DIY audio: what are you building?

[SBL 31]

Pylon racer? I've seen those things fly and they are unbelievable...

 

I used to fly model aircraft a lot

 

Regards,

 

SS

 

You should have said   I didn't want to confuse anyone with specifics. But yes, those props are being run on FAI F3D pylon models. These days they do over 340kmh and pull over 30g's through the turns. Awesome stuff.

[tuyen 715]

Another 3 DDDAC1794 modules to build.    Soldering iron is going to get a bit of workout tonight!

 

[Upfront 1,954]

Another 3 DDDAC1794 modules to build. Soldering iron is going to get a bit of workout tonight!

So how many total mate?

[tuyen 715]

So will be a total of 4 modules running in parallel.   Each module contains 2 BB PCM1794 DAC chips.

 

If there is a nice improvement from 1 to 4 boards, I would like to eventually try 8 paralleled boards like Triode Dick is currently using:

http://www.audio-creative.nl/hifi/dddac-1794-nos-diy-dac-deel-4-2/2/

 

[tuyen 715]

multitude of electrolytic caps populated! (I know you love it Mario). 

 

remaining bits soldered on.

 

Presto!  4 stacked boards paralleled and ready to be hooked back up to the power supply & xmos usb to i2s module.

[georgehifi 1,532]

Tuyen, one question.   Why didn't you use diy boards that have a proper multibit dac chip like the last one produced the PCM1704, instead of the PCM1794.? Was it a cost issue?

 

Cheers George

[tuyen 715]

Hi George.  Not sure why the PCM1704 was not chosen.  Maybe due to cost, maybe not..     I just went according to Doede Douma's recommendation of his current hi-res design, after enjoying his previous 16-bit TDA1543 based DDDAC1543MK2 NOS design for the past couple of years.

 

Only recently sold the 120chip paralleled modules to Jean Hiraga, who after hearing it somewhere, loved it.   I didn't really want to sell it, but he offered top dollar for it, so could not refuse.

 

Last photos of the modules before it left me.  So many hours putting these together.   Especially when having to tightly fit a custom aluminium heatsink layer between each of the 12 DAC chips per module.  Then having to solder all the legs together.  Fun times

 

 

 

 

 

Would you say the Audio GD Reference 7.1 with the 8x PCM1704UK chips, a quality implementation?  If so, I think I know someone local who has one.   Might be worthwhile and interesting GTG for DAC fanatics in WA...

Edited October 27, 2013 by tuyen

[henry218 1,359]

here's another WTF!!! moment from Tuyen 

 

love to hear the comparison between master7 and your dac tuyen, i saw on different forum the Master7 is a significant upgrade to the REF7.1.

 

cheers

henry

[lusk 174]

Have been attempting to build a 3-way speaker that seems to be going on forever. I have finally finished

the bass section.

 

It will have RAAL and custom Audiotechnology drivers in sealed enclosures.

 

 

 

 

 

 

 

[Gee Emm 1,918]

Impressive work!

 

Dimensions?

[henry218 1,359]

impressive indeed, not to mention one of the best drivers out there 

[lusk 174]

Impressive work!

 

Dimensions?

 

Thanks, when finished with the top section installed it will be 1100mm height, 400mm width and 480mm depth.

 

I didn't take many photos when building it but there are some more here of it going together.

 

http://s1236.photobucket.com/user/lusk888/library/DIY%203%20Way%20Speaker%20Build

[Krispy Audio 2,545]

Wow - looking fantastic! I especially like the machining work on the head. Hope you have a good trolley to move them around, they look like they'll weigh a tonne! 

[Peter_F 1,773]

Thanks, when finished with the top section installed it will be 1100mm height, 400mm width and 480mm depth.

 

I didn't take many photos when building it but there are some more here of it going together.

 

http://s1236.photobucket.com/user/lusk888/library/DIY%203%20Way%20Speaker%20Build

Outstanding work mate.  Is it made of marine ply?  

[Gee Emm 1,918]

Nice mill work.

 

Cabinets aint bad either!

[lusk 174]

Wow - looking fantastic! I especially like the machining work on the head. Hope you have a good trolley to move them around, they look like they'll weigh a tonne! 

 

Thanks Cam, they would be over 80kg which made it very difficult to move and paint. Need a fridge trolley to move them.

 

Outstanding work mate.  Is it made of marine ply?  

 

Yes, only because l planned to paint them, l wanted it to be stable and not move or crack. 

If l ever build another set of speakers they will be wood veneered 

Edited November 1, 2013 by lusk

[tuyen 715]

diy jbl crossovers according to this publication by Nelson Pass - JBL L300 CROSSOVER NETWORK MODIFICATION (http://www.firstwatt.com/pdf/art_l300.pdf)

 

[henry218 1,359]

tuyen,

 

looking at those $$$ components, perhaps invest measuring mic and DATS to measure impedance? just to be sure

 

cheers

henry

[tuyen 715]

hi henry,

 

I got a mic already     I'm not actually designing the crossover from scratch,  I'm just replicating the crossover nelson has published since the mid/tweeter is same.  The woofer the L300 uses is a the 15" JBL 136A (93db), while the speaker in question uses the 12" JBL 124A (89db).      Have been advised by jbl gurus on audioheritage forum that woofers are compatible, but may need to attenuate the mid driver a few extra db via fixed l-pad circuit if it is still to sensitive when adjusting with the l-pad.

[henry218 1,359]

DATS is essential, so when you have the freq response and impedance plot of every drivers, you will know the real transfer function happening between the drivers. As those are old drivers, i believe its even more important, where the parameters might differ slightly between units.

 

i measured few altecs, i remembered one of the compression drivers have impedance difference of 30-40%, which is ok for PA, but for home, with those prices and the components prices, i just think it is necessary.

 

however, yesterday i just found out my tweeter was connected inverted and introduce a dip on 1.5-4khz around 12db (crossover region), i can honestly said that i wouldnt even noticed if i didnt measure it. The correct connection definitely sounds better, where the soundstage become HUGE

[ArthurDent 2,946]

DATS is essential, so when you have the freq response and impedance plot of every drivers, you will know the real transfer function happening between the drivers. As those are old drivers, i believe its even more important, where the parameters might differ slightly between units.

 

Is this DATS as in Dayton Audio Test System @ US$130?

 

If so why not just use REW which is free plus a few bucks to build a breakout box.

[henry218 1,359]

i guess you can use that as well . i have not tried REW, since not many advanced users use REW for T/S parameter and Impedance plot measurements, mostly they used DATS, so i just think its a no brainer for me to spend $130 if i want to design speakers that used components more than $1k

[Gee Emm 1,918]

build a breakout box.

 

More info Ken?

 

Graham

[peterpan 278]

Have been attempting to build a 3-way speaker that seems to be going on forever. I have finally finished

the bass section.

 

It will have RAAL and custom Audiotechnology drivers in sealed enclosures

 

Lusk,

 

that is a very impressive build and quality components!  loved the pictures of the cabinet build. diy at its best.

 

good luck with it although I think you have that pretty well down pat! 

[ArthurDent 2,946]

More info Ken?

 

Graham

 

http://www.hometheatershack.com/roomeq/wizardhelpv5/help_en-GB/html/impedancemeasurement.html#top

 

The novelty of fiddling around with the leads soon wears off so you build one of these.

 

The extra set of leads is for attaching EQ components (something that the very similar and probably just as accurate DATS setup doesn't allow ).

Edited November 6, 2013 by KenTripp

[henry218 1,359]

the other leads to signal generator like amplifier?

[ArthurDent 2,946]

EQ components as in passive notch filters, rising impedance correction networks etc. so you fix problems like this...

 

 

[henry218 1,359]

EQ components as in passive notch filters, rising impedance correction networks etc. so you fix problems like this...

thats cool 

[tuyen 715]

Cool stuff guys. I will look into doing some sweeps/tests after I get the base crossover circuit up and running.

[tuyen 715]

Excerpt from the article by Pass that I found interesting:

 

"After I measured the stock crossover, I began to assemble a new one from scratch, wiring 

prospective components on a piece of plywood sitting on top of the speaker. I set the system 

up so that I could easily listen and also measure electrical and acoustic response.

 

This started a process that lasted several months. Some of you might imagine that we can 

just plug the measured numbers into the computer and it will spit out a perfect crossover 

network. Certainly it will spit out something, but it is unusual for the results to be the best 

possible sound. At best, the simulations will give you a good start. 

 

The revision procedure involves trying something, measuring and then listening a lot. With a 

soldering iron and lots of parts handy, the first two parts only take a few minutes. The last 

part is occasionally quick (when it sounds awful). If it sounds good or better, than it is 

necessary to spend more time with it, running through a series of familiar recordings, 

sometimes for days, deciding on whether the change stays or not.

 

And then you reiterate the process until you are happy with the sound and every change you 

make sounds worse.

 

Initially I worked with variations around the original in-phase wiring of the midrange, but I 

found it difficult to construct a filter which was flat at 1 KHz and also sounded as good as I 

wanted. I think the difficulty revolved around the distance time delay of the midrange, which 

can be clearly seen in a MLSSA pulse (that is to say there are two of them – one from the 

woofer and tweeter and another from the midrange). Finally I began working with the 

midrange phase flipped.

 

This is where I had to choose between faults. There is simply not an ideal complementary 

match between the character of these two drivers. After some more experimenting, I worked 

up my best compromise - a midrange high pass network which is different in the details and 

response curve, but delivers most of the warmth, intimacy and articulation I was looking for. 

This is where the instruments have a “float in the air quality†that takes the performance out of 

the box and into your room.

 

Very often you get something like this simply by trying everything and in the end depending 

on your ears to tell you when it's right. I continued playing with variations for several weeks, 

but kept coming back to this circuit"

 

 

What I'm wondering is, couldn't he of just use measurements to determine optimal config for freq/phase  and in turn produce best sense of soundstage, etc. ?

[henry218 1,359]

i dont know about his process, but from that article, he built his crossover with measurements, and simulate the closest target response to his liking.

after he built the first one, he wants to tailor "voicing" the speakers into something that he prefer, but probably those changes didnt have enough merit to sacrifice other aspects of the crossover.

[Gee Emm 1,918]

What I'm wondering is, couldn't he of just use measurements to determine optimal config for freq/phase  and in turn produce best sense of soundstage, etc. ?

 

You thought that too...

[henry218 1,359]

coorelation between freq response to the perceived sound is something that is not that easy to identify.

if it took him few months of experiment, that crossover must be something to be appreciated.

the impedance measurement i suggested to you earlier was just to make sure that your xo have the same transfer function with his

[tuyen 715]

Does that mean there are some things in speaker/crossover design that traditional measurements won't show?

[henry218 1,359]

measurement can only go as far as the measurement equipment and environment itself.

i dont think he used anechoic chamber to measure the drivers, so there will be room of error.

for the bass, in room response wont showed the real response, instead for us DIYers we need to come up with a solution where we need to splice the frequency at the threshold of the gating (minimizing refelctions) with dustcap measurement and port measurement(if ported).

as for midrange, it may have something to do with the distortion profile of the driver, relative xmax for its band pass responses and Le (inductance of voice coil)

[Newman 3,451]

It means Pass is just foolin' around. And that's how he likes to fool around. And he's entitled to.

[hochopeper 1,625]

Does that mean there are some things in speaker/crossover design that traditional measurements won't show?

 

No.

 

Many would have you believe that is the case but it isn't the case IMO.

 

It isn't to say that measurements are trivial.

 

A pedantic part of me would much rather not read the phrase 'traditional measurements' ... they're just measurements and while the general technique is reasonably well established there is room for movement and a lot of options for how to achieve measurements of different aspects of a speaker + crossover and frankly 'traditional measurements' doesn't add any information about how a measurement is performed. This isn't a comment about you in particular tuyen, you're far from alone in using the phrase and I know the phrase gets thrown around a lot, but really, when you think about it ... what on earth does it mean?

Edited November 6, 2013 by hochopeper

[tuyen 715]

Thanks for your replies.  Very interesting indeed.

[hochopeper 1,625]

coorelation between freq response to the perceived sound is something that is not that easy to identify.

 

I think this goes dangerously close to perpetuating a myth. 

 

 

 

if it took him few months of experiment, that crossover must be something to be appreciated.

 

Or he's tuned it to himself and his room, which is fine, but might not translate to you and your room .... 

[henry218 1,359]

I think this goes dangerously close to perpetuating a myth. 

 

 

its well documented actually, that is one of the reason why i put bbc dip to my speakers, where i want to experience why people from that study preferred it that way. if it was not in that paper, i would not know how to get the laidback and pleasant sound.

 

another thing is, IME the raal tend to rolloffs after 10khz (although i can alter this with foam position to correct it) but Scan R29 has a very flat response to 20khz, but the sound just better and more detailed on RAAL than on R29.

 

and the last, most speakers are enjoyable if +3-6db on the bass (including room gain)

 

although Nelson Pass is not that well known for his speakers design, his approach towards open baffle, active crossover (B4, B5 and XVR1) quite accepted as a good approach, not to mention hideously huge jordan jx92s horn. he may not be industry leader, but he got alot of experiences.

Edited November 6, 2013 by henry218