Attack of the Clone

After two years of solid research and reading of the diyaudio.com pages, I’m finally ready to begin my own project. I’m probably only going to do this once, so I want to get it as good as I can. Reading up on speakers, I was taken by various attempts to “clone” commercial products, namely the Wilson Audio WATT/puppy combo; this speaker became my benchmark in performance and looks. As my drivers are totally different, any resemblance will be in looks (and hopefully performance) but what the heck!

I really like what Tony Gee does – he puts a lot of design into his projects – and so want to use the drivers and crossover of the “Tempo” (Seas tweeter, Vifa PL18WO mid) but in a closed box of ~15L for the top half. For the bottom half I want to use the Seas L22RN4X/P ported in a ~30L enclosure. This would be a 2.5-way loudspeaker (like some of Tony Gee's other projects) with a 1st order low-pass filter and zobel correction on the woofer. So far so good if you're only using one woofer, but here's where it gets interesting. I want to use two woofers. Is this possible?

I have no idea how to wire this. If I run the 2 L22’s in parallel, the sensitivity goes up way above that of the Tempo. If I run them in series then I now have a 16ohm load vs. the 8ohm of the uppers and who knows what will happen. So what do I do? I need the help of the diyaudio.com community because I really want to make this work. Thanks for any help in advance,

Philip
 
philip said:
I’m probably only going to do this once

It is VERT hard to do it only once... not only is it addicting, with what you learn building the 1st one, you just know how to build the next one better

I want to use two woofers. Is this possible?

What XO point were you thinking of? If less than 200-250 Hz you really should be thinking active woofers -- then the relative sensitivities don't matter.

dave
 
Yeah, you're probably right, just reading about speakers is addicting. I mean build an all out project when I say do it once, this stuff isn't cheap. I've built a few kits so far but have been saving and planning for this special one so want to do it right.

The crossover point is not decided, but could be anywhere from ~150 like the real thing upwards. Tony Gee's Andromeda (1 x focal 13kx) uses 300Hz I think, and ~500Hz I think for the Geers EveII which he helped design (1 x ss 8"); these are examples of other series 2.5 way projects I know. Either way, two main considerations deciding the range of the crossover would be the roll off of the Vifa PL18 and the wierd cone action of the metal cone at higher Hz. I anticipate choosing a crossover point in the range of 150-300, maybe closer to 300. I could use the L26 to overcome this problem, but I really like the look of the 2x8's and they model a bit better for my purposes. I want to keep this project passive to maintain their versatility and because I like the concept of an integrated speaker i.e one input, speaker does the rest.

Philip
 
Maybe I need to back the train up a bit and ask a different question. This is about sensitivity. Before I knew much I would have said just run the two L22's in paralell and then cross them over to the upper speakers. However, I have heard that when you run two speakers in parallel their sensitivity increases by 6dB. That's what's stopping me. What are the effects of running two woofers in series (16ohm load) or parallel (4ohm load) on sensitivty?

The reason I am confused is because if you look at many mtm designs, you have 2 woofers in parallel being crossed over to a tweeter, often of similar sensitivity. If there is a massive sensitivity boost by running two woofers in paralell, then the speakers should not sound right, but mtm's sound great. What is going on in this situation?

Philip
 
p.s. I know this has been discussed before with the general conclusion that in theory running two woofers in paralell raises sensitivity by 6dB. However, previous posts discuss additional factors such as driver spacing and baffle step but are ambiguous as to their effects on sensitivity.

Philip
 
Hey man, yeah, I'm sure separate powered woofers would sound great - I've heard that biwiring can really make improvements. However, there are many factors, not all of them related to speakers, that have led me to a single-amp design. As for the 2 8's, part of that is just me.

Anyways, I've been doing a lot thinking about this design. I know that it isn't recommended that you attenuate a woofer because of the high power involved, but I had an idea: Instead of using one attenuation circuit for the two woofers combined, what if I was to attenuate each woofer separately, so that their overall circuit was attenuated. For example, the sensitivity of the upper system is 87dB/1W/1m. What if I was to attenuate each of the L22's from 86 to 81, so that when run in parallel they achieve 87dB. I don't have an excessively powered amplifier, nor would I be running them at loud volumes so I don't know if this would be a problem in my case. Would this work?

Philip
 
Philip, it sounds like quite a nice start, here are my 2 cents :

The 2 x L22s combined have a senstivity of 6+86 = 92dB.
However, if you lowpass them below 300Hz, baffle step will remove around 6dB of that sensitivity, hence bringing it back to 86dB.

That will make a nice match for the 87dB PL18.
However, I would recommend you choose a smaller midrange (like the Vifa PL11) with a much lower moving mass, to create a 3-way system and make the mid-tweeter transition much easier.

David

www.gattiweb.com
 
In my last project I did a 3 way using the Vifa PL11, and I am very happy with it, and they only cost 30 bucks, if I had to do it over I would go with two of them in a MTM configuration with two woofers.

If you have the amps, an active x-over will cost less and give definate better sonics, and greatly simplify your life, since you do not need to worry about sensitivity, etc, since you can change it at whim with a knob on your active x-over.

However, since you want to go with a passive design, I'd suggest getting like lsp-cad if you can get your hands on it for cheap so that you can model all of the different components in one program (it accounts for resistance of inductors/caps, baffle step, even the heating of the voice coil). Just a thought.

Have you concidered making the cloned version of the Watt/Puppy so that the x-over is already designed for you?

http://www.users.nac.net/markowitzgd/david/david.htm
 
Thanks for all the advice guys, this is what I need. I ordered Loudspeaker Design Cookbook yesterday so when that gets here I’m going to read all I can to understand this stuff a bit better.

It seems there’s some disagreement of crossover frequency: some say go above 300Hz to avoid nasty sonic effects of attenuation filtering, some say go above to avail of baffle step. A few questions:

I didn’t know an attenuation filter would have so many negative effects on the sound. Would it really be so bad? Also, and this may be completely unfounded, but I thought resistors didn’t affect phase?

About the baffle step, what if you were to only have one L22, would this mean that it would have an effective sensitivity of 80dB, or is baffle step loss a feature of 2 woofers? I’m assuming baffle step has something to do with sound waves bouncing off the front of the cabinet, being redirected towards the listener, and resulting in amplification; my understanding is that wavelength increases as frequency drops, and at a certain frequency related to the size of the front baffle this amplification is lost, resulting in a 6dB reduction in sensitivity. The concept of reflected sound waves giving amplification reminds me of room gain.

As for driver choice, I need to stick with the PL18 because I want to use this particular crossover. My original goal was a high level of reproduction of classical and jazz recordings, and although the Tempo does seem to excel at that with its low Q and revealing nature, I suppose I could use another design from the same guy; not only does he tend to use series crossovers which I would like to try, he puts a lot of effort into designing testing into them, so I know it will work well one way or another. I have heard, though, that the PL11 and PL13 are excellent midranges.

Yeah, I need to get a free XO sim; I’ll try lsp-cad, hopefully the demo will have what I need

Thanks for the suggestion about the Thorsten Loesch design; I remember when I read his project, it was the first time I had heard of someone trying to clone a set of super-speakers, and was amazed that someone would actually try; I read a post by a guy who built them and said they sound pretty good. On the other hand, I don’t know if they have the low-end capabilities I’m looking for. Not only do I like the metal cone of the L22, it has an f3 of 35Hz or less in a ported ~30L enclosure if I remember correctly. It also has good Xmax. Anyways, it’s good that you reminded me about it because I’m going to check out his crossover to see if I can get any ideas.

I appreciate this input so keep it coming! If I can get a plan together, I want to order these drivers this week!

Philip
 
Hey planet10, thanks so much for the link for baffle step, I've been reading that over for the last few days.

One quick question first, why does output jump by 6dB when you hook woofers up in paralell? From P=I^2/R, P is inversely proportional to R, so cutting the resistance in two should cause power to incease by a factor of 2, thus a 3dB increase. What have I missed?

Yeah, it seems the problem is sensitivities. I'm pretty set on the 2 L22's, so the question is finding an upper. The only place I'll ever use these speakers is against a wall, so I think 3dB loss for baffle step is most appropriate. Either way, that gives these 2 speakers a sensitivity of ~89dB. Unfortunately, the efficiency of the tempo system I'de like to use (http://home.hetnet.nl/~geenius/Tempo.html) is closer to 85dB, so I guess unless I do attenuation on the L22's, this plan is out the window. Apart from the difficulty in finding the appropriate high-power resistors to make this attenuation circuit, how bad would the sonic difference be if I was to do this?

On the other hand, why not take advantage of this high sensitivity? Most people aim for it, so why get rid of it? Therefore, I'm wondering if anyone has any ideas for high-quality completed crossovers. I'm looking for a high-sensitivity 2-way that sounds really good (i.e. 13-18cm woofers ~$50), with a good crossover. As I said, I don't want to chance this one so would like a proven design. Anybody have any suggestions? I would consider the thorsten loesch design, but I think you can't get the CA17 anymore, and the TD120 has gone up in price.

Philip
 
philip said:
One quick question first, why does output jump by 6dB when you hook woofers up in paralell? From P=I^2/R, P is inversely proportional to R, so cutting the resistance in two should cause power to incease by a factor of 2, thus a 3dB increase. What have I missed?

Can of worms that one... i for a long time held the belief that 2 woofers would incrase efficiency by 3 dB, but that has recently been brought into question by a couple posts here. Going back to 1st principles it seems that indeed 1 W in gives the same levels reguardless of 1 woofer or 2 in series or parallel. In the face of strong "convential wisdom" i'm doing some of my own research on this.

And you can't always count on getting 3 dB more power out of your amp (ie if a tube amp you could well get less)

Convential wisdom says you get 3 dB for having 2 woofers and if in parallel another 3 dB because your amp is noe putting out 2 W instead of 1.

In a situation like you are in i never bother with passive XOs so i just turn up the amp levels until everything is right.

dave
 
audiobomber said:
From the Loudspeaker Cookbook:
"Sensitivity will increase +3dB for a parallel connection and -3dB for a series connection compared to a single driver."

No further explanation given that I could see. I assume sensitivity means that one watt that produces 90 dB in a particular driver will produce 93 dB if you parallel with the same driver.

What he is saying -- assume 1 W into 8 ohms -- (and which is only correct if the amp delivers) is that 2.83 V in gives 3 dB more at half the impedance/parallel (because the amp is now putting 2W)
and -3 dB at twice the impedance/series (because the amp is putting out 1/2 a watt)....

Now if you have an OTL amp that puts out more power as the impedance rises (and can give you exactly opposite results) you can see that Vance is totally wrong. It is blanket statements like this that show Dickason is thinking in a pretty narrow space. Even when the amp is capable of doubling its power into half the impedance this is likely to only happen at low levels and as the amp is pushed will put out less than than double (and in the case of many amps just shut down)

You have to look at you specific amp/speaker/XO system to get real meaningful results

dave