Harmonic distortions of the unbalanced inputs
Edirol R-4, Zoom H4n, Sony PCM-D50
Ok, so there is not a big difference between the units when it comes to noise level. But what about harmonic distortions?
I connected the unbalanced 1/4" TRS (tip-ring-sleeve) line-inputs of the Zoom H4n and the Edirol R-4 to a pretty good Marantz CD-player and played back a test-CD with a pure 1 kHz sine-wave.
You might argue that the analog circuitry of the CD-player would introduce some harmonics as well, and that is probably correct. But the results of the Edirol proved that this wasn't much.
Before the recording, I reduced the output of the CD-player so that the Zoom at an input-level setting of 1 (the lowest setting) would just not trip. This was equivalent to a setting of Line-5 with the Edirol. Obviously, the Edirol has some reserves for even hotter signals while the Zoom is at its limit. The test with the Sony PCM-D50 was done later at full output level of the CD-player. The Sony's record level had to be set to 4.5 to maintain about -1 dBFS. Again, plenty of reserves for hot line signals.
All tests were done on battery power to make sure no distortion could creep in from an AC supply.
1 kHz, -0.5 dbFS sine wave on an Edirol R-4
The spectrum shows a first harmonics (k2) of 2 kHz at a level of -100 dB. Since the base frequency (k1) would peak at -10 dB this computes to a harmonic distortion of -90 dB, which is excellent. The second harmonics (k3) at 3 kHz is even lower.
It seems my CD-player isn't too bad after all ;-) You can download the wav-file of the Edirol's recording here.
I have also checked the noise level of this setting (Line-5) when the CD-player was off and it was -80 dBFS. Again a very good result.
1 kHz, -0.5 dBFS sine wave on a Sony PCM-D50
The Sony performs even better than the Edirol though the difference is marginal. The k2 harmonics is practically the same (-90 dB) but there are not further harmonics exceeding -100 dB.
1 kHz, -0.5 dBFS sine wave on a Zoom H4n (record-level 1)
This is really bad.
I've double and triple checked the results but there's no way around it: The Zoom H4n produces a harmonic distortion (k2) of -45 dB which is completely unacceptable. And k3 is about the same level.
By examining the waveform of the recorded file I could proof there was no overloading in the digital domain. The waveform never touches the brick-wall of full modulation. So what wee see here is either a nonlinearity of the analog preamps or there is a problem with the anti-alias filter. The second theory is supported further by the measurements with a 10kHz test-tone (see below).
For reference just remember this: -40 dB equals a distortion factor of 1%, -50 dB equals 0.3%. Both are pretty audible.
You can check for yourself by comparing the recording of the Edirol with that of the Zoom. In direct comparison the Edirol sounds cleaner.
Good devices are much below 0.1%. When testing the Sony PCM-D50, for example, I found a distortion factor of just 0.003%. That would also be a typical value for state-of-the-art hifi equipment.
To get to the bottom of the problem I did more tests to see if an increased record-level setting would improve the results. With a record-level of 10 (instead of 1 in the previous test) the output of the CD-player had to be cranked down considerably to avoid overloading the Zoom. But the result looked a little better. The first harmonics (k2) was still just 50 dB lower than the fundamental tone. And k3 was not much lower as well.
Obviously the Zoom H4n doesn't like hot signals so I checked different record-level settings (always adjusting the output of my CD-player to a constant -0.5 dBFS) - see the table on the right.
k2 and k3 are given in dBFS relative to the fundamental tone and as a percentage. The total harmonic distortion factor (THD) is the geometric sum of all the individual distortion factors. The first table row, for example, computes to a THD of about 0.7% (k4 and higher harmonics are small enough to be disregarded in this case).
The links take you to wav-files of the Zoom's recordings. Each is about 8-10 secs in length.
I did tests with even higher record-levels than 50 but the higher I got the more I had to reduce the output of my CD-player - and that would add extra harmonics. Even the last values of the table should be taken with a grain of salt.
1 kHz, -0.5 dBFS sine wave on a Zoom H4n (record-level 40)
We are slowly getting there... Raising the record-level will improve the results somewhat. But that is only a practical solution if your input signal can be turned down enough or you have a attenuator pad at hand because the Zoom doesn't have one built in.
I have done many more tests and found that this game also works the other way around.
You will get acceptable distortion at the lowest record level (1) as long as you input signal is kept below -12 dBFS. For a record level of 10, the input signal should not exceed -9 dBFS and so on.
For me, acceptable means k2 and k3 between -60 and -70 dBFS relative to the fundamental frequency.
But 1kHz is not all. Let's now look at a 10 kHz test tone:
10 kHz, -0.5 dBFS sine wave on the Edirol R-4 (line-8 with 18 dB pad)
Great result. k2 is at -85 dBFS relative to the fundamental frequency. Download the wave here.
10 kHz, -0.5 dBFS sine wave on the Sony PCM-D50 (rec-level 4.5)
The same great result as the Edirol. The k2 harmonics is even a tad lower. But at a level of -85 dBFS there is really nothing to worry about (after all, I am not using a professional signal generator and this peak may well come from my CD-player anyway...).
Now on to the same test with the Zoom H4n:
10 kHz, -0.5 dBFS sine wave on the Zoom H4n (rec-level 40 with 18 dB pad)
Rec-Level analog or digital ?
A reader suggested an explanation for all these distortions: What if the record level setting of the Zoom H4n is actually working on the digital side only, i.e. after the A/D-converter? I know the old Zoom H4 had a record level setting that worked that way. You always had to set it to 100, which was neutral. But the setting was buried in the menus and not available through switches or knobs on the outside.
On the new unit, Zoom has saved the little H-M-L switches that were used on the old H4 to control the analog input level.
Until recently, I was pretty sure the rec-level buttons of the new unit would now work on the analog side of the chain because there is no other way to control the analog signal.
But what if there is no control of the analog signal in the Zoom H4n at all? What if the rec-level is again only working on the digital side, i.e. after the converter?
In that case, any recording with a signal that almost hits full modulation (which is normal...) made in any other rec-level setting than neutral (100) must be heavily distorted because it overloads the A/D-converter - and that's exactly what I've found.
Again, I can not do any more verification tests since I have sold my H4n (... and I don't feel too sorry for that now ;-) but the conclusion seems pretty obvious.
Yes, this is real.
I've repeated the test a couple of times, tried with different sample rates (up to 96 kHz), with and without the attenuator pad (see next page). It doesn't get much better than this.
The setup was exactly the same as in the previous tests of the Edirol and the Sony. Actually the recording was done just a few moments after the Edirol's. I even had the same cables - just a different recorder. Here is the wave file of this mess.
Not to the surprise of anyone this distortion is audible in everyday recordings - not just in test setups. It may not be obvious at first but in direct comparison to the original or a good recording from another device the Zoom will sound harsh and unpleasant with overemphasized highs.
Remark added on 2010/2/10
The Zoom H4n offers a number of digital effects like limiter and preamp-simulation. Unfortunately, I can not verify any more what the effect settings were during the recordings since I no longer have access to the unit. I can, however, positively confirm that all effect settings were at their original factory default.
Line level ?
When I was complaining the Zoom H4n's inputs couldn't handle line-level signals a reader reminded me there are several definitions of 'line-level':
-10 dBV (or -7,8 dBu) for consumer audio (USA)
+ 4 dBu for professional audio (including electrical instruments like synthesizers, mixing consoles and the like)
+ 6 dBu for professional video (Europe)
The Zoom H4n's unbalanced inputs are rated +2 dBm (or +2 dBu or 1 V rms) and hence comply with the US-consumer standard (assuming the rating is correct) but with none of the professional standards.
Although my Marantz CD-player is a consumer device, it has a rated output of 2 V rms which is equal to +8.2 dBu. That is way above the input rating of the H4n and explains the overloading (but not the distortions as I have reduced the output level during those tests).
The output rating of the Marantz player is not exceptional, though. Today, the "2Vrms@0dBFS" ouput level is an industry standard for consumer electronics and was founded already in the 1980's with the very first CD-players. Nearly all CD-/DVD- and BluRay- players (from Panasonic/Technics, Sony, Denon, Pioneer, Kenwood, ...) and many other hifi components are using this output level.
A number of professional condenser microphones from well respected companies (like Neumann, Rode, Audio Technica, ...) also produce output voltages in the range of +8 ... +16 dBu at their maximum sound pressure level. None of these mics will work on the H4n when recording loud sounds without an attenuator (some mics already come with a built in attenuator).
And what about the other two field recorders?
The Edirol R4's inputs are rated +4 dBu but the unit still has 6 dBFS of headroom when recording a 2 Vrms signal.
The Sony PCM-D50's inputs are already rated 2 V rms but it's input level attenuator can be cranked down to deliver 40 dBFS of headroom when recording that +8.2 dBu signal.
In other words: Both recorders offer plenty of headroom well over their specification and are good for any signal.
The Tascam DR-100, which is probably the most direct competitor of the Zoom H4n, also has inputs which are rated for 2 V rms (+8.2 dBu).
What does this all mean?
In principle, Zoom is correct in calling the unbalanced inputs 'line level' - according to a depreciated consumer audio standard. But they are only good for some consumer gear. Most consumer hifi devices, professional audio equipment and professional microphones will overload the H4n's inputs pretty much.
Using the Zoom's 1/8" inputs
Well, maybe the 1/8" unbalanced inputs of the Zoom work better? They are located at the back of the unit and can be used as an alternative to the built in microphones. As such, these inputs should be much hotter than the 1/4" line-jacks at the bottom.
To make a long story short, I found the 1/8" inputs behave exactly the same as the 1/4" line-inputs. Same recording level, same distortions.
There isn't any difference between the two regarding signal level. If you call the 1/4" jacks line-level then the small 1/8" inputs are line-level as well. I would rather call both mic-level and use an external attenuator pad for hot line signals...
Also the distortions are exactly the same. I am not showing diagrams here because they are completely identical to the ones above which were recorded through the 1/4" jacks.
Conclusion - Part IIIt seems fair to conclude that the Zoom H4n, using its unbalanced 1/4" or 1/8" TRS inputs, should not be operated below a record level of 40 in order to avoid audible distortions even in the low frequency range. Although the manual calls these inputs line-level they are actually closer to mic-level and need an external attenuator pad to work with most line signals. If you want to build your own pad (which is easy), see the next page for details. Anything between 12 and 20 dB should be fine.
Unfortunately the Zoom H4n, using any of its unbalanced inputs, has big problems with harmonic and subharmonic distortions and I have not yet found a way to circumvent this. It might be a good idea to ignore these inputs altogether, at least for critical recordings.
Neither the Edirol R-4 nor the Sony PCM-D50 have such problems. Both produce excellent results even with the strongest input signals.
Some final remarks on the Zoom H4n (added 2012/01/07)
Recordings through the unbalanced inputs should never be made anywhere near the max. rating (+2dBm) or otherwise they will be distorted (THD of up to 1%).
A "REC LEVEL" setting of 50 or higher will give acceptable results (THD below 0.1%) provided the signal is muted enough not to overload the inputs.
I did not test the balanced inputs as I did not have the proper equipment to do so.
More "final" remarks (added 2012/02/20)
I have been in contact with an audio engineer recently who did some pretty thorough testing on the balanced and unbalanced inputs of the H4n. He also conducted an acoustic test using an Agilent/HP audio oscillator at 10 kHz, audio amp and tweeter directed at the H4n's built-in microphones. Not only did he confirm my findings but he also found something very disappointing about the microphones:
The balanced inputs are free of distortions but the microphones exhibit the same distortions as the unbalanced inputs.
To quote from his analysis of the Zoom H4N "All very interesting but shamefully poor design. I would call this a voice grade unit for non critical use at best."
The engineer opened his H4n to follow the signal path of the microphone signals. He was able to verify that the built-in microphones are using the same signal path as the unbalanced inputs and are subject to the same distortions with respect to the input level control setting. He points out the distortions mostly affect the high frequencies > 5 kHz and only at very high input levels. He does not recommend this recorder's microphones for very serious applications. It is not a professional level product. He also noticed that the balanced inputs are somewhat high level. They lack sufficient gain for using dynamic microphones without an additional mixer or preamp.
I also came across a very interesting Microphone Input Noise Comparison.
Not surprisingly the H4n ranks in the low end of all the handheld recorders tested.