T+A HiFi - DAC 200 - Official Discussion Thread
Nov 24, 2023 at 10:15 AM Post #316 of 454
I am still not sure I am understanding the actual signal path correctly. But I totally agree you seem to be right and I must be wrong. I am no specialist.

So, let’s just take one single canal, say “Right”, do you mean :
- PCM will go thru a “PCM to DSD” converting chip, then thru the “DSD to analog” 1-bit chip,
Whereas
- DSD will skip to first part and go directly to the “DSD to analog” 1-bit chip.

Is this what you mean ?
Not quite.

The PCM1795 DAC uses Burr Brown's 'advance segment' architecture.
The simplest way to explain this is that the top 6 bits are converted 'as is', but the lower 18 bits are converted using a 1-bit delta sigma modulator that modulates to the level of 1 LSB (least significant bit) of the top portion.

I've got a bit more of an in depth explanation about two thirds through this post on the Teac UD501 which uses the same chip and also has the same filter bypass option:
https://goldensound.audio/2021/12/02/teac-ud501-dac-review-measurements/

Though worth noting the DAC200 avoids some of the issues present in that unit, particularly when running with the unit in 'NOS'.

The chip does NOT modulate to 1-bit entirely, it's a multibit conversion just a bit of a hybrid approach.
Most DACs will modulate the entire signal to 5-bit usually, whereas this separates the 6 MSB (most significant bits) and the lower 18 LSB (least significant bits) and handles them separately

With PCM, the 1-bit DSD converter is not used.
With DSD, the data is fed un-altered to the 1-bit DSD converter and the PCM1795 chip is not in use. They are entirely separate DACs / signal paths.
 
Last edited:
Nov 24, 2023 at 11:02 AM Post #317 of 454
USB stuff and DDC can't be directly compared because they work in different ways.

With USB, your DAC is completely in charge of timing. There is no clock signal for audio on USB, and in fact it's a two way communication, the DAC itself asks the computer/source for more data when it needs it, and converts according to the timing dictated by its internal clock.
For this reason, with USB the only thing you need to worry about is noise, which an Intona completely addresses in that it both blocks all noise from the source, AND is incredibly low noise in and of itself.
USB 'reclockers' simply cannot provide any improvement in jitter/timing, it does not work that way and no one has ever demonstrated an improvement in jitter via the use of a USB reclocker.

DDCs that output I2S/AES/SPDIF are different because whilst they also seek to provide a very low noise source, they DO provide a clock signal to the DAC. This clock signal is what the DAC uses to determine when to convert samples and so the timing accuracy of the clock signal provided by the DDC measurably affects the performance of the DAC even though the data itself has not changed.

As to whether a DDC or using USB is better, it depends. A few years ago I would have said definitely a DDC, as many DACs simply didn't have very good USB implementations/internal clocking and so there were quite significant improvements to be had by using a high performance I2S DDC instead.

Nowadays though, many DACs even at lower pricepoints have exceptional jitter performance and it's a lot rarer that a DDC will actually improve things. In fact in many cases they may make it slightly worse. Though most good DDCs do also provide galvanic isolation so again blocking noise from the source is a benefit.

My default recommendation nowadays would be to get an intona, they make a cheaper USB 2.0 only variant as well, though the 7055-C is the lowest noise one if you want the best performance (also looks the best IMO).

thanks for the explanation. But then, products like innuos phoenix USB or from SotM USB ultra are rip-offs basically? More importantly and on topic, have you tried the intona with DAC200 specifically, and if so could your hear/measure any differences?
Im streaming from sotm or rendu to dac200 or holo spring 3 kte and wonder if the intona would improve things, if both these have vey good USB already.
I have the innuos Net and almost pulled the trigger on the innuos USB a couple times already, people usually tend to like it. Way back when my system was more simple, without any streamer, I used the entry level uptone usb regen from laptop to dac and heard zero difference at all, i understand th intona 3 is a different animal of course
 
Nov 24, 2023 at 11:13 AM Post #319 of 454
Hey - it’s a discussion - we are all learning here and my statement obviously wasn’t clear enough and what you inferred was probably what others did too - so your clarifications help me and everyone improve! You weren’t wrong - you just inferred something which my statement implied to you and likely others - so my statement wasn’t well crafted for which your comment was fair and appropriate and appreciated!
So, it seems none of us was right anyway :wink:
Thank you again for enlightening us, @GoldenOne
 
Nov 24, 2023 at 11:41 AM Post #320 of 454
thanks for the explanation. But then, products like innuos phoenix USB or from SotM USB ultra are rip-offs basically? More importantly and on topic, have you tried the intona with DAC200 specifically, and if so could your hear/measure any differences?
Im streaming from sotm or rendu to dac200 or holo spring 3 kte and wonder if the intona would improve things, if both these have vey good USB already.
I have the innuos Net and almost pulled the trigger on the innuos USB a couple times already, people usually tend to like it. Way back when my system was more simple, without any streamer, I used the entry level uptone usb regen from laptop to dac and heard zero difference at all, i understand th intona 3 is a different animal of course
It's not that they're rip offs, just that only part of what they're doing has any real basis in terms of theory or evidence for it.
Noise from a USB source can absolutely have a negative impact on the connected device, that's 100% true, and presumably the Sotm and innuos devices are very low noise sources.

BUT, USB 'reclocking' actually just means a USB repeater basically. That can help if you want to use a longer run, but isn't going to improve jitter performance or performance of your DAC. And as mentioned previously no one has ever shown any evidence that their reclocker product produced an improvement here, because quite frankly it just isn't how it works because of the buffered, asynchronous nature of USB.

Here's a bit of a physical analogy:

You are picking up balls from a bucket next to you and throwing them at a target. You throw one every 10 seconds exactly, going by your own watch.
If the bucket starts getting low, you ask the ball guy to bring some more balls and put them in your bucket. It doesn't actually matter whether he brings them one by one exactly as you throw them, whether he brings them in batches of 5 at regular intervals, or whether he dumps 20 in, goes and checks his phone for as long as he feels like and then comes back to dump some more in before you run out again.
As long as the next ball is always in the bucket by the time you reach for it, the timing consistency of when that bucket is filled is entirely irrelevant and you'll be able to throw a ball EXACTLY once every 10 seconds according to your own watch.

This is pretty much how USB audio works. Everything is buffered and the DAC converts according to its own clock. The timing precision with which stuff is put into the buffer simply doesn't matter. The only thing that matters is whether the buffer has the next sample in it by the time the DAC needs it.

You CAN encounter situations called buffer underruns, where the buffer actually runs empty, it's not been filled in time. And this will be super obvious cause it just manifests as audible clicks/pops as the signal stops momentarily. But besides that, your DAC is 100% in control of timing.

As to the DAC200, the galvanic isolation seems to be working properly, I wasn't able to observe any differences beyond run to run variation when isolated vs connected directly to my PC. So I wouldn't worry too much about USB treatment products.

Additionally, in some instances, you can actually end up with instability or problems if you use a galvanic isolator on a product that is already galvanically isolated. It'd be a bad thing to use an intona with the holo may for instance as it then leaves the USB connection ungrounded at both ends and more susceptible to dropouts.
 
Nov 24, 2023 at 9:40 PM Post #321 of 454
Not quite.

The PCM1795 DAC uses Burr Brown's 'advance segment' architecture.
The simplest way to explain this is that the top 6 bits are converted 'as is', but the lower 18 bits are converted using a 1-bit delta sigma modulator that modulates to the level of 1 LSB (least significant bit) of the top portion.
Note that the 8 x oversampling and digital filter is prior to the Advance Segment Architecture. So some oversampling is done on the PCM1795 on all PCM signals less than 8 x 44.1KHz. ESS chips use alternative technology to segment architecture to achieve similar results - they have 64 parallel 1 bit SDM on a chip and sum them up to get required 65 levels. They also randomize the exact elements used at each sample in order to reduce noise from inexact matching elements. The ESS methodology is similar to PCM1795 just done in a different way. Basically 1 bit SDM are perfectly linear but have poor S/N while 24 bit R-2R ladders are optimized for excellent S/N but suffer from very poor linearity - so most chips today are of the hybrid variety.

I maintain that all oversampling done on ALL modern chips is flawed as perusal of any chip documentation will show a problematic sinusoidal ripple (directly related to sample freq) in the pass band from the low latency simple upsampling filters used - below is the example for the PCM1795 chip. And that ripple produces the pre-echoes which causes the sound to be more edgy, defined or have harshness/glare ( main complaint analog listeners have against digital). Feed a DSD tru 1 bit converter high quality pc-processed upsampling to DSD256 and you eliminate the equiripple and the tiny but audible time domain echoes completely - resulting in a smoother sound and maximizing soundstage.
 

Attachments

  • IMG_5453.jpeg
    IMG_5453.jpeg
    109.8 KB · Views: 0
Last edited:
Nov 25, 2023 at 2:05 AM Post #322 of 454
@GoldenOne

Thanks for the measurements of the DAC200, extremely interesting results.

Now since I know you love Holo dacs, I’m dying to know how DAC200 compares to Holo Spring 3/May in your opinion? Were you able to compare them side by side? T+A and Holo are probably the two top designs used in combination with HQP so it’s always interesting to have a new view to their differences.
 
Nov 25, 2023 at 6:49 AM Post #323 of 454
Just a reminder.
I cannot recommend enough to everyone to complete their research on Audiophilestyle, where OE333 (The architect of DAC200 and the full T+A 200 series) is interacting constructively with @GoldenOne about his report. It is also enlightening about the commitment of this architect, and comforting about DAC200 serious bangs for bucks.
 
Nov 25, 2023 at 9:55 AM Post #324 of 454
To Anyone who has the Ha200 or DAC200 with HDMI Module, Is there any Server with i2S that works with that?

To my knowledge the T+A has no i2S but "just" HDMI and would only work with HDMI from a Source like Playstation, DVD Player etc
 
Nov 25, 2023 at 10:10 AM Post #325 of 454
@GoldenOne

Thanks for the measurements of the DAC200, extremely interesting results.

Now since I know you love Holo dacs, I’m dying to know how DAC200 compares to Holo Spring 3/May in your opinion? Were you able to compare them side by side? T+A and Holo are probably the two top designs used in combination with HQP so it’s always interesting to have a new view to their differences.
For DSD specifically, honestly they're so close that it's tough to tell them apart. Mayyybe the May had slightly better stage depth but beyond that they were near identical. That's somewhat to be expected, with a true 1-bit converter as long as the DAC doesn't have any clear hardware limitations (and both of these are very good so that's definitely the case), then the biggest factor in performance is the modulator you're using. Feed two good 1-bit DACs with the exact same data and they're going to be a hell of a lot closer to eachother than any two PCM DACs.

I'd say you've got a hair better SQ with the May but....so close you literally have to AB and listen intently to tell the difference, but then you have analog vol control with the DAC200 so that alone might be preferable.

For PCM (feeding both HQP upsampled PCM / identical data so just comparing the hardware itself) I definitely preferred the May. The DAC200 PCM section is good but to be honest I don't think it's worth buying for that. I'd happily take a Wandla for example over the DAC200 if planning to use it primarily with PCM.

Comparing un-altered PCM, it basically just depends if you like NOS or not, rather than the DACs themselves.
 
Nov 25, 2023 at 11:46 AM Post #326 of 454
For DSD specifically, honestly they're so close that it's tough to tell them apart. Mayyybe the May had slightly better stage depth but beyond that they were near identical. That's somewhat to be expected, with a true 1-bit converter as long as the DAC doesn't have any clear hardware limitations (and both of these are very good so that's definitely the case), then the biggest factor in performance is the modulator you're using. Feed two good 1-bit DACs with the exact same data and they're going to be a hell of a lot closer to eachother than any two PCM DACs.

I'd say you've got a hair better SQ with the May but....so close you literally have to AB and listen intently to tell the difference, but then you have analog vol control with the DAC200 so that alone might be preferable.

For PCM (feeding both HQP upsampled PCM / identical data so just comparing the hardware itself) I definitely preferred the May. The DAC200 PCM section is good but to be honest I don't think it's worth buying for that. I'd happily take a Wandla for example over the DAC200 if planning to use it primarily with PCM.

Comparing un-altered PCM, it basically just depends if you like NOS or not, rather than the DACs themselves.
So you feel that the Ferrum Wandla is superior to the DAC200 when using PCM primarily ? Is this with the Wandla connected to the Hypsos?
 
Nov 25, 2023 at 11:49 AM Post #327 of 454
So you feel that the Ferrum Wandla is superior to the DAC200 when using PCM primarily ? Is this with the Wandla connected to the Hypsos?
I do yes

For the Wandla specifically I don't find the Hypsos to make a huge improvement, it's a lot smaller than the upgrade when using it with the OOR.
If you happen to be using an OOR + HYPSOS already it's definitely worth getting the splitter so you can run both on the HYPSOS. But I wouldn't get a HYPSOS just for the Wandla
 
Nov 25, 2023 at 12:18 PM Post #328 of 454
I do yes

For the Wandla specifically I don't find the Hypsos to make a huge improvement, it's a lot smaller than the upgrade when using it with the OOR.
If you happen to be using an OOR + HYPSOS already it's definitely worth getting the splitter so you can run both on the HYPSOS. But I wouldn't get a HYPSOS just for the Wandla
So how is the build quality of the Ferrum Stack? I have not seen Ferrum units in real life, but I read that the build quality of the components could be better. From the pictures these unit don't look like much, but I am aware of course that one cannot judge the mechanical engineering quality by looking only at pictures.
So do you personally feel that Ferrum might have some issues in this regard?
(Maybe as German I like my components to have a great haptic feeling also) .
 
Nov 25, 2023 at 12:23 PM Post #329 of 454
So how is the build quality of the Ferrum Stack? I have not seen Ferrum units in real life, but I read that the build quality of the components could be better. From the pictures these unit don't look like much, but I am aware of course that one cannot judge the mechanical engineering quality by looking only at pictures.
So do you personally feel that Ferrum might have some issues in this regard?
(Maybe as German I like my components to have a great haptic feeling also) .
The engineering behind the Ferrum stuff is solid, they had an issue early on with the OOR which turned out to be because the way the gain switch was mounted by the automated manufacturing process was coming loose after some time, so they switched to manually soldering that part.

Other than that I'm not aware of any issues and they've been extremely responsive and welcoming to criticism, in fact both Ferrum and T+A are probably the two companies that so far have stood out to me the most in this regard.

A big thing with the Wandla is that the oversampling is significantly better than most DACs. They got Jussi Laako (the guy that developed HQPlayer) to write the filters and they have a fair bit of dedicated compute power for them too, so that's a big help.

T+A external build is better, but also more than twice the price so that's somewhat expected
 
Nov 25, 2023 at 1:25 PM Post #330 of 454
For DSD specifically, honestly they're so close that it's tough to tell them apart. Mayyybe the May had slightly better stage depth but beyond that they were near identical. .
.
Thank you for this. You saved many T+A DAC200 owners having to check out the May and vice-versa! Both DACs are obviously keepers - in fact the D200 may be my last DAC ever!

I always do rapid A to B testing for several weeks before drawing a conclusion (even though the D200 has an excellent Pre-amp I am keeping my amazing Benchmark LA4 pre-amp as I can always toggle A to B on anything new I try out).

For those who have not taken a serious look at upsampling everything to DSD256 on a PC - you are missing out no matter what DAC you own. And I suspect the next best thing to DSD256 is a good non upsampling ladder DAC.

@GoldenOne do you prefer May with DSD256 or the non-upsampled R-2R output?
 

Users who are viewing this thread

Back
Top