TEAC NT 503 or UD-503: Opinions?

[Jay-dub]

I just received a new UD-503 in the mail. The first order of business is to confirm that it is in fact the US model. The sticker on the box says 120V~,60Hz,16W, and it came with a US power cord, but the seat for the power cord reads "INL-13- / 10A 250V~" and then after some logos "15A 250V", and above this is printed "Apparaten skall anslutas till jordat uttag," which is Swedish for "The appliance must be connected to a grounded outlet." Is this normal for a US model? I plugged it in, and it seems to be working fine.

[Jay-dub]

My first impression: PCM playback on this device isn't all that pleasant. I tried listening with and without 4x upconversion on the DSP chip (FIR Sharp as the digital filter setting for the AK4490), and I experienced the same problem – seemingly with ultrasonic noise – that I did with the Bifrost 4490.

 

The Teac advertises -3dB at 80kHz, while the Bifrost gives -1dB at 100kHz; I was optimistic that this difference in analog LPF would mean that the Teac would sound fine to me, but alas the difference is not that great. The AK4490 has two switched-capacitor filters built in, and the one that it uses for all PCM input has a cutoff of 150kHz. But if I upconvert to DSD upstream of the chip, a 50kHz lowpass becomes available. When I set the UD-503 upconversion to DSD and the DSD filter to Cutoff 50kHz, the output is listenable and not fatiguing. As @Miska has pointed out, however, the DSD upconversion is not great. In particular, it allows high levels of imaging around low multiples of the input sample rate. As a consequence, I am letting iTunes convert all output to at least 192kHz sampling in order to minimize that imaging. Subjectively, native rate > DSD upconversion gives tone colors that are slightly dulled, while software upconversion to 192kHz makes for much purer colors.

 

So, iTunes > 192kHz > UD-503 > DSD > 50kHz filter sounds very good to me, but it certainly looks like a convoluted signal path and suboptimal compared to upconversion in software to DSD256. I'm looking into reasonably priced solutions that don't require me to convert my library from ALAC. Roon + HQPlayer is a kind of expensive route to go for that last little bit of sound quality. For experimenting, there is Mansr's SoX fork, the output of which I can play on the TEAC player, but if I'm going to follow that route with a significant number of files, I'll need to buy a very quiet external hard drive.

[Jay-dub]

Hello,

Does the upsampling, digital filtering etc is done by the AK4490 chip or is it done by a DSP chip with TEAC algorithm ?

 

In summary I want to know if I can send a PCM signal upsampled and filtered with HQplayer and bypass everything until the delta-signal modulator ?

The "Digital filter" options select settings of the AK4490; the "Upconversion" options select a separate DSP. Whether that DSP is a solution implemented by TEAC or an off-the-shelf chip that they incorporated into the 503 isn't clear.

 

When you send PCM at 352/384 kHz, the PCM filters on the separate DSP are bypassed, as are those on the 4490. As long as you don't select DSD upconversion, the settings don't matter; the DSP will do nothing, and the 4490 will only do DS modulation followed by the 150kHz switched-capacitor.

 

I'm not completely sure what selecting digital filter off does when feeding lower-rate PCM to the unit, but somebody posted a spectrum from the UD-501 with digitial filter off, and it showed huge amounts of aliasing above the 22kHz Nyquist frequency of the input. As a result of that graph, my working hypothesis is that "digital filter off" performs sample-and-hold to 8x rate (352/384 kHz), maybe for the benefit of people who like the sound of an NOS DAC running at 44.1kHz. I avoid that setting, and I recommend you do the same.

 

I must ask, if you're using HQPlayer, why you want to use it to upsample to PCM instead of DSD. The latter option eliminates aliasing around 352/384 kHz and multiples, bypasses the modulator on the 4490 and gives the option of higher-performance DS modulators, and allows you to select between two switched-capacitor lowpass filters. There's no downside, if your computer can handle it.

[Jay-dub]

There are other possibilities. Maybe they considered ways of splitting the signal to two separate loads and decided (rightly or wrongly) that any of them would either compromise sound quality or introduce a vulnerability. Maybe an order came from the business office that this DAC was not to have signal-splitting functionality.

 

I must say I find your presence on this forum very strange. One post to advertise a streaming event, and then two years later you came back to necropost this topic (?!) You are, I presume, the Groot who has made a number of recordings for Pentatone. I'm a fan of the surround mixes but not always of the stereo versions. (Orchestral balances often seem a little off.) Welcome to the community, Erdo, but if you want to use this forum as a place to make announcements and pronouncements, I do recommend participating. Reply to some active threads, and get to know some of the more active members.