5847, eh?
Glad one of us isn't too old to remember. Anyway, sounds right. I have so many different flavours of those filters around, depending on what version the local rep had in stock that they all start to sound alike.
Part number wise, that is.
Francois:
You are starting to sound like the guys who started Wadia. I met Don Moses, and chatted with him a great length. A smart guy, but out of his natural surroundings when it came to manufacturing. He had some interesting theories on why slow roll-offs sounded better. Problem was, his version had lots of ultrasonic crud coming through.
Jocko
Glad one of us isn't too old to remember. Anyway, sounds right. I have so many different flavours of those filters around, depending on what version the local rep had in stock that they all start to sound alike.
Part number wise, that is.
Francois:
You are starting to sound like the guys who started Wadia. I met Don Moses, and chatted with him a great length. A smart guy, but out of his natural surroundings when it came to manufacturing. He had some interesting theories on why slow roll-offs sounded better. Problem was, his version had lots of ultrasonic crud coming through.
Jocko
Re: 5847, eh?
Actually, you're pretty close to the mark: I got turned onto the possibilities of slow rolloff filters after reading a Wadia review in St*r*ophil*. I also put together 10 kHz filters (long story) as part of a dayjob project, and the steep rolloff filter sounded HORRIBLE compared to one with slower cutoff. It shouldn't really - it's not adding information - but we kept the gentler one.
As for the ultrasonic crud, one FIR structure saves about half the computrons by placing the -6 dB point at exactly half the sampling frequency (FIR mavens will recognize a half-rate filter). The rolloff is symmetric about the -6 dB point when measured in linear amplitude, which is OK if your rolloff is fast because the transition band is narrow, but with slow rolloff the broader transition band means you will get higher energy of aliasing junk past the Nyquist frequency. Not a problem if you run soft dome tweeters, but that could excite the usual ultrasonic resonance in metal domes, with possible intermodulation consequences in the audible range. The obvious solution is to avoid half-rate structures at the cost of doubling the required CPU power. Computrons are cheap these days, however.
Francois.
Jocko Homo said:
Actually, you're pretty close to the mark: I got turned onto the possibilities of slow rolloff filters after reading a Wadia review in St*r*ophil*. I also put together 10 kHz filters (long story) as part of a dayjob project, and the steep rolloff filter sounded HORRIBLE compared to one with slower cutoff. It shouldn't really - it's not adding information - but we kept the gentler one.
As for the ultrasonic crud, one FIR structure saves about half the computrons by placing the -6 dB point at exactly half the sampling frequency (FIR mavens will recognize a half-rate filter). The rolloff is symmetric about the -6 dB point when measured in linear amplitude, which is OK if your rolloff is fast because the transition band is narrow, but with slow rolloff the broader transition band means you will get higher energy of aliasing junk past the Nyquist frequency. Not a problem if you run soft dome tweeters, but that could excite the usual ultrasonic resonance in metal domes, with possible intermodulation consequences in the audible range. The obvious solution is to avoid half-rate structures at the cost of doubling the required CPU power. Computrons are cheap these days, however.
Francois.
If you take money and effort out of the eqation would you run with the 1704 or the newer 179x?
Does anyone know which DAC chip the Gold LINK III DAC uses, the data sheet says BB, 24/96 single dacs in dual diff.
As far as I am aware BB only have one chip that is single and 24/96 ie the 1704.
My dac sits directly coupled to the poweramp balance input with a passive I/V in between. I know this sounds and works great. I dont think it will get anybetter whatever dacs I use so I'll just stick with what I have but want a real pcb and not the prototype, hardwired mess I got now.
Does anyone know which DAC chip the Gold LINK III DAC uses, the data sheet says BB, 24/96 single dacs in dual diff.
As far as I am aware BB only have one chip that is single and 24/96 ie the 1704.
My dac sits directly coupled to the poweramp balance input with a passive I/V in between. I know this sounds and works great. I dont think it will get anybetter whatever dacs I use so I'll just stick with what I have but want a real pcb and not the prototype, hardwired mess I got now.
Actually, I can't find a single instance of anyone in the DIY community doing a PCB using a pair of PCM1794 in digitally balanced mode (i.e. using two '1794s, one L and one R, each operating in "monaural mode.") -- with or without an external filter.
Am I missing a posted and much-discussed design? Or is this a worthy (and not freakishly difficult) project?
Am I missing a posted and much-discussed design? Or is this a worthy (and not freakishly difficult) project?
Designing a decent I/V convertor, without using opamps (The datasheet specifies NE5534's) may be the biggest problem with the 'PCM1794.
In mono mode, these output almost +/-16mA peak, which is a considerable increase from the 1-2mA of other DAC's. How do you design a balanced I/V convertor which will cope with that with sufficiently low distortion?
In mono mode, these output almost +/-16mA peak, which is a considerable increase from the 1-2mA of other DAC's. How do you design a balanced I/V convertor which will cope with that with sufficiently low distortion?
A transformer would be the most obvious solution... A good-quality output transformer does I/V extremely and has the added merit of taking care of your HF filtering (because even great trafos roll off somewhere above 30khz). Also, it would allow for an optional single-ended output.
Here's a nice example:
http://www.raleighaudio.com/passive_output.htm
Of course, with that kind of current you could do well with a simple resistor-based I/V convertor -- a great approach, if you can get the value right. But I seem to recall reading somewhere that a simple resistor-based passive I/V solution doesn't work with the 1794, though I can't remember why. If it does work, it's by far the most elegant approach.
Here's a nice example:
http://www.raleighaudio.com/passive_output.htm
Of course, with that kind of current you could do well with a simple resistor-based I/V convertor -- a great approach, if you can get the value right. But I seem to recall reading somewhere that a simple resistor-based passive I/V solution doesn't work with the 1794, though I can't remember why. If it does work, it's by far the most elegant approach.
Why the 1704..?
The 1704 is the only dac that is multibit and can handle 24 bits ok. Running two dacs in balanced mode enables a shorter and more symmetrical analogue signal path (if you want to stay balanced) then converting to balanced after the I-V stage. In my (hands-on and subjective) experience this is more important then a +-3dB noise floor @ -120dB
I also believe being able to individually decouple the dacs is a benefit and should help with definition particularly with intense and complex signals.
It is also my experience (after subjective testing and in my system) that a balanced set up is sonically superior to a non balanced set up.
Also in my (agreed very subjective) experience that the delta sigma stuff is to "nice", fluffy and laid-back compared to the multibits bite, force, and definition.
16 bits are out as I also run dvd-audio and video with up to 24 bits.
So what else is there??
The 1704 is the only dac that is multibit and can handle 24 bits ok. Running two dacs in balanced mode enables a shorter and more symmetrical analogue signal path (if you want to stay balanced) then converting to balanced after the I-V stage. In my (hands-on and subjective) experience this is more important then a +-3dB noise floor @ -120dB
I also believe being able to individually decouple the dacs is a benefit and should help with definition particularly with intense and complex signals.
It is also my experience (after subjective testing and in my system) that a balanced set up is sonically superior to a non balanced set up.
Also in my (agreed very subjective) experience that the delta sigma stuff is to "nice", fluffy and laid-back compared to the multibits bite, force, and definition.
16 bits are out as I also run dvd-audio and video with up to 24 bits.
So what else is there??
sasha said:It looks like We can't agree on balanced or not balanced, but what
about reclocking? Is that something We agree it is beneficial?
And is it possible to do it with PCM1704?
Hi
Everyone taking digital audio seriously, should re-clock. Start with a clean clock though, otherwise the child goes with the bathwater.......
In my new upcoming DAC, extensive reclocking is one of the key features. It is unbalanced, for those wanting to know.
best
Extreme_Boky said:
Hi
Because you need more electronics, or a transformer. But when interfacing equipment in complex systems, balancing has much advantages, these may be bigger than the loss of quality in extra electronics. Whenever possible I'd use transformers (galvanic isolation too) rather then extra active electronics.
In "normal" consumer audiosystems, the interfacing is that simple that balancing is not needed, assumed proper cabling and common mode termination is applied.
best
balanced dac
I want in. I need something with balanced outs for my Aleph amp. There have been many other discussions about this in other threads. The Pass D1 uses 2 x pcm63 DACs with a typical Pass I/v output (balanced). I have not had time to analize the schematics and am not sure without much investigation how the dacs receive their signals. I think that the digital is inverted in one. This DAC has more parts than I would like to see. It would be interesting to listen to a simple, non- os system (like cs8412 + TDA1543) with differential output. I thint that it might be easier to determin the value of the diff output with a more primative circuit.
Equally as interesting would be to compare to simple non-os dacs, in which on used an os150 (think this is right) to decouple the receiver/dac. I mean if we are using the diff out to reject minute amounts of common mode noise. . . .
I want in. I need something with balanced outs for my Aleph amp. There have been many other discussions about this in other threads. The Pass D1 uses 2 x pcm63 DACs with a typical Pass I/v output (balanced). I have not had time to analize the schematics and am not sure without much investigation how the dacs receive their signals. I think that the digital is inverted in one. This DAC has more parts than I would like to see. It would be interesting to listen to a simple, non- os system (like cs8412 + TDA1543) with differential output. I thint that it might be easier to determin the value of the diff output with a more primative circuit.
Equally as interesting would be to compare to simple non-os dacs, in which on used an os150 (think this is right) to decouple the receiver/dac. I mean if we are using the diff out to reject minute amounts of common mode noise. . . .
response to rfbrw
The sarcasm is a little dull. Since there are so many discussions (and some interesting boards) on this topic already, what was the purpose of staring this one in the first place?
Answer: Some people were interested in talking about it.
What is the need to be so dismissive?
The sarcasm is a little dull. Since there are so many discussions (and some interesting boards) on this topic already, what was the purpose of staring this one in the first place?
Answer: Some people were interested in talking about it.
What is the need to be so dismissive?
The AD1955 incorporates diff outputs on chip for both L/R channels. But, I think it requires a controller. Not personally interested in using a controller, but if someone has experience with this, it will make balanced outputs easy. I don't have any experience with AD DACs although I like some of their analog stuff. The output section of the D1 would be perfect for this if the DAC sounds any good.
Anyone know of any chips like this that do not require a controller?
Anyone know of any chips like this that do not require a controller?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.