Acoustic / Electrical Roll off??

[Thooms]

I've heard these phrases thrown around quite a lot - I'm not sure I understand the acoustic roll off, people seem to mention it but I can't see it in the measurements?

What am I missing?

I'm building an active crossover, so would a 4th order crossover with a driver that exhibits 2nd order acoustic rolloff work as a 6th order slope? Or does it not work like that?

[eb15]

Acoustic roll off is what you measure with a microphone.
Electrical is what you measure with a voltmeter/oscilloscope.

Speaker drivers have natural roll-offs low and high which interact with roll-offs designed into their crossovers low pass or high pass electrical filter roll offs to produce the final acoustic response.

[Dave Bullet]

Yes your assumption is correct. All things being equal, natural driver rolloff + electrical rolloff = acoustic rolloff.

Put another way - if you had completely flat frequency / impedance response (eg. 0dB across the spectum and nominal impedance) then the electrical rolloff is the transfer function applied by the crossover.

Of course, when the crossover reacts with a driver that has an inductive rise (increasing voice coil impedance), resonance peak at the bottom end and variable frequency response, the electrical slope wont be textbook and neither will be the resulting acoustic slope. That's why people usually refer to acoustic slopes (for passive systems anyway) when they mention crossover order unqualified. Target 4th other acoustic slopes may require any given electrical order, depending on where the crossover point is (Fc) in relation to driver passband. Lower order electrical is required at the end, due to the increasing natural driver rolloff (as you've noted) to achieve the required acoustic slopes.

[penngray]

Acoustic roll off is what you measure with a microphone.
Electrical is what you measure with a voltmeter/oscilloscope.

That sounds confusing. We measure everything happening to a driver and XO design with a microphone. Correct?

[Thooms]

Well this is the thing. I can do the electrical stuff, but I don't have a measurement mic. I'm going for an active design as it sounds like a better option to me. That, and I enjoy making amps

So, lets say I'm planning on building a 2.5 way system. I'm using Peerless HDS Nomex drivers (provisionally...), and I've taken inspiration from Zaph's ZDT3.5, with the two woofers using common roll off components, but the BSC woofer with an aditional low pass filter in the form of an inductor. I'm going to do the same, but with an LR4 network and a simulated inductor for the second woofer.

The question is, I've not seen many 4th order electrical networks. Is there a reason for not doing this? Is the slope too steep, or would it just be a lot of big, expensive components and a shallower slope will do?

I could do an LR2 network I suppose, but there looks to be a bit of breakup and I wonder if I could get away without notching it with an LR4 alignment. I guess the proof is in the listening

Thanks for the replies.

[Dennis H]

Well this is the thing. I can do the electrical stuff, but I don't have a measurement mic.

Get one. You can't really do serious speaker design without one. It's like trying to troubleshoot a circuit without a voltmeter or a scope.

http://www.parts-express.com/pe/showdetl.cfm?Partnumber=390-801

[Dave Bullet]

That sounds confusing. We measure everything happening to a driver and XO design with a microphone. Correct?

No. You could just measure the crossover over a nominal resistive load to see what the transfer function is.

[Dave Bullet]

Well this is the thing. I can do the electrical stuff, but I don't have a measurement mic. I'm going for an active design as it sounds like a better option to me. That, and I enjoy making amps

So, lets say I'm planning on building a 2.5 way system. I'm using Peerless HDS Nomex drivers (provisionally...), and I've taken inspiration from Zaph's ZDT3.5, with the two woofers using common roll off components, but the BSC woofer with an aditional low pass filter in the form of an inductor. I'm going to do the same, but with an LR4 network and a simulated inductor for the second woofer.

The question is, I've not seen many 4th order electrical networks. Is there a reason for not doing this? Is the slope too steep, or would it just be a lot of big, expensive components and a shallower slope will do?

I could do an LR2 network I suppose, but there looks to be a bit of breakup and I wonder if I could get away without notching it with an LR4 alignment. I guess the proof is in the listening

Thanks for the replies.

Assuming you have an existing passive design you like (aka Zaph's ZDT3.5) - you can make it active by mirroring the transfer function for your active circuits.

[penngray]

No. You could just measure the crossover over a nominal resistive load to see what the transfer function is.

Sure but what does that have to do with specific driver XOs?

[eb15]

One might measure the frequency response at each drivers terminals, to verify operation, (perhaps before connecting $$$ drivers), but that says very little about what the final summed acoustic response is out in the room, since it totally ignores the drivers actual response in its cabinet/baffle and their alignment in regards to each other.

[Thooms]

Hmm - well the design I'm mirroring is this one:

http://www.troelsgravesen.dk/Peerless_HDS830875.htm

But, I don't really understand what he's done with one of the filters...I'm trying to combine the features I like in Zaphs crossover, with drivers in Troel's design. Or at least that's the theory...

I'm going to have to get a mic I think. I was hoping I could do it based on Zaphs measurements, set the xfer functions and the gain and it would all go together and work without fiddling, but I guess the enclosure is going to play a huge part so that's not really going to work. Maybe that's the first step before designing the crossovers and building more amps...

An electret mic shouldn't be too expensive to build. I'll see what I can do

So do I need to target a specific slope, or could I just go with a pair of set slopes? Are there any designs out there which uses a 4th order electrical roll off, or is it done using the two roll-offs to achieve the desired slope?

[Saurav]

I guess the enclosure is going to play a huge part so that's not really going to work.

Correct. The dimensions of the enclosure and the positioning of the drivers will affect the individual driver response. The crossovers designed by Zaph and Troels are based on the actual measurements in the cabinets, so they accommodate all those parameters in the design. If you start changing things around and mixing/matching, you will most likely not end up with flat response.

An electret mic shouldn't be too expensive to build. I'll see what I can do

You might be better off buying the Dayton mic from Parts Express, since they now come with individual calibration files that are reasonably accurate. A DIY mic isn't much use unless you know how accurate its response is. You could send your mic in to be calibrated, but that will probably cost more than the mic from PE.

Ultimately, you need to decide how far you want to go. Based on your questions, I'm guessing you're somewhat of a beginner. I would suggest spending your initial budget on buying some books and doing some reading. Once you have a better handle on what it takes to design an enclosure, measure drivers, design a crossover, etc. ... then you can decide if you want to tackle your own design, or build one of the proven designs. Or, you could assemble something by picking parts from various designs, but how good that will sound will probably be a matter of luck.

So do I need to target a specific slope, or could I just go with a pair of set slopes?

Depends on the drivers you pick, the intended use of the speakers, and other factors I don't understand. Like I said, I'd suggest doing some reading.

This is a good link for some initial reading:

http://sites.google.com/site/undefinition/diy-mfaq

Make sure you read Part 2 as well, that covers some of what you're thinking.

But, I don't really understand what he's done with one of the filters

Hazarding a guess - it's probably some kind of filter designed to shape the response of the driver in a frequency region where the driver isn't behaving the way he wants to. It could also be something designed to compensate for the driver's impedance curve. Or both of the above. As you can see, if you use a different driver, this crossover filter won't be very useful.

[Thooms]

The problem is that I'm in the UK - it's very expensive to order stuff from the US after customs etc. have had their paws on it. The cheapest option I've found is the Behringer equivalent, but that's still the best part of $120...

Assuming my response is reasonably linear with frequency, then I'm not too worried about the absolute value. I may be able to borrow a calibrated mic however.

Thanks for that link. I'll look up those books he mentions, that's exactly what I was after!

I understand the basics, but getting these funny slopes and equalisation right all sounds incredibly complicated. A book to guide me through it a bit would be really handy. I can see how measuring the thing would make it a heck of a lot easier...

To try to better understand these thing I've been modelling crossovers in Spice - which is my issue with the Peerless crossover, it doesn't look to be doing what I'd expect it to with my limited knowledge. Either that, or I wrote the source file wrong

I'm a complete beginner. I have a good understanding of the concepts (physics undergrad), but the way you bring it all together in a speaker system is a bit of a learning curve...

At the moment I don't even know if I'll have space for a floor stander, if anyone knows of a design using non-metal drivers (I just fancy trying something different from my current speakers) that uses a 'normal' electrical slope and doesn't work too badly, I'd be keen to have a read :B

[Saurav]

Assuming my response is reasonably linear with frequency, then I'm not too worried about the absolute value.

Search the forums, there are threads where people have posted data of several Behringer microphones. There's quite a wide unit-to-unit difference in how flat the FR is. And you're right, that's the data you need, the absolute mic sensitivity is less important.

I understand the basics, but getting these funny slopes and equalisation right all sounds incredibly complicated. A book to guide me through it a bit would be really handy. I can see how measuring the thing would make it a heck of a lot easier...

It is pretty complicated, and at the end of the day, I firmly believe part of it is a black art Again, there are threads on various forums with tips and tricks and guidelines and general how-to's for crossover design. Those should help.

Also, one thing to keep in mind when trying to convert a passive XO to active - the passive will interact with the driver's impedance, while the active doesn't (because in general you'll have an amp connected directly to the driver, and the amp's output impedance will be a lot lower than the driver's impedance). So you may go off-track if you try to model the passive XO into a flat resistive load. Find the impedance curve of the driver and put that as the load for the circuit (not sure if your Spice app can do that, been a long time since I messed with LTSpice). Likewise, pull in the driver's frequency response and combine that with the XO's FR. When you start combining drivers, make sure you factor in the relative delays between them (since they'll generally have acoustic centers that are at different distances from the listener). A good XO will take all of these into account, and make sure the drivers overlap with proper matching of amplitude and phase.

if anyone knows of a design...

Check the 'Mission Accomplished' sub-forum, those are completed designs. Maybe you can find something there. Someone else may suggest specific designs, I'm reallyh not up to speed on what the popular designs are and what drivers they use.

[5th element]

The problem is that I'm in the UK - it's very expensive to order stuff from the US after customs etc. have had their paws on it. The cheapest option I've found is the Behringer equivalent, but that's still the best part of $120...

Thooms, we can order Behringer stuff from Thomann.

http://www.thomann.de/gb/behringer_ecm_8000.htm

Pricing is reasonable and includes VAT.

[Thooms]

I've found the Behringer for about £45ish which isn't too bad

I would still need to build a preamp however, but that's a lot less than I thought they were (I've looked before and remember it being near double that)...

This is something that I've been giving a lot of thought to over the past couple of days, and I think it may actually be easier to get the slopes right in a 3way than a 2way system (bear with me here...)

My 'dream' driver combo would be a Seas ER18RNX, Seas MCA12RC and a Peerless HDS on top. I looked at using the HDS164 drivers as that would be a little cheaper, but I now think it would actually be harder to get it all to work as the crossover point would be quite close to where the driver starts to roll off.

Now, I know the box will have a big effect on things (and I am planing to measure the things before I build the x-over), but using the drivers listed above I was provisionally thinking of crossing at for example 1k and 3k - this would mean the W and the T would be very close to flat (+/- 2db or so) for about an octave above and below the x-over point, with the mid flat 2 octaves above and below. My understanding is that the baffle layout plays a bigger part at higher frequencies anyway, which is something I'm keen to experiment with nearer the time.

These numbers come from measurements made by other people, and seem to be backed up by the manufacturers data. As this would be my first full range build, I imagine this could make life a bit easier than having to modify slopes from 'standard' arrangements, and trying to sort out the phase craziness that my simulations seem to indicate can occur when doing this (although I'm sure this would be difficult to hear, if I can minimise it, I might as well ).

Any thoughts? I'm trying to figure out what I'm most likely to be able to get to work without having to buy different drivers after the event because I can't get it to work together. In this respect, the well-behaved nature of the drivers I've looked at seems like it could make life a bit easier.

Thanks for the Misson Accomplished pointer, Saurav. There's some seriously tasty stuff in there - although most of them use metal cones - something I'd like to avoid in pursuit of trying something I've never used before

Still, lots of good knowledge in those threads...