If one company uses inches and another meters, you're right, there is something fishy going on.
Brian this is really simple:
0.1% = -60dB
0.03% = -70dB
0.01% = -80dB
0.003% = -90dB
0.001% = -100dB
0.0003% = -110dB
0.0001% = -120dB
Hope this helps.
The sounds vs the two chips, I don't think it is a huge difference(CD, 44/16), never listened to them though.
Brian this is really simple:
0.1% = -60dB
0.03% = -70dB
0.01% = -80dB
0.003% = -90dB
0.001% = -100dB
0.0003% = -110dB
0.0001% = -120dB
Hope this helps.
The sounds vs the two chips, I don't think it is a huge difference(CD, 44/16), never listened to them though.
Ohhhh, Gigantic, BTW, I was comparing both DACs in Mono mode, 192Khz, 24 bit.
If I'm reading the graphs correctly:
TI @ 192 Khz THD+N = 0.001%, or - 100db.
AD @ 192 Khz THD+N = arrrrrrrg, ok, which graph do I use?
There doesn't seem to be a THD+N measurement for the 192Khz mode, unless it's still -110db.
However, both companies don't really 'fully' spec their mono mode.
Darn, gotta keep those .pdfs open... Ooops, just accidentally closed them again...
If I'm reading the graphs correctly:
TI @ 192 Khz THD+N = 0.001%, or - 100db.
AD @ 192 Khz THD+N = arrrrrrrg, ok, which graph do I use?
There doesn't seem to be a THD+N measurement for the 192Khz mode, unless it's still -110db.
However, both companies don't really 'fully' spec their mono mode.
Darn, gotta keep those .pdfs open... Ooops, just accidentally closed them again...
It seems that 192 kHz is nothing TI wants to talk about. Notice also _how_ the measurements are done. I can't see any detailed desciptions of that. TI for instance has NE5534 as recommend opamp (unbelievable!).
Th AD1955 seems to be substantially better at 192 kHz but I think this is not the whole truth.
Th AD1955 seems to be substantially better at 192 kHz but I think this is not the whole truth.
what I have noticed when comparing AD to TI DAC´s: both are using the AudioPrecision System 2, AD states that they are using RMS mode, while the use of Averaging mode gives about 2dB better results, guess what mode TI is using....
(Hope it´s not nonsens what I write here, think it was PCM1738 and AD1955 datasheets...)
(Hope it´s not nonsens what I write here, think it was PCM1738 and AD1955 datasheets...)
It's not complete nonsense, however, the super crucial graph on the AD1955 sheet is the bottom left on page #11.
'Wideband FFT plot...'
Here, mono mode will have a profoundly larger margin in difference, more like another + 4-6 db because of the way the DACs are matched in an inverted fashion. With a similar test on the TI, it would probably stink on this test. The really high performance uncompressed SACD band doesn't appear to be one of their concerns.
'Wideband FFT plot...'
Here, mono mode will have a profoundly larger margin in difference, more like another + 4-6 db because of the way the DACs are matched in an inverted fashion. With a similar test on the TI, it would probably stink on this test. The really high performance uncompressed SACD band doesn't appear to be one of their concerns.
this thread was i looking for... anybody knows whether the ad1955 needs a seperate int. filter in mono mode? or is it performing the filtering also in mono mode?...
im wondering, besides this... pcm 1792/4 or ad1955.. although i think i ll stick with the ad, for entirely technical reasons - i cant have 14 opamps (propably opa627) on a board to get an s.e. out grom the pcm.... itd be a rather populated board.. (+2*28 caps)..
anybody used an ad1896?? seems nice to get rid of low pass filtering....
im wondering, besides this... pcm 1792/4 or ad1955.. although i think i ll stick with the ad, for entirely technical reasons - i cant have 14 opamps (propably opa627) on a board to get an s.e. out grom the pcm.... itd be a rather populated board.. (+2*28 caps)..
anybody used an ad1896?? seems nice to get rid of low pass filtering....
peranders said:
It does seem rather odd, but I have been in contact with their application engineers and they highly recommended this approach to "enable datasheet level performance". Needless to say, I have my own view on how this can best be done.
Now, if you look at the PCM1794 which appears to be the same chip as the 1792, you will see that this chip is MUCH easier to use than the 1794 and the AD variant. My opinion is that these devices are all probably significantly better than previous chips (at least from data sheets) and whether one chooses one or the other is probably much less important than the actual implementation. Since the 1794 appears to be easy to use you can spend more time on the implementation and probably extract better results as a DIY'er
Petter
I'd rather see something like 20-20kHz band limited noise because A-weighting can certainly conceal some sins, whereas at best it contributes a little bit of psychoacoustic data.
That is, if two pieces of equipment are
20-20kHz, -100dB; A-weight -103dB
20-20kHz, -100dB; A-weight -106dB
Then maybe #2 has an edge. But if all you have to decide is A-weight then you could end up with
A-weight -103dB; 20-20kHz -100dB
A-weight -103dB; 20-20kHz -96dB
This is pretty extreme, but I could see it happening because A-weighting conceals both low-frequency noise/distortion and the rising noise/distortion that accompanies the higher frequencies. Unweighted results can't hide anything, as long as you keep an eye on the OOB noise (if the manufacturer will tell you!)
Analog supplies both weighted and unweighted results side-by-side on their earlier datasheets, not sure why they dropped that for AD1955.
That is, if two pieces of equipment are
20-20kHz, -100dB; A-weight -103dB
20-20kHz, -100dB; A-weight -106dB
Then maybe #2 has an edge. But if all you have to decide is A-weight then you could end up with
A-weight -103dB; 20-20kHz -100dB
A-weight -103dB; 20-20kHz -96dB
This is pretty extreme, but I could see it happening because A-weighting conceals both low-frequency noise/distortion and the rising noise/distortion that accompanies the higher frequencies. Unweighted results can't hide anything, as long as you keep an eye on the OOB noise (if the manufacturer will tell you!)
Analog supplies both weighted and unweighted results side-by-side on their earlier datasheets, not sure why they dropped that for AD1955.
In terms of pinouts, the pcm1792 is almost identical to the pcm1794. It might be possible to make a pcb that would take either chip, as the difference between the two is that on the pcm1794, the i2c/spi serial interface lines are replaced by control lines. I will try on my pcb for my pcm1792 dac to allow it to be interchanged with a pcm1794. I am sticking with the 1792, because I will already have a microcontroller in the project for controlling the cdpro2.
--
Brian
--
Brian
audiodesign said:
Interesting article.
Do you have any comments on the listening tests? What about stereo performance?
I have made a dac with pcm1798 and a pass lab D1 I/V stage. I'm however a bit dissapointed with the sound. It has better specs than pcm1730, which I have also tried, but the sound appears to be more dull. My dac with 1730 did not work completely - got some scratching noise sometimes, but between the cracks, the sound was much better IMO.
I don't really understand this obsession with specs and why do they tend to make the sound boring? Does the human ear prefer some background noise?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.