Picking a Woofer: When different versions exist, go for the more sensitive one?

[JonW]

I’m narrowing down my driver choices for my MT design project. The leading candidates are the ScanSpeak 6600 tweeter (4 ohms, 92.5 db) and the ScanSpeak 18W(X)531 Revelator 6.5” woofers. These woofers come in various versions. There are:

A 4 ohm, 90 db sensative variant:

http://www.tymphany.com/datasheet/printview.php?id=84

An 8 ohm, 87 db flavor:

http://www.tymphany.com/datasheet/printview.php?id=85

A 16 ohm, 83 db version:

http://www.tymphany.com/datasheet/printview.php?id=86

And a shielded 8 ohm, 86.5 db one:

http://www.madisound.com/cgi-bin/index.cgi?cart_id=8841481.26641&pid=1163

In general, do you just pick the driver that has the highest sensitivity? I realize how that helps in needing less amp power to achieve a given SPL. That would be the 4 ohm/90 db one here.

I’m guessing the 16 ohm sensitivity (83 db) is way too low. And for the shielded 8 ohm driver, you loose 0.5 db relative to the non-shielded version, so no reason to get that. I’m left with the 4 ohm/90 db and the 8 ohm/87 db. From my reading, high sensitivity is a good thing. So just use the 4 ohm/90 db choice.

I’m also figuring that the frequency output, impedance, and distortion characteristics of these drivers are all pretty similar. They look to be from the ScanSpeak plots. Zaph and MarkK tested different versions (one did the 8531, the other did 4531) and both looked excellent, hence my likely choice of these drivers. (Except for the unfortunate impedance warble at ~1000 Hz)

Looking back at the Modula MT, MTM, etc. projects it seems that Jon picked the 8 ohm/87.6 db flavor of the RS180 over the 4 ohm/91 db version. So maybe I’m not thinking about it correctly. Or perhaps I recall that only the 8 ohm version of the RS180 was available when he did his designs.

Thanks.

-Jon

[Jed]

You have to first simulate a design based on the imedance curves and reach your target response and design goals.

For example, what amp will you use? Is it a tube amp? If so I'd stay away from a final impedance below 4 ohms.

The 16 ohm woofers are nice so that you can wire 2 in parallel to have a final impedance that doens't dip too low.

Most people use 8 ohm woofers for home hifi because they have a nice balance of tradeoffs, but using a 4ohm driver isn't out of the question. In fact, I'm using the 12m midrange for my next 3-way speaker. With the series resistor to pad it down, the final system impedance doesn't dip too low. It's more about the system with crossover, than the individual parts. Once you get that figured out, all the rest falls into place.

I'd recommend you take a look at the FRD tools and work on a simulation using the rjbaudio.com tutorial.

Jed

[cotdt]

they are all the same efficiency actually. sensitivity is different, though.

[JonW]

Thanks for the input, fellas. As you might guess, this is my first go around designing a speaker. I didn’t realize I needed to do any modeling at this stage. I thought I’d pick the drivers, build the cabinets, play around with various settings on a Behringer DCX2496 to see what sounds good, make measurements of the drivers in the actual cabinets, and THEN start with the crossover modeling. I was figuring it might be a waste of time to try and design the crossover without real measurements in the actual cabinets I’ll be using. But if that’s not the case, I can start now.

For amps… Idunno, haven’t decided. It’ll be solid state. I need to buy something to run this project so I don’t have to dismantle my current system while the MT design is ongoing. I’ll be using the Behringer DCX2496, so I need to pick up 4 channels of amps (maybe 6 channels of amps if I try a 3 way later). A decent powered receiver like a Yamaha RX-V1700 might be an option. Although I could be susceptible to being talked into used separates. Idunno…

[mazurek]

I went with the 4531G because my amps are rated to deliver more power into 4 ohm loads. LM3886 chip amps, and the Hypex UCD amps, and a lot of pro amps are rated to deliver more wattage into 4 ohm loads.

[TurboFC3S]

If you choose a 4ohm woofer and 4ohm tweeter in a normal parallel network, you're going to be dropping into 2ohm territory likely ... and unless your amp has a massive power supply far bigger than needed for it's rating, you're going to lose headroom and dynamics into those loads - you simply won't be able to supply the necessary voltages. Not to mention damping factor is much lower, of course the effect of that is debatable. But me personally, I'll give up the 2db or so max SPL in exchange for better dynamics and control any day. I think the vast majority of amps available just sound better if you feed them higher impedances, 6ohm or higher.

Jon, you definately need to set some design goals BEFORE choosing a driver. All those drivers really are the same sensitivity, as cotdt said. So it really just matters what your physical driver deisgn is, and what network design you want to use.

[ThomasW]

Looking back at the Modula MT, MTM, etc. projects it seems that Jon picked the 8 ohm/87.6 db flavor of the RS180 over the 4 ohm/91 db version. So maybe I’m not thinking about it correctly. Or perhaps I recall that only the 8 ohm version of the RS180 was available when he did his designs.

Correct the 4 ohm versions didn't exist when Jon created his designs.

[Dennis H]

A decent powered receiver like a Yamaha RX-V1700 might be an option.

Definitely go with 8-ohm version then. Generally, 8-ohm (dip down to 6) speakers are more friendly with a wider variety of amps although most "good" amps can handle 4 ohms (dip down to 3) okay.

I'm sure you already know all this but, as cotdt mentioned, get clear on the difference between sensitivity (dB/2.83 volts) and efficiency (dB/watt). If the load is a perfect 8 ohms, efficiency and sensitivity are the same. Driver specs are almost always referenced to 2.83 volts although they may say it's per watt to keep from confusing us poor doofuses (doofi ?).

And, just to review (you already know this too) Ohm's law is your friend:

volts = amps * ohms
watts = volts * amps
and by substitution
watts = (volts * volts) / ohms
watts = (amps * amps) * ohms

[mazurek]

Ah, I didn't catch that you want to move to a passive crossover, impedence differences are less important when driving with a high quality amp channel per driver.

[JonW]

Hi Guys,

Thanks for all the help. :T I’m learning all this for the first time. Reading some books, web sites (the easy part) and trying to actually apply what I’ve read. And heck Thomas, my memory is better than I thought.

If you choose a 4ohm woofer and 4ohm tweeter in a normal parallel network, you're going to be dropping into 2ohm territory likely

An excellent point. Not having yet modeled my first crossover, I didn’t realize that, but it makes sense once you said it.

But me personally, I'll give up the 2db or so max SPL in exchange for better dynamics and control any day.

Certainly. I’m not after anything particularly high for SPL.

Jon, you definately need to set some design goals BEFORE choosing a driver. All those drivers really are the same sensitivity, as cotdt said. So it really just matters what your physical driver deisgn is, and what network design you want to use.

OK, well I don’t have too much for design goals apart from the usual. I want to make the best sounding speaker possible for someone at my early stage. I’m keeping it to a simple 2 way MT design. I was picking drivers from the testing done by Jon, MarkK, and Zaph, with an emphasis on low distortion drivers that have smooth frequency outputs. And that the woofer reaches high enough (or tweeter goes low enough) for a good overlap and crossover design. For power, I want the speakers to be run by “regular” solid state audio amps and not need some pro super amps that can drive a 10,000 W load at 0.1 ohms. So basically, just a regular speaker. Is there somehow you see a need tobe more specific than that?

[JonW]

Thanks Dennis. :T

Definitely go with 8-ohm version then. Generally, 8-ohm (dip down to 6) speakers are more friendly with a wider variety of amps although most "good" amps can handle 4 ohms (dip down to 3) okay.

Right, I thought that might have been the case.

I'm sure you already know all this but, as cotdt mentioned, get clear on the difference between sensitivity (dB/2.83 volts) and efficiency (dB/watt). If the load is a perfect 8 ohms, efficiency and sensitivity are the same. Driver specs are almost always referenced to 2.83 volts although they may say it's per watt to keep from confusing us poor doofuses (doofi ?).

And, just to review (you already know this too) Ohm's law is your friend:

volts = amps * ohms
watts = volts * amps
and by substitution
watts = (volts * volts) / ohms
watts = (amps * amps) * ohms

Got it. One of the books I was reading was pushing for choosing higher [XXX] drivers. And now that you and cotdt mention it, I must be mixing up 'sensitivity' and 'efficiency.' Oops. ops: I'll have to look at the book again later. I've never had a chance to learn electronics well. I guess that's part of the appeal here- really getting a working knowledge on something I'm pretty ignorant about.



OK, I can go with the 8 ohm version. No problem. And come to think of it, it seems that I see many more designs out there, in general, using 8 ohm drivers than 4 ohm ones.



So an MT with the 6600 tweeter and the 18W/8531 woofer. Sounds interesting to me.

[JonW]

Ah, I didn't catch that you want to move to a passive crossover, impedence differences are less important when driving with a high quality amp channel per driver.

Right. I'll play with the DCX2496 for a while and after it sounds OK, use that info as a starting point for the passive crossover design. At least that's what I gather is a good way to go about it.

[TurboFC3S]

Right. I'll play with the DCX2496 for a while and after it sounds OK, use that info as a starting point for the passive crossover design. At least that's what I gather is a good way to go about it.

I believe to do it that way, you'll also need some measuring gear.

And you can't buy the 6600 anywhere. It's not that nobody has stock, it's that Tymphany simply isn't making them. You could always go for the 7000 :T

MT is the important thing to know before chosing, sounds like you're on the path of a killer simple design. Have you dug through the troelsgravesen and humblehomemadehifi designs? They both use a LOT of Scan drivers, you can get some great ideas there - like Tony's Transmission Line Scan-Speak MT.

[chasw98]

Jon, you definately need to set some design goals BEFORE choosing a driver. All those drivers really are the same sensitivity, as cotdt said. So it really just matters what your physical driver deisgn is, and what network design you want to use.

Jon, you might want to build a sort of checklist/worksheet of design goals that you could compare one set of components/design against another. Of course you want a MT that will go to 10 Hz and extend to 40K off axis at 60 degrees. But realistically you might want to look over what is available and say that maybe the LF can only extend to 50 Hz. Maybe a generalized performance factor list that might make adapting a xover to the drivers easier the first time around. If you had 3 tweeters and the off axis response was different among them, given the same baffle, your choices might start to stand out. Then you can compare them using a different baffle design and get a feel for how one will act under different circumstamnces. As you know, this is one 'never designed a speaker in his life' noob giving advice to another 'never designed a speaker in his life' person, so take it with a grain of salt, maybe a couple grains.

Chuck

[cjd]

If you choose a 4ohm woofer and 4ohm tweeter in a normal parallel network, you're going to be dropping into 2ohm territory likely ...

I can't agree with this at all - In fact, I think it's flat-out wrong. With a properly chosen transfer function this is not the case. In fact, this would only happen if you simply parallel the drivers and have no crossover in there at all. For example, RS150/RS28A MTM so we have 4ohm on the mid-woofers (paralleled) and 4ohm on the tweeter.

All those drivers really are the same sensitivity, as cotdt said.

That's not what he said, and isn't correct. He said they're all the same efficiency.

Jon, to answer the question, it is about your design goals. You can get an 8ohm speaker with 1 8ohm, 2 4ohm in series, 2 16ohm in parallel (assuming an 8ohm tweeter). Or maybe you want to run 3 16ohm units in parallel. Or...

You don't need to be doing full-on modeling up front but if you don't know how to look at the graphs, the specs, and already start to get some of this kind of information, DO model before you choose components. I've done it for every project so far. It really helps put things in focus.

Good design takes lots of rigor in process. Used here the way it is used scientifically.

[JonW]

I believe to do it that way, you'll also need some measuring gear.

And you can't buy the 6600 anywhere. It's not that nobody has stock, it's that Tymphany simply isn't making them. You could always go for the 7000 :T

MT is the important thing to know before chosing, sounds like you're on the path of a killer simple design. Have you dug through the troelsgravesen and humblehomemadehifi designs? They both use a LOT of Scan drivers, you can get some great ideas there - like Tony's Transmission Line Scan-Speak MT.

Yup, got the measuring gear already. My understanding is the 6600 will be back soon. I haven’t seen much for testing of the 7000. And I do want to limit myself to drivers with independent testing, to help pick the better ones. Looks like MarkK and Zaph will be testing the 7100 soon. Could be another contender. And yes, those 2 web sites are, indeed, filled with some very nice projects.

Jon, you might want to build a sort of checklist/worksheet of design goals that you could compare one set of components/design against another. Of course you want a MT that will go to 10 Hz and extend to 40K off axis at 60 degrees. But realistically you might want to look over what is available and say that maybe the LF can only extend to 50 Hz. Maybe a generalized performance factor list that might make adapting a xover to the drivers easier the first time around. If you had 3 tweeters and the off axis response was different among them, given the same baffle, your choices might start to stand out. Then you can compare them using a different baffle design and get a feel for how one will act under different circumstamnces. As you know, this is one 'never designed a speaker in his life' noob giving advice to another 'never designed a speaker in his life' person, so take it with a grain of salt, maybe a couple grains.

Chuck

Hi Chuck,

Yup, 10-40 kHz, flat, off axis, no distorion, high SPL, from a single driver, in a cabinet the size of a ham sandwich.

I’ve been thinking of just picking the drivers based on low distortion, smooth FR output, etc. and just take whatever bass extension I get. If it’s 60 Hz, fine. Or 50 Hz even better. I haven’t yet done any WinISD modeling, but I’m looking forward to that because I’ll get a much better understanding of tuning and such. And that will now be easier if I take everyone’s advice and know the driver I’ll use (18W8531).

this is one 'never designed a speaker in his life' noob giving advice to another 'never designed a speaker in his life' person

Hey, I think I know acoupla knuckleheads like that! :

[JonW]

Chris-
Thanks for the perspectives. :T
Given that it’s only an MT I’m after at this point, there won’t be any wiring of multiple woofers in series or parallel. So I get what the woofer is. That should simplify the design (goals).

You don't need to be doing full-on modeling up front but if you don't know how to look at the graphs, the specs, and already start to get some of this kind of information, DO model before you choose components. I've done it for every project so far. It really helps put things in focus.

I (think I) understand the electrical aspects of things enough. Like what you said about multiple drivers in series, parallel, etc. So would you recommend just getting the drivers, putting them into the cabinet I build, then listen with the DCX2496, then measure, then model the crossover? Or perhaps before all that use the manufacturer’s published data and work through the FRD tools to mock up a speaker now?

Good design takes lots of rigor in process. Used here the way it is used scientifically.

No idea what you’re talking about. Actually, that’s why I’m trying to learn all these things up front. Before buying anything or cutting any wood.

[cjd]

It looks like I sort-of mis-typed. Until you can look at a spec and KNOW how you will deal with it, what you can do to manipulate the frequency response, etc. and whether that's going to cause impedance issues or whatever else in conjunction with everything else in the circuit... you start modeling before you buy.

I usually start picking drivers on spec, distortion and otherwise. Look for a good fit, at least 1 octave (if not 2) beyond your crossover point where things remain manageable and relatively linear, etc. So you get drivers. Now, I go with a box model. Get a grasp on volume requirements, which suggest some basic design parameters for the final size (I always guesstimate extra series impedance to get volumes right, based on my probable target slope and inductor impedance, plus some for wire...). Then I do diffraction simulations, put together a basic layout that is going to work for me for the baffle, making sure that I can achieve my intended final volume realistically. This can be a simplification of things where you just can't model exactly what you plan to build. It's to get you in the ballpark of "this is going to work".

Buy drivers. Build box.

MEASURE. Fix mic at tweeter plane, measure response of all drivers from there. Any working together (i.e. in series or parallel) measure together.

Model. Start putting together a probable crossover.

Start analysing off-axis, power response predictions, etc. Tweak. Tweak. Hem. Haw. Run around the block a few times. Commit. Build. Listen. Fix screw-ups. Listen. Tweak. Listen. Measure. Tweak. Listen.

In this last phase, measurements can get complex to really analyse performance adequately.

In your case, you'll get to do some sketching/listening with the DCX before buying components and may therefore avoid some mistakes in component purchases.

If that DCX enters the system before you've measured and modeled, you're giving in to the "I can do it by ear" - it may be enlightening how bad your ear is, or you might get lucky, but... my recommendation is: don't.

Rigor pretty much means "boring"

C

[JonW]

Chris-

Again, I appreciate all the thoughts. That sounds like a reasonable plan. Like you, I was picking drivers with good distortion, etc. that look to make for a good pair crossing over to each other.

I’ve already started the diffraction and volume sims. Nothing too rigorous yet, but trying to get a feel for what’s involved. At first I thought I could just buy a premade cabinet, but I’ve since decided I really want to minimize diffraction issues, so I’ll be making my own cabinets. [I can blame a conversation with Thomas for that. ] With a driver now in mind, I can do more of those sims and decide what I want the cabinet to look like. And start with a rough crossover design, using the manufacturer specs, and see if comes up with something reasonable looking. Then buy drivers and build cabinets.

And that’s an interesting point about measuring before using the DCX2496. Seems like a perfectly reasonable approach. I can try that. Although after building the cabinets and having a rough idea of what the crossover should look like, I can’t promise I won’t hook it all up just to see how it sounds.

-Jon