1st Order High Pass on Compression Driver?

[merlin]

I could do with some reassurance from some of our resident experts.

The situation is as follows. In a two way, I am using a compression driver, whose recommendations are for a 12bd per octave slope at 600hz or higher. Power handling is 30w/60w peak.

I am using an active crossover and employing a first order slope at 750hz. To complicate matters, I am then putting a 33uf cap in series with the driver to offer protection from switch on.

Now my main question is, in this setup, does that 33uf cap act as an additional first order slope below 375hz, making a summed 2nd order slope below that frequency with the two first order slopes (one active one passive)?

I really need to know if that's the case. It must be as far as I can see but some clarification would help as people are warning me against using this first order slope. If the passive cap does turn the existing 1st order into a second order then the power demands seem to be perfectly safe and within recommended guidelines. Or am I getting horribly confused?

[Hoerninger]

Good morning,

these two first order high passes will sum up to a second order high pass for very low frequencies.

But in the cross over region ( - 3 dB) this combined filter will never be like Butterworth or similar even if the two frequencies are the same. In the cross over region the slope is much alike that of a first order filter.
____________
Peter

[Ian Mackenzie]

Hi Merlin,

I'll jump in here first as you pm the question earlier.

In terms of best practise a capacitors should always be used for protection when biamping the compression driver.

Details of this type of protection are spelt out in the 5235 JBL crossover manual. (page 18 table 3.)

http://www.jblproservice.com/pdf/Vin...5%20manual.pdf


There are two ways of looking at the crossover transfer function.

You can (presumably) have a 2nd order active filter with your active crossover. The benefit being you can control and modify the transfer function ie filter Q/type to suit the design in terms of on axis and power response and phase response.

Planting a fixed value passive element right in the passband will make it very difficult to optimise the crossover filter. (A parrellel inductor in a passive hi pass filter also acts as a pole)

Select the protection capacitor to be well out of the passband of the driver. (see JBL 5235 manual above).

If your high frequency power amp is a valve unit it is unlikely the compression driver will see any dc offset but may still suffer damage from a low frequency pulse.

Technically the above approach will give you flexibility and control to optimise the crossover region. Your expensive driver will be protected from dc, pulses and any likely hood of pushing the drive beyond its limits at the low end of the spectrum which I sense you are concerned about.

In terms of protection looking at table 3 in the 5235 manual if your driver is 16 ohms 30uf should be fine, if its 16 ohms double that value.

The active crossover should be set up for 12 db slope operation at 600 hz. (or above)

This assumes your active filter can be tailored.

The other route (as used in the 4435) is to use a capacitor as one of the poles of the filter and also offers protection. In this instance they engineered an integrated solution. This would required siginificant design and trial and error testing to get it right.

If you are in the design or optimising phase I would recommend you adopt the former appoach.

In terms of crossover suitability to answer you question directly I dont think a 1st order filter is a wise choice unless you can model the overall voltage drive to the compression driver. I would treat that as a risk particularly if running the driver hard as it may exceed the X max of the suspension and will lead to distortion or damage.

The above is more about an approach to dealing with the issues of protection and design of the actual crossover slopes, not actual values as such. If you have a variable active crossover this would be practical and the approach I would recommend.

Hope this helps.

Ian

[merlin]

Hi Peter and Ian, thanks for that. I still don't understand though!

Particularly when you Peter say at "very low frequencies". My understanding would be that at 750hz with the 1st order, we have -3db at the crossover frequency . At 375hz, one octave below, we would be -9db but the series capacitor is coming in as protection at that frequency and acting as a second pole there so we have -12db. Anyhere below that, we have effectively two seperate first order slopes summing to one second order. Is that wrong?

If the series capacitor does act as a second pole from one octave below the crossover frequency then there is no problem using 1st order as I see it. As I understood it, a capacitor will attenuate the signal by 6db per octave below a certain frequency. Surely it does that regardless of whether the incoming signal is level in frequency or already attenuated in a similar manner by an active device? And finally, and this is where I am probably fouling up, if you attenuate a 6db per octave slope by 6db per octave, do you end up with a 12db per octave slope? If not are there any calculators out there?

If you do end up with a second order slope then some quick calculations sugggest something close to the following:

Order 1st 2nd


750hz -3db 0

600hz -4db -3db

450hz -8.5db -9db

375hz -12db -12db

300hz -16.8db -15db

150hz -28.8db -27db

75hz -40.8db -39db

[Hoerninger]

Anyhere below that, we have effectively two seperate first order slopes summing to one second order. Is that wrong?

You are right, that is what I have meant. Two independent filters sum up as you have done it.

If you have a 12 db/oct filter topology you have an influence on the cross over region (Butterworth, Bessel, Linkwitz, Tschebyscheff and others). This can't be achieved with two independ 6 dB/oct filters. Far off this cross over region there is always a fall off with 12 dB/oct, I mean at frequencies with about a tenth of the cross over frequency or below with high passes.

For higher order filters there can be added several independent filters with 6 dB/oct and 12 dB/oct respectively. (There are other ways as well.)
Thus the capacitor for protection can be part of a higher order filter, may be 18 dB/oct.

I did not want to go into filter design in deep. Ian has given you a vision.

The electric filter topologies alone are not a sufficient approach for a loudspeaker filter. Speaker responses and enclosure influence must be considered as well. Approved recommendations are always a good starting point.
___________
Peter

[merlin]

Sorry Peter I still don't understand what you are saying - apologies.

Are you saying that a capacitor in series will not attenuate an incoming signal that is already reducing by 6db per octave, by a further 6db per octave? How does the capacitor know that the signal has aready been actively filtered?

Maybe I'm wrongly describing what I have as 1st order. It is an active 1st order slope with a passive cap inserted an octave below the F3 point as recommended by JBL amongst others.

Are my calculations wrong? That's all I really need to know. If they are wrong I'll start again. If they are right - it sounds great and at no point is the pseudo 1st order attenuation less than the text book 2nd order

[Hoerninger]

Does that 33uf cap act as an additional first order slope below 375hz, making a summed 2nd order slope below that frequency with the two first order slopes (one active one passive)?

Yes.

All capacitors contribute to the attenuation. Their attenuations can be calculated independently from each other as they are separated by amp stage(s).

All I wanted to express that your second order filter has a very "soft" cross over region.
(May be details should be discussed by PM.)
__________
Peter

[merlin]

Thanks Peter - I've sent you a PM.

[Ian Mackenzie]

Hi Merlin

http://www.passlabs.com/downloads/xvr1_om.pdf

Please refer to page 10 of the manual and look at the curves.

The 1st order as Peter says is gradual..almost a straight line. The combined reactance of the 2 poles in the 2nd filter with a Q of 0.707 create the 2 pole filter per butterworth/ LR filter. From memory a 1st order filter does not have a Q factor.

What I am saying is you may get the required attenuation at a given frequency but not the the desired rolloff rate or Q factor for the crossover to work correctly by attempting to sum two independant 1st order filters.

Please explain why you are trying to go this route?

Ian

[merlin]

Sorry Ian you are losing me totally I'm afraid - I suspect we may be a crossed purposes. I am just trying to confirm the effect the series protection cap has on the roll of in the stop band.

A capacitor in series results in a 6db per octave attenuation below a set frequency as I have always undertood it. Therefore, if a signal comes in an octave below the F3 point and that signal was originally say 100db, it will be attenuated to 91db. Is that correct? Now if that signal was really originally 109db but has been attenuateed to 100db by an active filter an octave higher before the capacitor sees it, it's still going to end up at 91db isn't it? And that would be a total attenuation of the music signal of 18db.

My goal is to use a first order slope - specifically in the octave around the crossover. There is always going to have to be a protection cap an octave lower isn't there? So it's only a first order slope in the crossover area itself. Another reason is it sounds shootloads better than using 3rd order. I see Pass Labs recommend the lowest possible order too. In my case, the horn/compression driver starts to roll off at 6db per ocatave acoustically below 900hz, whilst the bass unit, a 1500AL has a rising response above 500hz of 6db per octave. I'm therefore trying to use a third order slope on the 1500al and a first order on the horn, resulting in matching second order slopes at the crossover point. I hope that makes sense!

Put another way, in thrying to answer my quandry, if we follow guidelines and insert a series cap 1 octave below the crossover point, does that not mean that a 1st order slope becomes a 2nd order slope below this lower point, and that a 2nd order would become a third order and 3rd a fourth?

[Zilch]

The answer is, "Yes."

I believe Ian is making two additional points:

1) The acoustic transfer function will be different because of the response curve of the driver, so it's important to measure and incorporate that into the design.

2) With the two filters being an octave apart, the poles will not interact to contour the knee at the crossover frequency, i.e., there's no "Q" control; it's just the active first order rolloff in that region....

[merlin]

Zilch, Ian and Peter thanks.

So it sounds great and it's safe. It was the safety aspect I was mainly concerned about. I'm only just starting out with these and did not want to do something silly. I appreciate the confirmation guys - people were telling me I was risking damaging the diaphrams and I couldn't for the life of me see why. I was starting to lose my sanity!

Cheers to all!

[Ian Mackenzie]

Good morning,

these two first order high passes will sum up to a second order high pass for very low frequencies.

But in the cross over region ( - 3 dB) this combined filter will never be like Butterworth or similar even if the two frequencies are the same. In the cross over region the slope is much alike that of a first order filter.
____________
Peter

Peter,

I agree and I suppose that was my point.

Merlin,

Sorry if I confused you!

Ian

[merlin]

No problem Ian and I appreciate the input.

My goal was to use a first order electrical slope but keep the compression driver safe. It seems that's OK. What was confusing me was everyone on AA and elsewhere were saying I could destroy the TD4003 diaghram - and I couldn't understand that!

The active crossover by the way is an XVR-1 by coincidence. Superb unit. And it does allow me to adjust the Q of the 1st order slope which I have set to medium. I use low Q on the 3rd order LP. Hopefully I'll have worked it all out by the time you take an autumnal stroll through London! Do you enjoy a beer at all?

[Mr. Widget]

Do you enjoy a beer at all?

I have it on good authority that Ian certainly does!


Widget