Hey guys,
First off, let me say what a wonderful resource this is. I have been interested in DIY audio for a long time, but only recently have I discovered the wealth of resources available on the web.
I have an idea, and I would like to throw it out and see how feasible it is.
First, I like the Pass ideal of having as few components in the signal path as possible.
I have also noticed that most speakers are at least 2 way systems, if not 3 way or greater. In general, most high fi systems use some form of EQ, in the form of crossovers and correction, etc.
In the case of a digitial source, why not move this part of signal path to the digital domain?
Another thing i have noticed is that it seems most setups perform all amplification and then split up the signal to the individual drivers afterwards. It seems it might be a good idea to split this up digitially first, then send ieach component to an amplifier, and then on to its driver. This would ease the problem of Pass like amps which use few gain stages and have very low efficiency. Also, this would be analogous in a way to moving the idea of bi-amping further up the signal path. If speaker wire can benefit from carrying a smaller portion of the musical spectrum, would it be reasonable to think an aimplifier would as well?
So, the idea can be summarised like so:
A computer serves as:
- the source,
- the crossovers,
- the eq correction required for each resulting portion of the signal path.
Each digital signal is sent to a DAC, then to a gain stage, and then to a driver.
There are a few problems:
- This requires a large amount of equipment. A DAC and gain stage for each driver of each speaker.
- This requires coming up with a good way to get multiple digital signals coming out of a PC in a format which a DAC can handle.
However, it does not seem unreasonable to think that the entire thing can be accomplished via DIY means.
Speakers and gain stages most certainly can be produced of very high quality via DIY means.
DIY DACs may be less common, but are certainly possible.
While I have not found a readily available open source EQ software package which would suit the needs of this idea, I am a computer science student and am sure it can be done given the proper pool of resources.
The main hindrance I see left is getting the digital signal from the computer to the DACs.
If we consider DIY DACs only, we have the freedom to modify their input interface to make this problem easier, however it might be desireable to keep the interface to the DAC compatible with existing commercial DACs. I think this would mean SPDIF?
The idea I have been kicking around is making some sort of hardware unit which will accept an Ethernet connection on one end, and output a digital SPDIF signal on the other end.
Drivers could be written to take SPDIF blocks of information and wrap them in TCP/IP packets, which would make it very easy and cheap to get the digitial signal out of the computer to an arbitrary number of external devices.
The big question I have is how realistic would it be to design a peice of hardware which is capable of accpeting some sort of Ethernet based protocol on one end which is sending packets of SPDIF information, strip off the ethernet protocol information internally, and send out the contained bits of SPDIF information to the DAC?
anyway, just somethig to think about. any and all feadback is welcome.
By the way, which forum should this discussion go under - it seems to overlap more than one topic.
-jason
jason@pepas.com
First off, let me say what a wonderful resource this is. I have been interested in DIY audio for a long time, but only recently have I discovered the wealth of resources available on the web.
I have an idea, and I would like to throw it out and see how feasible it is.
First, I like the Pass ideal of having as few components in the signal path as possible.
I have also noticed that most speakers are at least 2 way systems, if not 3 way or greater. In general, most high fi systems use some form of EQ, in the form of crossovers and correction, etc.
In the case of a digitial source, why not move this part of signal path to the digital domain?
Another thing i have noticed is that it seems most setups perform all amplification and then split up the signal to the individual drivers afterwards. It seems it might be a good idea to split this up digitially first, then send ieach component to an amplifier, and then on to its driver. This would ease the problem of Pass like amps which use few gain stages and have very low efficiency. Also, this would be analogous in a way to moving the idea of bi-amping further up the signal path. If speaker wire can benefit from carrying a smaller portion of the musical spectrum, would it be reasonable to think an aimplifier would as well?
So, the idea can be summarised like so:
A computer serves as:
- the source,
- the crossovers,
- the eq correction required for each resulting portion of the signal path.
Each digital signal is sent to a DAC, then to a gain stage, and then to a driver.
There are a few problems:
- This requires a large amount of equipment. A DAC and gain stage for each driver of each speaker.
- This requires coming up with a good way to get multiple digital signals coming out of a PC in a format which a DAC can handle.
However, it does not seem unreasonable to think that the entire thing can be accomplished via DIY means.
Speakers and gain stages most certainly can be produced of very high quality via DIY means.
DIY DACs may be less common, but are certainly possible.
While I have not found a readily available open source EQ software package which would suit the needs of this idea, I am a computer science student and am sure it can be done given the proper pool of resources.
The main hindrance I see left is getting the digital signal from the computer to the DACs.
If we consider DIY DACs only, we have the freedom to modify their input interface to make this problem easier, however it might be desireable to keep the interface to the DAC compatible with existing commercial DACs. I think this would mean SPDIF?
The idea I have been kicking around is making some sort of hardware unit which will accept an Ethernet connection on one end, and output a digital SPDIF signal on the other end.
Drivers could be written to take SPDIF blocks of information and wrap them in TCP/IP packets, which would make it very easy and cheap to get the digitial signal out of the computer to an arbitrary number of external devices.
The big question I have is how realistic would it be to design a peice of hardware which is capable of accpeting some sort of Ethernet based protocol on one end which is sending packets of SPDIF information, strip off the ethernet protocol information internally, and send out the contained bits of SPDIF information to the DAC?
anyway, just somethig to think about. any and all feadback is welcome.
By the way, which forum should this discussion go under - it seems to overlap more than one topic.
-jason
jason@pepas.com
Hello!
I had a similar idea, but buidling the digital circuitry into an amplifier (no computer involved) - resulting in an amplifier which will only work with SPDIF (or AES/EBU, Dolby AC-3, what-have-you) digital inputs. One could add an A/D on the front for analogue signals, but that really defeats the purpose...
So, with the digital input, you'dd have resampling and/or dithering as neccessary to get it into the format used in the DSP (say 24bit/96KHz), which performs the funtion of an electronic cross-over. The DSP outputs then are fed to D/A converters for each amplifier (2 or 3-way) and so on...
I thought about this to build a bi-amped arrangement that uses a muscle-head solid-state amp for the low frequencies and a quality tube amp for the mid and high's. If it's done right, I'm sure the results would be terrific, since the tube circuits are usually limited in LF performance by the ouptut transformer. An OPT for 500Hz and above would be much smaller too, and could have a bandwidth up to 100KHz if properly designed.
Overall, you could end up with a record/playback system involving only a single pair of A/D D/A converters, and amps matched specifically to speakers, with 100% accurate cross-overs that introduce no phase shifts or other garbage... I like!!
I had a similar idea, but buidling the digital circuitry into an amplifier (no computer involved) - resulting in an amplifier which will only work with SPDIF (or AES/EBU, Dolby AC-3, what-have-you) digital inputs. One could add an A/D on the front for analogue signals, but that really defeats the purpose...
So, with the digital input, you'dd have resampling and/or dithering as neccessary to get it into the format used in the DSP (say 24bit/96KHz), which performs the funtion of an electronic cross-over. The DSP outputs then are fed to D/A converters for each amplifier (2 or 3-way) and so on...
I thought about this to build a bi-amped arrangement that uses a muscle-head solid-state amp for the low frequencies and a quality tube amp for the mid and high's. If it's done right, I'm sure the results would be terrific, since the tube circuits are usually limited in LF performance by the ouptut transformer. An OPT for 500Hz and above would be much smaller too, and could have a bandwidth up to 100KHz if properly designed.
Overall, you could end up with a record/playback system involving only a single pair of A/D D/A converters, and amps matched specifically to speakers, with 100% accurate cross-overs that introduce no phase shifts or other garbage... I like!!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.