DirectStream Junior Dac


Note: Moderator please delete if appropriate or move to main page

I was thinking we should have a separate forum page just for the Junior?


Let’s give it a try!


Ok I had had mine for a few weeks hooked with an old DH Labs 75 ohm coax digital cable coming from a Oppo BDP 105 and the Junior just blows away the Oppo, everything is better, width, depth, air, vocals :slight_smile:

Anybody try different coax cables was thinking of picking up a wireworld here

Ultraviolet for the price worth a try on the other end I can get Audioquest Carbon at cost

Haven’t had a separate DAC since the Audio Alchemy days, remember them, but what a improvement the Jr is

Have a DMP on order but looking at that thread has me worried about operational issues


The DH Labs cables are good cables, the 110 which I own was considered a budget reference. I would go with the AQ Carbon, that would be a step up.

I use a Carbon USB that replaced a AQ Cinnamon. Different cable, but if I replaced my Music Metre Fidelus 75ohm digital cable the AQ Carbon would be the first cable I auditioned.


I guess this question is directed at Ted Smith, but I’d be interested to hear from others on this.

I’m putting my new DMP through its paces, initially through DSJ before I try it with my reference DAC. I’m assessing the connection methods, particularly I2S vs coax. I’m using the remote to flip between I2S input and coax input on the DSJ with the DMP playing redbook CD.

And this is my point. The transition between I2S and coax input is absolutely seamless, I can’t hear any perturbation in the music at the instant the selected input changes. I can’t tell one connection method apart sonically from the other. So much so that I’m inclined to believe the DSJ is not changing inputs at all, despite the display indicating it is. I’d put my money on the DSJ simply being source locked on I2S.

I have never experienced this before. On every musical switching device I am familiar with I have always been able to pick when the selected input changes (playing the same source, but connected in different ways). Firstly, with an ever apparent momentary perturbation in the music as the input changes, and secondly, different connection methods sound ever so slightly different.

If Ted could comment on this aspect, and maybe others could try the same experiment.

Given I can’t tell one connection method from the other, the next step is to send the DMP I2S to my reference DAC, and send its coax to the DSJ, and thus configure a comparison between the 2 DACs.

EDIT: Oops, I just discovered that the I2S input on my reference DAC has an RJ-45 connector. Is there such a thing as HDMI>RJ45 adapter?


Paul had the same reaction when he listened to the first DS prototype.

It’s expected behavior. In the DS and the DS Jr all inputs are “unpacked” to 24 bits and their sample rates, etc. measured in parallel. The input select then picks witch to pay attention to. In essence all inputs are locked all of the time if they have an input signal. As you switch inputs you are always seeing whole samples so there’s no glitch from partial samples. At most you might loose a sample or two (because of different bit delays in the unpacking algos) but you usually won’t hear that. Also, FWIW, the DS and DS Jr lock within a sample or two when an input starts sending bits.

When the inputs come from the same source all of the grounding issues are the same as you switch from one input to another so the sound is essentially the same: all inputs (including TOSLink if it’s bandwidth is high enough) are dejittered with the same algo and convey the same bits so the output doesn’t sound different (to a first approximation.)

You can prove that the DS (Jr) is paying attention to a given input by disconnecting other inputs while it’s playing.


So by virtue of the signal processing in the DS/DSJ tends to make them input agnostic?

The seamless switching of inputs in this way is new to me, it’s impossible to tell if a commanded input change has actually happened (when connected to the same source), short of seeing it on the display, because you just can’t hear it.

New feature request: have an option in the setup so that when changing input sources the output level can ramp down (say 10dB over 1 second) and then ramp up again to the output level that the new input was when it was last de-selected.

When changing input sources, does DS/DSJ recall the last volume level for that particular input. Or is volume a global setting common to all inputs?


The phase “signal processing” implies doing DSP or other processing. It’s simply that without changes related to jitter or grounding all inputs have to sound the same. By connecting multiple inputs at once from a given source to the DAC they all share the same grounding and the DS (Jr) does a good job of eliminating jitter so they have to sound the same…

Hearing changes without ramping makes A/B’ing a lot easier. To ramp DSD as well as PCM the ramping needs to happen near the output processing and at that place in he code there’s no record of which input was selected. Also, FWIW, I’m used to no ramping and it distinguishes the DS and DS Jr in that I don’t know of any other DAC that works this way (not that I’m familiar with all other DACs, but those that don’t do their own decoding in a processor would need separate hardware for each input and I don’t know of DACs that does that.)

Since you can’t tell which input is selected by ear, why does it matter which is being used? If you hear a difference pick the one you want. That’s why I almost always have my input select set to Auto, that way the DAC plays the last input that started, which is almost always the last one you hit play on.


brodericj - ramping sounds like a thing you would want while you’re “testing” stuff, but after that you wouldn’t want it.

Ted - are you saying that any differences we are hearing are due to grounding and jitter? e.g. - your coax sounds the same as I2S? Do you hear differences in HDMI cables to the I2S input? If so, why?


We need to be clear, I was talking about the specific situation of multiple interconnects between a single transport (or other source) and a single dac.

There are lots of things that affect jitter in a cable - how much it radiates/receives EMI, how fast the signal’s edges are, how solid of a ground, how bad any impedance discontinuities are, etc. But if you don’t care about jitter (like the DS), things are simpler: If the data gets thru uncorrupted it’s good enough.

Ignoring jitter: the differences we hear in digital cables are mostly due to how much noise they radiate/receive and how they affect ground loops in the system. The speed of the signal, edge rise times, cable materials, impedance discontinuities, shielding quality, connector quality, etc. all affect the radiation emission and reception, etc.

But when all of the cables in question come from one transport to one DAC in parallel and all of the cables are left connected when A/Bing - all of the electrical cable problems are combined together identically no matter which input you select - all that’s left are jitter differences.

In general when cables aren’t all connected in parallel between a source and a DS (where jitter doesn’t matter much) the other cable parameters matter the more. Here are the generalized (non jitter related) differences you might expect:

Optical interconnects avoid grounding / EMI issues so they are often the best sounding IF they have the bandwidth required for the signal in question. Often people don’t A/B them correctly - an optical cable has to be compared to a non-optical cable by disconnecting the non-optical cable(s) on both ends when listening to the optical cable.

I2S cables (of the HDMI variety) radiate little because HDMI is designed for high speed signals in a consumer environment. The high speed signals are balanced (and often done with twisted pairs.) They connectors and wires are fairly well impedance matched. There’s also pretty good shielding and connectors with a good shield/ground connections. (Tho I said we’re ignoring jitter, I2S has separate clock and data so the jitter is usually lower than AES/EBU or S/PDIF.)

AES/EBU has balanced signals and a separate ground so they are (all other things being equal) are next down on the pecking order from I2S.

S/PDIF is the worst design for multiple reasons: non-balanced signal, RCA connectors can’t be designed to match the speced 75 ohms impedance of the cable so there will be discontinuities that add jitter and radiation… (Jitter wise S/PDIF is typically the worst of AES/EBU, S/PDIF and I2S.)

When jitter matters optical can end up anywhere from best to worst depending on the properties of the cables in question and/or other system properties.


Thanks Ted. Very interesting indeed. Two quick (I hope) questions. Where does USB fall in the noise/jitter spectrum? And, when comparing cables, should the Ethernet cable connected to the Bridge be treated like other connectors and disconnected for the comparison or is it sufficiently isolated not to matter?


USB has inherent jitter sources that I2S, AES/EBU, S/PDIF, TOSLink don’t have. E.g. it forces a packetizing on the data that isn’t always convenient for a given sample rate. Still it also has more buffering, etc. that can be used to smooth this extra jitter back out. All in all (at least as far as the DS is concerned) it’s USB’s other failings (e.g. the unavoidable noise on the ground and VBUS lines) that seem to matter more in practice.

In a generalized sense the Ethernet cable can be the least intrusive electrical connection in the system, but that doesn’t mean it’s always benign.

As always, getting rid of as many variable as possible will make the things that are left the most audible, but you can be listening to differences that just don’t matter for typical use.

My advice is to listen to your system the way you normally would use it when comparing cables, e.g. when cable A is being listened to wire your system like you would if you had already chosen cable A (i.e. in all probability cable B wouldn’t be there.)


Thanks for all the info, Ted. Where does that leave USB with LANRover? From what I understand, most of the shortcomings of USB are attenuated with LANRover in terms of jitter and EMI/RFI. (I immediately noticed a difference with LANRover over USB alone.) In particular, compared to I2S and ethernet with bridge II while streaming from an optimized Mac Mini. I’m on the verge of getting the DS DAC but without the Bridge (for now). I understand that the LANRover benefits from a linear power supply, but is that really necessary if the power of the LANRover is fed by the P10? Sorry for all the questions, but these are the type of things that keep me up at night. Sad, but true!


Necessary? That’s a matter of personal preference. Switching supplies are harder to clean up than linear supplies. They generate noise at some frequency (say 100kHz) and then filtering is necessary to get rid of that noise. Linear supplies have to smooth out the bumps at 120Hz and they also have noise generated from the diodes switching to get rid of - that noise from switching diodes is higher frequency, but a much lower level than the noise from switching supplies and much easier to get rid of. Smoothing the bumps is usually taken care of by brute force - big caps… All of this is independent of where the supply is plugged in… Having a P10 will help to keep the supply from directly affecting the rest of your system but it doesn’t help attenuate most of the noise generated by the supply to what it feeds. All in all the LANRover can help USB quite a bit.

At the end of the day USB starts at a disadvantage compared to other inputs, but the degree of that disadvantage depends a lot on the rest of the system and the tweaks you apply. I maintain that you can get any input you have (or prefer) to sound great or at least good enough that something else in your system is the limiting factor :slight_smile:


amgradmd said: “I understand that the LANRover benefits from a linear power supply, but is that really necessary if the power of the LANRover is fed by the P10?”

My guess is that (as Ted alludes to) the wall wart switching supply is the culprit in that part of the chain, so a linear supply should help regardless. I got a $50 linear supply on eBay (which is perhaps 20 times the size, and is adjustable to a range of power needs) and this made an obvious difference in the noise floor (at least) in my USB chain. The only downside was it has a fan, and so I had to put it on the other side of a concrete wall in my basement, which conveniently had a hole in it, so I didn’t have to make any long cables.

I was a LANRover beta tester, but stayed with the similar device I had prior to the LANRover’s release, as it was more device chain/software agnostic (the LANRover didn’t work with my preferred setup).

badbeef said

I was a LANRover beta tester, but stayed with the similar device I had prior to the LANRover’s release, as it was more device chain/software agnostic (the LANRover didn’t work with my preferred setup).

Yes mine works fine with uRendu and Win 7 but not 10, anyone got theirs working or know of a fix ?


On the topic of linear power supplies – to what extent do they feed bad stuff back into the power line? I assume it’s less than with switching PSs but don’t really know. I have two LPSs, one for my Apple Airport Extreme and an HD-Plex that feeds four components. They are now plugged into a power strip (for surge protection) that is plugged into a household outlet (different from the dedicated line that feeds my audio components powered by a P5). I probably don’t need to isolate these further but am not sure.


magister, In case you haven’t seen this here is John Swenson’s, brains behind the Uptone LPS-1, thoughts on noise issues with power supplies.


Thank you, Roger – very interesting.


If interested/haven’t seen here are two of their CA threads on the LPS-1,�-linear-power-supply-1-launch-thread-29974/�-linear-power-supply-1-listening-impressions-thread-30172/

Im thinking of buying a second one for my LANRover.