And yet toslink is often panned for poor jitter performance. I am a single mode fibre fan from working with telco equipment so the bias against multimode fibre is high, but I need to research again to see if there is actually and quantifiable difference between coax and optical.
That reminds me --- Doesnt the filter used in recording have to be matched in characteristics on the playback side (reconstruction filter)??
How do we know if the two are complimentary filters for exact reconstruction?
THx-RNMarsh
We can have in France ~1Gb/s long distance internet connection via optical fiber (30€/month).
Files are not corrupted.
Tom Danley mentioned that you had to adjust every different manufacturers' DSP filter set differently for the same loudspeaker. You commented that slope bandwidth Q etc. completely defined a filter and they should be interchangeable. You might not have been thinking of notch filters, but that was inherent in any complete loudspeaker eq.
And that is most likely single mode with decent connectors. Toslink is nasty plastic fibre with nasty inaccurate connectors and nasty multimode. It can't possibly sound good. At least that is way some paint it. Like I say need to look for any measured data comparing the 2.
wave division multiplexed fibre setups are impressive beasties when you are plonking 64 high througput frequencies down a 9 micron diameter glass tube. Then half the bandwidth is wasted using MPLS. Sob.
wave division multiplexed fibre setups are impressive beasties when you are plonking 64 high througput frequencies down a 9 micron diameter glass tube. Then half the bandwidth is wasted using MPLS. Sob.
It is digital datas. No question of sound. Are they corrupted or not is the only question.
It is stupid to rely to "flow". FIFO (buffers) and reclock is the way.
Didnt think so. Was wondering - Why so many ways to do it suggested and written about? Wiki gives a good summary. Its all the variable filter suggestions that maked me wonder -- what for? Just use one most cost effective way that removes the junk freqs. Seems like it might make most difference with low sampling rate of CD/44.1 by slightly affecting upper end of audio.... depending on filter characteristics. But there have been work arounds for that, also.
just checking.
THx-RNMarsh
Last edited:
Filters at recording are there for avoid Aliasing. Playback side to remove sampling frequency. Why did they have to be matched, as long they do not introduce phase shifts in the audio range or level errors ?
It wouldnt be the f.o. itself but the interfacing...... reflections off the f.o. ends and gap between f.o. and recver and angle of f.o. cut........ clean square-cut ends, polished and tightly matted and aligned to receiver etc...... BTW -- the dB attenuation at each coupling is the same as in coax. You loose a fraction dB at each coupling. For a cable company that matters. But cheap toslink and f.o. connection are higher loss and more variable. The f.o. receiver may not be the best, either.... all leading to jitter.
THx-RNMarsh
Last edited:
You don't care about the edge or length of the bit (OK, in a limited range).
Just put each of bits in a registry. Then read this registry with a perfect local clock. Just the registry has to be large enough to compensate any frequency difference without running out of datas because you are reading too fast or two slow. That you can avoid with some PLL.
I don't remember saying that to Tom. He made a comment about loudspeaker design when I simply pointed out a simple correction to a standard peaking equalizer that is defined by gain and Q due to the frequency warping in the Z domain. The original comment was actually just pointing out how a person's correction to his own work can get lost in the shuffle. My comment had nothing to do at all with complex DSP speaker crossover design.
MiniDSP and SoX use Robert Bristow Johnson's Audio EQ Cookbook to compute the basic IIR biquad filter coefficients. He posted a correction for peaking EQ's that neither picked up on even though at 48K you might want to place eq at a substantial fraction of Nyquist.
peakingEQ: H(s) = (s^2 + s*(A/Q) + 1) / (s^2 + s/(A*Q) + 1)
Last edited:
Yeah. And as long as upstream jitter isn't sufficient to cause data errors, the only jitter that counts is when the DAC chip does the final conversion.
se
Sounds like a good work-around to minimize Toshlink jitter. But, is that what they do to a high standard in a $200-300 CD player?
THx-RNMarsh
Last edited:
Nowadays i use a stripped down 250gb netbook running on it's battery and connected to one of these great sounding low power consumption DAC's - HRT Music Streamer III | Hi-Fi+
I get around 6 hours running foobar before it needs charged again. Been there and this is the best solution (bang for buck) i've yet come across.
Worst solution was a desktop PC with it's inbuilt SMPS, a total nightmare of ground loop noise. If that was the only way open to me i'd never have abandoned my CD players and would be among the first to rubbish the whole concept of computer based audio.
I believe it is done in ANY transport - the wobble and flutter and wow of the mechanical turntable is horrendous compared to clock stability. I don't think a CD drive would be possible without an intermediate shift register to buffer the data.
This is another reason why many people are skeptical to reports that modifications to CD transports or the use of special drives and platter constructions would make any difference to the sound.
Jan
Does anyone make a pass through USB connector so one can use an external super battery for power to USB devices while still getting the audio data from the computer.
... but the battery on which that netbook runs just supplies a bunch of internal smps's. So I wonder if that is the big difference?
Jan
- Status
- Not open for further replies.