Review of a behringer U-PHORIA UMC404HD

front viewrear view
Pictures shamelessing nicked from Behringer website

I bought this recently (May 2017) from Harrison Music in Adelaide, on a whim.  I was in their store buying something else, and it caught my fancy.  It would have cost more than buying it online, but I'm supporting a local business by doing so, and it's there for me to have straight away.  No waiting for a delivery that could come any time within the next fortnight, but not actually deliver because I wasn't there at the minute the courier turns up.  And I've got a shop to go back to, if I have some problems, and for future purchases where I'd actually like to see the goods before purchase.  So, still a win, from my point of view.

What it is

It's an USB audio and MIDI bi-directional interface.  That's the nutshell definition, here's a piece-by-piece description:

It has four independent mike/instrument/line balanced-input analogue to digital converters (ADC), using combo XLR & ¼″ TRS jacks.  While line versus mike is obviously about some rather extreme differences in signal levels, the instrument input mode is very high-impedance, allowing things like electric guitars to be plugged directly in without being muffled (though I haven't been able to check whether they, also, have a suitable amount of gain to do that job, yet, but it's about 20 db more sensitive than the line input, so I'd expect it to be).  The input controls are a line/instrument switch (which alters gain and impedance), a pad switch (providing 20 db of signal level reduction), and a variable gain pot.  Each input is an individual channel, it's four-channel mono.  Each input has a LED to show signal activity of some kind (lighting up at around −18.5 dBFS, a reasonable nominal level for sound capture), and another LED to indicate peaking.  It's not defined how that's detected (analogue input levels, digital clipping, etc) but my measurements show that it lights up about −3.5 dBFS, and there seem to be a row of comparators at the front of the main circuitboard near the jacks.

Despite what some people advise, my opinion is that you should set up your gains so that clip lights never come on.  Your nominal levels should be sufficiently down from full-scale that it should be near-impossible for any peaks to reach clipping level.  With today's equipment, it's feasible, and reasonable, to reserve 20 dB of headroom, and it should be ample to prevent clipping.  That is, the equipment is able to handle 20 dB of signal above normal levels, it's not wasted, it's reserved to handle those peaks without distortion.  If you have inadequate metering, monitor your loudest audio, turn the gain up until the clip warnings begin to light up, then back it off until they go off and stay off.  With a dynamic source, you can expect the green signal lights to blink along with your sound.  And with a more consistent sound source (e.g. an organ with continual tones, as opposed to percussive voices, they may stay on while notes are sounding).  Or, if you have decent metering on your software, set up for a sensible nominal level with a proper amount of headroom.  If you have some odd need to produce material much closer to full-scale, then you should do that as a post-production exercise, where you have good metering, good level controls (limiters, compressors, etc), and the ability to undo mistakes.

As I've said, all channels are independent, four in, four out, none of them have anything to do with each other.  If your software handles four channels, then you get to do whatever you like with all of them, separately.  It is just an interface, it is not a mixer.  Mixing is something that you'd do, with your software, as a separate process when you're editing.  You can record four audio sources, as four tracks, mix and pan them how you like.  You can play back four audio sources, how you like.

Plugging a tone generator into the ¼″ jacks, I've noticed that switching to instrument mode adds about 20 db of gain, switching the pad in removed about 20 db of gain, giving a 40 dB range, overall.  The pad switch also works on the XLR input, but the instrument switch has no effect on it.

Connecting 0 dBu tone into a line jack produces −18dBFS, with the gain all the way down and the pad out.  So if you're going to hit it with +4 dBu nominal audio with +18 dB strong peaks, then you're going to need to use a pad to stop it going over full-scale (+4 dBu is a very common industry standard output level, and 18 to 20 dB of headroom is standard for initial capture/recordings, you should only push things closer to the limit during the mastering phases of production).

And it's a four-channel line-out digital to analogue converter (DAC), with ¼″ TRS and RCA jacks for balanced and unbalanced connections.  They're not actually true balanced outputs, they're what's known as “impedance balanced” (one leg of the output is driven (tip), the other leg (ring) is terminated to ground with a resistance similar to output-driver's impedance).

I can't say that I like impedance-balanced connections.  If not done precisely (and it rarely is—a simple resistor to ground does not have the same characteristics as the output stage of the driven half of the allegedly balanced output), CMRR will be poor.  You don't have the isolation that a fully-floating (non-ground-referenced) connections will have.  You get no signal if someone has made an unbalanced connection by grounding pin two instead of pin three.  And some balanced-input equipment doesn't work too well when only one leg is driven (I have one mixer like that).  While some may say I'm being picky, having the best possible noise-rejection is important when working in electrically noisy environments, especially when you have leads going all over the place.

There's a set of separate XLR & ¼″ TRS jacks for analogue outputs that it calls the “main outputs,” but I'd call them monitor outputs.  They can monitor the input signals directly, or the first two DAC output channels, with a mix control between input versus output monitoring, and an overall level control.  The input monitoring can be heard as two stereo pairs (channels one and two being left and right, with channels three and four being another left and right), or a mono button sums them all together.

The headphone monitor has an A/B switch which can alternate between the main (monitor) outputs and listening to the outputs from the DACs on channels three and four.  The only value I see in that would be if you were recording extra audio while listening to playback from channels three and four (which might be a pain to arrange).

Being able to directly listen to the input sources means you can hear yourself without any delay.  Most systems have too much signal propagation delay to be able to listen to a signal going all the way through ADC, mixing, and back out to DAC.

What it calls its “main outputs” are what I'd call monitor outputs, it's what you'd listen to in your recording booth through speakers or headphones.  And what it calls its “playback outputs” are what I'd call main (or direct) outputs (they are the four DAC outputs from each channel, without any switching or level controls).

There are ¼″ TRS insert points on all four input channels (like you'd have on a mixing desk; tip is send, ring is return, sleeve is ground), letting you put effects gadgets between input pre-amp and the ADC (such as compressors, reverbs, etc).

There's MIDI in and out (no thru), but I haven't any 5-pin DIN MIDI gear to test it with.

hand-drawn diagram of behringer UMC404HD
Block diagram of UMC404HD interface (updated 17 Jan 2024)

The good

I've plugged mike and line levels into it, and it seems to do a nice clean job of capturing them.  Monitoring sounds quite good, likewise with the headphone output, but all the line output levels are quite low and there's substantial high-frequency noise on all monitoring outputs (I have to turn my monitor amplifier up higher than I normally do, more about that in the bad section).

The gain pots do appear to be what they say they are, adjusting the gain of the stage, rather than simply being input fader level controls (which they can't be used for, as they do not go all the way down to “off”).  It's an interface, not a mixer, you're expected to do that kind of control at the computer.

It can supply +48 volt dc phantom power to the XLR inputs, and I've measured that to be sure that it does actually supply that voltage (some equipment's phantom supply is far less, causing problems with some mikes).  And, it doesn't appear to add any noticeable noise to the inputs (I've got a Soundcraft mixer that often adds a hum to any input when the phantom supply is on—it obviously has poor phantom supply filtering and poor CMRR on its inputs).  As with most small to middle level audio equipment, it has one master switch for all input channels.  It tends to only be expensive high-end gear which have individual phantom power switches, or separate switches for small groups of inputs.  I've also checked that the phantom power isn't supplied to the ¼″ jacks (I've encountered one mixing desk that did, and it's a very bad idea), and since this unit used combo XLR & ¼″ TRS jacks, I wanted to be sure it didn't do that.

The device has a 5 volt DC power input socket (and comes with a wall-wart supply), yet doesn't appear to need it.  I've got the whole thing running off the one USB port on a friend's Mac computer.  Though I suppose that some computers aren't up to the task, or introduce unwanted noises (though various wall-wart supplies do that, too), and having an external supply option is a good idea.  When an external supply is connected, it cuts off the supply from the USB port, it's not using both at the same time.  Note that the pin-polarity is the opposite from the usual, so don't mix up your plug packs.

It claims to be “built like a tank,” which is often a requirement of audio equipment, considering how people treat them.  And it's nice to have equipment that's not covered with sharp edges and corners.

I plugged it into a PC running CentOS 6.9, and everything worked the way I wanted, straight away (no extra drivers needed).  (NB:  I haven't any MIDI equipment to test that aspect).  I have four independent inputs, and four independent outputs.  How those outputs are driven depends on what software you're using.  Some programs duplicate channels 1 & 2 onto channels 3 & 4, some will ignore channels 3 & 4, other programs give you four-channel audio.

I was going to try this on Fedora 25, but I've been having no end of trouble with that installation on my old laptop, so I'll have to follow that up, later on.

I've plugged this into a friend's Mac computer, and everything appears to work without needing any extra drivers installed.  I have four independent inputs, and four independent outputs.  How those outputs are driven depends on what software you're using.

I don't really use Windows, anymore, and only have an old laptop with Vista on it, so I don't think I can do a currently worthwhile test on it for Windows.  And can't, at the moment, because Windows is refusing to run the software I'd use (Audacity), though I did get it to play some music out through the interface, using already installed music player software.

On both computers, I used it with the free Audacity progam, and various music player programs. Although the hardware supposedly comes with an option to download a free full copy of some software from Traktion, and Behringer (say they) also have various other free downloads, I didn't get anywhere trying to locate that software, and I have no desire to get mired in the mess of software licensing, nor have to contend with Operating System and program incompatibilities.

The hardware is firmware upgradable, and although I've seen people comment that they could upgrade the old UMC404 with the UMC404HD firmware, Behringer has commented that they, also, improved the hardware, and the older hardware wasn't stable at the higher sampling rate.  So persue that experiment at your own risk.

The bad

You can only use the XLRs inputs for mike signals, and only use the ¼″ jacks for line, they're not interchangeable (though, it does have XLR and ¼″ line outputs).  While I can think of one person who prefers separate mike and line input connectors, I don't.  My first mixer let me use either connector for either purpose, the only differences between them was the impedance, and that only the XLRs can have phantom power.  And for me, being able to use the XLR for any source is the most convenient way to do things.  I use professional video equipment, and every input and output is XLR, no matter what the signal level is.  It is a reliable connector, and only needs one type of cable to connect anything together.  Here, only patchbays use phone jacks.  Having to keep an array of leads with different ends is a nuisance, and adaptors are unreliable, not to mention the extra strain they put on connectors with their extra bulk.  And phone jack cables are a pest if you have to join two together to make a longer lead.  Joiners are unreliable, and an extra thing that you need to keep a supply of.

Related to that is a pet peeve of mine:  Different signal levels on inputs and outputs.  It can be a right pain to patch together equipment when you find they have configurations like one of my mixers has with:  0 dBu maximum input levels, −6 dBu nominal channel insert levels, +4 dBu nominal output levels.  And then you want to connect it to something else which is all over the place, too, and with incompatible levels between the bits you're patching together.  When you want to plug in a compressor, for example, you end up over-driving or under-driving one or both of them.

Monitoring sounds good, though is seriously on the low side (on all of the unit's line outputs), I have to turn my amplifier up much higher than I do when monitoring the +4  dBu output of other gear (this is with all the computer's software volume controls set up at maximum).  Normally, professional equipment has a +4 dBu output when producing a nominal output level (which is somewhere around 20 dB under full scale—different manufacturers, and different producers, have their own preferences for how much headroom to keep/what nominal level to use).  This equipment just about makes it up to 0 dBu when producing audio at full scale (but that's not how you operate professional equipment).  So, to produce signals with normal amounts of headroom, it's output is going to be more than 20 dB lower than everything else.  If I was to try and connect this unit's output into any professional equipment (such as any analogue recorder), it'd be seriously under-driving it, to the point where it wouldn't do the job (professional balanced-input equipment usually has a nominal input level of +4 dBu, sometimes an input level of 0 dBu, but very rarely a −20 dBu input level).  I'd have to put an audio mixer, or pre-amp, between them.  Even non-professional equipment usually has a nominal input level of around −10 dBu, so they'd be noticeably under-driven, too.  Having said that, I think the main use of this will be putting signals into a computer, probably monitored with a domestic amplifier (which will cope with that low output level), and maybe sending back to an analogue mixer (which only rubbish ones, without input gain controls, should have any real problems with the low level).

Subsequent testing in 2024, after I received an email discussing noisy outputs, shows that mine does have an awful lot of hash on the output (around 200 mV) with the situation not being helped by the very-low audio signal levels.  It's looks like white noise on my analogue oscilloscope, and is ever-present, even when the device is not producing audio signals.  Digital scopes often have a lot of input dither noise that they create, themselves, making a mess of what you're trying to measure, so having an analogue scope for this kind of work is definitely an advantage.  It'd be far less significant problem if the audio signals were as strong as expected with prosumer and professional equipment.

I don't know if it's the DACs, themselves, that are noisy.  It could be the internal power supply is unclean (the noise is the same if the device is powered from the USB port, or the supplied wall-wart 5 volt DC supply).  There's a certain amount of noise on the supply rails and outputs related to the internal supply's switching frequency.  It's also on the input stages of the output drivers, so I'd say they're not causing it.  The supply noise could be getting magnified within the DAC, and they could add their own noise.  The outputs with level controls (headphones and main outputs) will go quiet when the level is turned down.  And input monitoring is quiet.  But all the DAC output channels are noisy, by the same amount, even when mute, so I suspect noisy power, or bad design.

If I were still using it I'd be more inclined to experiment with quieting its supply, but I've since moved onto a U-Phoria UMC1820 (I needed more input channels for some live music recordings).  It also has some noise on its outputs, but far less.

These devices being made from tiny surface mount components, and not having a schematic, makes experimenting difficult.  But scoping around the internal high-voltage supply, shows its switching frequencies are related to the output noise (dual channel scope, synced to the switching supply—I can see that the noise is harmonically related).  I think the supply should have better filtering, and should have been put in a shielded can.

I know it's difficult to produce full standard output levels when using equipment running from a low voltage supply (the whole thing runs from 5 volts DC from the computer, or a wall-wart), but it obviously has at least one switch-mode power supply in it to produce the 48 volt phantom supply, so they could have done something similar for the main line-outputs (provide the output stages with a high-enough supply for producing industry-standard output levels).

Although (as an estimate, it looks to be) 1RU high, it's less than 19″ rack unit width, with no straight-forward solution to rack mount it.  There are no threaded holes on the sides, and it'd need spacers, as well as wings, or bolting to a tray.  You'd have to open it up to see if there was sufficient clearance to add some bolts into the chassis (I've seen people do dumb rack-mounting modifications to equipment, where they've just jammed self-tapping screws through the side, right into other things).  It would appear to be designed to be a box to sit on your desk near your computer, rather than something to be mounted with other audio gear.  That (one box by your computer) may be desireable for small home studios, but awkward for anything more complex (rack mounting allows you to organise your clutter, out of the way).

I discovered the highly unexpected behaviour that switching in the −20 dB pad adds hiss to the input.  I've never experienced that in any other audio equipment.  And it's the complete opposite of what I expect—a gain reduction of the pre-amp ought to reduce its noise level, too.  I suspect it's just an input attenuator pad, rather than a gain change.  I'm yet to discover how noticeable this is in actual use, since I haven't inputted any strong signals, yet, but it is completely obvious when monitoring silent inputs at normal monitoring levels.

The mike inputs are not particularly high gain, I think it's really designed with the idea of capturing singers, or miked instruments, rather than someone casually talking into a mike that's not virtually pressed against their lips.  But I've come across plenty of mixers that are like that, too.  If you were recording audio books or interviews, I think you'd want something else acting as a pre-amp between the mike and this unit, or you're going to have to post-process (it seemed quiet enough to do that).

I plugged this into a friend's Mac computer, and the system virtually ignores the channel 3 & 4 outputs.  Programs can give you four-channel audio (whether they do that, and how they do it, is up to them), but the system only bothers to use channels 1 & 2.

The booklet that comes with it is barely instructive.  It gives brief outlines of the controls, but most of that you can work out for yourself with a bit of experimenting.  And experimenting is required to work out various things that the book omits.

Additional comments after using this device for some time on a Mac

At some stage, I believe after one of the MacOS updates, I began to have audio break-up issues after one minute of playing any audio.  It would start to crackle, then get progressively worse until all it produced was utter distortion.  I could stop playback, wait a moment for the computer to reset something, start playback and have another failure start again about one minute in.  This was quite repeatable, yet also intermittent.  i.e. It wasn't a once-off, nor completely regular.  It might happen several times in a row, unplugging and replugging the USB lead might stop the behaviour for a while.  And it didn't make any difference whether the device was being powered by the Mac, or its own plug-pack.

Audio capture was fine, it was just playback that was a problem.

Behringer suggested I might have a faulty unit, but I haven't managed to return the device because I couldn't find the purchase receipt.  But I'm not the only one to experience this issue.  The consensus seems to be that it's a buffer size problem, but there is no way to adjust the buffer size on a Mac (system-wise), only some applications give you way to adjust their buffer size, and behringer does not supply their own driver for the Mac (where they could directly address this issue).  Adjusting various audio parameters that are available on the Mac, such as bit-rate and bit-depth, don't alleviate the problem.  It just resets it for a while.

The Mac is one of those multi-giga-hertz things with multi-processors and oodles of RAM, so it should be quite capable of running the device.  And it still fails after a completely new MacOS install, so it shouldn't be down to some random gubbins installed on the Mac.

Since the Mac is a major target for professional and enthusiast audio and video work, behringer really ought to collaborate with Apple, and either have Apple tweak something to support it better, or behringer release a Mac driver specifically for their device.

In conclusion

The unit is certainly a step above the usual audio inputs and outputs of most computers, in quality, in being able to handle four independent channels, and in being able do so on standard audio connectors, supporting balanced audio, and providing phantom power.  The amount of noise added to the signal when the pad switch is engaged is disturbing.  The low output-level is a disappointment, but not a show-stopper.  For most people, that's probably not even going to be an issue.  The hiss on the outputs is very disappointing, but again probably isn't going to be a showstopper for people using it as just a capture device.  Even with the noise, if you have a strongly recorded audio file it sounds good in playback, better than many PC's built in sound outputs.  You can certainly enjoy listening to your MP3 collection through it.

Although most of that output noise should be well above the frequencies that the human ears can hear, it can mix within equipment and produce other noises.  And noisy monitoring makes it hard to tell whether your recordings are good.

Price-wise, it seems a good deal, for what it is, and compared against other similar-seeming products.

The ability to use it, on some (*) operating systems, without having to load any special drivers is a definite bonus.  Apart from being less hassles to get it going, in the first place, it means that your hardware doesn't suddenly become a useless paperweight when you update or replace your computer system (unlike hardware that had special drivers that were only ever produced for particular versions of operating systems).

* There are Windows drivers, but I'm not sure whether the device is useable, partially or wholly, without installing them.  I don't use Windows.  I'm told it will work in a basic two-channel mode without installing drivers.

They make models with less, and more, inputs.  I really only had a need for a two-input device, but saw the virtue in buying something more than I currently need.  My main uses would be digitisation of pre-existing stereo recordings, or creation of new ones from two microphones.  If you intend doing multi-track recording, count up all the channels that you will need as a minimum, and buy something with more.

Later on, during the 2020s plague, I got into recording music using multiple microphones with friends, and upgraded to the UMC1820.  Since then, the UMC404HD has sat on the shelf.

Written by Tim Seifert on 17-May-2017, and last updated Friday, 01-Mar-2024 08:47:30 AEDT.