Sign in to follow this  
Master Mi

How to connect a headphone amp to an audio interface in the best possible way?

Recommended Posts

I know - it might be a funny question for the studio hardware experts out there.
But I want to go absolutely sure before buying some new equipment.

My intention is to buy a new audio interface with additional ports where I can connect an external high-end headphone amp for the best possible use of some high-impedance studio headphones (Beyerdynamic DT 880 Pro, 250 ohms) and some lower-impedance headphones (like my Sony MDR-7506, 63 ohms) at the same time (so, in the end a hardware solution with 2 separate volume controller units for 2 different headphones).

The headphone output of the audio interface should be used for the lower-impedance headphones (so, I don't want to connect the external headphone amp there).
And the headphone output(s) of the high-end headphone amp should be used for the high-impedance headphones.

I'm pretty sure I have to buy an audio interface with 2 additional line output ports and make a connection between both elements from there to the input ports of the external headphone amp.

1) Is it correct or do I have to choose an audio interface with 2 additional line input ports instead (so, that the signal goes the other way from the headphone amp to the audio interface?

If my first assumption is correct I'd go for the following studio equipment:

- audio interface >>> Roland Rubix 24 (4 line output ports, impedance balanced >>> guess it's to handle like normal balanced ports)

- headphone amp >>> Lake People G103-P (2 balanced XLR input ports)

For the connection between both elements I'd use 2 standard balanced male mono 6,3 mm jack to male mono XLR cables.

2) Is that okay or would you prefer some special cables for this purpose?


Edited by Master Mi

Share this post

Link to post
Share on other sites

Short concise answers first, and then I'll follow up with technical geeky mumbo jumbo.

1 - Technically, a line output from your audio interface should be able to drive many line inputs because they use Impedance Bridging, which relies on the core assumption that both the line output and line input are done properly. Though to do this does require have cables or a box that allows you to split the signal. However, if you have multiple outputs then it affords you more flexibility in how you can set things up. 

2 - I don't see any reason why not. Just well made cables I suppose. You know for cable longevity and reliability. 

All right time for the technical geeky mumbo jumbo if you want some further explanation on things. 

Lets talk about that headphone output on the Roland Rubix 24, which Roland actually gives a maximum power specification (well at least sorta kinda, but better than nothing). Anyway, so a headphone output is really just a mini power amplifier because at the end of the day all a pair of dynamic headphones (like the two you listed) are just tiny speakers. So, Roland specifies the output power as 20mW into 47 ohms. You can estimate the amount of SPL your headphones will have based on this. It isn't perfect and you could sit down and try and do fancy mathematics to get at a more concise answer, but this is just quick and dirty. However, simple cross multiplication should "ballpark it". So, for the Sony MDR-7506 headphones you can expect that about 27mW maximum will be on tap. This is calculated by the classic (20mW/47 ohms) = (X/63 ohms). From here you get (63 ohms * 20mW) / 47 ohms. The ohms cancel out and you're left with 26.8mW. At any rate the MDR-7506's have a specified sensitivity of 106dB/W/m. This would give a peak SPL in excess of 120dB SPL. So, I'd say there is probably plenty of room there. Next up are those Beyerdynamic DT 880 250 ohm headphones. So, these are rated at 96dB/W/m (not stated like that, but at any rate). So, doing the same trick as before the Roland unit should produce around 106mW into the DT 880s. This gives a peak SPL around 116-117dB. Now, these numbers are estimations on the maximum SPL that you can expect from the Roland and can be eaten up quite quickly by very big and powerful transients. So, unless you want more volume I've never really understood the point of headphone amps 99% of the time personally, but I digress. Most of the time headphone outputs simply use an inadequate device for the output stage without enough bulk energy storage (big capacitors on the power rails). My best guess is that the Roland is probably using an "okay" opamp as the output stage. I'd be willing to bet that the headphone output on the Roland is either a single JRC4558 or two JRC4558s that are paralleled up. I doubt that they're using anything better, and for a whole mess of reasons this part is basically barely & adequately passable part for the job, but again I digress.

This then I guess begs the question why would I say much about that? Well see the output is probably just an opamp anyway. This means that it'd have no problems driving any line input. It won't be balanced and it will probably be noisier than a dedicated line output given that it is being used as a power amp, BUT there is nothing "technically" wrong with using it that way. Not best practice of course. 

I also feel it necessary to say something about the impedance balanced output. Really the trick here is to understand what "impedance balanced" actually means. In a nuthsell this means that the cold lead or "- lead" is not actually driven by anything, but instead presents the same "impedance" to ground that the hot lead or "+ lead" presents. This will eliminate most things like ground loops and hum, but not much else, which for most things is an acceptable compromise in cost. The other common option is to drive both the hot and cold leads and this provides much better rejection of all sorts of things, but it is more costly. Why use two opamps, four bypass capacitors, and five resistors when you can use one opamp, two bypass capacitors, and six resistors. Takes less board space and costs less overall to do the single opamp. 

Hopefully, you found something useful in there.

Share this post

Link to post
Share on other sites

Haha, thanks for the in-depth guide. ^^

And yeah, I was really impressed of the obviously powerful headphone device of the Roland Rubix 24 as well.
40 mW (20 mW left + 20 mW right) into 47 ohms seems to be pretty much.

Even my already pretty powerful Steinberg UR 22 USB audio interface seems to have only an output level of around 12 mW (6 mW left + 6 mW right) into 40 ohms, compared to the Roland Rubix 24.

But yeah, in other words...
The set-up for the audio-interface+headphone-amp combination in my imagination would technically work well, right?

Share this post

Link to post
Share on other sites

It would totally work! If you're finding your DT 880s too quiet or perhaps maybe lacking then a proper dedicated headphone amp will be the ticket. Likewise with your MDR-7506s and moving to the Roland. With that said, if you're happy with the headphone output on the UR22 and you need to drive monitors as well then you can. However, it requires that you have the ability to "split" the line output signal. This can be done with something like this, and if you're handy with a soldering iron you can make them yourself :) The reason this works is because in transferring audio the thing of concern is the AC Voltage. So, as I mentioned in my first post audio systems operates on Impedance Bridging. The fundamental principle in action here is that your audio interface will have a very low output impedance, on the order of a few ohms or less, and the input to any given device will typically be on the order of at least 10 kilohms. This means that very little current actually flows into the receiving device and you're effectively measuring the voltage. Then since the devices receiving the AC Voltage are connected in parallel they will have the same voltage potential across them. It is actually quite an elegant solution honestly. 

Your upgrade plan has nothing "technically" wrong with it all. It is very much equipped for the intended purpose of giving you two headphone outputs with one of those outputs having tons more power than you do you now. Whether it sounds "better" is something only you can decide. If I may make one suggestion. Think about what you're going to need in the future in terms of your audio interface. If you want more ins and outs then buy it now. You may have plans in the future for an outboard synth or an outboard buss compressor and those will eat up valuable analog ins and outs. I'm not saying that the Rubix 24 is bad or the like, but it is a small interface without much actual I/O digital or analog. You may be working 100% ITB with maybe a need for a mic occasionally. For that it is perfectly equipped. I know personally I'd be too limited by it immediately. So, just think about what you actually want to do in the future and buy based off that ;)

I've got a fairly complicated setup so I've had to learn to deal with the sorts of issues that pop up. I've had to deal with ground loops, bad cables, not enough I/O, varying power voltages, etc . . .. This can be a complicated mess to sort out. So, if you ask me a question about the "technical details" I can probably explain it. However, when you start getting into subjective things like will it sound better or the like, well only really you can decide that. I mean there are some obvious things like if you want more detail in the bass region then you've got to do the whole room treatment things and get bigger/better speakers. There really is not an easy way around that sort of thing. However, if you want to know why your square waves don't look like square waves on an oscilloscope in the real world or connecting things up. I've got you covered.

Share this post

Link to post
Share on other sites

The main background of my whole intention to enlarge my studio equipment is basically that a guy from a music store told me that my 250 ohms Beyerdynamic DT 880 Pro headphones might not be driven properly on my Steinberg UR 22 USB audio interface.
He said that some frequencies (I think he named especially the top end frequencies und deeper bass frequencies) of these headphones might perish or won't sound as clean and truthful at my current audio interface - so, I should rather go for a good headphone amp.

Another customer in a product rating corner mentioned that the Beyerdynamic DT 880 Pro with the full frequency response range from 5 to 35000 Hz sound much better and cleaner on a good headphone amp.

And so, I looked for some higher end stuff of headphone amps, got stuck to the Lake People G103-P (has a very large frequency response range from 5 to 150000 Hz!!! - so, much more than the Presonus HP4, for example - at which the frequency response range goes only from 20 to 20000 Hz).

But I recognized that I would need more line outputs on my audio interface.
2 line output connections go already into my Fostex subwoofer - so, there 's nothing else left for my preferred headphone amp and I want to keep the heaphone output on my audio interface for the lower impendance headphones.

So, my decision was to buy a new audio interface first and sell my old one to a friend who wants to start with composition and music.

There are 2 other audio interfaces which are still in my mind and which have a good power support, lots of line input and line output ports and which have already 2 headphone output ports.

1) Tascam US-20x20 (also good ol' Japanese hightech stuff with USB 3.0 support!!!)

>>> Without the nasty stands (don't really know if you can demount them completely) which bring the interface obviously in a slightly skewed position, I could easily place my monitor screen on this pretty big interface.


2) Steinberg UR44 (the bigger brother of my momentary UR22 USB audio interface and highly established stuff, grown in a long-term German-Japanese development)

>>> I'm still thinking about getting this one, because with the Steinberg products I already know what I have and lots of people say that these are very good.
But even for this one I need to enhance my desktop with some further attachments to get enough place for my equipment.
I'm thinking about getting some attachments for my desktop whereon I place my Presonus Eris 3.5 studio monitor speakers and under which I can also place something (like the new audio interface on the left side and the Lake People headphone amp on the right side of the desktop).
Maybe I'll build some handcrafted stuff with the wood of my desktop design before I tend to buy some faltering stuff in a furniture store.

But, yeah - with the Roland Rubix 24 (which is also a high quality Japanese studio gadget, I guess - and I kinda love its highly stylish design, by the way) I could just concern about the audio stuff without caring too much about my desktop.
But there still some people who say that the Roland Rubix audio interfaces have a pretty prehistoric audio driver settings design where you should choose between levels with numbers from 1 to 7 or things like that... instead of choosing between spezific buffer size settings with the usual values.

Guess I'll have to think about this topic a bit longer before making too fast decisions.

Share this post

Link to post
Share on other sites

Well, that guy isn't "technically" wrong but there was a lot glossed over in that simplified explanation. So, now frequency response is just one specification of an amplifier that you should pay attention and really it is one of the most useless for actually telling you much about it. Beyond telling you what frequency ranges the amplifier will amplify it also implies the phase margin of the amplifier. Essentially, the amplifier has a more phase aligned output throughout the actual audible band in "theory".*

As I said before you don't "technically" need additional line outputs to use a headphone amplifier. There are disadvantages to using one set of line outputs, but it isn't necessary. If you're using the built in volume control on your audio interface then you've got two stages where you can introduce either gain or attenuation before you actually hear it. Right or wrong really is a semantics kind of thing on paper. In practice the fewer things in your monitoring signal path the better because there are less things to get in the way of what you're hearing. 

I know before I bought my current audio interface, RME FireFace UFX+, I considered a lot of audio interfaces, but I also had some different requirements than you did. However, I also don't use the audio interface for monitor control. That is done via an external unit in my case and I just pipe over an unattenuated signal to my monitor controller, which has a banging headphone amp lol. So, the Tascam is generally well liked for what it offers. The Steinberg stuff is always pretty highly liked too. I like Focusrite stuff too so something like the 6i6 or even 18i8 are also quite comparable. But another possibility is Presonus. There are lots of audio interfaces out there. You have to figure out your requirements to see what it is you actually need and may need in the future. I can't tell you what those are. You spell out your requirements and select an audio interface based on that. If for example you want to use Universal Audio plugins then you may actually consider getting something like the Universal Audio Arrow or splurging for an Apollo Twin mkII. I only brought this point up because I find it smarter to spend however much money you need at first to get what you actually need. 

* - Extra Geek talk. Now, I don't really need to read much into the specifications of the Lake People G103-P because I know what it is built around, a LM1876. That is the datasheet to the device at the heart of the G103-P. Now, this device is designed to drive 4-8 ohm speakers. However, because it is driving such high impedance loads the chip's performance will be much better in several aspects. If you look at the datasheet on page 10 for Figure 10 to Figure 15 you'll see a bunch of graphs that are titled THD + N vs Output Power (THD + N stands for Total Harmonic Distortion plus Noise). Now, I doubt you're going to be running this anywhere near 1W output power. More than likely you'll be running it around 50-75mW. So, you're going to want to pay attention to that first decade on each graph which represents 10mW to 100mW. It is also important to note that these are specified at particular loads and frequencies, so pay attention to what it says in the upper left of each graph. So, it looks like THD + N will probably be around 0.05%-ish, which is quite good actually. There is also Figure 19 on page 12 that specifies the amount of output power into a given load. It only goes 40 ohms, but it looks like it is starting to level out. The maximum output power is probably going to be ~4W. There are other important factors at play here that aren't specified often that you should always look for. These include the Slew Rate of the amplifier, which given that this is a power amplifier is quite high at 12V/us** (this is a worse case figure). One factor that is a bit lacking is the Crosstalk though, at a typical 80dB, but again should be much better because it is very unlikely that you're going to be driving the headphones with a nearly 11Vrms signal. That is just absurd. There are plenty of other things to keep in mind too. Figure 29 is a good one for showing how the phase of the amplifier varies, Figure 28 is also helpful for showing how good the amplifier is at rejecting common mode noise vs frequency. Like from this datasheet you can pretty much guess exactly how good this amplifier will be in terms of specifications ;) This is why I say specifications are important. Now, they could improve the CMRR of the LM1876 by using an external opamp and running the output of that into the LM1876's non-inverting input, but I can't really find much more on the G103-P other than it uses a LM1876. For further reading to better understand these specifications I highly suggest reading Rane's Note on Audio Specifications

So, why do I say the guy is "technically" correct? Well, because like most things in life there is a grain of truth to what he is saying, but it isn't the whole truth. See, by his logic then by extension higher impedance speakers need more powerful amplifiers, which isn't actually the case. See any power amplifier is really just trying to swing the output voltage to the voltage rails of the power supply. That is to say if you're running a 60V power rail then it can produce a voltage difference of 60V across the load, aka the speaker. So, then you may quite rightly ask, "Then why do power amplifiers need these massive heatsinks if all a power amplifier is doing is changing voltage?" Well, this is where Ohm's Law comes into play. In a nutshell Ohm's Law says that if you have resistance and there is a voltage difference applied across that resistance then current must flow as a consequence. See, you know the resistance of the speaker, which say for the sake of simplicity and the like is 8 ohms, then that 60V across that 8 ohms requires a certain amount of current to flow. According to Ohm's Law, the Current (I) is equal to the Voltage (V) divided by the Resistance (R) or I = V/R, I = 60V/8 ohms = I = 7.5 amps. That is why an amplifier needs all that heatsinking, because it has to be able to source A LOT of current as a consequence of Ohm's Law. Now, the main issue with many power amplifier is that they actually become current limited. This means that in order to sustain that 60V across the load the power amplifier can no longer source enough current to maintain the relationship between Voltage and Current as required by Ohm's Law. Headphones must obey this as well, and why I originally said that the problem with most headphone outs is that the output device is simply inadequate to drive the desired load. Looking at a common audio opamp, such as the NE5532, there is a specification called Output short-circuit current labeled as Ios. This spells out the ABSOLUTE maximum amount of current that the NE5532 can deliver. There are three figures listed for it, which are "Min", "Typ", and "Max". Assuming the "Typ" or Typical value of 38mA along with the Maximum peak-to-peak output-voltage swing (pretty self explanatory, but says the maximum voltage difference the opamp can produce) you can actually figure out the maximum amount of resistance that the NE5532 can drive using Ohm's Law. In this case R=V/I = R= 24V/0.038A = 631 ohms. Further, since we know both the voltage and the current, we can also estimate the power, which is about 912mW (basic estimates are easy as P = IxE, where E is actually Voltage but is based on slightly different terminology). So, for your DT 880s, a NE5532 could drive them quite well :) The problem is most headphone outs don't use a NE5532 because while it is a cheap opamp, it costs more than a NJM4558, which is often used. In 10,000 quantity the NE5532 is about $0.29 vs the NJM4558 at $0.10. Why spend an extra 19 cents, when you don't have to when the output is just "adequate"? The simple answer is they don't.

** - Super duper extra Geek Talk

Slew Rate is an interesting specification. It is basically how fast the amplifier can change the voltage or current in certain situations. So, for the LM1876 that is 12 volts per microsecond or V/us or for fancy folks V/μs. This means that if the LM1876 is at say 0V and all of a sudden it needs to jump to +12V then it will take it 1 us to do that. Now, say the LM1876 needs to move to -12V then it will take 2 us for it to move to that. 1 us to move to 0V and 1 us to move to -12V. This specification fundamentally limits the maximum frequency that amplifier can reproduce in a linear fashion. Assuming that the LM1876 is running on +/-15V (a common value for audio circuits and I'd be surprised if the G103-P is running much higher or lower than this) then that means the maximum peak voltage the LM1876 needs to be able to swing is 30Vpk. So, if it is quoted to be able to reproduce 150,000Hz then to do so in a linear fashion requires that the LM1876 have a Slew Rate of 28.278V/us. Now, the LM1876 could very be able to operate at that kind of slew rate depending on the load that it is driving. I'm just using 12V/us as a worst case scenario kind of thing. There also isn't a graph to show how the Slew Rate changes vs anything so I couldn't even hazard a guess at how it is performing. However, that worst case figure gives an upper bound of frequency right around 63,700Hz anyway. But it still pails in comparison to a lot of other parts. The NE5532 for example is about 9V/us, the TL07X series have 13V/us, and high performance opamps like the OPAx134 series are 20V/us or OPA161x is 27V/us. Also, super high slew rates aren't necessarily a good thing either as they can lead to opamp instability and may require frequency compensation to stabilize the opamp, typically in the form of a compensation capacitor. A classic example of this is with the NE5534 which requires a compensation capacitor whenever the gain is below three, typically a 22pF. And even then I'm not going to enter into the topic of Gain-bandwidth product or GBW and how it is also a factor to consider with amplifier stability and how all of this comes together to help explain whether an amplifier is actually stable. Definitely, a lot of good reading on the topic if you want to explore more into it though.

Edited by APZX
Fixing teh English because I is are dumb plus some better clarification on things.

Share this post

Link to post
Share on other sites

Haha, thanks, man. :DD

I guess that was more detailed stuff than I could technically digest at the moment. xD

But if you 're interested >>> I already went for the ultra-nice headphone-mixing-audio-combo:

- Steinberg UR 44 (audio interface)
- Lake People G109-P (headphone amp)
- Beyerdynamic DT 880 Pro (headphones)

Be sure, this combo might blow even your meticulously analytic mind away with a suprising smile.
But I'll start a new thread for this topic because it will contain much more information about the headphones and headphone amp stuff.

And thanks again for your great ambition to teach me about all those things which hold the world of audio technology together in its inmost folds. ))

Share this post

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.


×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Sign in to follow this