dorkus:
hm.. i figured the power supply might need some upgrading. I suppose if worse comes to worse, it's not a huge deal to build a new linear power supply for it, but one would probably have to tear apart the stock PSU to get the power switch and LED assemblies. Also, the stock PSU will likely have some power switching features and control features which would need to be duplicated in the replacement...
I suspect that the DC offset's you've seen on the outputs are a result of the cheapo opamps. This, if nothing else, is reason enough for me to upgrade the opamps. At least I can get rid of the awful 'lytics in the signal path (yuk!). Also, it may be an option to use a nice film coupling cap at the input to the opamp if there is still some DC offset from prior circuitry. I'll also have to scope the input to the muting transistors to see if the player uses them during operation... I'm kinda hoping they're used in power up and power down only, which would make their removal irrelevant to me, as I follow a strict power-up / power-down procedure anyway (my preamp is also DC coupled). Whatever happens, I think these are the three key mods that really need to be done for an immediate improvement, which should suffice until I'm using an external DAC.
Regarding output in PCM format, one of my main reasons is that I want to be able to perform signal processing on the data, which is virtually impossible with DSD streams. In fact, I learned something interesting today while reading some online articles... it seems DSD studio equipment actually employs a quasi-PCM format for editing(!)... it would seem that all the marketing hype about maintaining the purity of the data streams by avoiding PCM conversions is just propoganda! It is well known that in order to do any useful studio editing, some form of PCM conversion is almost essential, and of course it turns out that this is pretty much exactly what happens during the process of producing a master. There are only a very small percentage of recordings (maybe 5%) which can be produced without any conversion into PCM. Of course, Sony and Philips would never want the consumers to find out that the recordings they're buying have actually gone through DSD-PCM conversion, and then back again! Personally, I understand the mathematical underpinnings of the conversion processes quite well, and I'm not the least bit afraid of them. A single conversion into PCM isn't likely to do much if any audible damage to the music.
I don't want to get into a DVD-A vs SACD war, but for the purposes of evaluating whether or not it is a worthwhile proposal to convert SACD streams into 24/96 or 24/192 PCM data for processing or the use of an external DAC, it might be useful to do a little channel capacity comparison. So, as a crude start, we can look at the raw bitrate per channel:
2.3 Mbps for 24/96
4.6 Mbps for 24/192
2.8 Mbps for SACD
Of course, this is just the raw bitrate, and doesn't account for whatever redundancies exist in the actual data streams. Both SACD and DVD-A use lossless compression schemes to remove as much redundant information as possible. Given the similarity and sophistication of the compression schemes (both are predictive encoders), I think it is safe to assume that the compressed bitstream should contain fairly close to 100% relevant data (perhaps within 10%?). Thus, it is perhaps more useful to look at the compressed data rates to see just how much actual information is really there. Conveniently, MLP (Meridian Lossless Packing for DVD-A) and SACD predictive run-length encoding both acheive about 2:1 compression, so we can just compare the raw bitrates anyway.
Based on this simplistic analysis, we could conclude that SACD should be capable of providing slightly better resolution than 24/96, but certainly less than 24/192PCM audio. There is, however, one thing which bugs me: the fact that 24/96 audio will only code signal components at or below 48kHz (most likely filtered to remove most stuff above 20 or 30kHz), while SACD's encoded spectral content is less clearly defined... how much of the non-redundant compressed data is useless stuff above 20kHz? SACD claims a bandwidth of 100kHz, and basic sigma-delta and noise-shaping theory tell us that there should be a tremendous amount of noise (and therefore uncompressible data) in the ultrasonic region of the spectrum. Now, here it would be helpful to know more about the SACD prediction algorithms, in order to determine more about how much of the compressed DSD data stream is wasted on irrelevant ultrasonic noise...
Anyway, I have basically concluded that there should be little difference between SACD and 24/96 in terms of absolute resolution capability, and my gut feeling gives the edge to 24/96. Whatever the reality is, I think that converting DSD into 24/96 should be an acceptable compromise in order to gain the benefits of all-digital processing for things like speaker crossovers and so on... the other benefit to me will be the ability to use a single high-quality DAC for all of the various audio sources I will be using (CDDA, MP3 CD-Rs, DTS audio discs, DTS and DD movies, SACD and of course, DVD-A). Oh yeah, and of course, if I am using a separate DAC, I won't need to worry about the noisy power supply
Heh, there goes yet another verbose post... oops. Oh well.
Jim:
I used a BUF-03FJ preamp for many years, and enjoyed it immensely. However, I have recently replaced the BUF-03s with a different circuit which is much better... the circuit I designed is actually based on an OPA134, and uses a high-current SE class-a mosfet output stage inside the opamp feedback loop. The improvement was startling, given how good the BUF-03 is... improved transparency and detail, as well as warmth and smoothness. The noise floor was also noticeably improved (perhaps responsible for some of the apparent increase in detail and resolution), since the BUF-03 is *very* noisy at something like 50nV/rt(Hz)! The replacement circuit is also much better at driving difficult loads - with the high current output stage, it will push 5Vp-p into 10ohms or less, or a 10uF cap without so much as a hiccup! Bandwidth is flat to 5MHz+ (couldn't measure past that at the time I built it). I don't think I'll ever go back to the BUF-03. It was a great device in it's day, but I think it has been surpassed by more modern devices.
hm.. i figured the power supply might need some upgrading. I suppose if worse comes to worse, it's not a huge deal to build a new linear power supply for it, but one would probably have to tear apart the stock PSU to get the power switch and LED assemblies. Also, the stock PSU will likely have some power switching features and control features which would need to be duplicated in the replacement...
I suspect that the DC offset's you've seen on the outputs are a result of the cheapo opamps. This, if nothing else, is reason enough for me to upgrade the opamps. At least I can get rid of the awful 'lytics in the signal path (yuk!). Also, it may be an option to use a nice film coupling cap at the input to the opamp if there is still some DC offset from prior circuitry. I'll also have to scope the input to the muting transistors to see if the player uses them during operation... I'm kinda hoping they're used in power up and power down only, which would make their removal irrelevant to me, as I follow a strict power-up / power-down procedure anyway (my preamp is also DC coupled). Whatever happens, I think these are the three key mods that really need to be done for an immediate improvement, which should suffice until I'm using an external DAC.
Regarding output in PCM format, one of my main reasons is that I want to be able to perform signal processing on the data, which is virtually impossible with DSD streams. In fact, I learned something interesting today while reading some online articles... it seems DSD studio equipment actually employs a quasi-PCM format for editing(!)... it would seem that all the marketing hype about maintaining the purity of the data streams by avoiding PCM conversions is just propoganda! It is well known that in order to do any useful studio editing, some form of PCM conversion is almost essential, and of course it turns out that this is pretty much exactly what happens during the process of producing a master. There are only a very small percentage of recordings (maybe 5%) which can be produced without any conversion into PCM. Of course, Sony and Philips would never want the consumers to find out that the recordings they're buying have actually gone through DSD-PCM conversion, and then back again! Personally, I understand the mathematical underpinnings of the conversion processes quite well, and I'm not the least bit afraid of them. A single conversion into PCM isn't likely to do much if any audible damage to the music.
I don't want to get into a DVD-A vs SACD war, but for the purposes of evaluating whether or not it is a worthwhile proposal to convert SACD streams into 24/96 or 24/192 PCM data for processing or the use of an external DAC, it might be useful to do a little channel capacity comparison. So, as a crude start, we can look at the raw bitrate per channel:
2.3 Mbps for 24/96
4.6 Mbps for 24/192
2.8 Mbps for SACD
Of course, this is just the raw bitrate, and doesn't account for whatever redundancies exist in the actual data streams. Both SACD and DVD-A use lossless compression schemes to remove as much redundant information as possible. Given the similarity and sophistication of the compression schemes (both are predictive encoders), I think it is safe to assume that the compressed bitstream should contain fairly close to 100% relevant data (perhaps within 10%?). Thus, it is perhaps more useful to look at the compressed data rates to see just how much actual information is really there. Conveniently, MLP (Meridian Lossless Packing for DVD-A) and SACD predictive run-length encoding both acheive about 2:1 compression, so we can just compare the raw bitrates anyway.
Based on this simplistic analysis, we could conclude that SACD should be capable of providing slightly better resolution than 24/96, but certainly less than 24/192PCM audio. There is, however, one thing which bugs me: the fact that 24/96 audio will only code signal components at or below 48kHz (most likely filtered to remove most stuff above 20 or 30kHz), while SACD's encoded spectral content is less clearly defined... how much of the non-redundant compressed data is useless stuff above 20kHz? SACD claims a bandwidth of 100kHz, and basic sigma-delta and noise-shaping theory tell us that there should be a tremendous amount of noise (and therefore uncompressible data) in the ultrasonic region of the spectrum. Now, here it would be helpful to know more about the SACD prediction algorithms, in order to determine more about how much of the compressed DSD data stream is wasted on irrelevant ultrasonic noise...
Anyway, I have basically concluded that there should be little difference between SACD and 24/96 in terms of absolute resolution capability, and my gut feeling gives the edge to 24/96. Whatever the reality is, I think that converting DSD into 24/96 should be an acceptable compromise in order to gain the benefits of all-digital processing for things like speaker crossovers and so on... the other benefit to me will be the ability to use a single high-quality DAC for all of the various audio sources I will be using (CDDA, MP3 CD-Rs, DTS audio discs, DTS and DD movies, SACD and of course, DVD-A). Oh yeah, and of course, if I am using a separate DAC, I won't need to worry about the noisy power supply
Heh, there goes yet another verbose post... oops. Oh well.
Jim:
I used a BUF-03FJ preamp for many years, and enjoyed it immensely. However, I have recently replaced the BUF-03s with a different circuit which is much better... the circuit I designed is actually based on an OPA134, and uses a high-current SE class-a mosfet output stage inside the opamp feedback loop. The improvement was startling, given how good the BUF-03 is... improved transparency and detail, as well as warmth and smoothness. The noise floor was also noticeably improved (perhaps responsible for some of the apparent increase in detail and resolution), since the BUF-03 is *very* noisy at something like 50nV/rt(Hz)! The replacement circuit is also much better at driving difficult loads - with the high current output stage, it will push 5Vp-p into 10ohms or less, or a 10uF cap without so much as a hiccup! Bandwidth is flat to 5MHz+ (couldn't measure past that at the time I built it). I don't think I'll ever go back to the BUF-03. It was a great device in it's day, but I think it has been surpassed by more modern devices.