Inexpensive Crossover Upgrade for the OSD Nero MB5 Bookshelf Speakers

So I picked up a pair of OSD Nero MB5 bookshelf speakers about a month ago and after giving them a listen and measuring their frequency response, I decided these speakers need an upgrade! In a previous post I updated a pair of Jamo Concert 9 Series speakers where we tackled internal damping, port tuning and the crossover. In this update we’re just going to be tackling the crossover with just one very minor change to the damping. First off I think the Nero MB5’s are great little speakers. I got them on sale over at Crutchfield for only $99. Amazon typically has them for around $159. As of this writing they are still on sale at Crutchfield and for this price, in combination with the updates we’re about to go over, these speakers honestly cannot be beat. These are essentially the passive version of the Nero AB5 powered speakers, ditching the internal amplification (and any DSP) for a simple passive solution with a passive crossover. They even left the LED for the logo on the front baffle, but it isn’t connected to anything.

They have the look of a nice studio monitor, but without the finesse or feel. The box is light, lacks any real bracing and any significant damping, save a single piece of poly fill that covers about half of each side and the top. The soft dome tweeter is framed in a very good-looking waveguide which provides a classic 1st order gain response as the frequency decreases, but any benefit here is not really realized due to the paltry 6 dB/octave filter on the woofer which lets an excessive amount of cone break-up modes to enter the equation. The dual ports provide a tuning frequency of about 60 Hz which is fantastic for a speaker this size. The little 5-1/4″ woofer can produce an impressive amount of bass and depending on how close you set these against a wall, the bass can either be just right or way too much. Don’t expect miracles here, but the bass response is quite punchy. Some port noise (chuffing) can be heard for bass-heavy tracks that don’t have a lot of other music to mask it. But is bearable.

At first listen I really enjoyed these speakers, there was nothing that appeared completely wrong or unmanageable. After a quick measurement with REW and my Behringer ECM8000 mic, we start to see some of the issues this speaker actually has. Measured on axis at about 0.5m directly in front of the speaker midway between the woofer and tweeter, we see a 6-7 dB peak around 5 kHz that screws up any hope of expecting these speakers to sound flat or neutral. Note that this response will look different depending on listening position. Move up or down or left or right and this peak is less predominant. However for purposes of this crossover upgrade, the intent was to make the speaker flat on-axis and ear-level being directly between the woofer and tweeter. As it stands, these speakers will sound a little better either above or below the midpoint of this speaker. But for now let’s open this speaker up and see what makes it tick and see if we can figure out what is causing this uneven frequency response and what it might take to make it better.

Disassembly of the Nero MB5 is pretty straightforward. There are a total of (12) 4mm hex wood screws holding the plastic front baffle and terminal cup/crossover. The only real trick is there are two smaller phillips-head screws hidden underneath the 3M EVA vibe isolation pad in the front left and right corners. You have peel back the foam over about a 1″ x 1″ area to access the screw to remove it, thus releasing a plastic tab attached to the front baffle. This seems holey unnecessary and just makes removal of the front baffle trickier. See pics below for an idea of what this looks like. Once that’s out of the way, the front baffle and as it turns out, the tweeter assembly, come off together. The 5-1/4″ woofer is revealed and can be removed as well. After disconnecting the drivers we can remove the terminal cup and crossover assembly and get our first look at what kind of magic the folks at OSD came up with to tame this beast. Turns out to be just another typical cheap crossover consistent with this type of speaker at this price point. The low-pass filter duties are handled by a basic iron-core inductor creating a simple 1st order, 6dB/octave filter network. Boring, and, ineffective.

I have, in my experience, never found a 1st order filter to be sufficient for a low-pass design to suppress cone break-up modes to a manageable level. The best you might get is with a poly cone as they tend to have slightly better break-up modes, but OSD touts this is a “composite fiberglass woofer” and you aren’t breaking any laws of physics here. Once you get above the fundamental 1st mode of this material, it’s going to resonate, its going to break up, it’s going to like being excited at these frequencies and will respond in kind. If you’re gonna design a speaker and expect to get away with a 1st order filter (which by the way can be turned into a 2nd order filter with a cheap-o 79-cent capacitor), you better be prepared to work out some exotic cone material with specialized composites for better damping to control those break-up modes or you’re going to be fighting a losing battle. 1st order filters just don’t have what it takes to be effective in this regard. So that’s the first thing we will fix. The woofer measured about 3.6 ohms at DC and therefore represents a solid 4 ohm speaker. See the impedance plot of the woofer measured in the box below.

Moving on to the tweeter, we have a 2nd order high-pass filter made up of a 1.2 uF poly cap, a 0.24 mH inductor and an 8 ohm series resistor. This actually sets the crossover frequency to be fairly high, you’d expect a gap going from the woofer to the tweeter, however that’s where the waveguide comes in and does its magic. The waveguide is providing about 6 dB/octave gain all the way down to about 2 kHz. When combined with the filter response, you end up with a decent overall frequency response that actually combines really well with the woofer. And the benefit is that this gain comes acoustically, so the tweeter is able to do less, i.e., less distortion, and still provide a decent output. The only adjustments needed here will be to match the crossover point and level with the woofer once we add the 2nd order filter to the woofer. We’re going to make only a small adjustment here, as I felt like the raw tweeter response with the waveguide was quite good with no serious defects. In which case all we need to do is set the crossover point so it sums well with the woofer and adjust its level so it’s not too bright but not too dull. Special notch or other shaping filters won’t be needed here.

So there’s a couple of approaches to designing (or redesigning) a crossover. You can either wing it real-time with crossover parts and measurement gear and just start changing things to see what happens. Initially I like to do this because I’ve played around with enough crossovers to already have a general idea of what changes different values and topologies will make. Plus it’s quick and easy, but can be limiting if you don’t have a stock-pile of caps, inductors and resistors on-hand. The second way is to measure the frequency response and impedance of each driver separately in the cabinet (without any crossover) and then model or simulate the crossover using a crossover simulation tool, such as Passive Crossover Designer or XSim. I’ve been accustomed to using PCD 8.0 so it’s my preferred tool of choice. Once you’re in your simulation space, you can play around with thousands of crossover options to your heart’s content. To validate the simulation all you need to do is design the crossover as-built and compare it to the baseline stock measurements of the speaker. In this case we can easily replicate the uneven response of this speaker with the basic crossover it came with. And within the simulation we can see why – that fiberglass woofer has a hard break-up mode at 5 kHz that is about 9 dB above the nominal response of the driver. With the cheap 1st order filter the response is only down about 8 dB at 5 kHz, which means the break-up mode is still very much in the audible pass-band of the speaker. Combine that with the fact that the tweeter response is now starting to roll in, you end up with the peaking response you see here.

So after playing around in PCD for a while, we come away with a few reasonable options for this speaker. First and foremost, we gotta get to a 2nd order on the woofer. I initially started at a 10 uF cap, then a 12 uF cap, then jumped to 16 uF and settled on 15 uF as it’s a standard value. I bought a pair of Dayton Audio poly caps from Parts Express to tackle this job. We’re not going to change the inductor, you could go to a larger air-core inductor, decrease the series resistance, maybe by some extra low-end, but I wanted to make this mod pretty simple and more importantly as cheap as possible. This speaker is already pretty cheap, and it doesn’t make sense to dump a bunch more money into to make it better, when that money could have been used to just buy a better speaker to begin with. So with that in mind I just played around with the parallel cap value and some series damping via a resistor to go in-line with the cap. In most cases a crossover will benefit from some damping here since moving to a 2nd order response will bring the gain up just before the cutoff point as compared to a 1st order filter. That can be see in the modeled response plots shown below. This peaking is not always desirable so we can add just a small amount of resistance to tone that peak down. Anything between 0.5 ohms to a couple of ohms is usually enough. Going much bigger than this you will start to sacrifice the high-end roll-off response, and it ends up looking less and less like a 2nd order filter. I landed on 1.5 ohms as it brought the response effectively flat in this region when measured. Though 1 ohm and 2 ohms were reasonable options too which either increase (1 ohm) or decrease (2 ohms) the presence in the 1 kHz to 4 kHz region. See some of the plots below that showcase this effect. You can play around with this value a bit depending on how sensitive you are to this frequency region, whether you like it a little hotter or a little more subdued. I was basically just targeting flat. In the end it ended up being a touch light than my initial measurements with my makeshift crossover but am perfectly happy with where it turned out. I bought a bunch of Dayton Audio’s newer audio-grade resistors (the green ones) and think they turned out great and look great too.

For the tweeter, all we’re going to do is drop the crossover frequency a tad by going from the standard 1.2 uF cap to a 1.5 uF cap. Even this small adjustment made a huge difference in filling that gap now left at the original crossover frequency since the woofer’s massive 5 kHz peak is no longer providing any significant output here. I went up to a 1.8 uF cap and it was just too much, overlapping the woofer creating another peak, but this time just due to having too much overlap at the crossover point. Dropping back to 1.5 uF was perfect. From here the last thing I played with was the leveling. I tweaked a lot of options here, up and down, going to a standard L-pad, and just keeping the series resistor. Basically any value from 4 ohms up to the standard 8 ohms looked reasonable. I had initially settled on 4 ohms, but increased it to 5.1 ohms in the final crossover as I was worried about making the speakers too bright. At 6 or 7 ohms the treble is knocked down just a touch, but this is easily up for personal taste. I like where it’s at with the 5.1 ohms, it measures well, it sounds balanced, I think it’s the sweet spot here, at least for me. [UPDATE: After a couple of months of listening, I ended up swapping the series 5.1-ohm on the tweeter for a 3.0-ohm resistor. The speakers just sounded too dull for my everyday liking. While they might have been more flat before, I appreciate more having a somewhat lifted top end, and dropping the series resistor to 3.0 ohms seems to have done the trick. I did a quick measurement on axis and we picked up about a dB or so through the high-frequency range and so it sounds just that much better. Your personal taste may dictate otherwise.] I picked up two sets of poly caps from Parts Express, the Dayton Audio ones and the Jantzen Audio Cross-caps. I ended up using the Jantzen ones and they fit perfectly, but the Dayton Audio ones would have worked just as well. And just like for the woofer, we’ve got some of the new Dayton Audio audio-grade green resistors which worked out great.

Alright so here’s links to all the parts. Plus if you want to get extra values for some of the resistors, I’d recommend it, as they are fairly cheap and you never know when just one more ohm or one less ohm is going to make the difference for you. The capacitor values are pretty well set though and shouldn’t need too much tweaking on their own. If you take a look at the pictures you can see how I mounted each part to the existing crossover such that you can just drop it back into the cabinet like normal. I removed the original resistor and capacitor and added the new ones in their place. Well sort of. The resistors are way longer, so I rotated them lengthwise and soldered one end to the cap and the other end to the normal location in the circuit board. The massive 15 uF poly cap was hot-glued to the top of the inductor, soldered to the resistor, which was hot-glued next to it, and then connected across the output connections to the woofer. Take a look at the pictures to see how I put it together. I did one crossover better than the other, as I thought it would be easy enough to attach the crossover to the terminal cup after it was installed into the box, but this proved to be more difficult than I would have liked. So I improved the layout on the second crossover to make sure it could be installed along with the terminal cup in the back. This worked out much better and just required me to move the one resistor up a bit.

So that’s about it! Check out that sweet flat frequency response plot now. Reversing the tweeter results in a nice deep null when measured on-axis and 0° with the microphone directly between the woofer and tweeter. No more polar shift up or down. These speakers will sound great when right at about ear level. Now all of this is measured with a single mic and in a room, so don’t expect Klippel-level accuracy here, but for doing direct comparisons of the speaker against itself, with all the other room and measurement faults being the same, you can still get a pretty good indication of what this speaker once sounded like compared to what it sounds like with this new crossover. I also added a small piece of poly-fil to the back panel which was completely bare. I could do more to improve the damping here, but for now this is where we’re going to leave it. Panel bracing would be the next thing too, but we’ll wait on that for now. I’ve listened to a handful of songs and so far have been pretty impressed. I’m driving the speakers with an AIYIMA A07 amplifier. This amp will push about 80 watts into 4 ohms so it’s got plenty of power to drive these speakers to some decent levels. Especially if you intend to use them as near-field monitors. I’ve got them hooked up temporarily as part of my small recording setup and I’ll leave them here for a while to keep testing them out. They honestly sounds great though and you’d be hard-pressed to build a comparable pair of speakers for as little money. Anyway, check out the pictures below as well as a bunch of plots/graphs of various measurements I took that provide some insight into what each driver is doing, what the crossover is doing and how different changes effect the frequency response. And thanks for stopping by!

Graphs and Plots from REW

Teardown and Crossover Build Pics

About Dan

For a complete biography, stop by www.danmarx.org and click around for a while.
This entry was posted in DIY Audio and tagged , , , , , , , , , , . Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *