New dipole project - Neo3, CHR-70, DPL-10

JohnR · « **on:** 5 Sep 2010, 09:18 am »

I've started a new dipole project, using the BG Neo3 PDR, Mark Audio CHR-70 (the Mark I version, my local distributor has it on sale), and a pair of Exodus Audio DPL-10s (no longer in production, sadly). I've dubbed it "Francesca" for no particularly good reason. This project is an exercise for me to learn more about speaker measurement. Also, I want to see what can be done to make a dipole speaker for use in small rooms - or more specifically my study, which is 2.7m wide by 4.3m deep. I was inspired to do this by the "Diminui" project that was posted recently. I also wanted to experiment with getting dipole radiation from 30/40 Hz up to 20,0000 Hz. High output is not one of the primary goals

Here is the prototype baffle. As you can see, it's made quickly and cheaply. The baffle is a 12x36" piece of 1/2" chipboard. Nuff said - it's a prototype to be thrown away when a better one is built.

My current thinking is that the crossover between the woofers and the mid/tweet will be active using a miniDSP, with a passive crossover between the mid and tweeter. At present though I am listening to them using a very simple passive crossover made with parts I had on hand. The bass and mid/tweet are bi-amped so I can have gain control over the two sections. The woofers are low-passed with 5.5mH, and there is a first-order crossover at around 1700 Hz between the mid and the tweeter. There is no high-pass on the mid - I didn't have enough capacitance and there's no point in buying it since this will be active before long anyway. Plus I figure it's a good way to break in the CHR-70s

The crossover thus relies on the dipole rolloff of the mids which starts around 400Hz. Here is the in-room response of this setup:

Here is the positioning of the prototype in the room. Initially there were pulled further out from the rear wall, but I was getting a big null at 100 Hz, this positioning gives a smoother response. The amps are on the floor - funny story about that, this is a pair of AKSA 55s that I built back in 2001-2002 with the idea of building a bi-amped loudspeaker, and I only just now realized that I've finally used them for that!

Considering how flimsy the baffle is, bass is surprisingly solid and tight, when it's on the recording. There are some obvious problems, some of which should be cured with the miniDSP when it gets here, and some with a more solid baffle when I decide what the best shape is. Others - we will see

mcgsxr · « **Reply #1 on:** 5 Sep 2010, 12:19 pm »

Interesting project John, I like the skunkwerks approach!

The use of those woofers will ensure good bass, and with tuning via the electronics, I am sure you will be able to sort out something that makes you happy.

I have looked at the specs of those Mark Audio's for a while now, nice to see someone using them on OB in a 3 way!

JohnR · « **Reply #2 on:** 5 Sep 2010, 12:51 pm »

Hi Mark, to be honest I'm not sure the CHRs are the best choice for dipole application but they were available relatively quickly (and my reasoning was they could always be built into an enclosure better designed for them later). I'm wondering if there are other suitable drivers available that are more open at the back. I have a feeling the mid is the hardest to get right for dipole.

Blackmore · « **Reply #3 on:** 5 Sep 2010, 12:54 pm »

Nice,John! That's a pretty smooth response from 600 on up. Will you use the dsp to smooth the mid bass region or will it do mainly crossover work?

JohnR · « **Reply #4 on:** 5 Sep 2010, 01:08 pm »

Hi Mark, thanks

Hope all is well over there - it must be cooling down for you as it's warming up over here (no fire needed the last two days!). I'm expecting that the miniDSP will be used for equalisation and lowpass on the woofers, which at present means up to around 400 Hz. For the mids/tweets, it's an open question. But I think to answer the question, if the miniDSP is in circuit for a channel then I will use it for eq as well as crossover. You can choose either a graphic EQ or paramteric EQ, it's only 10 bucks for each software module, it looks like a great little tool. (Let's see what I say after I actually get it!! I'm not sure how long, I only ordered it a couple of days ago).

JohnR · « **Reply #5 on:** 9 Sep 2010, 03:24 am »

The minidsp is up and running in the system. So far the 48dB/octave crossover really helps in cleaning up the mids, and I've used it to shelve down the treble a little. Takes a few seconds to make adjustments. More to come.

lowtech · « **Reply #6 on:** 9 Sep 2010, 04:47 am »

Nice work. Glad to see that someone "gets it".

You might want to rethink using an overly thick baffle since it will introduce more problems than it solves. (YMMV).

The ScanSpeak 10F would be a worthy replacement mid. The relatively low cost is just a bonus.

JohnR · « **Reply #7 on:** 9 Sep 2010, 09:23 am »

Interesting you should mention the 10F, as I have just been looking at the new 15M... lessee... 15M/4624G Discovery - https://www.madisound.com/store/product_info.php?products_id=8926

I'm expecting to keep the top section thin and reinforce the bottom section somehow. Dunno how - we'll see

poseidonsvoice · « **Reply #8 on:** 9 Sep 2010, 02:36 pm »

Wow John, from 600 Hz on up that is really, really sweet! I'm pretty sure room modal issues are of a concern below 600Hz though...overall I think you should be proud

Anand.

Danny Richie · « **Reply #9 on:** 9 Sep 2010, 02:40 pm »

Hey John,

I have designed a lot of speakers in the last four or five years that are of this type and have used those tweeters extensively.

I'll list a few tips for you.

The tweeter is very sensitive to the way it mounts. And little surface variation will show up in the response. If you can use one of the BG face plates with it then it will smooth out the response. It is also sensitive to baffle width and surface reflections especially with the low crossover points that can be used with it. So I'd minimize the baffle width around it as much as possible.

I also did extensive testing with the rear wave and how it is effected by the through hole. You can really disrupt the rear wave output or really smooth it out based on how you shape the through hole. I can share some data and post pics of what worked and what didn't if you like.

I'd recommend a minimal amount of surface area around the mid as well.

On your lower crossover point. I found the heart of the vocal region to be in the 300Hz to 500Hz range. So splitting up that range into dissimilar drivers with different voice coil offsets really can disrupt that range. I found the best thing to do is to get under that range. Try crossing at 200Hz or a little lower if you can. That will give you a cleaner midrange.

Also, try to keep the support structure as shallow as possible and as solid as possible. They can be caused to resonant quite easily. And if they are too deep or if the woofers play up to high then they easily cause a resonance cavity to be created within that area. You really want the length of those supports to be shorter than the wavelength that the woofers are producing to avoid that.

lowtech · « **Reply #10 on:** 9 Sep 2010, 02:59 pm »

Quote from: JohnR on 9 Sep 2010, 09:23 am

Interesting you should mention the 10F, as I have just been looking at the new 15M... lessee... 15M/4624G Discovery - https://www.madisound.com/store/product_info.php?products_id=8926

I'm expecting to keep the top section thin and reinforce the bottom section somehow. Dunno how - we'll see

Yeah, it really comes down to where you are going to cross the drivers at. The 10F will work better if you're crossing high. In my new design, I'm crossing it to an open-back Neo3 at 4kHz. JohnK runs it up to 5kHz in his new Note design.

I was going to comment on your flush mounting of the Neo3, but your response curve looks pretty good as-is. This is how I mounted mine.

Have you taken any off-axis measurements?

P.S. If you wish to eliminate the three octaves of bumps/dips below 800Hz try this simple experiment. Disconnect your bottom woofers, adjust the gain to compensate for the output loss (6db) and re-measure your response from the same point as your original measurement. (I suspect you will find that the irregularities are caused by cancellation of the two drivers covering the same range. Ultimately, you will find that you'll need to modify your design to operate as a 3.5-way so that your bottom woofer is operational mainly below 150Hz or so).

JohnR · « **Reply #11 on:** 10 Sep 2010, 05:10 am »

Oh. That's interesting, I was going to run the tweeter as low as possible, assuming that it will maintain a dipole pattern better than the mid. What's your lower crossover on the mid?

I haven't done any gated measurements yet. Today was messing with the miniDSP, it's pretty easy to use, once you get it into "sync" mode you just change things on the screen and it updates in real time. Here's the in-room response after eq of the woofer section. Highpass at 30 Hz, lowpass at 300 hz.

This is the miniDSP parametric eq screen. Now that I look at the curve it looks a bit extreme but that's what I ended up with today:

Lots more to learn

JohnR · « **Reply #12 on:** 10 Sep 2010, 05:18 am »

Quote from: Danny on 9 Sep 2010, 02:40 pm

Try crossing at 200Hz or a little lower if you can.

Hi Danny, I'll give it a try but I think I'll probably need a wider baffle to do that.

lowtech · « **Reply #13 on:** 10 Sep 2010, 05:24 am »

I'd recommend you try my little experiment with your original setup...

The mini-DSP interface looks nice (I've been interested in their modules). Your smoothed response looks really good, but you're measuring both the speakers and the room. There is surprising little EQ required to offset the dipole loss which means those DPL-10s are well suited for dipole use.

My design is still in flux but I suspect the high-pass on the mid will be between 700Hz and 900Hz. I'm currently using a DEQX, so the slopes will be high-order, linear-phase. My woofers will be running in a 3.5-way configuration to avoid the cancellation issues I previously mentioned. The "real woofers" (W-frame dipoles) will be crossed at 80Hz.

JohnR · « **Reply #14 on:** 10 Sep 2010, 07:13 am »

Quote from: lowtech on 10 Sep 2010, 05:24 am

I'd recommend you try my little experiment with your original setup...

Oh, I missed that. OK, will do.

JohnR · « **Reply #15 on:** 11 Sep 2010, 12:54 pm »

A non-update for today - I struggled with the theory of gated swept-sine measurements but I think I've figured it out now. (I need to understand what the software is doing before I can make sense of the results.)

Danny Richie · « **Reply #16 on:** 11 Sep 2010, 03:07 pm »

Quote from: JohnR on 10 Sep 2010, 05:18 am

Hi Danny, I'll give it a try but I think I'll probably need a wider baffle to do that.

I use 5.25" woofers on an 8" wide open baffle and get enough output to cross at 200Hz, and only use a first order high pass filter on it.

Worse case scenario, with that little woofer, is that you add a little gain in the 250Hz to 300Hz range to keep it linear to the 200Hz crossover point.

JohnR · « **Reply #17 on:** 11 Sep 2010, 04:34 pm »

OK, great. I'll look into it.

Rocket · « **Reply #18 on:** 12 Sep 2010, 05:19 am »

Hi John,

Great looking project. I hope it works out well for you.

Regards

rod

poseidonsvoice · « **Reply #19 on:** 13 Sep 2010, 02:23 pm »

Looks like long time veteran of OB designs, John K. is also building a new speaker using a Neo-3, Scan Speak 10F midrange, dual Seas ER18's in the lower midrange and of course dual woofer U-frame/N-frame subs.

Here is the link and his research.

Best,
Anand.