1st order Passive Line Level Crossover + 2nd order Highpass?

[alexclaber]

I'm trying to design a bass cab with fairly flat frequency response from 50Hz - 6kHz using an Eminence Kappalite 3015LF in 100 litres ported cab tuned to 50Hz plus an 18sound 6ND410 in 2 litre closed cab.

My longterm plan is to use a digital active 24dB @ 500Hz crossover with compression and limiting to maximise output at high SPL but finances do not currently permit this. I've looked into a full passive crossover but once impedance compensation is included it gets very expensive, especially bearing in mind the amount of power it will have to handle.

As my amp has two channels I'd like to try biamping with a passive line level crossover. Research so far suggests that a 6dB lowpass @ 500Hz on the woofer will work ok. I've looked into a 12dB highpass @ 500Hz on the midrange but it appears that the filter knee will be a bit strange as it's cascaded 6dB filters. I gather that there is some voltage loss but I have quite a bit of gain available on my preamp to compensate. So It looks like a 6dB highpass @ 500Hz on the midrange will be ok.

However, as this is for a pro bass rig all the connections are balanced. I read a comment on here suggesting that the balanced wiring can be used to change the filter slope - can anyone expand on that? Otherwise should I simply build identical filters for both sides of the balanced wiring?

Finally, to protect the midrange I'm looking to add a 12dB passive speaker level highpass @ 500Hz to give a total slope of 18dB. What Q should I design this highpass for to get the combined line level and speaker level filter to act like a true 3rd order butterworth filter.

One more thing - is there anything else I need to consider when combining a 1st order lowpass and a 3rd order highpass, both at 500Hz?

Oh, and another thing - I have a plot of the impedance magnitude for the midrange speaker. Based on this data how do I design an impedance compensation circuit to flatten what the crossover sees at resonance?

Thanks ever so much for any help!

Alex

 

[SemperFi]

I dont have answers to all your questions, but I've just built something similar and have a few observations.

Because of the different phase response of different filter slopes, it is common to use same filter slopes in both high- and lowpass filters. 2nd order gives 180degree phase at cutoff, so one of the drivers are connected in opposite phase to compensate. In most pro audio speakers with passive filters I think 2nd order is the norm.

If impedances become problematic with a 2nd order passive at line level, you can always use one 1st order at line level, and another 1st order at speaker level which will result in 2nd order slopes.

You may not need to filter out the highs to the bass drivers...
I just finished a similar bass system for the band I do the sound engineering for. We had 4 active 2-way speakers with 15" bass units and horn tweeters. When I tried to bring the bass and kickdrum up the whole sound went groggy on us. So I built two cabs with 2x15" jensen neo150 bass drivers and a single hornsub with 2x12" bass drivers. Now I can run the bass and drumkit separate into the bass system, and everything else thru the 4 active speakers. The sound is punchy and clear even at much too loud levels.
Because of time and budget I havent built a crossover yet, but the sound is actually spot on for these instruments. The Jensen 15" bass drivers have a pretty flat response to a little over 2kHz, and the horn sub goes from about 40-250Hz. For the drumkit and bass this is pretty good. The bass is plenty articulate and clear without a tweeter.
Of course the location where we play has a big impact on the low end, but so far we are very happy and are thinking of using our money on other things than the digital crossover we are looking at.

 

[richie00boy]

The midrange will have a 12dB/oct roll off in the sealed box, therefore you need only to add a 6dB/oct filter to achieve your 18dB/oct target. If you set the resonance frequency of the filter to the same as that of the speaker/box resonance, the total Q will be the product (multiplication) of the speaker/box Qts and 0.7071 which is the filter Q.

You don't need any impedance compensation on the speakers as you are using line level filters.

 

[alexclaber]

The midrange will have a 12dB/oct roll off in the sealed box, therefore you need only to add a 6dB/oct filter to achieve your 18dB/oct target. If you set the resonance frequency of the filter to the same as that of the speaker/box resonance, the total Q will be the product (multiplication) of the speaker/box Qts and 0.7071 which is the filter Q.

As this is to be used in high SPL environment with high power amps (550W per channel) I'd like to give the midrange more protection from low frequencies than a simple 6dB filter, even though the sealed box provides an 18dB slope in response. Also I'm not sure I can get the box small enough to raise the Fs to 500Hz whilst fitting in the speaker and enough insulation to prevent reflections. At the moment Fs is 262Hz in 2 litres closed.

You don't need any impedance compensation on the speakers as you are using line level filters.

I still think the speaker needs at least a 1st order highpass at speaker level to reduce excursion from unwanted lows, if not a 2nd order highpass, to give an 18dB electrical slope and 32dB acoustic slope.

I found some info on designing an LCR circuit where you stick the Qes, Qms, Fs and Re into the calculator and it gives you the result. However as placing the speaker in a small sealed box raises the Fs hugely and changes the Q I tweaked the parameters in WinISD pro until the Q, Fs and Re for the adjusted raw speaker matched the real speaker in a 2 litre box, and then plugged those figures into the calculator. Would this have been the right thing to do?

The results were:

C = 173.97uF
L = 2.12 mH
R = 6.82 Ohms

Alex

 

[richie00boy]

If 6dB/oct is not enough then make a 12dB/oct filter giving you 24dB/oct acoustic on the midrange. Same principles apply.

I don't see why you can't make the 6dB/oct highpass 'protection' filter passive line level. Trying to do that at speaker level means huge expensive inductors and an absolute nightmare trying to get the impedance smoothed out.

Have you considered just making the whole thing using analogue active filters and forgetting the digital lark? You could do it for about £50.

 

[alexclaber]

If 6dB/oct is not enough then make a 12dB/oct filter giving you 24dB/oct acoustic on the midrange. Same principles apply.

How does one go about desiging a 2nd order passive line level highpass? Is it as simple as cascading two 1st order filters? I have a suspicion that the load impedance could cause issues.

I don't see why you can't make the 6dB/oct highpass 'protection' filter passive line level. Trying to do that at speaker level means huge expensive inductors and an absolute nightmare trying to get the impedance smoothed out.

Yes, the LCR circuit is the expensive bit. If I could just put cap inline with the midrange it would be easier.

Have you considered just making the whole thing using analogue active filters and forgetting the digital lark? You could do it for about £50.

Not really, this was just a stop gap. But maybe I should...

I'm now considering another option - if I raise the crossover point to 750Hz then I reduce the load on the midrange and can use the PLLXO 1st order on both woofer and midrange, plus a 1st order passive speaker level highpass on the midrange, and as the midrange resonance is at 250Hz the crossover point should be far enough above it to not worry about impedance compensation. Correct?

On the other hand, if the PLLXO can be done 1st order on the woofer and 2nd order on the midrange then that's even better. I do need about 4dB of attenuation on the midrange so maybe the voltage loss in the filter can do that for me?

Alex

 

[jerko]

Wouldn't an active crossover make more sense for what you are doing?
http://www.partsexpress.com/webpage.cfm?webpage_id=3&CAT_ID=46&ObjectGroup_ID=790

Of course I don't really know what I'm talking about, so...

 

[richie00boy]

I'm getting confused, lets try and clear some things up. You say you are bi-amping but then you go on about needing a passive speaker level filter. Bi-amping means each driver has it's own amp channel so there is no need for speaker level filters you can do it all at line level.

And yes a passive 12dB/oct line level filter can be as simple as cascading two 6dB/oct ones.

 

[alexclaber]

I'm getting confused, lets try and clear some things up. You say you are bi-amping but then you go on about needing a passive speaker level filter. Bi-amping means each driver has it's own amp channel so there is no need for speaker level filters you can do it all at line level.

Sorry, I am being confusing! As the information I've so far found on passive line level filters suggested that any more than a 6dB slope was unlikely to respond well (low Q etc) I thought I could obtain 6dB of filtering from the PLLXO and then add an extra 6 or 12dB of slope on the high side of the amp with a speaker level as the response would be more accurate. However that does require impedance compensation and even then voice coil temperatures will affect things.

And yes a passive 12dB/oct line level filter can be as simple as cascading two 6dB/oct ones.

Ok right, I'm sold! Could you help me out with the numbers on this?

My preamp has an output impedance of 600 ohms balanced - it's an Avalon U5 so should have tons of current capacity and be untroubled by low impedance loads. My power amp has an input impedance of 12000 ohms balanced, 6000 ohms unbalanced. I'd like to drive it balanced due to the extra gain (6dB I think).

I'm going to shift the crossover point to 750Hz.

According to my sums, assuming that I have to use one filter on each side of a balanced lead, then for the lowpass 6dB filter:

The filter sees the amp's input impedance as 6000 ohms (half the balanced impedance). The resistor in the circuit will be a 6000 ohm resistor. So the capacitor needs to be 0.0707uF to get the 750Hz lowpass point. This gives a voltage loss of 0.5 or 3dB - does that give a power loss of 6dB?

So how to cascade two filters so they respond correctly? The second filter should be straightforward as it sees the amp's input impedance. But what impedance will the first filter see?

Thanks again!

Alex

 

[richie00boy]

The Q of cascaded 6dB/oct passive filters is droopy but not disastrously so, in fact it creates a 2nd-order Linkwitz-Riley crossover when used with the corresponding filter on other driver.

However, with 600 ohm output impedance and 6000 ohm input impedance you cannot implement any kind of passive filter whatsoever, so you will have to go active. Can you confirm the output impedance really is 600 ohms and not the heaviest load is 600 ohms? I would expect the output impedance to be in the tens of ohms for a good quality output stage of pro equipment with electronic balanced outputs.

 

[alexclaber]

The Q of cascaded 6dB/oct passive filters is droopy but not disastrously so, in fact it creates a 2nd-order Linkwitz-Riley crossover when used with the corresponding filter on other driver.

Oh ok, that's good.

However, with 600 ohm output impedance and 6000 ohm input impedance you cannot implement any kind of passive filter whatsoever, so you will have to go active. Can you confirm the output impedance really is 600 ohms and not the heaviest load is 600 ohms? I would expect the output impedance to be in the tens of ohms for a good quality output stage of pro equipment with electronic balanced outputs.

Ah, you may be right! Here's what Avalon's rather bare specs say:

Maximum Output Level (Class A): +30dBu DC coupled, balanced 600 ohms (mic and line out)

So does that mean the heaviest load it can drive is 600 ohms and therefore the output impedance is likely to be a tenth of that?

Thanks!

Alex

 

[richie00boy]

It would seem that the spec is quoted driving 600 ohms, whether the output impedance is lower than that I can't say. I would imagine it is if it's electronically balanced.