The outputs are balanced TRS. If you are using TS cables, I suspect that is your voltage drop.
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Focusrite Scarlett 2i2 Audio Interface Gen 3 Review
- Thread starter amirm
- Start date
AnalogSteph
Major Contributor
2.4 V sounds consistent with one half of the balanced output not being used (Amir wrote ~4.6 V balanced, consistent with the +15.5 dBu spec). I hope you didn't just stuff in some TS to RCA cables, the output driver may not particularly appreciate it long-term if one half is shorted to ground. I vaguely remember distortion going way up if you did an unbalanced loopback on the gen2, like -70 dB or something? Doing it right would require either soldering a cable up yourself or kludging together a bunch of adapters. Plan B, the headphone out isn't a bad source either.
Measurements for the Dragonfly Red indicate a dynamic range of about 112 dB, which is decent (the Scarlett is actually noisier!). At high levels we are seeing dominant 2nd order distortion making its way to -90 dB at 1 kHz and -80 dB in the SMPTE IMD test by the time you're reaching full-scale. As mediocre as that is in measured terms, you would be very hard-pressed to be hearing it - a high-frequency IMD test (e.g. 19 + 20 kHz) may be showing up a hint of difference tone at high level (80 dB down is 80 dB down after all!), but I wouldn't be holding my breath. Our hearing is quite tolerant of H2, besides in playback average levels tend to be quite a bit below 0 dBFS anyway.
I looked up a teardown of the device, which revealed an ESS ES9010K2M DAC and TPA6130A2 headphone amplifier. Most of the extra distortion is actually likely to be stemming from the latter, these little driver ICs are rarely doing too well at the extremes of the frequency spectrum in particular. The DAC itself is rated at a THD+N of -106 dB (and DR = 116 dB), which would be good by any standard.
Not sure what you were expecting, DACs perform at a high enough level that even those that don't measure all that great will generally still sound fine. Line-level operation further reduces dynamic range demands as levels can be dialed in freely.
Measurements for the Dragonfly Red indicate a dynamic range of about 112 dB, which is decent (the Scarlett is actually noisier!). At high levels we are seeing dominant 2nd order distortion making its way to -90 dB at 1 kHz and -80 dB in the SMPTE IMD test by the time you're reaching full-scale. As mediocre as that is in measured terms, you would be very hard-pressed to be hearing it - a high-frequency IMD test (e.g. 19 + 20 kHz) may be showing up a hint of difference tone at high level (80 dB down is 80 dB down after all!), but I wouldn't be holding my breath. Our hearing is quite tolerant of H2, besides in playback average levels tend to be quite a bit below 0 dBFS anyway.
I looked up a teardown of the device, which revealed an ESS ES9010K2M DAC and TPA6130A2 headphone amplifier. Most of the extra distortion is actually likely to be stemming from the latter, these little driver ICs are rarely doing too well at the extremes of the frequency spectrum in particular. The DAC itself is rated at a THD+N of -106 dB (and DR = 116 dB), which would be good by any standard.
Thanks for your informative answer.
i am in fact using RCA to TS, but the Scarlett manual specifically says it’s ok to use either TS or TRS, so I should be ok?
I am not sure what I was expecting (certainly not a night and day change) just looking for best performance based on the gear at hand (even if it’s theoretical performance and is hardly heard) while learning more about DACs and Amps.
I’ll have to lookup some of the concepts you have used (like unbalanced loop back) to understand better.
From what you said the TPA6130A2 might be the weakest link at high volume, so theoretically, am I better off dialing down the volume on the Scarlett and boosting the L30 a bit more?
i am in fact using RCA to TS, but the Scarlett manual specifically says it’s ok to use either TS or TRS, so I should be ok?
I am not sure what I was expecting (certainly not a night and day change) just looking for best performance based on the gear at hand (even if it’s theoretical performance and is hardly heard) while learning more about DACs and Amps.
I’ll have to lookup some of the concepts you have used (like unbalanced loop back) to understand better.
From what you said the TPA6130A2 might be the weakest link at high volume, so theoretically, am I better off dialing down the volume on the Scarlett and boosting the L30 a bit more?
If the manual says TS is OK on the outputs then you should be fine. TRS would give ground loop protection and more voltage which might improve things everywhere. The Scarlett will perform better at higher gain and the L30 will do better with higher inpit to a point before clipping.
Last edited:
AnalogSteph
Major Contributor
The important point about doing a loopback test with unbalanced cabling was that it would probably have done the same you have, i.e. TS plug in TRS output.
The Dragonfly Red, not the Scarlett. It has so much dynamic range that you can dial it down quite a bit before noise intrudes.
The Scarlett is basically noise-limited in the 1 kHz test. Not quite so in the SMPTE IMD test but it's still good.
That said, it appears that you may not even have to crank up the Scarlett all the way, I think the monitor control is an analog pot and the output stage has low enough noise to give you some leeway. I imagine this may improve high-frequency distortion performance a bit more. But since even performance at full tilt is still good, there's less to be gained than with the Dragonfly Red.
I have a 2i4 2nd gen on which i tried measuring line out Vrms with my little oscilloscope, but i cant get anywhere close to the voltage measured here, always landing around 1.28 Vrms.
I know its a bit different device, but maybe im doing something wrong. I used a 1khz 0dbfs sine for that straight from REW, but also changed the frequency without it doing anything. You guys have any idea what im doing wrong?
I know its a bit different device, but maybe im doing something wrong. I used a 1khz 0dbfs sine for that straight from REW, but also changed the frequency without it doing anything. You guys have any idea what im doing wrong?
restorer-john
Grand Contributor
Are you using a TRS 1/4" and connecting to the hot and the cold? It sounds like you are just getting half the voltage swing. That will happen with a 2 connection 1/4" jack- you will short on output to ground. (Actually the focusrite has 100R resistors to prevent damage from that)
I used RCA to 1/4, but also just measured ground to tip with the oscilloscope, so it was only one channel that i measured on the unbalanced line out. It also does look like a perfect sine wave on the scope. I mean, it is around the line level standard when i understand wikipedia right, but im just confused how amirm got such a hot signal out of the line out
restorer-john
Grand Contributor
That's why you are only getting half the voltage. Inside each 1/4" jack there are three connections (tip, ring and sleeve) Tip and ring are + and -. Sleeve is ground. This is the same for each of the two outputs on the 2i2.
You are just measuring tip to ground, with sleeve likely shorted to ground too.
When used for balance single channels (like the 2i2/4)
The other thing to bear in mind is your PC and your scope may be both sitting on the same common ground. By connecting the 2i2 to the PC, the output (1/4 TRS) ground connection would be the same. But Focusrite put a 100R approx resistor to common (input/USB gnd).
Measure between your hot and your cold on the 1/4" TRS (not the RCA) with a non grounded scope to see the correct max voltage.
Last edited:
AnalogSteph
Major Contributor
Maximum 0 dBFS output level on the RCA outputs for the 2i4 2nd gen is specified as +5.5 dBu (1.46 Vrms). For balanced TRS it's +10 dBu (2.45 Vrms), using the latter unbalanced is likely to yield 6 dB less or 1.225 Vrms. Sounds close enough to 1.28 to me.
I'm not an EE so I'm curious what folks think of this blog post by JDS Labs.
In short, they're addressing the issue of how to connect the balanced TRS line outs of a Scarlett 2i2 to the unbalanced RCA inputs of a headphone amp. (In this example, an Objective2.) Their proposed solution is to use 2 x TRS to RCA adapters (one per channel), connecting positive (+) and ignoring the inverting (-) signal. Is this a valid approach?
Here is their diagram ....
In short, they're addressing the issue of how to connect the balanced TRS line outs of a Scarlett 2i2 to the unbalanced RCA inputs of a headphone amp. (In this example, an Objective2.) Their proposed solution is to use 2 x TRS to RCA adapters (one per channel), connecting positive (+) and ignoring the inverting (-) signal. Is this a valid approach?
Here is their diagram ....
somebodyelse
Major Contributor
- Joined
- Dec 5, 2018
- Messages
- 3,739
- Likes
- 3,024
If you're trying to make do with readily available cables it's a reasonably safe approach - the unused RCA sockets don't expose the signal pins so there's no risk of accidentally connecting the cold signal to ground or anywhere else, which would be the risk if you used TRS to RCA Male. Most interfaces won't come to any harm if you use TS to RCA, which shorts the cold to ground, but performance may be compromised. If you can find or solder up a pair of TRS to RCA that don't use the ring you'd have the correct cables for the job.
restorer-john
Grand Contributor
Messing around with the 2i2v2 yesterday to find the sweet spot for lowest THD. 1M TRS to XLR cables made up specially- shield connection at send end only.
382.1mV out and no gain on the gain pots.
Just picked up a 2i2v3 at a price I couldn't resist, so I'll do a comparison.
382.1mV out and no gain on the gain pots.
Just picked up a 2i2v3 at a price I couldn't resist, so I'll do a comparison.
Dumb question: why not connect shield on both ends of a cable that has a connection for it on each end?
restorer-john
Grand Contributor
Not a dumb question at all.
The cable set pair is part of six cables I built yesterday specifically to allow various combinations of send and receive in both balanced (XLR) and unbalanced (RCA) to and from the 2i2v2 (or the other interfaces). As it has 6.35mm and XLR in as well as 6.35mm TRS out. I wanted the full swing to be available if I needed it at each end, but to also allow chassis reference shielding on the DUT if it was unearthed. (otherwise you essentially short the cold to earth (IIRC via an internal 100R res) and can get distortion)
Complicated, yes. But all my test gear sits on a common ground/safety earth. The 2i2v2 is also sitting at the same potential via the USB connection to the PC. Yes, I could isolate it, but the second I connect a 'scope or other piece of proper earthed test gear, I can kill half the swing. It's better to stick to a common ground and make the differential stuff to suit.
If I go portable (laptop on battery) the interface can 'float' with respect to the DUT, even if it is safety earthed or not. Makes a massive difference (>10-20dB) in the low end mains and harmonics.
Original 1980s NOS metal made in Japan RCAs. Down to my last few. Core/hot goes right to the tip and soldered at the very end. Best way to do it IMO.
Last edited:
telecode101
Member
- Joined
- Apr 9, 2021
- Messages
- 26
- Likes
- 19
Hi, just wanted to ask, has anyone been able to see any hardware breakdown of the Scarlett 8i6 model? Can they confirm the headphone and internal DAC is same as Solo or 2i2? I am assuming the internals are a little different as it has to provide much more output and input.
Gorgeous pro cables. I love the black rubber / nickel plate aesthetic. Cool RCA connectors.Not a dumb question at all.
The cable set pair is part of six cables I built yesterday specifically to allow various combinations of send and receive in both balanced (XLR) and unbalanced (RCA) to and from the 2i2v2 (or the other interfaces). As it has 6.35mm and XLR in as well as 6.35mm TRS out. I wanted the full swing to be available if I needed it at each end, but to also allow chassis reference shielding on the DUT if it was unearthed. (otherwise you essentially short the cold to earth (IIRC via an internal 100R res) and can get distortion)
Complicated, yes. But all my test gear sits on a common ground/safety earth. The 2i2v2 is also sitting at the same potential via the USB connection to the PC. Yes, I could isolate it, but the second I connect a 'scope or other piece of proper earthed test gear, I can kill half the swing. It's better to stick to a common ground and make the differential stuff to suit.
If I go portable (laptop on battery) the interface can 'float' with respect to the DUT, even if it is safety earthed or not. Makes a massive difference (>10-20dB) in the low end mains and harmonics.
View attachment 121924
Original 1980s NOS metal made in Japan RCAs. Down to my last few. Core/hot goes right to the tip and soldered at the very end. Best way to do it IMO.
View attachment 121925
Here are the input cables I made for mine. Mil-Spec Type E silver-plated cables; 550 paracord sleeve; Amphenol and Neutrik gold/black connectors.
Similar threads
