Knotty Office Speakers

[mdocod]

Hello Tech Talk!

I built these speakers during the week between Christmas and New Years 2021. I wanted to do something interesting for our office. I saw this project "pine" board at lowes made from pine boards of various length all glued on edge and thought it looked really cool. It inspired a new speaker build.

Design Goals:

• Build my first 3-way speaker.
• Try to do a "wide-band-ish" midrange so that most of the mid and treble comes from a "point-source."
• Light weight, reasonably small so it can be wall mounted with dry-wall anchors in an office without tearing down the wall.
• Good looking! Conversation piece.
• Easy to finish! I don't want to spend days sanding and applying coats of varnish or stain or paint only to be disappointed by the result.
• Deep bass extension.
• Low to moderate output levels, low sensitivity is fine for this intended use.
• I guess they should sound good and have reasonably low distortion? hehe...
• Select components that are actually in-stock! (this isn't always easy!)

----------------------

​Driver Selection

To keep weight low I wanted to use all neo-motor drivers.



Woofer: Tang Band W5 W5-1138SM

I was originally planning to use all ND series drivers but the simulations just weren't panning out as well with ND140's in a small box. I'm sure I could have hammered something out with the ND140's and it would have sounded fine, but the W5's characteristics produce a better looking transfer function in a tiny box.


Midrange: Dayton Audio ND91-4

I tinkered with simulations on the intended baffle width with lots of different placement of ND64/65/90/91 series drivers and landed on this as a good compromise. I was actually hoping to use the ND64-4 for this project but I wasn't happy with the way the response was turning out with it on baffle and with filter components in-play. I'm sure I could have "hammered" it into submission with more filter components but the way the ND91 was simulating just looked a lot better on paper with fewer parts thrown at the problem. I also liked the fact that this driver claims to have had special attention payed to making a very low distortion motor. Indeed they do sound very "easy on the ears" in my opinion, even when driven hard/loud, suggesting a low distortion performance.


Tweeter: Dayton Audio ND16FA-4

I just wanted something small to pick up the top couple octaves that would be easy to install close to the midrange driver. After reading about a lot of different little neo-motor tweeters all indications are that for crossover points above 4K these offer phenomenally low distortion for a dirt-cheap light weight driver. The even smaller flange ND16FA-6 would have also probably worked well for this but I felt like the larger flange of this driver would be more forgiving to install.




----------------------

Simulations / Crossover Design:

I have taken a liking to "SpeakerSim" as of late. It has some quirks and bugs, but is easy for me to use. I used PE published data for the ND series drivers and a tracing tool to generate frd/zma data for the W5.

This is what I came up with after lots of tinkering and hammering and compromising. It's not a perfect result but it's a usable starting point. Crossover points are approximately 570hz and 7700hz. The "component detail" windows are overlayed here to show that there's also a 3ohm and 2ohm resistor in the band-bass circuit at critical locations to "shape" the response.



The woofer circuit uses electrolytics and iron cores to keep cost/weight down. The cap values purchased are 100uF and 125uF. The "10% tolerance" on these cheap caps almost always errs on the side of what would cost less to manufacture so the simulation is done with 90uF and 115uF responsively to give a more realistic representation of what to expect.

I will probably experiment with lower crossover points between the W5 and ND91 in the future. As I recall, many of the higher value inductors and capacitors I was going to need to pull down the crossover point were back-ordered at the time I was ordering parts for this build, so I was sort of simulating "around" what was actually available from PE. I prefer to do business with PE for this stuff when I can, I've been ordering stuff from them for occasional speaker projects since I was a teenager, (over 20 years) so there's a nostalgia factor here. They also tend to have good prices but this build really wasn't intended to be cost efficient, just elegant where possible.

The dip at ~150hz is the "bottom" of the valley created by baffle step loss intersecting falling room boundary reinforcement. The dip causes the speakers to sound slightly "thin." EQ helps. If I revise the crossover I intend to pull the crossover point down to around 200-300hz, which will give some tweaking leverage against that dip.

The dip at ~2500hz is an uncorrected baffle diffraction issue. My ears aren't sensitive enough to be able to care in this range.

Minimum Impedance is about 3.6ohm @ 84hz. One of the challenges when trying to hammer a response flat is to do so without inadvertently causing a ton of electrical phase shift or deep kinks in the impedance. The ND91 was easier to work with in this regard as a mid-range than other ND series drivers I experimented with in simulation.


Box is about (woops not 14) 7 liters of remaining volume tuned to around (woops not 28hz) 35hz. These require solid floor/ceiling/wall boundary reinforcement to counteract the falling response below 60hz and a touch of bass boost also helps. These are intended to be placed close to a wall, not out in the room like some speakers. Unfortunately, the predicted boundary reinforcement in the office was way off. More on this later, but in short, a drop-ceiling is not a boundary for bass! The bass extension in the office is lost due to a number of factors (the room shape is terrible and the ceiling is acoustically transparent at low frequency)

The sim below is wrong... I'm not sure why my sim file had 14 liters saved in it, maybe I was planning a revision floor standing version of this or something?


Baffle step/diffraction is accounted for in SpeakerSim.... This is just a "combined" baffle response example, but each driver has its own simulation based on the position on the baffle. The ND91 and ND16 drivers were pushed off-center to combat that baffle diffraction peaking response around 1-1.5khz.



The simulated listening position for this speaker is actually in front of the Woofer. This may seem odd, and normally, this would be the WRONG way to do this type of simulation, but I knew I was going to be mounting these up HIGH on the office wall. Our listening position is actually almost always BELOW the speakers. Even when standing in the middle of the room with the speakers tilted down a bit I'm still looking basically "at" the woofer as my listening position. The mid/tweeter drivers are above my head when standing.

Continued....

[mdocod]

​ ...Continued Simulation / Design:

Predicted Room/boundary reinforcement:

The boundary reinforcement prediction turned out to have some flaws. Primarily, I failed to account for the fact that the drop ceiling in the office wouldn't actually act like a boundary, and there's some reflections destructively coupling some of the bass energy as well in our office.


Here's the predicted listening window response and power response:

I intentionally tilted the on-axis response of this speaker design a bit "up" knowing that these speakers were never going to be listened to on-axis. In a future crossover revision I may tilt up the high frequency even slightly more. If you look at the on-axis response in my previous post, you may notice there's a "bump" in the response just below the crossover between the ND91 and ND16 (around 5Khz). The ND91's power response is obviously falling off at 5Khz, so when compromising on crossover component values there it makes more sense to go with a value that leaves the bump rather than shaves the bump up there on the ND91. The ND16 comes in around a dB lower at 10khz than where the ND91 peaks around 5Khz, yet the power response still shows a bit of a "bump" at 10Khz. I think I could have blended these with even more "tilt" between them and been fine, despite an uglier looking on-axis.

Continued... ​

[mdocod]

Working the Wood:

I'm not a wood-worker but I've been known to work the wood once and awhile... ;)


Since I'm not very good at this I try to make things easier on myself by using some rabbet jointing to help me align the woody bits. I use a little table router for this sort of thing. This photo pretty well captures my "old" ways of drawing up a cut sheet.


This pretty well tells the story of how they go together....


Lots of test fitting and adjustments to get it to actually "fit" since my cuts are all pretty trash.




If you're a DIY speaker builder you're familiar with the "clamp-o-rama" part of the build:

My advice.. buy more clamps than you think you need, and don't bother with the really cheap black and grey ones from HF, they are used in this photo but they provide almost no clamping force before slipping. They are just on here to make the clamping situation look more ridiculous. I should probably throw those out.

The shelves on the garage wall shown in the background of this image are made from the same cheap project wood that I made these speakers from, and was selected for the same reason; It's light weight and looks cool! When mounting stuff on walls I like to keep the weight down as much as possible. This stuff is ~5/8" thick and probably weighs about half what 3/4" MDF weighs.

Continued.... ​

[mdocod]

The Easy Way to "finish" a box:

In the photo below you can see things starting to take shape. I believe this photo was taken after some light sanding and a quick light coat of teak oil. I did not want to spend a lot of time "finishing" the speakers and was going for a rustic look anyway. I think I may have spent about 30 seconds per side with a sander then maybe a minute or 2 to rub some oil on each box. This is the fastest way to "finish" speaker cabinets I have ever tried and will certainly being keeping this in mind for future projects. I hate sanding/painting/sealing/varnishing/veneering/etc. All those processes are time consuming and fraught with opportunity for failure! I figure if I use a material that looks kinda good from the get go then I would save a ton of time and effort....

The photo above also reveals how some "clearance" was opened up on the back of the baffle around the ND91 to breath. I'm not sure if you can see it but the back-side of the hole for the W5 was also given a bit of a radius for good measure.

Sorry I didn't take a whole ton of build photos, but you can use your imagination. The usual suspects were all invited to the party. A router, router table, router hole saw jig, drill, saw, clamps, jigsaw, etc were all involved as expected. I tried a forstner bit for the tweeter flush mount before drilling the through hole. It didn't work very well because I used a cheap dull bit and there were knots in the path. lol.

---------------------------

Crossover Assembly:

For "rev 1" of a crossover design I usually just zip tie and wire-nut everything. I expect I'll probably be going back and making changes later so this makes re-using components in a different arrangement easier. I'm sure there's an audiophile out there who would absolutely hate this.... I'm just a nerd with tools. I don't have golden ears so I'm not going to be lifting any vails with better connections. Mountain Dew is also a critical component to all garage projects.

If you're wondering about those fancy pants 20W premium grade resistors.... That was not planned. I was hoping to use the DPR10 series resistors, which are about 1/3rd the price, but again, ran into inventory issues. Several of the values needed at the time were out of stock in that series, so I figured I'd splurge on these. I probably could have just used regular sandblock resistors here and not been able to tell the diff but to be honest these 20W resistors remind me of some of the ultra-hgh-voltage resistors we use at work in pulsers. Based purely on liking that they look like something from a cool thing somewhere else I decided I go for the better resistors.

Oh... these wound up taking up a LOT MORE space than I anticipated. I'm probably loosing a liter in my boxes to all this junk. haha. They are just piled in behind the woofer. I used the polyfill and accoustic foam jammed in there to basically "jam" the crossovers into place.

--------------------------

Missing Photos / Use your imagination:

A few things going on inside the box that aren't shown.

• I applied a few small pieces of adhesive-backed butyl rubber sound deadening sheet on the sides of the interior of the boxes. This adds some weight but I figured this pine could use some help in the resonance department.
• I installed both "spiked" acoustic foam panels and polyfill material inside the boxes. The small chamber for the ND91 is basically stuffed front to back top to bottom with it. The woofer chamber has the back and bottom wall lined with the foam, then polyfill down across the bottom to basically envelope all the crossover and wire rats nest. I tried to keep the path to the port entry clear.
• I also simulated the low frequency/box with hornresp to get a better idea of what the actual tuning frequency and response would be with these drivers in this box. There were differences but they aren't significant enough to worry about here.

-----------------------

Finished Speakers:



I set them up at home on top of my HT speakers to have a listen after they were finished. At home, with proper boundary reinforcement these sound absolutely awesome. I configured my HT receiver with the sub turned off, fronts set to "full range/large," and a bit of EQ fill for known response dips. I kept going over to the sub to make sure it was actually not playing. These produce a startling amount of deep bass for large bookshelf speakers, and the mid-range on these ND91's sounds great to me. These have a remarkably non-fatiguing sound and they do the "imaging/disappearing act" tricks I was hoping for with most of the mid and treble coming from a small point source. The crossover point to the tweeter is up so high that the tweeter is really just picking up the top octave "air" in things, and those frequencies are difficult to pinpoint a source or issues with anyway.

In the photo above, you may also notice my HT speakers, which are on the opposite end of the spectrum from the "knotty office" speakers. Those are built with an 8" pro-sound midbass and are not intended to be full range. Those are EBP optimized LOUD speakers. They only dig to 60-80hz depending on boundary reinforcement and are crossed to a dome at ~1500hz. They are about a full 10dB more sensitive than the "knotty" speakers while also being closer to an 8-ohm nominal load, but require a sub to be even remotely useful. I do have some photos of that build process also. I may post another build thread about my HT setup at some point.

Mounted on the wall at the office:


-----------------------

Regrets, planned changes:

These have been fun to have at the office but the bottom end sounds terrible in this room. The wall vibrates when the speakers drop the bass, and I believe the lousy wall mounting counteracts a lot of the bass energy from the speakers. Furthermore, the drop ceiling (of which we have some tiles removed for better ventilation from our hot office loaded with electronics) does not provide any of the predicted boundary reinforcement, in fact, the space above the office is "clear/open" in many directions. Unfortunately, the bass/treble controls of the little aiyma integrated amp aren't' really voiced in a manner that can correct for the response dropoff correctly, it just makes it sound boomier. I don't think any amount of EQ could fix the bottom end here.

I want to do a rev-2 crossover that brings the crossover between the W5 and ND91 to ~250hz in order to get even more midrange out of the ND91 and also to correct the dip in response at 150hz. I may also see if I can extend the crossover to the ND16 a bit higher as well, while also "tightening" up that crossover with sharper knees so that the null when listening off axis(vertical) is less pronounced.

I want to see if there's a way to reach up through the woofer hole to remove the first-leg of the port at the top of the box, reducing the port length to just the "L" shape of the box. This would increase the tuning frequency slightly and give a bit more gain at the tuning frequency. I got too greedy for sub-30hz response with these. Tuned to around 32-35hz would be an improvement.

I will probably retire these out of the office and bring them home at some point. I intend to replace these speakers in the office with a 2.1 project, moving the LF to the floor/corner of the office where bass energy propagates from so much more efficiently. I'm looking at maybe doing a CX120 build, or perhaps a 1.5 or 2.5-way with RS100's for the tops. Perhaps a RSS210 series driver in a horn for the sub.

Here's a rough draft of changes, with the room simulation and listening position to represent our great room (dining/kitchen/entertainment space).



More components required to flatten out the mid-range here, but the ND91 would be covering basically 200hz-10Khz in this configuration.

----------------------

I think that's all I have to share on this project at the moment. I'll try to check in on this thread once and awhile to respond to any questions or concerns. I have a number of speaker builds planned in the coming years, including a possible HT redo. I'll try to post them here when I do.

-Eric​​

[djg]

None of your pictures show for me after post number one.

Note to self, don't use hashtag symbol.

[mdocod]

None of your pictures show for me after post Home

I did hand to hand combat with the forums interface last night for many hours to try to inject the photos. They are showing for me. I wonder if I'm the only one that can see them? That seems weird...

[djg]

hopefully, people will comment.

[Steve Lee]

Nice write-up but a lot of pics don't show up for me either, sir.

[mattk]

No pics

[Wolf]

No mo photo
Wolf

[mdocod]

I just viewed this thread from another browser with no site data or my login, and I believe I am seeing what yall are seeing.... "empty" photos.

I tried uploading the missing files again, and now it's just attaching them all to the end of the post, I then "injected" them over the missing photos, and it's still insisting on putting them at the end of the post as well as where I inserted them.

This is absolutely abysmal.... Almost as bad as sanding and painting over and over again only to see differently bad results.

If there are any site admins or devs reading this thread, please, I beg you, fix the photo handling situation here. This is terrible!

[Chris Roemer]

Well, today I can see a couple dozen images (posts Home - #4), but I have to view with "Posts" clicked, instead of "Latest Activity" (which is my default view).
But MANY threads act like this (where I have to click on "Posts" to see any images).

[Steve Lee]

Yay! All pics showed up today and what a nice job you did on these speakers, sir.

[chrisn]

Pics work for me, now I want a cold Mountain Dew... I don't know the specs of the amp you are using with them, but some of the small ones fart out with something power hungry. +1 on buying more clamps than you'll ever need. I've got dozens and end up needing more sometimes

[mdocod]

We're using a little Aiyima A05 to power them. It's a no-frills nothing special dirt cheap chip-amp with a bluetooth receiver/DAC. Even when I let it rip after hours it only gets slightly warm to the touch after running a long time. It's remarkably efficient....

Review: https://www.audiosciencereview.com/f...plifier.26440/

I may look into a better amplifier at some point. Maybe after Rev-2 of the crossover. For now I think most of the disappointing aspects of these speakers can be traced to where they are mounted and some of the crossover design compromises.

[djg]

Nice job, nice documentation, thanks.

[mdocod]

Made some changes:

Pulled the crossover to the outside and made it a conversation piece.

Replaced the W5's with E150HE-44's.

Added a lot of driver weight but the crossovers moving outside roughly compensates.

The front baffles on these aren't thick enough to be flush mounting these heavy drivers so they are just surface mounted.

Changed some crossover components to work with new woofer. Decided to ditch the iron cores and electrolytic caps in favor of air/poly. Yikes those 90uF caps are YUGE! pretty cool looking! Pulled the crossover down to around 400hz.






The woofer swap fixed the response hiccup centered around 150hz and really cleaned up the critical low mid-range. The W5's just don't take well to passive crossover and really needed a bigger box, and just didn't present the same low-distortion that the other drivers in the design were capable of. The E150's blend so much smoother and honestly sound noticeably better in the bad room, even wall mounted. That strange shaped hook in the response couldn't be solved with the usual bass adjustment knob found on many integrated amps and such. The E150 on the other hand has a smooth rolloff that takes well to basic bass adjustment. The bass went from sounding kinda boomy on the top end and non-existant on the bottom end (in this room) to balanced/rich sounding, and the low mid-range from 100-600hz, where tons of vocal energy resides now sounds like it's coming from a truly high-end speaker.