Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Viitalahde]

I've been keeping myself from starting this topic for so long, until the project is truly finished and I can take a bunch of pretty pictures for you.. But I can't control myself because I feel like telling the story.

You've seen them:

Still standing on a temporary stand. The proper ones should be coming in within two weeks.

The project started years ago - 2006, I think. I was figuring out what I like and what I don't like, and slowly I just was not interested in any commercially available speaker. I'm sure many of them them would have sounded good to me, but I just did not know where to go.

I knew I wanted three-ways, I knew I wanted a really accurate midrange, and I also knew that I had trouble with the sound of bass of bass reflex enclosures I had heard.

The choices of speaker drivers narrowed down and I was getting interested in ATC drivers more and more, knowing that I like the sound of dome mids. I had a certain type of transient impact in my mind (partially influenced by an Eastlake system I once heard, honky and coloured but I liked the "speed" it delivered in some parts) and I figured ATC's would be a good way to get some of that.

I decided to try out sealed alignment for the loading of the bass driver. Since it's a pretty inefficient method and the cone excursion would be easy to get out of the hand, I needed a big driver, and a 12" seemed like a good compromise. ATC had a model that fitted well in a big sealed cabinet, and the good match to the mid wouldn't hurt.

For the tweeter I went for a Morel MDT-33, simply because of my working history with Dynaudio & Morel tweeters. I just like them, and I read from the internets the ATC mid isn't too fussy about mid-tweeter pairing.

The speaker cabinets are a 100L in inner volume, and my estimation is that they weigh about 80kg, each. The cabinet is made out of 18mm plywood and 22mm MDF sandwich, heavily braced and damped with lead bitumen sheets from the inside. The front baffle is a triple layer construction, one sheet of MDF and two of plywood.

For the amplifiers I decided to try out Hypex amps - I'm using 2x 400HG's and 4x 180HG's. There's a pair of custom made toroids for them and a big bank of Siemens Sikorel series capacitors. I also have a soft start and a DC protection system installed.

I'll try to take a picture of the 5HE thing tomorrow!

Crossover: The crossover is active, an old ATC model I bought from a chap in the UK. I was going to use it for testing only, but frankly, the old thing works perfectly with the system once I got it set up properly. It uses 2nd order slopes with all pass filters for phase alignment. I still need to service and modify it a little, and it's a job for this spring.

This has been an interesting ride for me, and while I was already mentally prepared to have problems installing the system in my new room (I didn't use it at the previous place), things went smoothly and the system does sound excellent.

There's something in the sealed cabinet bass alignment that really appeals to me. It doesn't hit you in the chest, it is just politely there and tells you what things sound like. Since the rolloff is 12dB/oct, they go pretty deep.

As a whole, the system is a big window to the sound at hand. It tells me what is there, but doesn't act like I'm working with a microscope - there's actually a bit of a party to them, which I like.

The driver integration is perfect, and while it's dumb to evaluate a part of the system alone, the midrange and transient information from the ATC driver is really, really great. What's even greater is the big wide sweet spot I have here. Good for me, since I like to stand up and/or move around.

Personally, I think it's kind of nuts to build your own speakers to work with. It's actually pretty stressful to figure everything out and measure objectively, and to translate these measurements to an opinion on how things actually sound, and then go back to tweak. The possibilities for fatal errors are just too great.

I might have just lucked out here, but right now something's locked in very well and my work seems to be translating like never before.

Maybe I'll manage to tweak even more out of the system. Or maybe I'll just mess it up. Hope not. The point of "it's done" has to be found somewhere.

I don't know if this is the last speaker system I'll ever own, probably not. But I predict many happy working years with it.

[DOMC]

That is so awesome, Congratz on getting it done and successfully as well  - ive got some pics I took of a mates studio who just went through a similar process - ill try and find out some more information for you on hi speakers as well but here is a teaser.

[SafeandSoundMastering]

They look good. 

[bblackwood]

That's some really impressive DIY, Jaakko. I've long wanted to get back into building speakers (played around with speaker design years ago) but haven't the guts to DIY my working mains. I'm sure they sound good, would love to hear them.

I am really intrigued by the concept of a well designed actively-crossed system - it looks like you covered most of the things I'd like to see in a 'perfect' system, though I wonder: does the crossover allows for 'time-alignment' of the drivers? Also, how well do they throw the stereo image?

Thanks for posting this - your DIY is crazy cool.

[Viitalahde]

Dom, crazy looking speakers, those!

That's some really impressive DIY, Jaakko. I've long wanted to get back into building speakers (played around with speaker design years ago) but haven't the guts to DIY my working mains. I'm sure they sound good, would love to hear them.

Thanks! It indeed takes some nerve to go through the project, but it has been one of the most rewarding ones to go through. Nevertheless, if the design or my capabilities had failed, I would probably had ordered something from Quested for starters (as I worked with VS2108's for a couple of years) and then figure out where to go to.

Recently, one of the speakers I'd be interested to try out are the Me-Geithains. On the other hand, it'd be interesting to have a listen at modified Tannoy ML10's.

I am really intrigued by the concept of a well designed actively-crossed system

It's well known that theoretically, active systems should dance around in circles compared to any passive system. Yet, there are excellent sounding passive systems. Passive can be perfection when tweaked to the last detail, but it's obviously a game of making everything work together perfectly. My "simpler is better" mind tells me that ultimately, a good passive system would be the best choice, but I really have nothing to complain about a good active system either. A couple of op-amps on the way are nothing if the crossover slopes go well with the drivers.

When you consider the essential basics of a mastering speaker, I'd say it comes down to an extended, flat frequency/phase response and excellent dynamics/power handling. From there, it's all a matter of taste. Pick your distortion.

it looks like you covered most of the things I'd like to see in a 'perfect' system, though I wonder: does the crossover allows for 'time-alignment' of the drivers?

There's no way to time align drivers in an analog crossover, not without huge amounts of distortion. To get the delays needed, I think you'd need L/C delay lines from here to moon and back, and delay chips are for effects only.

The bass/mid are not time aligned, but as I have understood it, it is less essential of an issue when going down in frequency range (the extreme example being a subwoofer placed in a whole different place in a room). The crossover frequency here is 380Hz-ish. As the drivers are not time aligned, the midrange is already rolling off on its own and the crossover slopes are 2nd order, the all pass filter really does help in evening out the response. It can also be heard in better integration of the drivers and as more information in, say, vocal reverbs.

The mid/highs are time aligned, or near it. The difference between the tweeter and the mid isn't big as the domes are nearly at the same vertical axle, and I have further compensated for it by simply having the tweeter higher than the ear level. This was done simply by the ear.

Also, how well do they throw the stereo image?

Now here's one of the design points. The front baffle is wide and the whole lay-out is pretty old school.

For absolute pinpoint accuracy, narrow, tall baffles and minimum diffraction are the way to go. I suppose Dunlavys and similar are the extreme of this sort of a design.

Myself, while designing, I really began to think how important it is in mastering to evaluate pinpoint accuracy? We're listening at the whole picture, and I personally like how a bigger baffle projects the sound towards the listener, creating a large, but not necessarily absolutely sharp picture.

The stereo image I get from these is wide and large, and I can clearly hear/see where things are located. But they're not pinpoints, they're more like thumb-points, each clearly sitting where they are at.

This is where the "party" I earlier mentioned comes from. I'm terrible in analyzing anything. I go by the instinct, and when things are right, I don't know they're right, I feel they're right. Speakers like these help me to work as they don't shovel the information down my throat, they just represent it and let me decide whether to grab on it or not.

My choice of having a slightly more live room than usual was also a part of this. Put me in a dead room with a pair of the most accurate speakers in the world, and I'll EQ every nuance out. 

[bblackwood]

Nevertheless, if the design or my capabilities had failed, I would probably had ordered something from Quested for starters (as I worked with VS2108's for a couple of years) and then figure out where to go to.

Wouldn't be a bad choice, I also worked on some Quested (3208's) for a number of years when I was @ Ardent.

Recently, one of the speakers I'd be interested to try out are the Me-Geithain

I've heard these at Dave Greenberg's place, they sound very good. I think my top choice, were I to switch from B&W's (though I can't imagine doing that at this point) would be ATC's, though the Me-Geithain's would be high on the 'try out' list.

It's well known that theoretically, active systems should dance around in circles compared to any passive system. Yet, there are excellent sounding passive systems. Passive can be perfection when tweaked to the last detail, but it's obviously a game of making everything work together perfectly. My "simpler is better" mind tells me that ultimately, a good passive system would be the best choice, but I really have nothing to complain about a good active system either. A couple of op-amps on the way are nothing if the crossover slopes go well with the drivers.

Yah, @ B&W they literally design everything in house so they have the ability to control every detail, something most manufacturers can't do. That said, I've always found it interesting that B&W's ne plus ultra design (the Nautilus speaker) utilizes an active crossover...

I have further compensated for it by simply having the tweeter higher than the ear level. This was done simply by the ear.

Very nice. Simple yet effective design.

Now here's one of the design points. The front baffle is wide and the whole lay-out is pretty old school.

For absolute pinpoint accuracy, narrow, tall baffles and minimum diffraction are the way to go. I suppose Dunlavys and similar are the extreme of this sort of a design.

Yah, I noticed that, yet ATC's sound great, and image well, with similarly wide front baffles. While I agree that pinpoint imaging isn't a huge deal in a mastering monitor, it's definitely a strong point with the near baffle-less N802s, which I'm a huge fan of. *grins*

My choice of having a slightly more live room than usual was also a part of this. Put me in a dead room with a pair of the most accurate speakers in the world, and I'll EQ every nuance out.

Hahaha, sooo true. I made the point with thomas when doing my room that it needed to remain somewhat 'live' as I hate the 'headphone' like response so many mastering rooms have. Luckily, he hears things in a similar way and drew it up perfectly.

kudos.

[Viitalahde]

Small update before a bigger one next week or so.

I received my custom Towersonics stands today, and quickly put one together to see how they turned out. They seem excellent, and I just can't wait to install them this weekend. Besides far less pain in the eye, I'm expecting some degree of sonic improvement, but I really have no idea how much. Next week, I'll be in this topic either praising about more palpability in sonics or saying "Meh, but they do look cool". Just wait.

What I wanted to tell is that last week I felt I had gone through enough of different types of music with them, from church choirs to black metal. My work seems to be translating well, and what I especially like here is that I don't hate stuff that sounds bad, it just sounds bad and I can approach it in an objective manner.

During the summertime there will be a complete service+upgrade to the existing crossover. I mean, the coupling caps are from the 80's. If getting rid of them altogther doesn't increase the already great resolution, I don't know what will. The current settings will stay in, until I can somehow match L/R even better. The mono is tight, though, so I don't there's much room for improvement.

[Ben F]

Has anyone build a K100 kit from Wilmslow Audio? I've always thought that they would be a good start for a ATC speaker project. http://www.wilmslow-audio.co.uk/k100-5-p.asp

[Viitalahde]

OK, I'm almost ready to take a few more pictures.

A little over two weeks ago I completely serviced the old ATC crossover. I put it in a new chassis, hooked the crossover boards to a better PSU, swapped the op-amps to modern ones and added rail bypassing & feedback compensation capacitors, since there were none. That alone did an improvement. I also replaced some of the trimmers and put fixed resistors in place of the level potentionmeters.

Yesterday I took out an op amp stage from the input side, configured as a Sallen-Key filter, 4Hz cutoff if I calculated correctly. I also figured out how to get rid of the remaining hiss and a very slight buzz. There's a FET limiting circuit - the FET acts as a shunting resistor in an L-pad. The other end of the FET is hooked up to a funky looking virtual ground amplifier. The levels are padded way down before the FET and gained back after. This FET also acts as a level control for the bands, it's the "rough" adjustment (the pots were operating in a smaller gain range).

Since this FET is an obvious source for hiss, and I have been able to mute the hiss (and the audio, of course) by shunting the L-pad output to ground, I've been trying to figure an alternative place for the level control if I bypassed the FET stage altogether.

Yesterday I saw it, it's there and ready to be used! Just before the band's output amplifiers, there's an empty spot for shunting resistors. The output amp is fed via 18k resistor, so just about any kind of a pad can be put there.

I'm definately going to try this on Wednesday. I can just temporarily go around the FET stage and play with socketed pad resistors. I very much presume the thing will be dead silent after this, and I expect the sound to improve too, since to my knowledge, such FET attenuators are also a source for distortion.

[Thomas W. Bethel]

Your "home built" speakers look GREAT and from what you have told us also sound GREAT as well. Contgrats on doing an exceptional job all around.

[Viitalahde]

Thank you, Thomas! It feels great to finally see an end to this project, at least on the major stuff level.

Yesterday was a really good day. Bypassing the FET stage altogether went smoothly, and it killed the noise like I expected it to do. An added bonus was that the sound got cleaner, too. I suspect the FETs did contribute some of their distortion to the signal.

I still need to do some tidying up after all this, and I would very much like to use a Jung regulator for the PSU (right now I have a basic LM3x7 supply there). I have one built, but I couldn't get it act well under a load. Must be a small mistake from my side.