Hi Steven - thanks for the PM. Not sure that my own particular views are going to fit in here as I'm no longer a fan of the traditional ways of building an active speaker....
Let me explain more - from all the commercial active speakers I've bought they've been limited in the sound quality mostly by the electronics being constrained to fit inside the cabinet. In my view its way too much of a compromise to go that way, so now I'm starting to build active speakers where only the drive units sit in the cabinet itself and all the active stuff sits in an 'activator' box connected by an umbilical.
My long term aim is to do enough design work to make the 'activator' a generic box to as large a degree as possible, so it can interface with as many speakers as possible. I think eventually the crossover and frequency response characteristics can just be software configuration.
Let me explain more - from all the commercial active speakers I've bought they've been limited in the sound quality mostly by the electronics being constrained to fit inside the cabinet. In my view its way too much of a compromise to go that way, so now I'm starting to build active speakers where only the drive units sit in the cabinet itself and all the active stuff sits in an 'activator' box connected by an umbilical.
My long term aim is to do enough design work to make the 'activator' a generic box to as large a degree as possible, so it can interface with as many speakers as possible. I think eventually the crossover and frequency response characteristics can just be software configuration.
Richard bring your ideas. They are more than welcome here. Perhaps what you are doing can be reduced in size with many surface mount devices and design? I can still consider your idea of an external system but I would still like to put it inside if it can be done well.
Yes - certainly the semiconductor parts of the electronics can be made suitably miniscule. But power supplies for good sound tend to be rather bulky and its the power supply requirements which have led me to 'externalize' the electronics.
My focus on power supplies has led to some rather radical notions. For example since capacitors are the foundation of power supplies and they store more energy per unit volume (and per unit cost) as their working voltage goes up, the logical route is to run amplifiers at as high voltages as possible and step down their output voltage with a transformer. This is fine for the midrange frequencies and up, and at lower powers for LF, but trafos can't handle low frequencies very effectively at higher powers. ClassD amps then for bass duty make the most sense.
My focus on power supplies has led to some rather radical notions. For example since capacitors are the foundation of power supplies and they store more energy per unit volume (and per unit cost) as their working voltage goes up, the logical route is to run amplifiers at as high voltages as possible and step down their output voltage with a transformer. This is fine for the midrange frequencies and up, and at lower powers for LF, but trafos can't handle low frequencies very effectively at higher powers. ClassD amps then for bass duty make the most sense.
I was recently looking at some in-line smps that had up in the range of 1000 watt output, something I didn't see the last time I looked quit a few years ago. Perhaps this is a viable option to power up these circuits.
Yes it seems to me that SMPSUs should be used as liberally as possible. With the high voltage requirement of my approach I reckon only the front-end of a switcher will be called for - the PFC (power factor correction) circuit. This delivers a DC rail close to 400V. Isolation from mains will be provided by the output transformers, not the input (power) transformer.
Yes - I have an article about OPTs (output transformers) over on LinkedIn. They're called for by the desire to optimize the energy storage of the power supply - but it turned out they improved the SQ of the amp (a simple chipamp) when I first tried them. To my knowledge no-one else is using OPTs between amps and the drive units of active speakers - OPTs tend to be only used with valve amps and valve amps haven't been used in the active setups I've so far seen.
I knew you would be thinking outside the box. I'll take a look and read what you posted on your blog.
While we are talking about dsp and dsc's with this type of integration do we need more than one dac for all these functions or is one going to be able to handle all the different input types we are talking about. If we use the dsp for the xo and FR equalization and any phase corrections what else can we use the dsp for? Can we develop a basic block diagram and then fill in the details as we go?
'More than one dac' - are you referring to boxes or chips? DAC chips most certainly are required at a rate of one per (stereo) channel - so two or three will be needed to handle a 2 or 3-way setup. I imagine the inputs to the 'activator' I'm proposing will be digital ones - S/PDIF or AES/EBU.
DSPs can be used for additional features beyond the basic FR, phase and crossover functions - for example room correction and dynamics control plus internal housekeeping duties like amplifier bias control or even power supply noise cancellation.
DSPs can be used for additional features beyond the basic FR, phase and crossover functions - for example room correction and dynamics control plus internal housekeeping duties like amplifier bias control or even power supply noise cancellation.
Modular options
In the Pro monitor market they use a modular system to solve this issue:
(1) You can buy the passive speaker with no crossover or amplifiers and use your own.
(2) At any point you can "bolt on" the electronics pack to the back of the passive speaker and use the manufacturers amps.
So the electronics are "external" and connected via an umbilical cord.
I add in sorbothane pads to reduce vibration transmission.
Also making the electronics casework from stainless steel lined with ESR cloth ( or similar) will satisfy even the most "difficult" of customers!
The modular approach can be expanded further to include DSP /DAC's / A to D/ WiSA and mix n match class D with lower powered class A / B for mid / top.
One "Six Pack" we are we are working on is an A to D / WiSA transmission board in a "send box" and the matching receive box containing the WiSA receiver / DSP / DAC / power amplifiers ..... 6 different elements combined to great effect.
This means we can upgrade existing loudspeakers (passive or active) giving far superior results plus the freedom from no loudspeaker cables.
The customer can email in a "raw driver " measurement file and we can pre set crossover / Eq data, or the customer can measure in situ and the DSP will set everything automatically.....
The DSP firmware can be updated via USB.
Lots of flexibility and good stuff just around the corner....!
Hope this helps
Cheers
Derek.
In the Pro monitor market they use a modular system to solve this issue:
(1) You can buy the passive speaker with no crossover or amplifiers and use your own.
(2) At any point you can "bolt on" the electronics pack to the back of the passive speaker and use the manufacturers amps.
So the electronics are "external" and connected via an umbilical cord.
I add in sorbothane pads to reduce vibration transmission.
Also making the electronics casework from stainless steel lined with ESR cloth ( or similar) will satisfy even the most "difficult" of customers!
The modular approach can be expanded further to include DSP /DAC's / A to D/ WiSA and mix n match class D with lower powered class A / B for mid / top.
One "Six Pack" we are we are working on is an A to D / WiSA transmission board in a "send box" and the matching receive box containing the WiSA receiver / DSP / DAC / power amplifiers ..... 6 different elements combined to great effect.
This means we can upgrade existing loudspeakers (passive or active) giving far superior results plus the freedom from no loudspeaker cables.
The customer can email in a "raw driver " measurement file and we can pre set crossover / Eq data, or the customer can measure in situ and the DSP will set everything automatically.....
The DSP firmware can be updated via USB.
Lots of flexibility and good stuff just around the corner....!
Hope this helps
Cheers
Derek.
I too agree.
The amplifiers must be capable of driving the speakers.
In cabinet amplifiers would need to be small and send their heat outside.
Separate amplifiers, maybe in a single box per active speaker with an integrated PSU, that are capable of meeting the lesser current demand of actively driven drive units would then be near the active speaker. Two, three or maybe 4, twisted pairs connect the amplifiers to the active speaker box.
Only when the drive units don't need high power can the PSU+Amplifiers be small enough to hang on the back of the active speaker box. But there is a small downside to this, namely vibration from the drivers affecting amplifier performance. This is what I adopted for desktop computer speakers covering the range 140Hz to ~20kHz with a +20dB amplifier gain, since they are effectively heavy near-field monitors.
*Derek, just read your post above and it looks like a lot of what I just typed below was part of your prior art. Apologies - but will leave them below as a "Thumbs Up". Bravo.
Am I reading this right - the idea of discrete dsp's in the unit? I wonder if they could even be made to "talk" to each other if you were using wireless technology anyway - master clock and all.
Perhaps with the right implementation, a setup mode could be used to have them calculate the distance/time between them to setup some sort of check bit system in the clocking that is only intermittently confirmed through a signal in the "noise" spectrum.
In a non-ideal placement what of a reinforced "mono" mode instead of stereo. The two channels could be combined, with one monitor primary, the second one fill in.
Ooh I love a good rotary encoder...
Am I reading this right - the idea of discrete dsp's in the unit? I wonder if they could even be made to "talk" to each other if you were using wireless technology anyway - master clock and all.
Perhaps with the right implementation, a setup mode could be used to have them calculate the distance/time between them to setup some sort of check bit system in the clocking that is only intermittently confirmed through a signal in the "noise" spectrum.
In a non-ideal placement what of a reinforced "mono" mode instead of stereo. The two channels could be combined, with one monitor primary, the second one fill in.
Ooh I love a good rotary encoder...
Last edited:
Thanks guys I will consider your thoughts here. This is all welcome, the more information sharing the better.
Overkill,
I take it from your reply that you are in the business on the pro-audio side of things. I have worked on that side myself but it has been a long time. I do understand that the real pros want to do these things themselves and mix and match components for their own needs and personal taste. I worked on the PA side of things but basically stayed out of the studio side of things as far as speakers go.
Overkill,
I take it from your reply that you are in the business on the pro-audio side of things. I have worked on that side myself but it has been a long time. I do understand that the real pros want to do these things themselves and mix and match components for their own needs and personal taste. I worked on the PA side of things but basically stayed out of the studio side of things as far as speakers go.
The number of diverse ideas getting kicked around in this thread makes it clear that there will be more than one block diagram, or else you will need a diagram with lots of alternative paths.
The active speaker input can be analog audio, SPDIF, open-standard Ethernet (either wired or wireless), Bluetooth, or some proprietary networking standard (e.g. Sonos). So the input can be analog, real-time digital or packetized digital. Once the audio data is received, it can follow an all-digital path through the DSP to class D amps, or it can get converted to analog or digital in several different places depending on the type of DSP and class of amps. When you start to look at all of the possible input, mixing, processing and amplification paths, the problem gets very complex.
I think you are going to have to focus on one specific implementation or at least set some boundary rules to keep this thread focused.
One of my own interests has been for low-cost active speakers for which the DSP is done in class D amplifier chips, using embedded biquads. This link is to a briefing we had at DCDIY2009. It's a bit dated now, but it's got a couple of block diagrams that you might be able to use as a starting point.
More recently, I've been using SPDIF as the input for active speakers. The SPDIF standard allows sending messaging data through the User Data bits in the SPDIF frame, and I've got a simple protocol for sending volume, amplifier control information, filter specs, etc. to the active speakers using the same cable as the digital audio. That means I don't have to have a wireless control channel for sending this data, and I don't have to include a control panel with a display on each speaker. That 20-channel amp shown in Post #2 uses this approach--the User Data is injected into the SPDIF stream on the preamp side with an SRC4382 chip, and extracted in the amp by the SRC chip.
Another complicating issue for this thread is software. You will probably need microcontroller code for programming DSP chips, and probably PC code to off-load the microcontroller for calculating filter coefficients. The software is an essential part of active speaker design, and there are many ways to partition and allocate the software. So once again, I think you will have to focus on a specific implementation or else keep this discussion very generic.
My own opinion is that there are many, many topics that can be explored in an active loudspeaker forum. I'd like to see these discussions take place here at diyAudio because of the high level of DIY expertise that this board attracts. But I'm not seeing a good way to address this huge problem space with a single thread. I'll leave it at that--I don't want to be overly annoying.
Last edited:
Neil,
Thanks for the comments. I downloaded your link but need to read it now. Your comment about the control of volume and other aspects are well taken and came up in a conversation I was having that wasn't on the forum. It was in relationship to the fact someone may want to use a memory stick with their music on it and how you can control that and even pick a song out of a list. Then the question about a screen of some sort to do that, using an external device to do that seems the better approach to that rather than an all in one type of solution. Boy it is so much easier when all I have to do is design the speaker side of things!
Thanks for the comments. I downloaded your link but need to read it now. Your comment about the control of volume and other aspects are well taken and came up in a conversation I was having that wasn't on the forum. It was in relationship to the fact someone may want to use a memory stick with their music on it and how you can control that and even pick a song out of a list. Then the question about a screen of some sort to do that, using an external device to do that seems the better approach to that rather than an all in one type of solution. Boy it is so much easier when all I have to do is design the speaker side of things!
Here is (just) an example of integrated system of a commercial crossover (DSP) and power amp on the same pcb from the Aurelia Aniara DSP active 2-way:
https://www.facebook.com/2910018309...1001830930984/696337817064048/?type=1&theater
The Class-D amp chip is from ST and the DSP from VLSI who usually makes MP3 SoC's and room correction devices (DSPeaker brand), it's this chip:
VLSI Solution-VS1005 – All-In-One MP3 Audio System-on-a-Chip
No Bluetooth but there seems to be an ethernet interface mentioned. If you just could also power the speaker using ethernet it would be fine to use the ethernet (well, synchronisation between the stereo pair could require some hard to implement protocol).
https://www.facebook.com/2910018309...1001830930984/696337817064048/?type=1&theater
The Class-D amp chip is from ST and the DSP from VLSI who usually makes MP3 SoC's and room correction devices (DSPeaker brand), it's this chip:
VLSI Solution-VS1005 – All-In-One MP3 Audio System-on-a-Chip
No Bluetooth but there seems to be an ethernet interface mentioned. If you just could also power the speaker using ethernet it would be fine to use the ethernet (well, synchronisation between the stereo pair could require some hard to implement protocol).
I want everyone to know that though I started this thread I really do think it should be an open forum about different ways to implement a powered loudspeaker system. I love the personal answers here and will keep that up but I am sure others will have different ideas that may be applicable or not. How we keep it all straight and make is easy to find the information in the end will be something I will have to think about.
In that spirit, please have a look at the contraption I am working on in preparation to start marketing a novel loudspeaker design.
In all, it contains 14 LM3886 chip amps. Four are bridged and paralleled for the bass driver, two are bridged for the mids, and one will serve the tweeter (per side). The analog crossover is mounted against the rear panel and will be shielded.
Power supply consists of 2 MeanWell 360 Watt supplies. This is what they can deliver continuously, combined peak power is well over 1000 Watts. A linear power supply would become unwieldy. It would take over 10 Kg to get the same kind of power from transformers.
The problem with these SMPS´s is that they have pretty loud fans that switch on when their temperature gets above 40 degrees C. To solve it I ordered some very nice low noise PC case fans, which will be running all the time at variable speed. They will suck in air through the bottom panel, and exhaust through holes drilled into the heat sinks.
Besides the analog xover, it is pretty much IKEA screw it together kind of stuff, intellectually speaking, except for, of course, the grounding scheme. That gets really intricate in a setup like this.
And the other problem is cables and connectors. Speakon has a banana sized 8 way plug, and VanDamme has very nice supple 8 core matching cable, as thick as a fat mans thumb. Great for pro stuff, but widely over dimensioned for home stuff. So I lowered myself to making my own cables out of servo extension wire and found some appropriate multi-way plugs at RS.
In all, it contains 14 LM3886 chip amps. Four are bridged and paralleled for the bass driver, two are bridged for the mids, and one will serve the tweeter (per side). The analog crossover is mounted against the rear panel and will be shielded.
Power supply consists of 2 MeanWell 360 Watt supplies. This is what they can deliver continuously, combined peak power is well over 1000 Watts. A linear power supply would become unwieldy. It would take over 10 Kg to get the same kind of power from transformers.
The problem with these SMPS´s is that they have pretty loud fans that switch on when their temperature gets above 40 degrees C. To solve it I ordered some very nice low noise PC case fans, which will be running all the time at variable speed. They will suck in air through the bottom panel, and exhaust through holes drilled into the heat sinks.
Besides the analog xover, it is pretty much IKEA screw it together kind of stuff, intellectually speaking, except for, of course, the grounding scheme. That gets really intricate in a setup like this.
And the other problem is cables and connectors. Speakon has a banana sized 8 way plug, and VanDamme has very nice supple 8 core matching cable, as thick as a fat mans thumb. Great for pro stuff, but widely over dimensioned for home stuff. So I lowered myself to making my own cables out of servo extension wire and found some appropriate multi-way plugs at RS.
Last edited:
Wow Paul, can't believe you drilled all those holes in the heatsinks. Still, I think it's a good way to exhaust the heated air - the sides of the amp will be unobstructed in normal use anyway, and you can still place a pot of (tropical) flowers on top 🙂
BTW The 'friet-kot' owner doesn't want to rent the place out....
Jan
BTW The 'friet-kot' owner doesn't want to rent the place out....
Jan
- Status
- Not open for further replies.