

Cheers,

Dave.

[andyr]

Andy,

You can't cut off the LF extension of the driver because it's the woofer (primarily) that we're talking about here.  The resonant peak is in the area of interest. 

Dave.

Aah, thanks, Dave.     But I presume for mid and tweeter we would (cut off the LF extension to avoid the resonant peak)?

Regards,

Andy

[Kevin Haskins]

Thanks Daryl,

And this introduces the next topic......  impedance of speakers, and how it varies with frequency.

This is REALLY important, and when you look into it, you find many speaker manufactures who compromise the impedance curve so severely that the load becomes very difficult to drive with conventional amplifiers.

This is a hobby horse of mine;  the speaker with an impedance peak of 45R and a dip to 1.5R.  Don't laugh. These things are common, and they play merry hell with amps.

Hugh

With modern measurement and modeling software, you don't need to apply impedance compensation before designing the crossover.   In the old days it helped greatly to have a flat impedance so the network would behave as the "text-book" filter says it should.   These days, the software handles all the number crunching so its just easier and quicker to leave them out, until your done with the crossover design.

I use my impedance compensation network as a last step in design.   And, I agree with Hugh.   Its just bad design to have a loudspeaker that has an ugly impedance curve.  In terms of back EMF, as Daryl said, its part of the impedance of the driver and caught in the measurements.    The non-linearities in transducers are such that inductance, BL & suspension non-linearities are the dominant forms of distortion.    For most home audio loudspeakers, BL & eddy currents from the VC & motor interaction dominate power compression.     


 
 

I'm going from memory but I think this took a 30uF cap, 0.025mH inductor and a 8 ohm resistor across the binding post.    It brought the impedance peak at crossover down from 25 ohms and lowered the high frequency impedance to match the target 5-8 Ohm impedance.   

[AKSA]

Hi Davey,

FWIW, your 2 cents appears to be happily moving in the opposite direction to the USD.......  this is very good news for you, Sire.

In the design of the VSonics, we used Xover Pro, latest version, but I don't recall the values.  I think they might have been 10mH and 4,700uF (we used a couple of 10,000uF in series to get the non-polar effect), but my memory is poor on this one as it's a while back.  We tried it out with and without, and couldn't pick the difference in listening tests, so dropped it.  That inductor was pretty big, and at this low frequency we would otherwise have been tempted to shove in a few strips of M4 lamination.

We later checked with an experienced tech at Tymphany in Denmark, and he confirmed that most don't bother with impedance correction at the resonant peak.  It's just not worth it, and good design elsewhere in the crossover can control this peak appreciably anyway.

The amp is indeed tolerant of phase shift.  The latest iteration, the Soraya, will drive electrostatics (Acoustats) to almost destructive levels of sound pressure without raising a sweat.  I have discovered a secret way of compensating an amp for a range of reactive loading without sacrificing sonics.  This is pretty trick stuff, and I'm actually very proud of it.  I must gratefully acknowledge help from a very smart young Canadian, CB (you know who you are, Sir!) who greatly assisted with this recent work.

Hi Andy,

"Instead of using a Zobel to flatten the LF resonance peak of a driver (which you said would typically involve a very big inductor and 10,000uF of capacitance), why not simply cut off the LF extension of the driver with an active high-pass filter, well above the resonant frequency?"

As I believe was stated, you can't do this on a woofer, because you need appreciable output at resonance and a bit below.  And I'm always wary of introducing additional active or passive filters;  they damage the music by introducing losses and phase shift.  You gotta keep it simple if possible.....   

Kevin,

Thanks for your post.  We certainly agree on most details, but your knowledge of speakers is much deeper than mine.  I tend to concentrate on the amp, but of course this is an integrated system, all too often forgotten.  I commissioned a guy to designed me a good TL two way speaker, I didn't do the work, all I did was chime in at the end of the crossover design phase, and my input helped to narrow the impedance range and improve the imaging, nothing more.  I will say, by way of bragging, that the same guy at Tymphany (who has now since retired to UK) uses our speaker in his home system, he loves it, and loved the slightly unconventional way we did the crossover (which is third order, crossover at 2,700Hz).

Cheers,

Hugh

[Agisthos]

Can you tech guys explain what are the pros and cons to the different crossover types.

eg 1st order, 2nd order, 3rd order, 4th order..... what does it all mean?

[DSK]

Agisthos, you might want to start a new thread for that question as it is off topic and the answers (and differences of opinion) could go on for many, many pages.

You might also want to google and have a look at some of the speaker building websites.

Cheers!

[AKSA]

Hi Agisthos,

A few bullet points:

1st Order:  gentle slope of 6dB/octave on all filters, that is, half the amplitude gets through at double the crossover frequency.  Theoretically the best because it does not much phase shift the signal, giving very good amplitude and transient response.  Disadvantage:  considerable out of band energy passed to the drivers, so needs VERY good drivers which are expensive.  Difficult to handle the out of band driver output, since it mixes with the in band signal from a very different driver, causing phase shift and lobing and beat problems.

2nd Order:  stronger 12dB/octave slope on all filters, that is, one quarter the amplitude gets through at double the crossover frequency.  However, there is a 180 degree phase inversion, which means drivers must be reverse wired, no big deal.  Quite a good approach, but criticised by some for poor imaging performance.

3rd Order:  vigorous 18dB/octave slope on all filters, that is, one eighth the amplitude gets through at double the crossover frequency.  Widely used, and probably the most popular arrangement since out of band frequencies are minimised without too much cost on transient performance.  No unhappy phase isues, and often praised for good imaging.  Can use inexpensive drivers.  Dimensioning a good 3rd order is quite tricky.

4th Order:  Very strong 24dB/octave slope on all filters, that is, one sixteenth the amplitude gets through at double the crossover frequency.  Rarely used on passive speakers (exceptions are the Ellis 1801B, and some of the B&W speakers) but widely used on active systems, such as the Orion and most pro-audio sound reinforcement systems.  Excellent for high power, driver rating can be fully exploited because few out of band frequencies are passed on.  No phase issues, but transient response can be compromised, and if configured for good transient response, ringing can be an issue.  Multiple, series connected filters are complex mathematically and the effect with audio is to slightly dull the impact and slam.  Some say detail is also compromised with the fourth order;  I doubt this.

The order of the filter is a simple choice.  The dimensioning of the filter, however, is problematic, and not even good software can make the design of a premium crossover an easy exercise.  Further, the quality of the filter components is important, too, with inferior caps in particular highly audible.

Hope this helps,

Cheers,

Hugh

[bhobba]

'However, it seems to me that an impedance-correction circuit is just as necessary for an active setup.     (Hence that adjustment, Hugh, which you made for marcus's Orion mid-range driver?)  Otherwise the amount of power which the amp is able to put out will vary widely with frequency, if the impedance ranges widely (like from 1.5R to 45R)?  Which won't sound any good at all!'

Sure.  But as with anything in life you rarely get something for nothing.  One of the advantages of active setups is you have the amplifiers directly driving the speakers - it is more immediate.  Impedance correction may get in the way of that - inductors and capacitors all introduce signal losses and distortions - but then again the improvement may outweigh the disadvantages - especially if the impedance correction is not in the signal path.  Personally, as a guy that is looking at an active setup using a DEQX, I like the idea of direct drive.  The idea of amplifiers damping factor directly controlling the speaker appeals.

Thanks
Bill

[andyr]

Sure.  But as with anything in life you rarely get something for nothing.  One of the advantages of active setups is you have the amplifiers directly driving the speakers - it is more immediate.  Impedance correction may get in the way of that - inductors and capacitors all introduce signal losses and distortions - but then again the improvement may outweigh the disadvantages - especially if the impedance correction is not in the signal path.  Personally, as a guy that is looking at an active setup using a DEQX, I like the idea of direct drive.  The idea of amplifiers damping factor directly controlling the speaker appeals.

Thanks
Bill

Hi Bill,

A Zobel circuit is in parallel with the driver - so it still lets the amplifier control the driver directly in an active system.

Regards,

Andy

[bhobba]

Hi Hugh and All

Great background to passive crossover design Hugh.  For my money though, even though it is at present the expensive route, the future is digital approaches like the DEQX.  Individually tailored drive and room correction with no distortions introduced by capactors and inductors.  The issue of course is cost - it is currently rather expensive.  The cost will come down in time of course.  Also, as Andy said, even getting rid of crossovers you can possibly still gain an improvement impedance correcting the speakers.  I suspect some experimentation is required to see if it results in sonic improvements.  Any takers?

Thanks
Bill

[bhobba]

A Zobel circuit is in parallel with the driver - so it still lets the amplifier control the driver directly in an active system.

Hi Andy

Thanks for the quick reply.  Of course.  The only issue in that case is if it leads to sonic improvements.

Thanks
Bill

[andyr]

Hi Andy

Thanks for the quick reply.  Of course.  The only issue in that case is if it leads to sonic improvements.

Thanks
Bill

As I understand it, according to Hugh an marcus it did (on the mid driver of marcus's 3-way active Orion system).

Regards,

Andy

[gerado]

Hi Agisthos,


The order of the filter is a simple choice.  The dimensioning of the filter, however, is problematic, and not even good software can make the design of a premium crossover an easy exercise.  Further, the quality of the filter components is important, too, with inferior caps in particular highly audible.



Cheers,

Hugh

OK I dont know and was hoping someone else asked so I did not have to show my ignorance.
What is Dimensioning of the filter
assuming a first order cross over is used -which is just a cap or coil , what else does dimensioning involve/mean

[AKSA]

Ah, that's easy,

Fixing the values of the components......  nothing more!

Bill,

Welcome back, and good luck with your studies!

CB105 is on the website by description and price and is presently selling.  More detail does need to be added, but the website has just been revamped and there is always more work to do......


Hugh

[bhobba]

Welcome back, and good luck with your studies!

Thanks.  Getting back into the advanced math is happening slowly - but it is happening.  Just as an aside, if you want to post formula in your explanations - that's fine with me.  But I understand if you don't want to - it turns most people off.

CB105 is on the website by description and price and is presently selling.  More detail does need to be added, but the website has just been revamped and there is always more work to do......

Thanks Hugh.  Found it.  Good to see you are offering it as a kit as well.  For me, with time to spare, that looks the go.  Like the AKSA will it also include detailed assembly instructions?  Is it still true that everyone who has attempted to build an AKSA has succeeded?

Thanks
Bill

[gerado]

Bill
I started off with only knowledge how to turn on a soldering iron
have both a GK and LF from kits.
The LF was more like an assembly task rather than building.
Real easy

Theo