Beastaudio's 2015 NC Get Together...

[3ll3d00d]

That sounds like a very nice system. I've been considering swapping my JBL compression drivers for the 4550's but keep putting it off. They are run with vented TD15M-A's.

 

I do enjoy the design/build phase but I am v much looking forward to getting through that to the listening stage I had finishing designing a 2 way passive TD10H/BMS4550 which I thought worked really nicely but the opportunity came up to part ex my 3 TD10s for a load of TD12M's so I figured it would be rude not to. I just need to work out whether I can manage to go active or have finish the passive xo design. It does get expensive if I go active mind you, I'll need a bigger DAC to handle the extra channels (a 16 channel motu maybe is probably the obvious choice) and a whole bank of amps (thinking of getting a load of hypex based amps made). 

 

 

Not sure if it is of any help, maybe you have figured out how to do it.

But a few months ago I ditched the analogue multichannel AVR I used solely for master volume duty in favor for digital master volume within the PC, going straight to power amps.

 

that sounds like a pretty slick solution. My kit is all in a rack out of the room under the stairs and I have an RTi remote that has a MCE profile programmed in. It cost me a bomb but it's a mother-in-law friendly solution so was worth the money This isn't the blocker though, that is getting my cable box through the PC. The whole cablecard thing is non existent in the UK so, to go active, I need to get jriver working slickly with an hdmi capture card. I'm not optimistic that I can get that working completely seamlessly but we'll see. 

[Spacebug]

 

that sounds like a pretty slick solution. My kit is all in a rack out of the room under the stairs and I have an RTi remote that has a MCE profile programmed in. It cost me a bomb but it's a mother-in-law friendly solution so was worth the money This isn't the blocker though, that is getting my cable box through the PC. The whole cablecard thing is non existent in the UK so, to go active, I need to get jriver working slickly with an hdmi capture card. I'm not optimistic that I can get that working completely seamlessly but we'll see. 

 

Ouch, wish you the best getting hdmi capture to work good. 

I have a terrestrial DVB-T2 USB stick that works ok for TV,  pretty slow on channel changes in HD though. 

But I recently got fibre optics to my neighborhood and with that an IPTV box with hdmi out. 

If that box would ever be used for watching TV with decent audio I have to route that into the pc somehow... 

It has a stereo analogue out that I can use,  works well for getting music from phone or vinyl to the pc but then I would be missing multichannel audio... 

 

Stuff like this are never just that easy are they? 

Cause where would the fun be then

[3ll3d00d]

a PC as the source has always been the dream solution but it's never actually completely worked due to all the DRM borkage. I remember battling to compile mythtv and custom linux kernels way back in the day just to get TV to work, it did eventually but a few years of that is enough to turn you grey

 

https://www.blackmagicdesign.com/gb/products/intensitypro4k is probably the ideal solution, if it ever works properly that is (afaik it doesn't, not sure if desertdome/mojave is reading but he had one to try out)

[SME]

I think that is the question really, do analogue signal (preamp) paths really sound materially different? Of course there are lots of people who say yes but those people are also the ones who tend to use stuff like cable lifts  It doesn't seem particularly feasible to actually test this either.

 

I say yes, but not for the reasons believed by most.  If competently implemented, an analog signal path should be largely transparent relative to the damage that is done by the speaker transducers and the speaker-room interaction.  Even the ambient noise floor of a quiet room probably has a bigger impact.  Now, if all of these other things were some how perfect, then it might be possible to find some content that differentiates their performance, but I have yet to see evidence of such.

 

However, the real problem is that so many products are not implemented competently.  By this, I mean that mistakes were made in the engineering that substantially hinder or limit sound quality or otherwise introduce other inconveniences that make the equipment undesirable to own.  To pick the glaringly obvious example: the bass management is completely broken on the Oppo player.  How many audiophile magazines bother to test whether the bass management sections on the systems they measure can cleanly pass WCS material?  At least the D&M products appear to be capable of handling WCS signals and can output a healthy 4V on their pre-outs, but to do so the products must be configured in a non-obvious way.

 

Another consideration is reliability.  Cheap cables work just as well as expensive ones, until they don't.  And if your cheaply-made cable has a manufacturing flaw and gets nudged into a short, you may find yourself quickly having a bad day.  I spent about $250 on cable for my "new build" because I wanted some assurance of build quality rather than any kind of exotic alloys or fancy insulation.  Then there's my MiniDSP units.  The 2x4 unit behaves predictably, but emits full-scale DC thumps on shutdown, which is very startling if not damaging.  The OpenDRC-AN units have more problems.  The electronics are finicky and the firmware is very buggy.  From time to time, my units fail to pass signal after booting up, and I have to power cycle them again to fix it.  Worse still, I've had them occasionally pass garbage to their outputs during these incidents.  I've also observed that minor (even imperceptible) static electricity discharges to the outside of the aluminium chassis are sufficient to corrupt the memory state of the DSP, again leading to garbage on the outputs.

 

Of course, you don't *always* get what you pay for, *especially* in the audio world, unfortunately.

 

I'm considering going active with the LCR speakers I'm in the middle of designing/building (a 3way using a pair of TD12Ms under a BMS4550/SEOS) so that would see me move to a purely digital volume control. Of course so much else would change at the same time (would need new DAC with more channels, no prepro in the path, new amps) that a comparison would be impossible.

 

Didn't you just finish a 2-way with the TD12M and BMS4550s/SEOS?  Or by 3-way, do you mean "including the subs"?  I know you seemed pretty satisfied with your passive crossover in the recent past.  Why are you now considering going active?

 

I have 3 TD12M-4As sitting in my basement waiting for boxes to be mounted in.  I also have two super-secret compression drivers that are to be evaluated inside of SEOS-15 horns when I finish their boxes.  It's pretty slow going due to me being super busy, fighting illness, and most especially the fact that I'm a total woodworking newb.  I'm hoping to finish the speakers by the end of the year, and then I will go back to my software, which will be used for crossover and room correction.  I also have new racks to install along with a new 5 channel amp (as I'm going fully active), and it'll take me a while to get everything wired up, tested and configured.  The speaker boxes are essentially "first-pass".  I anticipate rebuilding the horn boxes with a slanted face for elevated mounting, and I anticipate adding an additional layer of plywood with a constrained layer to the woofer boxes before I'm ready to call them "done".

 

Edit: Err, just saw your later post about upgrading from TD10s to 12s.  So what is your third driver?  Or am I still misunderstanding something?

[MemX]

All of you guys make me feel inadequate, both intellectually and manual skills-wise  

[3ll3d00d]

Didn't you just finish a 2-way with the TD12M and BMS4550s/SEOS?  Or by 3-way, do you mean "including the subs"?  I know you seemed pretty satisfied with your passive crossover in the recent past.  Why are you now considering going active?

 

I have 3 TD12M-4As sitting in my basement waiting for boxes to be mounted in.  I also have two super-secret compression drivers that are to be evaluated inside of SEOS-15 horns when I finish their boxes.  It's pretty slow going due to me being super busy, fighting illness, and most especially the fact that I'm a total woodworking newb.  I'm hoping to finish the speakers by the end of the year, and then I will go back to my software, which will be used for crossover and room correction.  I also have new racks to install along with a new 5 channel amp (as I'm going fully active), and it'll take me a while to get everything wired up, tested and configured.  The speaker boxes are essentially "first-pass".  I anticipate rebuilding the horn boxes with a slanted face for elevated mounting, and I anticipate adding an additional layer of plywood with a constrained layer to the woofer boxes before I'm ready to call them "done".

 

Edit: Err, just saw your later post about upgrading from TD10s to 12s.  So what is your third driver?  Or am I still misunderstanding something?

I did and I was happy with that 2 way, it sounded v nice to my ears. I was also a woodwork noob so it took me quite a while to build the test box and even longer to get the crossover right, got there in the end though

 

During that build I realised that a floorstander would actually work better and this meant I had space for a 3rd driver, I modelled some 2.5 ways (didn't work well) and then started on some 3 ways (seemed more promising). I was going to leave it at that until the opportunity arose to part ex my 3 TD10H for 7 TD12M so I'm not going to use a TD12M as a woofer and another one as a mid in each LCR. It is not absolutely perfect in a model as a woofer but the sensitivity and the floor loading mean that it should still work nicely, at least all my models indicate it will deliver sufficient output down to ~70-80Hz without any issues at all. So that's the path I'm on at the moment, I have knocked up a model which crosses at ~350Hz and ~1200Hz  but need to get some directivity dataon the TD12 before I go further.

 

I'm considering active as my current amp is a bit noisy with the higher sensitivity drivers so I need new amps anyway & I've learnt that crossover components get quite expensive quite quickly. The capture card problem makes me doubtful though so we'll see what happens.

[dgage]

Gage, this is a really good value that I would have hit up if it was available when I got my Rigol: http://www.circuitspecialists.com/hantek-100mhz-digital-storage-oscilloscope-dso5102p.html?otaid=gpl&gclid=CNqi2brcrMkCFYI7aQodNTEBmg

 

Thanks Paul!  I just ordered another one of my Xmas presents today.

[Shredhead]

 

This goes back to the earlier point about whether the device can deal with WCS and which bit is clipping. Given this confirmation that the MV is after processing & DAC, it's not obvious to me how turning the MV down fixes the problem. Obviously it does fix it but still seems confusing to me.

 

_________________________________________________________________________________________________________________________________________________________

Each electronic volume control consists of a resistor string to attenuate the incoming signal, and a block of transistor switches. Each switch selects a different volume level at each tap in the resistor string.

 

Each volume control in an AVR must have a large gain adjustment range, for example -96 dB to +32dB. When half dB steps are desired, 256 resistor segments and 512 transistors (2 transistors per switch) are required. The LSI chip has ten volume controls, eight for adjusting the level to the power amp and two, with a smaller range, for adjusting the level to the ADC to prevent overload. In total over 2000 resistors and 4000 MOS transistors are needed. More transistors are on the chip as digital gates to allow an external microcontroller to close the correct switch.

 

The number of pins for these parts runs between 80 and 100. Integrated circuits incorporating this quantity of electrical components are called Large Scale Integrated (LSI) circuit.

 

I see what you mean about the MV and the WCS clipping being confusing.  I never thought about the volume control being IC based like that.  Makes it more feasible than I thought at first.  It's pretty asinine that it's that hard to talk to someone in the company for a technical question like that and when you do get a hold of someone, they make a guess up.  I just did some dumpster diving this week and scored a Marantz 6000 series.  Probably something wrong with it so when I get into the teardown, I'll see what I can figure out as far as the volume and dacs.  I'll post if I find anything interesting. 

[SME]

I did and I was happy with that 2 way, it sounded v nice to my ears. I was also a woodwork noob so it took me quite a while to build the test box and even longer to get the crossover right, got there in the end though

Sweet!  I'm so excited about getting mine done.  I've been sick the last two weeks, which is a big drag.  I did manage to glue up my first horn box.    It's not perfect, but way better than I thought I would achieve on my first try. Sadly those builds are stalled because I need a router bit/bearing combo to fix up the face plates before gluing them on, and my Monday order still hasn't shipped yet.    Tomorrow I'll hopefully feel up to routing all the woofer box braces.

 

During that build I realised that a floorstander would actually work better and this meant I had space for a 3rd driver, I modelled some 2.5 ways (didn't work well) and then started on some 3 ways (seemed more promising). I was going to leave it at that until the opportunity arose to part ex my 3 TD10H for 7 TD12M so I'm not going to use a TD12M as a woofer and another one as a mid in each LCR. It is not absolutely perfect in a model as a woofer but the sensitivity and the floor loading mean that it should still work nicely, at least all my models indicate it will deliver sufficient output down to ~70-80Hz without any issues at all. So that's the path I'm on at the moment, I have knocked up a model which crosses at ~350Hz and ~1200Hz  but need to get some directivity dataon the TD12 before I go further.

 

That sounds awesome.  I did consider options for a 3-way or 2.5 way but opted to stick with one TD12M.  I'm going active but don't want to have to buy 3 amps per speaker.    I would have rather put the active crossover between mid and tweeter, but I couldn't be sold on trying to implement a satisfactory passive frequency crossover.  Then I figured out that going ported would let me achieve my design goal with a single TD12M.  Indeed, this thing looks like these things should kick some serious butt in the 70-150 Hz range while keeping the mid range clean through controlled excursion.  Are you going to go to with vented cabinets?

 

About TD12 directivity, have you seen this?

 

https://sites.google.com/site/drivervault/driver-measurements/tang-band-75-1558se/ae-speakers-td12m/frequency-response

 

I'm considering active as my current amp is a bit noisy with the higher sensitivity drivers so I need new amps anyway & I've learnt that crossover components get quite expensive quite quickly. The capture card problem makes me doubtful though so we'll see what happens.

 

Interesting.  I'm hoping my Emotiva amps and Motu are quiet enough to not be a nuisance.  I can see dropping an 8-16 ohm resistor in series with the compression driver to lose some sensitivity and some noise, if need be.  We'll see how it works out when I get there.

[3ll3d00d]

I see what you mean about the MV and the WCS clipping being confusing. I never thought about the volume control being IC based like that. Makes it more feasible than I thought at first. It's pretty asinine that it's that hard to talk to someone in the company for a technical question like that and when you do get a hold of someone, they make a guess up. I just did some dumpster diving this week and scored a Marantz 6000 series. Probably something wrong with it so when I get into the teardown, I'll see what I can figure out as far as the volume and dacs. I'll post if I find anything interesting.

The only explanation I can think of is that they have a basic gain structure problem, ie the dsp has plenty of headroom to sum the input channels into the SW output without clipping but this peak output is larger than the input of the DAC (or electronic volume control) is designed (or configured) to take hence it is clipping on the input to either of those chips.

 

This explanation does fit the facts but also points to basic incompetence of the engineers/qa in that organisation if they can send a device out in that state. One key point of an integrated solution is that all that work is done for you after all so it should just work. Apparently not though.

[3ll3d00d]

 

That sounds awesome.  I did consider options for a 3-way or 2.5 way but opted to stick with one TD12M.  I'm going active but don't want to have to buy 3 amps per speaker.    I would have rather put the active crossover between mid and tweeter, but I couldn't be sold on trying to implement a satisfactory passive frequency crossover.  Then I figured out that going ported would let me achieve my design goal with a single TD12M.  Indeed, this thing looks like these things should kick some serious butt in the 70-150 Hz range while keeping the mid range clean through controlled excursion.  Are you going to go to with vented cabinets?

 

I am aiming for sealed. I've have some measurements that have tried to confirm the effects of the floor loading and I think it is loading it enough to make that work. We'll see though.

 

3 way active certainly does mean a load of amps are required. You could apply a relatively simple crossover and then use EQ to shape the signal though, a hybrid passive/active setup basically.

 

 

About TD12 directivity, have you seen this?

 

https://sites.google.com/site/drivervault/driver-measurements/tang-band-75-1558se/ae-speakers-td12m/frequency-response

 

 

I have yes, I wanted more detailed data though and data that was comparable with my data for my seos. I got round to that this morning which you can find at http://www.avsforum.com/forum/155-diy-speakers-subs/2188265-attempting-3way-seos10.html#post39684234 

[SME]

The only explanation I can think of is that they have a basic gain structure problem, ie the dsp has plenty of headroom to sum the input channels into the SW output without clipping but this peak output is larger than the input of the DAC (or electronic volume control) is designed (or configured) to take hence it is clipping on the input to either of those chips.

 

This explanation does fit the facts but also points to basic incompetence of the engineers/qa in that organisation if they can send a device out in that state. One key point of an integrated solution is that all that work is done for you after all so it should just work. Apparently not though.

 

In a good implementation you almost always provide a lot more "internal headroom" in a good DSP implementation than is available in the analog output stage.  This allows the signal to temporarily exceed "full scale" between multiple filters in a series without clipping if the action of a later filter reduces that the higher-than-full-scale level produced by the earlier filter.  This is very desirable when cascading multiple filters.  It is *not* a gain structure error because in most cases, the noise introduced in the digital domain is negligible.

 

Note also that the clipping may still happen digitally or in the DAC if the DAC output and MV control are identically matched.  What is most likely to happen is that the DSP must convert from its internal data format to 24-bit (or 32-bit, depending on the DAC) integers that the DAC can ingest, and that > "full scale" signals are clipped in this stage or limited in the stage immediately prior.

 

Does this make sense?

[3ll3d00d]

In a good implementation you almost always provide a lot more "internal headroom" in a good DSP implementation than is available in the analog output stage. This allows the signal to temporarily exceed "full scale" between multiple filters in a series without clipping if the action of a later filter reduces that the higher-than-full-scale level produced by the earlier filter. This is very desirable when cascading multiple filters. It is *not* a gain structure error because in most cases, the noise introduced in the digital domain is negligible.

 

Note also that the clipping may still happen digitally or in the DAC if the DAC output and MV control are identically matched. What is most likely to happen is that the DSP must convert from its internal data format to 24-bit (or 32-bit, depending on the DAC) integers that the DAC can ingest, and that > "full scale" signals are clipped in this stage or limited in the stage immediately prior.

 

Does this make sense?

Yes because this is what I was talking about, obviously not very clearly though I was using gain structure in the sense of mismatched signal ranges (min/max) as you pass from one stage in the signal chain to the next. I wasn't talking about exceeding a boundary within a single stage (the DSP).

[SME]

I am aiming for sealed. I've have some measurements that have tried to confirm the effects of the floor loading and I think it is loading it enough to make that work. We'll see though.

The sealed design is a lot more compelling when you have a dedicated woofer for mid-range duty.  I don't know what your output goals are, but you'll probably do just fine even without the floor loading.  Your DSP can re-shape the roll-off it into an in-room 4th order Linkwitz at 80 Hz if you so desire, and you'll probably still be able to do reference level without issue.  I was going to go sealed at first, but when I considered the fact that the speakers are meant to be limited bandwidth and that the port can reduce excursion a lot at the bottom end of the range, it was kind of a no-brainer.  In a sense, I'm doing ported to reduce excursion and improve sound quality rather than to play lower or louder.

 

I have yes, I wanted more detailed data though and data that was comparable with my data for my seos. I got round to that this morning which you can find at http://www.avsforum.com/forum/155-diy-speakers-subs/2188265-attempting-3way-seos10.html#post39684234 

 

Very interesting thread!  I need to take more time to read it.  Your latest posts are interesting as is the measurement data.  I made a post there under the alias "awediophile".

 

I am interested in the outcome of your build.  I actually chose to build with a horn larger than woofer *on purpose*.  My rough modelling suggested I would get a smoother directivity transition for two major reasons.  First, the directivty of the SEOS horns drop off much faster in the vertical dimension than they do in the horizontal dimension.  Second, directivity increases with the size of the radiating surface.  Two nearly-adjacent drivers in the crossover region will usually have higher directivity than either driver alone.  The relative crossover slopes also have a big impact on the directivity.  Overall in my calculations, I was surprised to see that the directivity effects of combining two drivers were significant a long way from even a 4th-order LR crossover point.  At the same time, there is plenty of opportunity to optimize crossover response and directivity both when using active crossovers.  The particular slopes used become less important, and the particular *shapes* used become more interesting.

 

That's not to say that I know what I'm doing.  Unfortunately, my models are very rough and mostly ad-hoc.  I started in on some quick code to more accurately model multiple drivers in 2-D polar far-field space.  It got "not so quick" when I realized I didn't know a good way to interpolate the horizontal and vertical polar data posted publicly into that 2-D space.  In fact, I'm not all that comfortable going through the trouble of doing that interpolation if it fails to accurately model reality.  Instead, I will have to measure 2-D space polars, and that will prove to be a major pain.  I think I might try to do it anyway.

[3ll3d00d]

The sealed design is a lot more compelling when you have a dedicated woofer for mid-range duty.  I don't know what your output goals are, but you'll probably do just fine even without the floor loading.  Your DSP can re-shape the roll-off it into an in-room 4th order Linkwitz at 80 Hz if you so desire, and you'll probably still be able to do reference level without issue.  I was going to go sealed at first, but when I considered the fact that the speakers are meant to be limited bandwidth and that the port can reduce excursion a lot at the bottom end of the range, it was kind of a no-brainer.  In a sense, I'm doing ported to reduce excursion and improve sound quality rather than to play lower or louder.

that's the reason I decided to go for a 3 way tbh. I have space constraints so it's either a bigger 2 way (hence could go ported and have a bigger wg) or a smaller wg but separate mid/woofer. I am gambling the latter will give the better result (and also I fancied trying to design a 3 way xo...)

 

 

 

Very interesting thread!  I need to take more time to read it.  Your latest posts are interesting as is the measurement data.  I made a post there under the alias "awediophile".

 

I am interested in the outcome of your build.  I actually chose to build with a horn larger than woofer *on purpose*.  My rough modelling suggested I would get a smoother directivity transition for two major reasons.  First, the directivty of the SEOS horns drop off much faster in the vertical dimension than they do in the horizontal dimension.  Second, directivity increases with the size of the radiating surface.  Two nearly-adjacent drivers in the crossover region will usually have higher directivity than either driver alone.  The relative crossover slopes also have a big impact on the directivity.  Overall in my calculations, I was surprised to see that the directivity effects of combining two drivers were significant a long way from even a 4th-order LR crossover point.  At the same time, there is plenty of opportunity to optimize crossover response and directivity both when using active crossovers.  The particular slopes used become less important, and the particular *shapes* used become more interesting.

 

That's not to say that I know what I'm doing.  Unfortunately, my models are very rough and mostly ad-hoc.  I started in on some quick code to more accurately model multiple drivers in 2-D polar far-field space.  It got "not so quick" when I realized I didn't know a good way to interpolate the horizontal and vertical polar data posted publicly into that 2-D space.  In fact, I'm not all that comfortable going through the trouble of doing that interpolation if it fails to accurately model reality.  Instead, I will have to measure 2-D space polars, and that will prove to be a major pain.  I think I might try to do it anyway.

 

AIUI there isn't a generally applicable model (e.g. like there is for a piston or baffle diffraction) for calculating such as it completely depends on the waveguide and the way in which the driver mates to that waveguide/horn. I am talking in terms of available speaker design software here not mathematical models of course. As such real world data is the way forward. I find I can do the measurements for polars pretty quickly these days. The lightening bolt was to stop faffing around with trying to build a rig for measuring precise angles and just use my phone in compass/gps mode (stuck to a particular point on my stand) instead.

[SME]

AIUI there isn't a generally applicable model (e.g. like there is for a piston or baffle diffraction) for calculating such as it completely depends on the waveguide and the way in which the driver mates to that waveguide/horn. I am talking in terms of available speaker design software here not mathematical models of course. As such real world data is the way forward. I find I can do the measurements for polars pretty quickly these days. The lightening bolt was to stop faffing around with trying to build a rig for measuring precise angles and just use my phone in compass/gps mode (stuck to a particular point on my stand) instead.

 

You are right.  I wasn't trying to model the horn but model the combination of horn and woofer using measurement data for the horn and either measurement data or a piston model for the woofer.  In the far-field approximation, each may be treated as point source with a certain 2D polar radiation pattern, and the response at any location in the room can be estimated by summing with appropriate delays, depending on distances.

[BeastAudio]

Perhaps a new thread would be good for this information? Ah, what do I care, the g2g is over and I think everyone here has read through the thread, so.... gung-ho mates! 

[3ll3d00d]

actually I did think to myself "I really am a million miles OT at this point" before my last reply , clearly your hospitality extends to your threads too 

[SME]

I'd be happy to create another thread.  Actually, I really should create a build thread for the work I'm doing.  Do you think people would be offended if I did a build-thread for full-range speakers here at data-bass?  Otherwise, I'll have to find another forum.  Maybe AVS is a better place for that sort of thing?  I just always find AVS to be irritating, what with all the ads and whatnot, plus the fact that the mobile website doesn't work at all, a lot of the time.

[maxmercy]

I'll follow a thread done here.  Full range includes bass.

 

JSS

[Ricci]

Infractions will be handed out for any further mentions of the words: full, range, fullrange,tweeter, midrange, Tom Brady, Patriots or Prius.

 

Posts will be removed that do not pertain to bass, beer, women, sports, bare knuckle boxing, moustaches or vintage pleated slacks.

 

Thanks

[mlah384]

LOL

[BeastAudio]

full, range, fullrange,tweeter, midrange, Tom Brady, Patriots or Prius. There. Ban Hammer

[Ricci]

The Panthers might be undefeated this year but they still suck.

[BeastAudio]

The Panthers might be undefeated this year but they still suck.

 

I have reported this post to the moderators.