Xt1086+bms 4540nd

[augerpro]

Glad to see this avenue has not died. I finally got around to purchasing some BMS 4540's and picked up a pair of B&C DE250s too based on feedback on other forums. This is so I can compare.

Any observations yet? I was considering buying a pair of DE250's for design that I had planned for this summer after I finish two other ones first.

[JoshK]

No observations.... I haven't built baffles because my shop is currently buried in the basement due to a remodel. I have to wait till it is a bit warmer to cut baffles in the backyard or when the remodel is done and my shop is unburied.

[Saurav]

Resurrecting an old thread...

I have a pair of 4540NDs coming, and since I haven't heard back from Assistance Audio, I went ahead and ordered a pair of XT1086, and the P-Audio throat adapter (which is hopefully better than the Selenium one). They only had one adapter in stock so I won't be listening to this combination in stereo until some time in July, but I should be able to put one set on a baffle and take some measurements in a week or so.

I'll see how it goes. If I don't like it, the next step would be to try the B&C DE250, since that's a bolt-on driver, and I've seen many accounts of that combination working well. Or maybe one of the BMS bolt-on drivers.

[A9X]

Resurrecting an old thread...

I have a pair of 4540NDs coming, and since I haven't heard back from Assistance Audio, I went ahead and ordered a pair of XT1086, and the P-Audio throat adapter (which is hopefully better than the Selenium one). They only had one adapter in stock so I won't be listening to this combination in stereo until some time in July, but I should be able to put one set on a baffle and take some measurements in a week or so.

I'll see how it goes. If I don't like it, the next step would be to try the B&C DE250, since that's a bolt-on driver, and I've seen many accounts of that combination working well. Or maybe one of the BMS bolt-on drivers.

Every adaptor I have ever tried for home has been a sonic disaster compared to a BO with decent matching into the throat. Best of luck though.

The DE250 is a very nice driver: mine will soon be on some Geddes WG's, I hope.

[Saurav]

What would a poor throat match look like in the frequency/impedance response? It'll be a while before I'll have crossovers dialed in, and this will be my first attempt at handling the impedance and EQ requirements of a compression driver / waveguide, so I won't really trust my ears on this right away.

[A9X]

Ripples in the Z plot, and perhaps they will show up on a waterfall - I didn't have a measuring system I trusted for this last time I did it, so I have suspicions I can't confirm with plots.
However, they are clearly audible in many cases.

[ThomasW]

What would a poor throat match look like in the frequency/impedance response?

Yes, the impedance plot for the bolt-on DDS ENG-1 with a screw-on adapter for the 4540 ND, was quite a bit different compared to that of the screw-on horn with the same driver.

As a result of the adapter there was a visible gap between the protection screen on the compression driver and the beginning of the horn.

[Saurav]

The WGs and adapters showed up today (they somehow found an extra one in their warehouse, so I got both). The XT1086 looks pretty decently made, in terms of fit and finish. It's metal, and rings quite nicely if I hold it in my hand and tap it with a fingernail. Not sure how much it'll be damped when mounted on a baffle. Ludvig, what did you do to damp yours? Once I decide I'm keeping them, I'll probably try duct seal or some kind of putty or something on the back.

The P-audio adapters are... interesting. They're metal too, and the match to the WG seems OK. I can feel a slight ridge at the intersection, but it's well centered, and it mostly feels like a change in profile (the inside is straight, and the end of the waveguide has a flare, I think). The other end isn't so good. I tightened the compression driver down as far as I could, and there's a gap between the end of the driver and the waveguide, just like you guys described with the Selenium adapter. There's some thread remaining on the driver and it looks like some on the inside of the adapter, so it could probably screw in further, maybe all the way to the end. But that'll take pliers or something, and there's nothing to hold on to really since both parts are round. I tightened it down to the point where I had a really hard time unscrewing them, so after that I just left it 'hand tight'.

Once I start measuring them I'll try some polyfill batting in that gap or something. But long term, I think I do need a bolt-on driver. The throat area isn't a good match at all, and it seems silly to spend all this money on driver and waveguide and not get that right.

No meaningful listening impressions yet. I placed them on top of my baffles and hooked them up, same XO as I have with my current tweeters, pressed the 'CD EQ' button on my CX3400 XO, added some arbitrary resistor values as an L-pad until the treble wasn't overpowering, and listened to a couple of songs. The treble does sound... clean. But that's all I can really say at this point. I hope to make a test baffle and get some measurements this weekend, and then see how close I can get to a reasonable XO with the parts I have on hand.

[Saurav]

Put the waveguides on a temporary baffle and ran some measurements today. I trust the impedance graphs more than the acoustic graphs... I was checking with my midrange driver, and the impedance matches the manufacturer's data, but the FR has some wiggles in it that I haven't seen in prior measurements.

Impedance of BMS 4540ND on XT1086 (which has a gap between the end of the CD and the WG):

Image not available

Stuffing some polyfill into the gap at the end of the throat adapter:

Image not available

I tried polyfill further out into the throat, but that didn't really make any difference.

FR curves. Like I said, I'm not too sure about these, but FWIW:

Image not available

Unsmoothed, gated to (I think) just before the floor reflection. The ripples might be diffraction from something, or reflections off the mic tripod (I don't have a boom stand). I didn't flush mount the WG, and I basically just cut a hole on the top of my current speaker's baffle. So the top of the WG extends a little bit above the top of the baffle, it's only held on by 2 screws... I'm sure all of that contributes.

Anyway... black = no stuffing, red = stuffing in the adapter, blue = stuffing in the adapter and the throat.

Stuffing the adapter gap does smooth out the 10KHz+ response, but at the cost of almost 10dB drop in output. Polyfill in the throat drops the level further, but doesn't seem to help the FR much. The 10dB drop is probably more than I can handle, as the next picture shows:

Image not available

That's the adapter-only stuffing (black) and the midrange driver (red), with the same settings on the amp/mic preamp/soundcard (to see what the relative levels are like).

Hmm... I just noticed that many of the FR wiggles in both drivers are at the same frequencies. What does that mean? Is this still likely to be a baffle issue, or a reflection off the mic stand?

Anyway... the compression driver needs to be EQ'd for the falling response, so the 2kHz - 10kHz range eventually needs to end up at the 10kHz level. Which is about 9dB lower than the midrange. If I go with a passive crossover, that's not going to work for me, that's giving up too much efficiency for the amp I'll be driving these with.

Another option that occurred to me yesterday - I have 3 amps and 2 crossovers now, so I could try going fully active. The CX3400 has a 'CD EQ' button that AFAIK lifts response above 3kHz or something like that, which may work for a temporary solution. This also bypasses the issue of dealing with that impedance curve. The crossovers are all textbook LR4, so a final active implementation will probably require me doing custom XOs... or adding an external box that handles the shaping so that the Behringer LR4 XOs work right... or doing some shaping passively after the amps... those are some of the options I can think of.

OK, that was quite a ramble. Any advice is welcome And I think I need a bolt-on driver.

[augerpro]

Interesting saurav. What is the CD again and it's nominal sensitivity? You might try some of that 30 ppi fish filter foam that Geddes uses to help with the diffraction you're getting without losing so much SPL. That throat match is pretty critical apparently, and is one reason that I finally spent the ducats and bought Geddes' waveguide.

The impedance looks reasonably easy to make resonance traps for. An Lpad arrangement might also do a decent job. Lynn O made some mention of using autoformers in the new project of his for the CD. I haven't sourced any yet but BudP apparently can (or has) made some that were quite good. Probably teh way to go if you don't want tune resonance traps or the Lpad doesn't do teh job well enough.

Hard to say much about the responses since you do seem to have some noise. Maybe electrical or reflections. I had a problem with my first mic stand that caused a lot of unevenness above 2khz. The little plastic sleeve that actually holds the mic was causing reflections. I ended up taking off and put some foam strips on the arm of the mic stand and just used a velcro strap to hold the mic to to it.

[Saurav]

I'm playing around with XO sims, and an Lpad makes a big difference. The 1kHz - 10kHz impedance range is now between 7.5 and 9.1 ohms. I'll play with notch filters too, but I have this (untested) notion that my tube amp won't be too happy with too many traps in the XO.

The tweeter is a BMS 4540ND, nominal sensitivity is 113-114dB on a CD horn. Figure about 9dB drop from the polyfill, plus 6dB from 3.5kHz to 7kHz, and the 7kHz number comes out to about 98dB, which is right where my midrange is supposed to be (even though Audax likes to say 100). So that math matches the measurements.

Hard to say much about the responses since you do seem to have some noise. Maybe electrical or reflections.

Hmm... I did turn off the reference input to Speaker Workshop, because it was clipping (I had the levels set for the impedance measurements). Now that I think about it, what I should have done is turned down the soundcard levels, and turned up the gain on the amp. I've seen SW work OK without a reference (my laptop's built-in soundcard has a mono input), but I can do better.

I'll also have to do something about the stand. I'm using a camera tripod, and the connector thread size is different, so the mic holder clip doesn't screw onto the stand. So I've just been laying the mic on top of the stand (it ends in a flat top surface). Which is similar to you foam+velcro, just not as stable. Again, not ideal, but I've used it in my living room before, measuring the same driver at the same distance, and didn't get those response dips. This was in my garage, and I don't think there was anything else nearby that should have caused reflections.

Wayna Parham (Pi Speakers) has written a tutorial of sorts on designing crossovers, and he goes into EQing a compression driver on a CD waveguide. I found a couple other suggestions for EQ circuits. Dr. Geddes also suggested that the Pi Speakers approach was a decent place to start. So I'll be playing with some simulations to see where I get, and/or buying some XLR cables and trying an all-active setup.

[augerpro]

So I've just been laying the mic on top of the stand (it ends in a flat top surface). Which is similar to you foam+velcro, just not as stable.

My mic is at the very end of the 2 foot arm and I just replaced clip with a velcro strap. Very little left of any part of the stand to cause reflections. If I'm visualizing your setup correctly you just put the mic where the camera would be? If so I see a lot of potential reflection problems there.

Wayna Parham (Pi Speakers) has written a tutorial of sorts on designing crossovers, and he goes into EQing a compression driver on a CD waveguide. I found a couple other suggestions for EQ circuits. Dr. Geddes also suggested that the Pi Speakers approach was a decent place to start. So I'll be playing with some simulations to see where I get, and/or buying some XLR cables and trying an all-active setup.

I'll check that out too.

[Saurav]

My mic is at the very end of the 2 foot arm and I just replaced clip with a velcro strap.

Ah, I see. Yep, mine is just sitting on top of a tripod. I need to fix that, it's been OK for the rough experiments I've been doing so far, but this time I hope to do a more serious job.

[Dennis H]

Ah, I see. Yep, mine is just sitting on top of a tripod. I need to fix that, it's been OK for the rough experiments I've been doing so far, but this time I hope to do a more serious job.

Maybe for now poke a hole in a piece of foam and slip it over the mic so any reflections from the tripod's head have to go through the foam.

[ThomasW]

Sorry the measurements are scaled differently, but smart people should be able to figure out what's going on since the first big spike on both plots is ~1.3kHz ...

A while back Jon and I chatted about the need for a zobel if an active XO were being used. His recommendation was yes if the goal was optimal performance, a zobel should be used with even when an active XO was being used.

Here's the BMS 4540ND with the DDS threaded waveguide.




Here's the BMS 4540ND with the $10 econo-JBL waveguide/horn

[Saurav]

Maybe for now poke a hole in a piece of foam and slip it over the mic so any reflections from the tripod's head have to go through the foam.

That's a good idea, I'll have to try that. The ECM8000 comes with a foam cap, but I don't see how that could be transparent to 20kHz from the front and attenuate from the sides.

You might try some of that 30 ppi fish filter foam that Geddes uses

I was trying to dig up info on what foam he used. So this is something that goes in an aquarium? That you buy at a... sports shop? Hobby shop? I'll go look up fish filter foam. Edit: Duh, PetSmart.

I scaled the "no polyfill" graph down to the same level as the "with polyfill" graph, and not only is it dropping the overall level, it's also attenuating the HF faster. So that makes the EQ harder.

A while back Jon and I chatted about the need for a zobel if an active XO were being used. His recommendation was yes if the goal was optimal performance, a zobel should be used with even when an active XO was being used.

Agreed, and even more so when using a tube amp. But it's still much less sensitive than with a passive XO, so it might work OK for a "put it together and see if there's any point in exploring this further" thing.

[Saurav]

Ran some more tests yesterday...

Poking a hole through some foam and putting that around the mic worked pretty well. I also discovered that one of the drivers screws in all the way into one of the adapters. Neither driver screws into the other adapter. There's something not lined up right with the threads... or some other problem I can't figure out. And this morning I had a pretty hard time unscrewing the driver that I'd managed to get all the way in. So now the 4540's are up for sale, and I'll be getting some bolt-on drivers.

The measurements improved with the driver screwed in all the way (which still left a small gap in front of the driver). I didn't take any impedance measurements, but here are a couple of FR graphs:

Image not available

Black is the driver screwed in all the way. Comparing with the old results, the 11kHz dip is still there, but the response above it is cleaner.

Blue is with some fish filter foam in the space between the end of the driver and the adapter. I went to PetSmart, none of the packaging said anything about ppi, so I picked a couple of small pieces to try. This removes that resonance dip, with barely any drop in overall SPL level. The 10kHz level also matches my midrange level, so if I could EQ this response correctly in the crossover, I wouldn't have to pad the midrange down. If I wasn't having a hard time with the threaded adapters, I'd probably have given it a shot with this setup to see how it sounds.

Not shown - adding more foam in the throat dropped the level by ~1dB from 2kHz to 8kHz, and slightly more above that. It also removed the 900Hz peak, which means that's probably a resonance in the throat.

That's probably all I'll do with this driver/waveguide combination. I expect a bolt-on driver should be a better starting point.

[JoshK]

A step down transformer is another way to tackle the impedance issue. It will attenuate the CD, which is likely needed anyway and help guard against back EMF from the CD (probably only needed if using SET amps). Just another option, although probably more expensive then a zobel. I however, like the idea very much.

[Saurav]

Got any examples/links? This needs a transformer that has flat response in the tweeter's passband, right. And the turns ratio will set the impedance reduction factor (and the attenuation factor). I forget if both are 1:N^2, or one is 1:N and the other is 1:N^2. Also, I'm assuming the simplest way to model this would be to measure the driver + transformer, and use that as the 'driver' in Speaker Workshop.

I use a pair of autoformers as my passive linestage, so I'm somewhat aware of the improvements of using transformers instead of resistors for attenuation.

I will most likely have a SET driving the mid + tweeter, unless I decide to try an active 3-way XO.

[Saurav]

Well, I figured I'd make a final post to finish off this set of measurements. I'm now using the BMS 4538, which is a bolt-on driver. Nominal specs are very close to the 4540, except it uses a ceramic magnet (not neo), so it's bigger and weighs more. You can compare the specs here:

http://www.assistanceaudio.com/02_bms.html

Measurements:

Image not available

Image not available

I was surprised that the impedance peaks are so much smaller. As expected/hoped, the 11kHz resonance is gone. And this response also looks like it'll be easier to EQ. I think this will work much better than trying to use a throat adapter.

The driver isn't centered perfectly on the WG. On one side I can feel a slight ridge and on the opposite side there's a dip where the WG and driver throat don't quite line up. I'll see if I can loosen the bolts and adjust that better. I'd expect that the Geddes WG and DE250 would match better, since he seems to be very particular about this kind of stuff.

Adding foam to the throat now only drops the on-axis response a dB or so. It probably helps with HOM, I'm pretty sure I don't have the ability to measure that. And I still have some reflections/noise in these measurements... I might try them again outdoors once I have a proper baffle.