Jan, you need your own place to drill holes in heat sinks too! Too bad the frietkot didn't go through, but you are welcome in my workshop at any time.
Vacuphile,
My very first conception was to use the 3886 chips in bridged parallel for the cone driver but there was so much push-back by many that the NI information on doing that was flawed and it just wasn't a practical solution. Believe me I looked hard at that option. I have chips sitting somewhere in storage and have all the NI notes on how to do it, I just couldn't find any support for that solution. I thought to use the chips that way in a push pull inverted mode around the speaker.
My very first conception was to use the 3886 chips in bridged parallel for the cone driver but there was so much push-back by many that the NI information on doing that was flawed and it just wasn't a practical solution. Believe me I looked hard at that option. I have chips sitting somewhere in storage and have all the NI notes on how to do it, I just couldn't find any support for that solution. I thought to use the chips that way in a push pull inverted mode around the speaker.
The reason I am using them bridged, except for the tweeter, is mainly to keep the currents into ground limited, and paralleled based on thermal considerations. Doesn't seem to cause any problem (and why would it?). How much power do you need?
(NI=TI?).
(NI=TI?).
Vacuphile,
I really don't need all that much power under normal listening conditions, the cone driver is about 86db per watt so they surely aren't HE drivers but I am after sound quality and not max output per watt. My thinking was extra power for overhead and also so I can electronically push up the bottom end without causing the amp to clip on low notes. I guess it really depends on what the maximum output anyone would want. My prototypes would get louder than most would ever consider reasonable in a normal sized room. I didn't measure that level but it was loud enough to get to the point you wouldn't want to listen at that level for long, no distortion, just to damned loud.
I really don't need all that much power under normal listening conditions, the cone driver is about 86db per watt so they surely aren't HE drivers but I am after sound quality and not max output per watt. My thinking was extra power for overhead and also so I can electronically push up the bottom end without causing the amp to clip on low notes. I guess it really depends on what the maximum output anyone would want. My prototypes would get louder than most would ever consider reasonable in a normal sized room. I didn't measure that level but it was loud enough to get to the point you wouldn't want to listen at that level for long, no distortion, just to damned loud.
The more I think about this, a computing platform is going to give you the most flexibility by being able to interface with samba shares from network mounted drives, flexibility to receive encoded audio via I2S and be flexible to have an established library to extend for interfaces with BT 4.0 and build in dsp processing in addition to any kind of sampling capability for tuning for the room. Then it is easy as "pi" to just plug the thing directly into an HDMI monitor for interfacing with via keyboard and mouse. Instead of hookup and download of plugin settings, just plug it in, make the changes you want and unplug it if you want... or not.
To my untrained ears bt 4.0 seems to work well enough, even 3.0 is fine.
It seems that the monitors will be where all the heavy lifting is needed. There are already media browsers like XBMC now, also known as KODI that can run a on a Pi B+ so the extensibility and polish of the shares is already there.
A "modern" loudspeaker with this configuration will seem foreign in some ways. Some people still look for the amp and speakers, some go for the soundbar, some have picked up the handheld boom boxes. In all, simplicity is what most of the 'kids these days' are going for... and a soundbar does that because they don't know any better. And, a soundbar might have a place... as a center channel anyway depending on any channel extensibility.
Given that major computing manufacturers are largely ignoring the media confluence to set up their own walled gardens of profit, this could be an opportunity for a return to the media hub. They thought a unified structure of server/backup/media storage would be good. But, fast forward a decade and Windows Home Server is largely dead. Samba UPnP shares are still ubiquitous, but will be mostly developed by the DIY'ers like Kodi for extensibility and function.
I think what MiSFiT and 4APL figured out was that despite their desire to create order from entropy, most people are just fine with their menagerie of disordered media tastes.
Ergo, the return of the all source signal distribution structure. The box you plug in, put your audio source in, tune the system and press play.
Hardware needs I see are accurate reproduction, flexibility, portability.
In many situations, with the wifi capability and a robust antenna moving the speaker outside for a gathering would be pretty simple. Just take the thing outside, plug it in, go back inside, adjust a preset and interact with your source as you like.
What most buyers get hung up on is if they are getting strung up on a software platform. Why not just avoid the problem and give them what they want. They want a box with a switch. And make them real looking, satisfyingly substantial knobs, even if they work like encoders. I have not yet found anyone who said "I just love working with this flat interface." "I love trying to imagine a menu system with no tactile haptic feedback. Even if the system isn't amazing to the 4th decimal place, if they feel good about it, it won't matter if it is only 3 or 2 dec. places amazing. Which if you are breaking ground, seems like a nice ver. 1.0 release candidate.
For power users, let them plug in the hdmi.
Sorry for the long post. Just sharing some thoughts even though I'm sure I typed them out in a way that doesn't sound like it, that's just me thinking out loud.
usul27 appears to have been doing this time, but it may be his own software/IP. website is hifiberry.com - Crazy Audio ? Do know him personally.
To my untrained ears bt 4.0 seems to work well enough, even 3.0 is fine.
It seems that the monitors will be where all the heavy lifting is needed. There are already media browsers like XBMC now, also known as KODI that can run a on a Pi B+ so the extensibility and polish of the shares is already there.
A "modern" loudspeaker with this configuration will seem foreign in some ways. Some people still look for the amp and speakers, some go for the soundbar, some have picked up the handheld boom boxes. In all, simplicity is what most of the 'kids these days' are going for... and a soundbar does that because they don't know any better. And, a soundbar might have a place... as a center channel anyway depending on any channel extensibility.
Given that major computing manufacturers are largely ignoring the media confluence to set up their own walled gardens of profit, this could be an opportunity for a return to the media hub. They thought a unified structure of server/backup/media storage would be good. But, fast forward a decade and Windows Home Server is largely dead. Samba UPnP shares are still ubiquitous, but will be mostly developed by the DIY'ers like Kodi for extensibility and function.
I think what MiSFiT and 4APL figured out was that despite their desire to create order from entropy, most people are just fine with their menagerie of disordered media tastes.
Ergo, the return of the all source signal distribution structure. The box you plug in, put your audio source in, tune the system and press play.
Hardware needs I see are accurate reproduction, flexibility, portability.
In many situations, with the wifi capability and a robust antenna moving the speaker outside for a gathering would be pretty simple. Just take the thing outside, plug it in, go back inside, adjust a preset and interact with your source as you like.
What most buyers get hung up on is if they are getting strung up on a software platform. Why not just avoid the problem and give them what they want. They want a box with a switch. And make them real looking, satisfyingly substantial knobs, even if they work like encoders. I have not yet found anyone who said "I just love working with this flat interface." "I love trying to imagine a menu system with no tactile haptic feedback. Even if the system isn't amazing to the 4th decimal place, if they feel good about it, it won't matter if it is only 3 or 2 dec. places amazing. Which if you are breaking ground, seems like a nice ver. 1.0 release candidate.
For power users, let them plug in the hdmi.
Sorry for the long post. Just sharing some thoughts even though I'm sure I typed them out in a way that doesn't sound like it, that's just me thinking out loud.
usul27 appears to have been doing this time, but it may be his own software/IP. website is hifiberry.com - Crazy Audio ? Do know him personally.
Last edited:
I'm liking this idea. A Pi will run MPD, Ecasound and such. LADSPA plugins are free, as is the software. OS is free, too. Shouldn't be that hard to put a GUI front end on "tuning" the settings for crossover, and most of what you need is right there. a 7.1 sound card will handle up to a stereo 4-way system....
You need 5.1 or 7.1? Add more sound cards, just slave the clocks together...
You need 5.1 or 7.1? Add more sound cards, just slave the clocks together...
Very interested in this topic/ subject, no matter what it's called or where it belongs.
To me it's all witchcraft at the moment.
Would love to build an active crossover with easily configured DSP/ room correction.
Down to basics such as minimum spec of PC for room correction, programming e.g Nadja, FreeDSP.
To me it's all witchcraft at the moment.
Would love to build an active crossover with easily configured DSP/ room correction.
Down to basics such as minimum spec of PC for room correction, programming e.g Nadja, FreeDSP.
How much power do you need? Bridged without paralleling is a highly practical option with output transformer and I think it'll do almost 200W (not that I need anything near that). Without the trafo the output load is too demanding. I'm currently building an amp with this configuration, but its just one of many projects on the go so progress is rather sedate.
How can you possibly get 200 watts out of a pair of 3886 chips in bridged mode? From what I remember the real output of one of those chips at a reasonable 1% distortion is only about 36-45 watts depending on the rail voltage. So for any audiophile type sound level you had to derate that even lower to get less than 1% distortion. Then you had to deal with the Spike protection circuit which seems to rear it ugly head.
I don't think they'll do 200WRMS on a continuous sinewave test - but with music they'll do nearly 400W peak power in bridged. I haven't tried it yet but I did sit down with the DS and examine the capabilities quite carefully, looking at the output voltage swing and peak current capability.
The trafo is what's helping the chips to get more power out - if you look at the graph of output saturation voltage versus supply, providing a lighter load (and running at higher supply) makes better use of the chip's capabilities. I seem to recall a load around 12ohms for the bridge was around optimum.
The trafo is what's helping the chips to get more power out - if you look at the graph of output saturation voltage versus supply, providing a lighter load (and running at higher supply) makes better use of the chip's capabilities. I seem to recall a load around 12ohms for the bridge was around optimum.
Would you get similar results from a speaker then with a minimum impedance of 12 ohms then or do the transformers bring something else to the party? And how large are these output transformers, something like you would see on a vacuum tube amplifier?
You'd get similar results with an optimized voice coil impedance yes, the trafos aren't adding more than impedance transformation. I'm currently using 15VA EI core trafos (roughly 50mm across) but I'm not aiming for 400W peaks - the determinant of trafo size seems to be what damping factor you're aiming for. Though of course if you need an amp to pass approvals tests running with a continuous sinewave the trafos will have to be bigger not to overheat.
Sorry - what do you mean by 'sink' ? The place where the power is dissipated? If so yes, but the voice coil is the sink anyway even with the trafo.
That would be an advantage of building the driver, I can change the impedance of the coil but I would imagine that a 12ohm could would start to get heavy if it has to handle the same current.
Dead right - yep I was worried about heat with the voice coil. So line matching with the trafo doesn't help that. 🙁 Humm... well with higher voltage, and a lighter load... may not present as much of a weight issue?
Richard would increasing the impedance not also cause much high current draw on the chip amp? Would it start operating in current mode?
A higher resistance voice coil is wound with a thinner wire. So there are more turns in the gap than with a lower resistance one. More turns in the gap means for the same shove, the current is lower. So no, a 12R coil needs 3X less current than a 4R coil so no overall loss by going to a higher resistance.
There is though the issue of the number of layers in the voice coil - too many layers and the inductance rises and losses go up at higher frequency.. In general lower impedance makes better use of the available resource (copper in the gap) because less of that total volume is wire insulation when the wire's thicker.
There is though the issue of the number of layers in the voice coil - too many layers and the inductance rises and losses go up at higher frequency.. In general lower impedance makes better use of the available resource (copper in the gap) because less of that total volume is wire insulation when the wire's thicker.
- Status
- Not open for further replies.