RMAA Test results for Audio-gd DAC-19MK3 and Reference 1
Jul 21, 2009 at 8:23 PM Post #16 of 22

Originally Posted by slim.a /img/forum/go_quote.gif
Fyi, I did some RMAA measurements of the DAC-19mk3, DAC-100 and EMU 0404 USB that I posted here.

Even the results do not mean much usually, they are still much better than those published in quote.
I used the EMU 0404 usb to do the measurements.

I haven't read that review in a while. Good job there!

Jan 23, 2010 at 11:27 AM Post #17 of 22
It's interesting that a DAC with a little rolloff in the extreme highs seems to sound pretty good. I suspect this rolloff is real, not a measurement artifact. Kingwa's explanation that the lowpass filter is doing this makes good sense. He's right that PCM has a TON of high frequency noise in it, going way up into the mHz range. That is inherent in the bitstream coming off your digital source, and unless you take steps to filter it out, and filter it GOOD, this crap is going to end up being fed to your audio amp stages. You can't hear this noise, but your audio amp stages won't like it and this supersonic noise can cause your audio stages to distort the audio you CAN hear. Feed a little 1 mHz signal to a typical audio stage and it can get saturated and do weird things to the nice audio you're trying to listen to. So a filter that aggressively keeps this stuff out of your analog stages is a great thing, and a really good filter might have a little rolloff starting at 15 kHz.

Note that most people over 25 years old cannot hear to 20,000 Hz anyway. Many people over 12 begin to show a rolloff in upper treble hearing. Over 40? Forget it!

Audiophiles might do well to have their hearing tested once in a while, and tell the audiologist you want to have the upper treble range tested as well (normal audiology tests don't test up very high.) Don't forget- your ears are your most important component!

Most adult audiophiles are surprised that their hearing above 12,000 Hz isn't really that acute.

SO: a little loss above 15 kHz is usually not audible, but a really good supersonic filter often IS something that you can hear as a sonic improvement.

Measurements don't lie- you just have to understand them. What you want are "good" measurements and "good" doesn't always mean "flat up into the broadcast band." Especially when it comes to DACs.
Jan 23, 2010 at 1:30 PM Post #18 of 22
Absolutely I will give $100 to anyone who can hear the difference between 82db and 85db at 18k, these measurements are meaningless. There really seams to be a lot of folks against what AudioGD is doing. Let me tell you these DAC's are the best I have heard ever. They compete with $5k DAC's. The 19mk3 w/ the PMD100 is a real giant killer. Much better than the cheap S-D DAC's everyone is using, and it has a discrete zero feedback analog stage. People have to understand that Burr-Brown/TI were forced to bring back the PCM1704 because of the poor performance of the S_D 179x series.
Jan 23, 2010 at 1:40 PM Post #19 of 22
those photos tell that even Dr.DAC2 has better RMAA results
Jan 23, 2010 at 2:28 PM Post #20 of 22
Isn't this the result of what milosz says? You want (/need) to filter all frequencies above the 20 kHz out of your resulting signal as they mess with your amp. To do that you need a very good analog filter (as Kingwa said). That low-pass filter needs to let all frequencies below the 20kHz to go through uninterrupted, but block(/attenuate) all that's above that threshold. The problem with such a filter is phase shifting, if you build a filter that would result in perfect RMAA results than that would mess too much with the signal's phase and that can be heard (loss of detail). Filters that don't mess too much with a signal's phase will not have perfectly flat RMAA results though and that can be seen from the REF 1 measurements for example.

That's how I understand it at least. Next time I'll have to pay more attention during Electronic Instrumentation and Signals and Systems classes from the EE faculty here. If I do that then I'll be able to figure it out myself.
Jan 23, 2010 at 4:22 PM Post #21 of 22

Originally Posted by DarKu /img/forum/go_quote.gif
those photos tell that even Dr.DAC2 has better RMAA results

Stick a graph in each ear and call me in the morning!
Jan 24, 2010 at 3:23 AM Post #22 of 22
Jeez...not this pointless graph/measurement BS again. When are people going to learn to ignore measurements and judge equipment with their ears and not their eyes ?



PS: That high frequency noise is from quantization (error) and has to do with algorithms used to covert digital to analog being imperfect representations (mathematically speaking) to their analog sine wave counterparts IIRC. The reason why first gen CDP's were almost universally rejected was because of the lack of decent filtering, no oversampling and strict adherence to the stupid 20-20khz spec which is fine for analog but terrible for pcm (any digital actually speaking). Until gentle roll off filters were implemented and oversampling applied to get the artifacts being filtered well out of the audible range we were stuck with ice pick highs and a bad case of listener fatigue from rampant digititis. Since filtering effects travel downward as well as upward in the freq spectrum you can well imagine why oversampling became one of the big break throughs in 85-87. That and the whole sale abandonment of NOS dacs and the requisite super steep brick wall analog filtering they employed which caused all kinds of problems well into the audible range. My first high end CDP I bought in 85 for 700USD, Carver unit with a special digital time lens circuit. It was worlds better than my first gen Sony machine was. Care to guess what was different vs the 1st gen SONY ? It also measured worse than the 1st gen machines in the highs and yet sounded worlds better. Go figure. I find it amazing that 37 years on we are still looking at 20-20khz and taking it seriously. As long as the freq response is good between 20-14Khz your pretty much good to go. Most instruments even synth stuff does not reach anywhere near 5K. Listen to test tones and try to correlate anything that is musical above 10Khz made by any instrument. I can only think of 2nd or 3rd order harmonics on something like triangle or a splash cymbal...even then that's a bit of a stretch.

Users who are viewing this thread