You just told-me you had a Sabre ESS9038 DAC evaluation board. What do you need more ?
You can find finished products with SOTA DACS for around 300$ that just wait your mods.
It is like to say you cannot find a car at a reasonable price, and point to Rolls Royce or Bugatti.
My wifi streamer ? < 40 $ with internal DAC Asahi Kasei AK4430.
NO EXCUSE ;-)
Last edited:
I don't want to.
Single ended distortion is at a few tens of ppm and its only 2nds, 3rds and a bit of 4th.
You don't need fully complimentary to cancel seconds, any diff pair has no seconds you only need to preserve that transfer function to the output. It's not hard.
The AK4430 is an exceptional sleeper DAC. Not in the premium class but sounds quite good, especially for a DAC that needs almost no external parts or handholding and costs less than $2 in large volume. Mostly unknown because they are in set top boxes (where the volume is).
Making a better DAC needs some real understanding and effort. A lot of the Asia DAC boards are not well executed demo board clones. They can be tweaked to be pretty good (I had some success with the Breeze DAC) but still a project.
Single ended complementary does essentially what differential does but with a 6dB improvement in input noise. Beat that!
I built a diff pair input stage for MC that worked fine and was quiet. A single IF 3602 kept the noise down. Most of the THD+N curve is the N portion at the bottom. Not a production piece but still interesting.
Well Wayne, that is a good solution for you, in fact it would have been my first choice, 50 years ago. Even 40 years ago, I did about the same thing with a 2SK146. It was quiet enough but I had to cap couple to the output. Then in 1981, I added a folded cascode pchannel jfet to make a direct coupled input stage for the JC-80. Then, maybe 1 year later I came out with the Vendetta input stage, and that reduced the input noise about 9dB. Instead of having 2 devices in noise series, I got 4 devices in noise parallel, and push-pull too! That is what one should strive for.
Last edited:
I just listened to-it on my PC speakers. Can find differences at fist sight.
On 4 things. Impact and realism of the charleston in the beginning. The voice that can be very disagreeable on some, and more fluid on others, confusion or not in the horn section, and overall dynamic (some sound like slightly limited).
Very interesting. Thanks a lot Max.
I will listen on my big system tomorrow.
Are-they different DACs ? Different digital formats ? And all exact same level (I had the feeling some were a little under) ?
Last edited:
Your old ears are still working perfectly fine, perhaps better than most ears especially considering that you are using PC 'Auratones'.
Funny typo, did you mean first sight or fist fight....these files and the reasons for differences may well cause the latter around here lol.
Yes, there are interesting things going on, FR errors/ripple, noise, .thd, imd etc.....but that ain't all and one result is changes in perceived dynamics, there are a bunch of others also as you have noted.
it is always a pleasure to discuss with sensible audience, the results I post here are very interesting, perhaps curious, but is actually just the way of the universe and along the way some occult hidden truths are revealed and nakedly so.
You will have some new fun I am sure, perhaps you can suggest another/ better recording to 'process'.
Loop gain is near as dammit to unity (by 10 turn trimpots output level modification, inputs are into 'insert' sockets of Tascam US-122).
All files are 44k 24bit wav files using the same usb soundcard, same custom 'BQP' loopback cable, and no digital effects whatsoever or other changes except for two types of additional materials placed within the soundcard in particular places.
"I Got The Music.wav" is 24bit copy of original downloaded 16bit file.
"I Got The Music - TS.wav" is loopback recording of original and is 'the reference' in this set of comparison files.....the fact that no recordings exactly match the original is irrelevant in this discussion..
"I Got The Music - TS - TS.wav" is loopbacked "I Got The Music - TS.wav", ie second generation recording.
The files TSV AND TSV2 are result of different "extras" in the system, and are also each re-recorded as TSV - TSV, and TSV2 - TSV2.
The six loopback files are each different to the original, this is to be expected of course despite Rightmark reporting v low distortion but some noise..... -80dB on DAW channel inputs level meters.
Differences between first gen files and second gen files is to be expected of course, what is not expected is differences between the first gen files, or differences between the second gen files, and also the nature of these changes.
I can tell you much more privately, for now I await your report of your listen on your big system, enjoy.
Dan.
How big of an issue is power supply rejection in a cost-no-object gadget like the Blowtorch? Is it practical to get to (for example) a 1dB "noise figure" (ratio of actual to idealized perfect) with all topologies, or does PSRR always need to be included at the design level?
Thanks,
Chris
Thanks,
Chris
3dB. As much as lost with a differential topology. So differential + complementary = single ended (from a noise perspective).
A pair of ZTX851 in the classic Demrow INA circuit is differential and balanced in and in the 0.4nV/rtHz region. I know pro audio circuits horrify some and 8 legs anywhere other than in the servo horrify others, but using current production parts rather than unobtanium with no huge matching requirements has its attractions.
After all you never see people claiming that old school 3Ohm ortofons are quieter than the newer 6 Ohm ones and coil resistance very quickly dominates noise performance...
After all you never see people claiming that old school 3Ohm ortofons are quieter than the newer 6 Ohm ones and coil resistance very quickly dominates noise performance...
I would say, makes sense. Otherwise, unobtanium has certain marketing attractions “carefully hand matched input devices” and other economic advantages, if you are one of the very few stocking them. It allows also adding a story to a product that otherwise would have nothing outstanding, from a performance perspective.
If John's point is that for the same number of non-obtainium Toshiba FET's (for you peasants out there - cool and bitchin people like me have a good stock - "We always take my car, cause it's never been beat, And we've never missed yet, with the girls we meet".) distributed in alternative ways, you get one answer. Otherwise: otherwise. But, is the power supply's contribution always potentially negligible?
To put it another way, are there performance issues related to the power supply significant enough (in cost-no-object projects) to warrant something beyond Sulzer/Jung/Didden levels?
Thanks,
Chris
To put it another way, are there performance issues related to the power supply significant enough (in cost-no-object projects) to warrant something beyond Sulzer/Jung/Didden levels?
Thanks,
Chris
- Status
- Not open for further replies.