Basically, in Italy they listen to the DS dac first. Its amazingly detailed, yet smooth and refined. People are impressed. Then they play the R2R dac which is more distorted and ask which one sounds more real. Everyone says the R2R sounds more "real." Why? because the lower midrange frequencies around 200-300Hz or so are warm, full, and nicely rounded. It sounds more like if you sit in a quiet room and listen to the sounds of house, cars driving by outside, etc. Its a warm, natural sound, in the low midrange. Then listen to the DS dac again and it doesn't have that effect. Instead it sounds great except it doesn't have that pronunced full, warm low midrange sound.
Just goes to show people listen to FR first (or what sounds like FR). Low distortion, fine details, smooth sound, etc. are all secondary.
OTOH, here we regard the low midrange sound of R2R as a quirk of the technology. Most mics don't pick up room sound the same way human ears do.
So the question is, is the R2R dac more accurate, or is it an effects box to make the hi-fi sound from the speakers more like the way the ear hears the sound of a room and or instruments in the room?
Also, I would say there is some tendency for DS dacs to sound detailed, measure pretty well, yet still not sound quite right. Modulator design and everything else about dac design seems to have a lot of bearing on it. IME getting exceptionally good sound is not easy, but its very easy to damage once you have it.
Just goes to show people listen to FR first (or what sounds like FR). Low distortion, fine details, smooth sound, etc. are all secondary.
OTOH, here we regard the low midrange sound of R2R as a quirk of the technology. Most mics don't pick up room sound the same way human ears do.
So the question is, is the R2R dac more accurate, or is it an effects box to make the hi-fi sound from the speakers more like the way the ear hears the sound of a room and or instruments in the room?
Also, I would say there is some tendency for DS dacs to sound detailed, measure pretty well, yet still not sound quite right. Modulator design and everything else about dac design seems to have a lot of bearing on it. IME getting exceptionally good sound is not easy, but its very easy to damage once you have it.
Last edited:
I believe they will both measure flat. If its anything in the DS dac, maybe its that the ear is very sensitive in low midrange. If there is what looks like a noise floor on an FFT, for a DS dac it may not sound like resistor noise. It may sound more like little squeeks, sweeps, chirps, etc. Although it may be too low level to make sense out of, it may mask some low level detail the ear expects to hear in the low midrange. That would be my guess if its the DS dac causing the perceived difference.
One thing I found out a long time ago is there can be digital audio distortion (not necessarily conventional distortion) that makes the midrange sound recessed. Trying to fix it with EQ didn't work, so I knew it wasn't really FR that was the problem.
Also as an historical aside, some years ago there was a big market for "warmer" VST plugins to make digital audio sound warm, more like analog. One of the more ubiquitous ones was "Vintage Warmer" which always made a track sound "better" the instant it was applied. It was very popular for years. However it was scam. What it really did by default was raise the signal level by 3dB. Of course louder always sounds better, and nobody apparently suspected the trick except me. I noticed that applying the plugin twice in a row made the sound worse, so I got suspicious. A quick check showed it added some distortion and 3dB of gain. Louder sounded better, but the distortion sounded worse if there was too much of it.
One thing I found out a long time ago is there can be digital audio distortion (not necessarily conventional distortion) that makes the midrange sound recessed. Trying to fix it with EQ didn't work, so I knew it wasn't really FR that was the problem.
Also as an historical aside, some years ago there was a big market for "warmer" VST plugins to make digital audio sound warm, more like analog. One of the more ubiquitous ones was "Vintage Warmer" which always made a track sound "better" the instant it was applied. It was very popular for years. However it was scam. What it really did by default was raise the signal level by 3dB. Of course louder always sounds better, and nobody apparently suspected the trick except me. I noticed that applying the plugin twice in a row made the sound worse, so I got suspicious. A quick check showed it added some distortion and 3dB of gain. Louder sounded better, but the distortion sounded worse if there was too much of it.
Last edited:
Much more likely reason is the high level of higher order distortions most R2R dacs have even at low levels. E.g. as shown in post #2982. Also other fan favorite diy dacs (AD1862, PCM1702/04, PCM63, ...) have very similar distortion profiles. Given the popularity of such dacs among audiophiles it is quite likely that such distortions can be appealing to some. But not all.
Well there is always @TNT and his field recorder, if you like that sort of thing.
Also, maybe something at places like: https://www.trackseventeen.com/mic_rigs.html
Also, maybe something at places like: https://www.trackseventeen.com/mic_rigs.html
I finally made some progress with my attempts to synthesize a combined LC and MFB filter with a Butterworth response and an input impedance that is not worse than the original filter. You can read all about it in section 6 of the attachment, but if you are only interested in the result, this is a fourth-order 83.72 kHz Butterworth filter with a smaller input impedance peak than the original filter (129 ohm instead of 198 ohm excluding Rs):
"Basically, in Italy they listento the DS dac first..."
Mark. From Your description it looks like You are talking about the event that I had been fortunate enough to partecipate. For me You try to convey the information that you could gather from Andrea. Your 'transcription' sounds quite faithful, maybe with little corrections. Or rather, let me add my own impression too..
First of all, the setup of the host of the event. I hope I am not offending him (Il Gavro) if I show a photo
These are JBL bass, with JBL compression drivers on the Mid and Tw sections with horns. The amplification is Audio Analogue Maestro 200 monoblock+A.A. pre. These are zero global feedback, 200W/8ohm, 800W 2ohm etc. etc. oversized, 'gentle giants'.
The JBL system is ~ 104db sensitivity. (going by memory..)
I went into details, to make it clear that it's a very high resolution, low distortion, ~infinite dynamic capability chain, with a quite merciless approach towards the feeding chain. It's totally capable to throw a big, wide and deep soundstage and execute a 'disappearing act', all those massive mobiles and horns notwithstanding.
Mark. From Your description it looks like You are talking about the event that I had been fortunate enough to partecipate. For me You try to convey the information that you could gather from Andrea. Your 'transcription' sounds quite faithful, maybe with little corrections. Or rather, let me add my own impression too..
First of all, the setup of the host of the event. I hope I am not offending him (Il Gavro) if I show a photo
These are JBL bass, with JBL compression drivers on the Mid and Tw sections with horns. The amplification is Audio Analogue Maestro 200 monoblock+A.A. pre. These are zero global feedback, 200W/8ohm, 800W 2ohm etc. etc. oversized, 'gentle giants'.
The JBL system is ~ 104db sensitivity. (going by memory..)
I went into details, to make it clear that it's a very high resolution, low distortion, ~infinite dynamic capability chain, with a quite merciless approach towards the feeding chain. It's totally capable to throw a big, wide and deep soundstage and execute a 'disappearing act', all those massive mobiles and horns notwithstanding.
Last edited:
So.. there had been not two but three dac implementations from Andrea. The 'classic' R2R, (twsdac-LT), their DSD dac (Twsdac-DSD), and the latest development Single Ended DSD dac (twsdac-dsd-se).
The 'DSD' sound that you had described came from this latter. Your description is quite good, would like complete it with a very nice soundstage developing in depth, too.
The other two dacs (yes, both DSd and R2R) are sounding different, and more close to each other than this above described smooth, analog, neutral sound from the DSD -SE model.
And yes, the difference is the color, impact, weight, reality..
Let me describe like this:
Dsd-SE dac:
The 'DSD' sound that you had described came from this latter. Your description is quite good, would like complete it with a very nice soundstage developing in depth, too.
The other two dacs (yes, both DSd and R2R) are sounding different, and more close to each other than this above described smooth, analog, neutral sound from the DSD -SE model.
And yes, the difference is the color, impact, weight, reality..
Let me describe like this:
Dsd-SE dac:
Last edited:
The point is that there is no such thing as recording with natural or real sounds. Whether or not the sound originate from nature or from musical instruments. Recording is always a compromise and deviation from the original.
But assuming there is a recording that closely resembles "real" sound, it is possible to quantify differences. Using a good ADC and DeltaWave make re-recordings of the original with both R2R dac and well performing DS dac. Assuming the R2R dac is of decent quality the differences against the original in both cases will likely be small but still it is very likely that the R2R dac deviates more from original. So what you refer to more "real" sound of R2R dac is actually quite the opposite: more artificial sound or in other words an effects box. Some like it, some don't. Neither has the "correct" opinion.
It is also possible to listen to the difference recordings made with DeltaWave amplified. I would be amazed if you heard any little squeeks, sweeps, chirps, etc. from either one.
Some like the sound of horns.... as with using transistor amplifier gear... so any planar & tube based gear provided the real beef. It depends also on the used recordings (acoustic based as not boom boom based) & required 0° / 180° phase tests before doing the real listening tests.
My 2 cents on used gear
Hi Joseph ... What a wonderful and (to me) very meaningful allegory on the differences between an R2R DAC and a DSD DAC (in this case Andrea's).
Looking at the two paintings gives me an almost instant feel and impression of how these two DACs may/could sound. Indeed much more so than "a thousand words" likely would do. Many thanks for sharing this
Jesper
Looking at the two paintings gives me an almost instant feel and impression of how these two DACs may/could sound. Indeed much more so than "a thousand words" likely would do. Many thanks for sharing this
Jesper
Marcel,
This seems like a step forward in getting rid of the alias products.
The input impedance , Red existing filter and blue the new proposal are in the first attachment.
Below are the FR in blue compared to a 4th order passive LC filter.
Above a few Mhz, the active filter is slightly deviating from the ideal filter curve.
Not happy with all the op-amps that were misbehaving in the double clocked RTZ Firdac, I tried to make a copy of the AD797 with discrete components with interesting results.
The response was now exactly the same as with a passive filter, although LTSpice really calculates distortion when discrete transistors are used.
So for the RTZ Firdac such a discrete solution could be an option.
Hans
Attachments
That is excellent.
Next stop, add a second Op-Amp at the inverting input followed by an Inverter (low noise of course) or perhaps just a low noise J-Fet with the output connected in front of Ra, as additional AC error servo.
Thor
I used the Opa2210 in the newly designed combined LC-MFB filter.
And my 797 like op amp was used in the original RTZ Firdac filter, replacing the Opa2210 while the digital circuit was driven with twice the clock speed, thereby using each .dsf sample twice.
As shown in #3013, no op-amp came even close to faithfully proces the Firdac’s bitstream.
But this discrete 797 like version passed the same test with flying colors.
Hans