Teac UD501 DAC Review and Measurements

Content and analysis made possible thanks to support from https://headphones.com, and https://patreon.com/goldensound supporters.

ChitChat:

The Teac UD501 is a now discontinued device, but one that retains something of a cult following in some online groups, and so I wanted to provide a brief review and some measurements of it as there are not many available. (Archimago has some here) . It does have a few design aspects which set it apart from other options, and so regardless of if you’re interested in buying this DAC or not, you might find it interesting to hear a bit about it.

The UD501 is a DAC utilising two burr-brown PCM1795 chips, and an output stage based on the MUSES8920 opamp. It features true 1-bit DSD support (unlike many modern DACs which convert to PCM internally) and also has an upsampler separate to the DAC chip itself, as well as the interesting option to ‘bypass’ the DAC chip’s internal oversampling filters (mostly) and run it in something akin to NOS on an R2R DAC. More on this later.
It also has some rather nice design aspects such as a dual-mono linear power supply to power the left and right channels independently. Not to mention a chassis build quality and aesthetic that still feels premium today.

But the question is:

Is this a hidden treasure worth seeking out? Or a product made redundant by modern options?

Review:

Note: This DAC was loaned to me for review by my friend ProfessorP3PP3R. I have also owned a UD501 in the past. My subjective impressions were formed BEFORE measuring, and in fact I owned the DAC long before I acquired the APx555 analyzer.

Build:

I really like the external build of the UD501. The somewhat retro but clean and tasteful aesthetic, combined with high quality machining and overall construction makes this a device which strikes a perfect balance between being subtle in appearance whilst still looking great.
The tactile response of the switch and control knobs is excellent, and honestly I could not really think of much to critique. It looks great, feels great, and even the physical size is a nice balance between practicality and being satisfyingly substantial.

Internally however, things are a little bit of a mixed bag. It is a mess of wires, connectors and crowded PCBs, with some design concerns such as having electrolytic capacitors practically touching heatsinks in some areas. It does not particularly scream ‘quality’ and I’d be concerned about longevity.

Features:

The UD501 offers an inbuilt headphone amp which is a nice feature to have, though it’s not particularly powerful and also does not sound as good as even many ‘budget’ dedicated amps. So it’s a nice feature if this unit is going to be in your 2ch system but you’d like to use headphones now and then.
However for dedicated headphone use I’d look for something better.

In terms of control, you have a variety of inputs and outputs which are all selectable. The DAC can output to XLR, RCA, or both simultaneously, and also features multiple coax and multiple optical inputs so that you won’t have any issue feeding this DAC from multiple sources, streamers, CD transports or DDCs without having to fumble about changing cables all the time.

The UD501 features selectable digital filters for PCM, and selectable analog filters for DSD. Allowing you to tune the sound slightly to taste. This is practically universal on modern DACs, but was not always a common sight.

And lastly, it has the option to apply digital oversampling prior to the DAC itself, which provides slightly better attenuation before the nyquist frequency than the PCM1795 internal filter does, however it comes at the cost of being minimum phase which is objectionable to many due to the phase shift it introduces.

Sound:

This DAC is a little tricky to describe in terms of sound due to how many configuration options there are which might change it.

The biggest change that the user can make is turning on the internal upsampler. I personally did not like the sound that resulted from this as I found it made many elements sound harsh and aggressive. Timbre was etchy, and instruments such as cymbals had notably more ‘glare’ than when the upsampler was turned off.
It turns out that there may be some objective explanations for this which will be discussed in the measurement section below. But the summary is that with the upsampler turned on, I did not enjoy the UD501 at all.

Because of this, I kept the upsampler turned off for all of my evaluation.

The second biggest change is that the user can change the digital filter between ‘Sharp’, ‘Slow’, and ‘Off’. With sharp and slow being similar to the filter options available on most DACs, and the ‘Off’ option being a sort of ‘semi-NOS’ operation.

I personally found myself switching between ‘Sharp’ and ‘Slow’ for most of my listening, with perhaps 65% of my time spent on the ‘Slow’ option.

With the digital filter set to ‘Slow’ or ‘Sharp’, the UD501 is quite a warm sounding DAC. Its biggest strength lies in the timbral ‘richness’ that many people are fond of Burr Brown chips in general for displaying, which gives both vocals and instruments a sense of weight, density and presence that some more clinical DACs can often lack.
It is not a strictly speaking accurate timbre, as instruments do sound denser and warmer than they do in real life, however it is a signature that will appeal to many listeners and provides a nice flavour option for those not interested in the more clinical sound offered by many budget DACs.

However this is somewhat held back by the fact that this DAC is not a particularly resolving one. Detail retrieval on this DAC is simply ‘ok’, with other price competitive options beating it quite comfortably.
This is not really helped by changing to the ‘Sharp’ filter. This slightly reduces the overall warmth of the DAC, but also makes it a bit more prone to harshness or ‘digital glare’.
I found that for overall softer tracks, the Sharp filter was better, but for most content, the Slow filter held back the treble aggression without changing things too much.

This DAC also has quite an interesting soundstage. It’s more expansive than some other DACs I’ve heard at this pricepoint, however it is also a fair bit more diffused sounding. With musical elements coming from a general area or direction rather than a precise point in space.

This can be quite enjoyable for certain orchestral and acoustic tracks, but in busier rock, pop and electronic tracks it leads to the overall sound lacking coherency and definition.

When the DAC is set to DF=OFF, it becomes too warm, veiled, and closed in for my taste. It removes almost all of the remaining harshness, but at far too great a cost to just about all other areas. Music is softer, lacks resolution, and with a spatial presentation that one could almost call ‘smeared’.

Summary:

So overall, this DAC has a few configurations that I personally would avoid. The upsampler, and DF=OFF options make the UD501 sound quite unimpressive.
However with the upsampler off, and DF=SHARP or DF=SLOW, it presents a warm, laid back, rich sound that will be exactly what many listeners will be wanting particularly if their playback chain could use some added tonal density and warmth, but one that will also leave many listeners wanting more resolution, more macrodynamic impact, and a more coherent soundstage. This isn’t a ‘bad’ DAC, and for $250 or so it’s an interesting option, but I would not pay more than that for it.

I should mention however, that the combination of HQPlayer’s DSD upsampling with this DAC’s true DSD capability, makes for a much more enjoyable listen than the DAC in stock form.
Doing this bypasses both the internal oversampling and delta sigma modulator, and replaces it with HQPlayer’s higher performance options.
This resulted in a clearer, better defined, more convincing and overall improved sound.

But the big issue for me, is that this DAC sells for around $250 on the used market, and when excellent options like the SMSL SU9n exist for the same money, and in my opinion sound significantly better, it’s hard to recommend this other than if you are explicitly looking for a more laid back sound.

Measurements:

Test Setup:

– Audio Precision APx555 B-Series Analyzer with 200kOhm input impedance set unless otherwise specified
– Neurochrome Headphone Dummy Load
– USB Source: Intel PC via ifi iGalvanic 3.0 and iUSB 3.0
– Measurement setup and device under test are running on an AudioQuest Niagara 5000 power supply
– Audioquest Mackenzie XLR and RCA interconnects
– Japanese UD501 driver (though I did check and found no difference at all with driver used)
– Intona Reference Impedance Characterized USB Cable
– All measurements shown in this post are taken using the Sharp filter with upsampling off unless otherwise specified
– All measurements shown are using the balanced output unless otherwise specified
– Full reports containing additional data and test configurations are attached
– Exact analyzer/filter configurations for each measurement are detailed in the full reports

Full Measurement Reports:

Reports available here:

Dynamic Range (AES17): 115.3dB

SNR: 115.8dB

Noise Level RMS (20-20khz): 6.65uVrms

Noise Level RMS (20-90khz): 606uVrms

DC Offset: 258uV active, 248uV idle

1khz 0dBFS Balanced Out:

Left channel has a fair bit more mains leakage and harmonic distortion than the right. Possibly due to physical proximity to the power supply. The board is also asymmetrical in design so there could be various reasons for this. Regardless, not ideal.

1khz 0dBFS RCA Out:

RCA performance is quite close to the XLR output which is good, however, it does seem that it is quite sensitive or picky with grounding or power.
Previous tests I did on this specific unit showed quite problematic behaviour on the RCA out whilst the XLR out was fine:

This measurement was from this same unit of the UD501, but on an ADI-2 Pro FS R and at a different location.
Whatever caused the issue with RCA has not repeated itself here, but I can only assume it was a grounding or power related, but it did not show up on any other devices I tested on the ADI-2 Pro so was specific to the UD501. I additionally encountered much worse crosstalk previously.
I am currently running the UD501 and the measurement setup on an Audioquest Niagara 5000 power conditioner with the setup properly grounded, but note that the UD501 may not be ideal in systems with a floating ground or isolation transformer.

1khz Headphone Amp Out (700mV):

Quite a bit of mains leakage on the HP out, but distortion itself is mostly ok.

1khz Headphone Amp Out (Max output):

The amplifier does indeed reach and in fact slightly exceed the specced 100mW with a 32 ohm load.

1khz Scope view:

-90.31dBFS 1khz sine (96khz capture bandwidth):

Messy due to high levels of high-frequency noise.

Filter Ultrasonic Attenuation:

Idle Noise FFT:

THD+N vs Frequency:

IMD:

Linearity:

Crosstalk:

Jitter:

Jitter is ok, but not great, and certainly not to the same quality as many modern DACs. Additionally the SPDIF coax and optical inputs show quite poor performance:

Multitone:

Wideband:

Quite a significant amount of ultrasonic noise regardless of DAC configuration or activity.

Measurement summary:

Overall, the UD501 measures ok. Not great, not bad, but certainly not upto the standard of many options available today and also has a few physical design elements which would be considered questionable.

For clarification: I’m not so concerned about THD itself for example, plenty of products with higher THD or other measured characteristics that are in theory ‘sub-optimal’ sound awesome. The Schiit Yggdrasil has some funky THD+N vs level behaviour and doesn’t measure amazingly, but it IS consistent and both channels behave the same. What concerns me with objective performance is when there is clipping, channel behaviour mismatches as seen here, significant power sensitivity concerns, or problems that simply shouldn’t exist like very high levels of ultrasonic noise. Some of these issues may be audible and some may not, but when there are quite a few of these issues present on a device it’s a little tough to take the view that this is a particularly high quality or well thought through design.

Further Discussion:

‘NOS’ on the UD501:

The UD501 has the ability to bypass the oversampling filter of the PCM1795 chips, and run in a mode that in many ways is similar to NOS on an R2R DAC, however not quite the same.

This is made possible by the physical ‘advanced segment’ architecture of the PCM1795. This chip has a 6-bit current-segment DAC which operates at high speed. Analog devices has a document discussing a few primary types of segmented DAC designs here if you’re interested. The design of this DAC is an interesting way of combining the benefits of both R2R (or other ‘native’ converters like a thermometer DAC) and Delta-Sigma.

First, the digital data is separated into the top 6 bits and lower 18 bits.
The six upper bits are then converted to Inverted Complementary Offset Binary (ICOB) code.
The lower eighteen bits are put through a delta-sigma modulator, with a 3-bit output that has a maximum value of 5 (not the normal max 3-bit integer of 7).
The 1-level of the modulator is equivalent to one least-significant-bit of the upper six bits.
This is then summed together (in the digital domain) with the upper 6-bits to produce a value with a maximum value of 66.

This is then processed by data-weighted-averaging to reduce the noise produced by element mismatch, and the final digital result (still with 66 possible values excluding 0), is then converted using the current segment converter.
It is not stated what particular type of current segment dac this is (it could be various approaches such as a thermometer (ring) dac or string dac), but that part is less important than how the data itself is prepared and what bit depth it is converted at in my opinion.

This also means you can SORT OF run the DAC in ‘NOS’, and many products like the ifi iDSD black label/signature or the UD501 allow you to disable the initial digital filter.
However this is not actually NOS, as the converter itself is not 16 bit, and can only convert at 66 different values. Meaning without any form of digital filtering, oversampling or delta sigma modulation you’d be limited to ridiculously low accuracy.
The DAC still does delta-sigma processing and does not convert the PCM information natively like a NOS R2R DAC would.

Running the DAC in this mode as you might expect does bring some problems.
For example the output differs from what you’d expect to see on either a NOS R2R DAC or a ‘Pseudo-NOS’ oversampling filter on a delta-sigma DAC like the ADI-2. This could potentially be due to the data weighted averaging still being applied, the delta sigma modulator being forced to run at a lower speed than intended, or simply the increased noise due to lack of noise shaping.

Because there is no longer any noise shaping being done as part of the initial oversampling, there is a significant increase in high frequency noise that reduces linearity.

Additionally, there is a sudden drop in SINAD at -24dB, implying that something around the 5th bit does not behave normally in this mode.
It’s possible that the advance segment DAC in the PCM1795 uses 5-bits for the top level converter, not 6 as stated in the PCM1792A’s description of the advance segment DAC architecture.

Upsampler:

The upsampler when enabled upsamples all audio to 192khz prior to handing it off to the DAC.
This provides slightly faster ultrasonic attenuation than the stock filter as shown below:

However it comes at the expense of being minimum phase, vs the phase linear stock filter.
This means that higher frequency content passes through the filter sooner than lower frequency content and introduces phase shift.
Whether linear or minimum phase filters are better is a matter of debate and there is plenty of excellent information available from resources such as Mans Rullgard’s post here.
I personally find that subjectively minimum phase filters sound more artificial and overall worse.

The other issue is that when the Upsampler is turned on, both IMD and THD start to rise around -15dB, whereas this does not happen with it off.

Jitter however sees a notable improvement:

Overall, I’d say to avoid using the Upsampler. If you want to do upsampling without the same problems seen here, use a tool such as HQPlayer.