NEW mid-top for 3 way!

Hi all! I am creating this thread for documenting progress on the mid-top section of my 3 way design.

The design will use Hi-Vi M8a woofers, Dayton RS52 mids and Hi-Vi RT2E-A tweeters all configured in a tri-amplified active crossover setup. The plan is to have seperate top units and floorstanding woofer enclosures; this is my progress on the top section.

Progress will be slow as I spend a lot of time away from home at uni and don't have my building equiptment there.

Anyhow, here is my starting point. I'll first post a rough sketch of the front of the mid-top section to give an idea as to what is being aimed toward:



A real wood front would be nice and so would brass fixtures (as the tweeter has a gold/brass element) but we will see about these. I am undecided on whether the drivers will be recessed in as normal (very difficult with this shape!) or recessed in but with felt around the drivers, a more triangular overall cutout can be used then.

Construction will be laminated, with basically numerous baffles glued together to make up depth. No actual enclosure is required per-se as both drivers are self enclosed, however a baffle on it's own wouldn't look right imo. Quick question, would having a solid wooden front baffle then MDF "baffles" behind it cause any issues with expansion rates being different, after being glued together?

Onto the first progress. I have been using my router to cut the aluminium faceplate of the Dayton RS52, party for aesthetics and partly to reduce the CTC spacing (at around 4.7khz crossover, this is important to minimise). I used a TCT 6mm bit and a home made circle guide to cut the arc form. This is my router:

http://www.machinemart.co.uk/shop/p...ter-kit-230v/path/11-routers-biscuit-jointers

Here is the cut shown:



And here it is butted up against the tweeter. A fairly seamless joint!




There of course was a problem. Not in routing, that went well, but the grille fell off of my other unit (I have a pair naturally) and when I tried to re-attach it with superglue a small amount leaked onto the faceplate. I made the error of trying to immediately wipe with whatever was to hand (tissues) when really leaving the blob to set would have been better. Needless to say, after failed attempts with acetone and hot soapy water the glue hasn't come off and the finish is spoiled further . I still have not decided if I will be replacing the unit or trying matte paint to hide the damaged area. I shall continue anyhow.

I hope to update after about 2 weeks. In the meantime I can try to answer any questions!

Hi,

I've had the tweeters a while and originally had a different mid. To me these domes seemed like a good choice to go with them. This way both the tweeters and mid are low mass drivers with similar transient characteristics. The dome mid also allows for a suitably high crossover frequency (tweeter needs at least 3khz, ideally higher to lower distortion) without breakup issues or large driver spacing issues. So far listening tests have found them to be a very detailed combination, which I like. Here is a design using very similar drivers to mine, although differently implemented:

http://www.lonesaguaro.com/speakers/cyclops/CYC.htm

For the crossover I will be using 24db LR, with a breakup notch on the aluminium dome. The tweeter will have it's protection capacitor forming the 4th filter pole with 18db actually active

The dispersion is different for the drivers however. Both have a wide horizontal dispersion, but the tweeters have a very narrow vertical dispersion. I hope this wouldn't cause too much issue

Sorry for lack of updates on this! A few reasons, firstly I don't have my camera back home, it's here at uni! Also, there is less to show than I had hoped

I went to get some more layers done this weekend but annoyingly the router guided trimmer bit fell apart (bearing broke open) nearly as soon as I started . It did damage the workpiece a little when it came apart but not too seriously. Woodfiller should fix it but only once it is all finished otherwise.

So, I have just 3 layers on each so far of about 20 required . 2 new TCT bits are on order and I will have a break for easter when I can hopefully get some done

Hello again!

This is still in progress! It's Easter break so I am getting a bit done. I've been working on a prototype crossover to go with a test speaker.

The test speaker is a baffle of correct shape (though poorly cut) holding the mid and tweeter, perched atop my mini computer subwoofer. It is the correct woofer (M8a) but is ported instead of sealed and does not have enough stuffing inside for midrange duties really.



Crossover is nearly as I intended. It features 24db slopes for all drivers (tweeter has 18db active, 6db protection cap), a notch filter for the woofer breakup and another for a mild peak in the response caused by the baffle. A notch for the mid breakup was experimented with but abandoned, it added a lot of noise for unknown reasons but it also caused phase shifts at the crossover point since the reduction required is quite dramatic. It's at 13khz and the crossover is 24db at 4.65khz so it'll be significantly attenuated by then regardless; it's right near the top end of my hearing anyhow so isn't that offensive!

The 24db electrical slopes should work well as all drivers overlap nicely. For instance, the woofer rolls off about an octave above Xover (once the peak is nulled) and the mid rolls off an octave below. A similar situation with the mid to tweeter but with more overlap.




What I want to experiment with is phase correction between the mid and tweeter, although they are very close to each other depth wise, mabye it isn't required?

It does play music and the result so far is promising. Obviously hard to judge in mono with one channel missing though! Some recordings from the 40s I have that are in mono sound great


On the actual speaker build the laminated layers are still going together, it's a slow process but I have my new router bits and they are much better!

Thanks Harry!

Per, to cut I just lowered the RPM to minimum (about 7000RPM on mine) and sprayed WD40 into the grooves I was cutting in the material before each pass. I used about 4 passes, so 1mm a time roughly. To cut I used a 6mm twin flute TCT bit. Most difficult was keeping a constant feed rate, it tended to move suddenly as I had to follow a guide to cut it (the radius was too small do do directly with the router compass).

Thanks for the tip, I will see how my new bits hold up then decide what to do. I think finding a bushing and bit of matching size might be tricky as the router bit won't fit through the bushing? Perhaps one with a smaller shank and larger cutter could be fitted with the router plunged...


My problem now is flush mounting those mid and tweeter units! The tweeter shape is a nightmare . I just tried and failed making a cardboard guide by scanning in the shape and adjusting the size up before printing, what I forgot was that preserving aspect ratio would elongate the shape, I just needed a 1cm border all around! I may try this again or buy a big enough bit of board to swing all the arcs on, but then those corners are a problem . Any ideas appreciacted here!

Hi, thanks for the comments! I do like the look of the mid and tweeter blended together, under some lighting the difference in texture shows up a bit (tweeter finish is rougher), but overall it certainly seems more attractive than two drivers spaced, expecially with that kind of excessive faceplate on the RS52!

I certainly want to do a flush mounting, I am going to try my cardboard template idea again, mabye tomorrow. The faceplates are 5mm on the tweeter and 4mm on the mid, it doesn't seem much difference in practise though.

I had considered felt and it would certainly be easier! The tweeter does have a loss of energy toward the very top though (12k+) and using felt might confound this? Aluminium is a nice idea too, it could look nice but I expect cost might be prohibitive here?

In this test crossover I have just cheap parts, TL074 for most bits, the buffer and BSC is NE5532 (better current drive I believe?). Seems a regular op-amp will do, but something with good drive capability will be best, no low power devices. I followed the guide here for my construction and developed notch filters in Switcher CAD:

http://sound.westhost.com/project09.htm

I have his PCBs so that is where the crossover will end up (along with additional bits for notch filters etc).


Here is a picture of the actual units in construction:



for better aesthetics they will go back a bit further, so more layers needed; actually one has been glued on since this and is drying now! Construction is naturally very dense and inert, though debatably not that essential with these enclosed drivers! No bad thing anyhow. I still plan on a real wood front, about 25mm thick. Not sure where to source it from or which wood to use though. Any ideas on good wood types? Needs to be a hardwood, easy to machine with a good looking edge and in a colour complementing the brassy tweeter element (I want brass securing screws too)

A quick update again. Just adding more layers to the construction, 2 more MDF ones to go (hopefully that'll give enough mounting depth for the connector panel and capacitor on the back!). The other one is at the same point though not fully flush routered yet, the hoover I use for dust extraction keeps cutting out oddly




Looks a bit wonky in the second pic I know! It's just confusing angles and the rough cut of that back opening that do it, it's all flat on the outsides really.

Still need the wooden front panel as well as the 2 more MDF layers. Also not decided on the finish for the sides . Still working on the flush mounting!

If anyone knows about Dyson hoovers too then let me know! It keeps cutting out, seems like a thermal cutout perhaps, it works again after a peroid powered down. Sucktion seems a little weaker than usual so am thinking there is an obstruction causing increased motor load but the filters look pretty clean. I cleaned all the other pipes out already.

Hello again,

These have ended up with 3 more layers than that last photo shows, which with the thick wooden front as well will give reasonable weight and stability (and not look too disproportionate!)

I'll have one ready for sanding and sealing soon, mabye today. I still haven't sorted the real wood , but am thinking of using elm. Anyone here used it? Will the cut edges be "shiny" (unlike with, ie, pine), will I need to be concerned with expansion if it is glued straight to the MDF? It will be aged and kiln dried wood I believe. The sides will be finished with vinyl wood effect most likely.

My main reason for posting was to present these CAD drawings (sketchup) I did:




It's my first time using any CAD so forgive the poor driver textures and constructional lines! The top section will actually be isolated with shallow profile spikes and rubber pads, I just couldn't figure out how to add these

Not decided which finish I prefer, originally was going for the light fronts (less staining of real wood). The wood texture is taken from a sample of the vinyl wood I wanted, so hopefully is fairly true to life (real grain texture will surely be finer though, I didn't want my textures obviously repeating in this).


Hopefully will have more real progress to show soon, things are moving along, just slowly!

About time for an update I think!

I have been sanding and sealing these ready to accept the "fablon" around the sides. Here's a pic of 2 completed sides with 7 coats of MDF sealer applied:



This one is at full depth now, minus the real wood front (32mm or so).

I haven't been able to get the wood effect covering I wanted (and shown on those CAD pictures), but have ordered and recieved a different one. I'm not sure about it really, It's not nearly as dark as the one I intended nor is it light enough to do the "inverted" scheme.

Another concern is that the speaker shrinks a bit towards the back, about 2mm difference from the front to back overall. This may make applying the covering difficult. Should have took all the layers from template really, rather than from each other

I have now my M8A woofers ready, I am undecided on if I will begin building bass cabs now or when these are finished. I also have a new RS52 to replace the superglue damaged one. I have cut this one already and took some photos in case any are interested. Bear in mind it is easier if you can swing a smaller arc in one go than to do this following a guide technique! I think the pictures are faily self explanatory but feel free to ask on any details! Annoyingly, the paint (powedercoat?) finish chipped a little this time, either the finish has been made poorer since my others (which were fine) or my cutting bit has bluntened too much



I am taking measurements with HOLMImpulse to perfect the crossover too

Been working on recessing the drivers. The mid is not a problem being round, but the tweeter is awkward! I've ended up running the router around the edge of the tweeter faceplate to create a template (the faceplate can be detached naturally!). Works quite well, now I've done this test I feel more confident about doing a "real" one.

What makes this harder still is the the 2 cutouts must be made in exact locations so the drivers sit closely together! I'll probably produce an overall template for use on the real solid wooden baffles rather than trying to line the 2 seperate cutouts together on the baffle.

This test baffle is made from (cheap) 12mm ply and 18mm MDF, giving around the thickness I want from the real wood eventually. I will probably buy a larger roundover bit too, this is a bit shallow. I also cut the recess a bit deep on this and the tweeter cutout is a bit ropey. This baffle will do for my measurements I hope

Well yesterday was measuring day. A day filled with confusion and frustration for a noob to it like me

I have changed several things in accordance with what I saw. I've only really be trying to "fix" these mid-top sections. Integration with the woofer is still off (more on that later).

Firstly, the effect the baffle has on the mid unit's response is more pronounced than the sim showed:



And measured:


So I employed a broader and deeper notch filter than prviously used, adjusted in accordance with feedback from measuring:



This seems to work faily well. It still leaves a relative dip at 4kHz, but I don't think it's significant enough to warrant further intervention. Problem, the circuit introduces a lot of hiss through the mid. Are the resistor values simply too large?

The crossover between mid and tweeter is now much better. Before the mid was effectively rolling off early. Inverting tweeter polarity now reveales a deep, sharp notch (which I stupidly forgot to save an image of, it's around 25-30db anyhow).

The tweeter seems to exhibit a mystery 6kHz peak. It's not on the datasheet, nor are there any simulated artifacts at this frequency. I'm hoping for possibly an acoustic fix for this, some carefully used felt perhaps.

To reduce it's relative effect I've added a treble shelving filter to the tweeter. It;s natural response drops by 6db after ~10kHz, so the filter is designed to combat this then allow me to lower the tweeter level (which I haven't actually done yet) and hence the 6kHz peak.

Here is the mid and treble combined output. It's probably better than it looks here, observe the scale:



I have yet to lower the treble level. It sounds quite hot to me even though it's still down in the very top frequencies.

Here is combined with the woofer. This is with the woofer technically incorrect phase . With correct phase a notch is revealed, around 1kHz oddly, not the 820Hz Xover point. Don't know what's going on here really, I think all measurements below 1kHz look dodgy. Shows some 50Hz noise too, not from the speaker itself mind



I don't wish to repeat this process, moving 2 amps, a desktop PC, crossover, speaker, DAC, cabling, mixer, mic, stand etc downstairs for the day

I've "undone" most the measurement based adjustments as I think there's something quite radically amiss with it. I've got the prototype crossover set around the simulations now and it just sounds much better.

It's causing me some issues now though. Basically, the woofer cuts in and out seemingly randomly and for that matter it even seems to fade in and out, the change is seamless. To make it more bizzarre, it'll invariably come back on if I increase the signal level through it (currently using PC as source and hence level control, though the 3 xover outputs are fixed attenuated to a more reasonable level). No problems when the volume is set fairly high. Any ideas what on earth is happening here? It's got me baffled and is getting rather annoying. Using quad TL074 Op-amps, NE5532 input buffer+BSC and a TL072:



Any ideas? Must be a dodgy chip I figure, I can't see how it could fade in and out otherwise