Taperssection.com
Gear / Technical Help => Recording Gear => Topic started by: jerryfreak on December 06, 2019, 09:04:47 AM
-
this recorder is the baby brother to the beloved A-10. has a similar interface
this is designed for voice dictation and can record 16/44.1 .wav
MSRP $80, found mine for $50 NIB/NOS on ebay
key differences from A-10:
-about half the size and weight
-built in mics but they are much smaller and not adjustable in direction (who really uses these anyway? ...did not test)
-no bluetooth/remote app
-4GB internal memory, takes a microSD card (32GB sdhc tested, claims sdxc support, untested)
-plays mp3s and 16/44.1 wavs. cannot play FLAC, 48K, or 24-bit wavs
does not have level control on line-in! needs level adjusted by whatever is feeding it. Its line-in input sensitivity is approximately equivalent to level 12 (of 30) on the A-10, which is in the neighborhood of where i usually run my A-10 line-in. when running it to -0.1dB (FSD), no clipping is apparent on a 10-22KHz logsweep file, see image below
frequency response is claimed as 50-20KHz in the specs, however i think they are referring to the internal mics. when tested line-in with generated frequencies it actually has more low-frequency sensitivity than the A-10, see images below. A10 was actually worst of the bunch tested. DR100 was ruler flat
some rolloff tests, using test setup of soundforge playing generated sine waves>usb>grace sdac>line out>recorders
all recorders 24/44.1 except ICD-UX560@ 16/44.1
Tascam DR100 MKIII: flat to 20 Hz, -0.02dB@15 Hz, -0.1dB@10Hz
Sony A-10: -0.2 dB@100 HZ, -1.1dB@45Hz, -2.9dB@25Hz, -4.1Hz@20Hz, -6.0dB@15Hz, -9.6dB@10Hz
Sony ICD-UX560: -0.2dB@55Hz, -1dB@30Hz, -2.3dB@15Hz, -4dB@10Hz
Roland R07: -0.2dB@80Hz, -1.0dB@35Hz, -2.75dB@20Hz, -4.3dB@15Hz, -7.5dB@10Hz
noise tests in silent portion of above tests:
Tascam DR100 MKIII(minijack line-in, gain set to 10.5dB):peak signal =-2.1dB, peak noise= -86dB, rms noise = -100.5dB (-84dB/-98dB relative to peak signal)
Sony A-10 (line in, level set to 11 out of 30):peak signal =-1.1dB, peak noise= -79dB, rms noise = -92.5dB (-78dB/-91.5dB relative to peak signal)
Sony ICD-UX560: (line in)::peak signal =-0.0dB, peak noise= -74dB, rms noise = -90dB (-74dB/-90dB relative to peak signal)
Roland R07 (line in level set to 28 out of 100):peak signal =-0.1dB, peak noise= -79dB, rms noise = -92.5.5dB (-79dB/-92.5dB relative to peak signal)
noise floor of line-in on ICD is a few dB higher than the A-10 and R-07 at the same input sensitivity. Which is a slight hinderance considering its only 16-bit. DR100MKIII is in another league but was a good reference for my test setup.
finally an 8ch polywave. of 2 live sbd tracks played thru the same setup above. i purposely attenuated the input, so the resulting recordings would have levels peaking around -12dB, in a manner similar to how i would normally run a 24-bit field recorder. Giving no mercy to the 16bit recorder. wavs were all normalized in post to same levels then the 24-bit files were dithered to 16bit before all were combined into an 8ch 16bit wav. theyre not perfectly synched but you can audition various tracks
can you guess which recorder was used on each pair of tracks? im kinda more interested in what your EARS think, as anyone can really pick out most of the tracks from statistics using the numbers i gave above
8 ch poly wav: https://drive.google.com/file/d/1er8Z1Y4DTaVMVd4lJG4Ib_Q9olpCaFaD/view?usp=sharing
key to which tracks are which (in hidden text, highlight to see:)1-2:PCM-A10, 3-4:ICD-UX560, 5-6:R07, 7-8:DR100MKiii
pics and links:
ICD-UX560 not clipping at FSD:
(https://i.imgur.com/j1MSel6.png)
low frequency tests. white numbers in top row are Hz.
top to bottom: Tascam DR100mkIII, A-10, UCD-IX560, R07
(https://i.imgur.com/vrnsCvn.png)
side by side with A-10
(https://i.imgur.com/q0CKSCA.png)
(https://i.imgur.com/YMA6gOQ.png)
(https://i.imgur.com/IhF4SRj.png)
sony page:
https://www.sony.com/electronics/voice-recorders/icd-ux560
quickstart guide and full manual
https://www.sony.com/electronics/support/res/manuals/4573/45739336M.pdf
https://helpguide.sony.net/icd/u56/v1/en2/index.html
specs:
3 model variations:
ICD-UX560 - 4GB internal storage (the one i have and tested)
ICD-UX560F - 4GB internal storage+ FM receiver
ICD-UX565F - 8GB internal storage+ FM receiver
my impression: this tiny recorder definitely could have some application for:
-those on a serious rig budget
-those who just want a cheap backup recorder to kick around the gearbag as a spare or to record a backup off of headphone out
-people who are into the uber minimalist >:D rigs
-anybody who wants a cheap tiny mp3 player to beat around, active sports, etc
this is from 2016, would be nice if sony came out with a successor version with 24-bit wav and FLAC support, in the same form factor.
-
This is great stuff, thanks very much for sharing this.
I love the A10 the way I record, but I do wish the low end roll-off wasn't so aggressive. I'd rather take care of that in post (if necessary). In some cases, it works out well with my mics depending on the situation.
-
Man, this recorder is really how I would like a proper deck to be. Very minimal in layout and simple. Add bluetooth app capability and level adjustment and I would take this thing.
Appreciate the solid write up. Always nice to see comparison of rigs.
-
Great write up. Thanks!
-
Just think in a few years we will probably have a 32 bit float version of this
-
runs on an ARM LC823450TA SoC capable of 32-bit/192K, no wonder it sounds so good. too bad its crippled to 16/44
https://www.onsemi.com/pub/Collateral/LC823450-D.PDF
-
no takers on the listening tests, huh?
-
runs on an ARM LC823450TA SoC capable of 32-bit/192K, no wonder it sounds so good. too bad its crippled to 16/44
https://www.onsemi.com/pub/Collateral/LC823450-D.PDF
A new model was released late last year, ICD-UX570. Unfortunately, it looks to have the same 16/44 limitations...
page 170: https://www.manualslib.com/manual/1669042/Sony-Icd-Ux570.html?page=170#manual
-
This audio recorder is small and works great for me! It literally just the features I need. As I get audio from a line-in signal from preamp and external mics, this player is great for stealth recordings.
-
570 is $100 on ebay and amazon. 575 seems not to be available in US (theres some $140 japanese ones on ebay)
https://www.amazon.com/dp/B082QL6KLG/ref=asc_df_B082QL6KLG1580036400000
one feature it offers over the 560 series is the ability to change input sens to hi/med/lo on the minijack inout, which might be of use for fixed-gain/no-gain bat boxes
actually, scratch that, it appears that adjustable sensitivity settings are only available for 'mic in' (which supplies PIP), not "audio in" (aka line level). so it appears to offer no additional features over the 560 series (though it could have better specs/performance, given its a few years newer, hard to tell what chip they used without opening one up)
-
i just got this back after leaving it behind at TXR (thanks Jon Hart!)
that evening i set it on a table a few feet behind where my mics were set up at stage lip. recording via the internal mics. id call it almost passable and a better-than-nothing EQ-able backup in event of a catastrophic failure. it may not be the most hi-fi ever, but the internal mics didnt overload..... sounds ok on the quieter parts. might fly for an acoustic gig
https://soundcloud.com/aurorabluevalentine/damn-skippy-icd-ux560
for comparison, heres the 4011s a few feet in front (and a few feet farther from the crowd)
https://drive.google.com/file/d/1RQwkmZAuN_d8JQyX68j8g9DDl6bNse3D/view
-
that evening i set it on a table a few feet behind where my mics were set up at stage lip. recording via the internal mics. id call it almost passable and a better-than-nothing EQ-able backup in event of a catastrophic failure. it may not be the most hi-fi ever, but the internal mics didnt overload..... sounds ok on the quieter parts. might fly for an acoustic gig
https://soundcloud.com/aurorabluevalentine/damn-skippy-icd-ux560
I think it's quite passable for internal mics. I was expecting some distortion from the recording. Thanks for posting.
-
look carefully in the first few seconds of this vid, you can see it hiding on the other side of that M10+table stand rig
https://www.youtube.com/watch?v=0ka2riRdRyY
-
a friend needed a compact mp3 player so i gave the 560 to her, then grabbed a 570 new off amazon for $70 to check out. i dont have the other one to be side by side, but for all intensive porpoises it seems to be identical in specs and function for our purpose. I'll tear it open sometime to see if it uses a different DSP chip.
The 570 appears to have different built in mics ("S-mic" which is touted to have higher sensitivity), which in general is the opposite of what we want - however, it might have better capsules or something that makes the built-in mics better overall
one thing i dont like vs the 560 is that the display is now missing peak level margin... which is particularly useful for 16-bit recorders, but not a deal-killer imo.
it has some guides for min and max level targets, which are useful.
there are 10 bars between the two 'guide dots'. the 'guide dots' correspond approximately to -24 and -3 dBFS by my testing and there is one more bar (~-2dB) above the -3dB dot. So, at full scale on the level meters there is still 2dB to spare
the 10 bars between the guide dots are roughly
-24 -20 -16 -14 -12 -10 -8 -6 -5 -4 -3 dBFS
so if recording, kinda shoot for levels consistently above the lower dot with peaks past the middle point between the dots
heres a quick and dirty test of line in with a GD SBD. just eyeballing the levels per the method above, and they peak just above 6dB with plenty of headroom. did nothing but FLAC the raw file and upload it
https://drive.google.com/open?id=12Weh6RKaYG8LZmNUj6iKkUIJdE0LdThH
input noise seems about the same as the -560, about -75dB peak, -90 dB RMS, more than acceptable for aud recording.
ill try to run them side-by-side again if i get the 560 back, if not ill run it against the A10
-
Because of this post, I picked up the ICD-UX560 for ~$50 on ebay in January/February and carry it around practically everywhere. It's a pretty cool recorder for being so cheap and idiot proof, while still able to get good recordings. At one office meeting, I had the recorder tucked in my shirt and picked up everyone, even though some were behind me a good distance. I most often use it to record my children without being too invasive.
I'm happy to know the 570 is basically the same, which isn't all that surprising. $70 from Amazon is pretty good (ebay seems to be $80-$85), and I do see what you mean about the meters:
https://www.amazon.com/Sony-ICD-UX570-Digital-Recorder-ICDUX570BLK/dp/B082QL6KLG/ref=sr_1_1?dchild=1&keywords=Sony+ICD-UX570&link_code=qs&qid=1587948981&sourceid=Mozilla-search&sr=8-1
Kind of a weird thing to change from the 560.
Are the mics on the 570 movable? They look like miniature mics on the A10.
Anyway, thanks for bringing our attention to the 560 record and now the 570!
-
Are the mics on the 570 movable? They look like miniature mics on the A10.
no, and i cant say they are the same as A10, they seem to be smaller diameter based on housing size
-
there are software settings on the 570 for 'narrow'/'wide'/etc type pickups. cant remember if thats on the 560 as well. not sure whats going on there, if its a sensitivity/noise cancellation thing or if its actually altering the pattern of the mics
-
my friend sent me back my 560, so now i have 560 and 570 in hand. ill do some side-by side testing soon. as far as a size comparison, the 570, while 3 g lighter than the 560 (48 vs 51g), is larger in all 3 dimensions. it has the lighter-weight plastic-y feel of the A10. the older 560 is more dense, and 'brick-like', just has a sturdier feel to me
-
ok so i did some more tests of UX560 vs UX570 vs A10 at 16 and 24 bit
all tests playing sine waves and program material out of a mytek brooklyn dac being fed 32bit/96K files from soundforge
beginning of file was a generated 1kHz sine wave at an amplitude of -0.5dBFS, followed by a portion of generated silence
after the silent portion, the following program material was used: 'Hey Nineteen' by Steely Dan Gaucho 24/96 allbum bounced into a 32/96 file and reduced by -20dB, -40dB, and -60dB to simulate low-level 'mic-level' program material
remember that both ICD-UX5x0 unis have no level control on line in, so i am feeding it the maximum signal level possible to peak between -1.0 dBFS and FSD. This corresponds to an input level of 12 (out of 30) on the PCM-A10. This corresponded to the volume control on the Mytek set to -20 dB. the resulting signal level was measured at -4dBU using my AD2K
The maximum input level the PCM-A10 can take (at input level 01) is 16 dB higher than the ICD units (-4dB on the Mytek volume control, +12dBU as measured by AD2K). This high-level signal was tested later to see if we gained any advantage in dynamic range with the hotter signal to the A10 (see next post)
by all intents and purposes UX560 and 570 appear to act identically on line-in
they have the same input sensititivity (+/- 0.2dB) - both correspond roughly to an input level setting of 12 on the A10
they have about the same noise floor in the silent portions in both cases , -74ish dB peak, -90.309 dB RMS
the A10 at 16/44, input level 12, had a noise floor of -76-78ish dB peak but the same -90.309 dB RMS*
the A10 at 24/44, input level 12, had a noise floor of -75-76ish dB peak and a similar 90.2ish dB RMS**
when gain was applied to the A10 in 24-bit mode (by adjusting input level), the noise floor raised in a similar fashion. Adding 18dB of gain added about 17 dB of noise. Adding 27dB of gain (input level 30) added approximately 26dB of noise
*notice the 16 bit files all have RMS noise of -90.309. Is this a function of 16bit noise/dither? in other words is the RMS input noise lower than the LSB data? This seems to be the case, as when i go into the AD2K at 16/44.1, the peak noise is -inf, but the RMS is again -90.309 dB. since RMS noise is usually 15-20 dB below peak, obviously when the RMS noise gets below 90dB in a 16-bit file it is reported as -90.309 dB
**notice that the A10 at this input level sees essentially no advantage in dynamic range at 16 vs 24 bit
raw data of noise floors from test files 'peak level' in each test is the recorded level of the -0.5dB sine wave in the test file
icd-ux560, peak at -0.9 dBFS
Left Channel Right Channel
Maximum sample value (dB) -74.746 -73.407
RMS level (dB) -90.309 -90.309
Integrated Loudness (LUFS) -Inf.
Loudness Range (LU) 0.00
Maximum True Peak Loudness (dBTP) -71.22
Maximum Short-Term Loudness (LUFS) -81.48
Maximum Momentary Loudness (LUFS) -81.42
-----------------
icd-ux570, peak at -0.8 and -0.7dBFS (L/R)
Left Channel Right Channel
Maximum sample value (dB) -74.746 -74.746
RMS level (dB) -90.309 -90.309
Integrated Loudness (LUFS) -Inf.
Loudness Range (LU) 0.00
Maximum True Peak Loudness (dBTP) -73.41
Maximum Short-Term Loudness (LUFS) -81.50
Maximum Momentary Loudness (LUFS) -81.44
----------------------
pcm-A10 , 16/44 line in level 12, peak -0.4, -0.6dBFS (L/R)
Left Channel Right Channel
Maximum sample value (dB) -78.268 -76.330
RMS level (dB) -90.309 -90.309
Integrated Loudness (LUFS) -Inf.
Loudness Range (LU) 0.00
Maximum True Peak Loudness (dBTP) -74.75
Maximum Short-Term Loudness (LUFS) -85.21
Maximum Momentary Loudness (LUFS) -85.15
----------------------------
pcm-A10 , 24/44 line in level 12, peak -0.4, -0.6dBFS (L/R)
Left Channel Right Channel
Maximum sample value (dB) -75.570 -76.336
RMS level (dB) -90.141 -90.275
Integrated Loudness (LUFS) -Inf.
Loudness Range (LU) 0.00
Maximum True Peak Loudness (dBTP) -75.44
Maximum Short-Term Loudness (LUFS) -85.62
Maximum Momentary Loudness (LUFS) -85.53
----------------------------
pcm-A10 , 24/44 line in - input level 24, about +18dB of gain more than original test at input level 12
Left Channel Right Channel
Maximum sample value (dB) -59.417 -58.953
RMS level (dB) -73.051 -73.263
Integrated Loudness (LUFS) -69.24
Loudness Range (LU) 0.10
Maximum True Peak Loudness (dBTP) -58.88
Maximum Short-Term Loudness (LUFS) -69.22
Maximum Momentary Loudness (LUFS) -69.04
----------------------------
pcm-A10 , 24/44 line in - input level 30 (max), about 27dB of gain more than original test at input level 12
Left Channel Right Channel
Maximum sample value (dB) -50.915 -50.675
RMS level (dB) -64.137 -64.164
Integrated Loudness (LUFS) -60.27
Loudness Range (LU) 0.10
Maximum True Peak Loudness (dBTP) -50.12
Maximum Short-Term Loudness (LUFS) -60.21
Maximum Momentary Loudness (LUFS) -60.11
-
i just got this back after leaving it behind at TXR (thanks Jon Hart!)
that evening i set it on a table a few feet behind where my mics were set up at stage lip. recording via the internal mics. id call it almost passable and a better-than-nothing EQ-able backup in event of a catastrophic failure. it may not be the most hi-fi ever, but the internal mics didnt overload..... sounds ok on the quieter parts. might fly for an acoustic gig
https://soundcloud.com/aurorabluevalentine/damn-skippy-icd-ux560
for comparison, heres the 4011s a few feet in front (and a few feet farther from the crowd)
https://drive.google.com/file/d/1RQwkmZAuN_d8JQyX68j8g9DDl6bNse3D/view
Just checked this out - chief problem I hear with the sound is that it's effectively mono - there seems to be no separation between the mics and I imagine that they are closely spaced omnis, M10 style, which is hardly the best way to capture a stereo image. So I wonder whether the 570 with its width control attempts to do something about that, maybe by a simple XY to MS to XY matrix? With the M10 that could help with the higher frequencies but the low frequencies were always clumped in the middle (due to the laws of physics!). Of course at the price you have to temper expectations.
-
yeah that was more of a test of overload. i didnt expect a real stereo image from (presumably) omnis a few inches apart, but the 'width' is something i can experiment with on the 570
-
the next test was a control on the PCM-A10 to see if the tests above (which showed the ICD-UX voice recorders to not be that far off the A10 at the maximum input level of the voice recorders), actually had the A10 operating it at its maximum dynamic range. Since the A10 can actually take a 16dB higher line in than the voice recorders at its least sensitive setting (input level 01 , 0dBFS = +12dBU), tests were conducted in both 16 and 24-bit with the same program material, and the mytek volume adjusted 16dB higher
+12 dBU is a pretty hot level for an unbalanced signal... lets see if it clips.
lookin good!:
(https://i.imgur.com/9powtYB.png)
the statistics on the noise floor at this level were improved quite a bit
pcm-A10 , 24/44 line in level 01 (+11 dBU), peak -0.9, -1.1 dBFS (L/R)
Left Channel Right Channel
Maximum sample value (dB) -82.969 -82.473
RMS level (dB) -96.128 -96.398
Integrated Loudness (LUFS) -Inf.
Loudness Range (LU) 0.00
Maximum True Peak Loudness (dBTP) -82.16
Maximum Short-Term Loudness (LUFS) -90.84
Maximum Momentary Loudness (LUFS) -90.79
-------------------------
pcm-A10 , 16/44 line in level 01 (+11 dBU), peak -0.9, -1.1 dBFS (L/R)
Left Channel Right Channel
Maximum sample value (dB) -84.288 -84.288
RMS level (dB) -90.309 -90.309
Integrated Loudness (LUFS) -Inf.
Loudness Range (LU) 0.00
Maximum True Peak Loudness (dBTP) -80.77
Maximum Short-Term Loudness (LUFS) -89.32
Maximum Momentary Loudness (LUFS) -89.26
---------
again we see the RMS noise measured below the limit of the 16-bit container
again we see the 16-bit file with a slightly lower peak noise floor than the 24-bit file (!) odd....
the hotter levels into the A10 definitely seemed to pay off.... we gained 5-7dB of headroom.
Impressive that the A10's input couldn't be brickwalled at its lowest input sensitivity . Its best performance apparently will always be seen with an input hot enough to get high levels with the input level set in the single digits.
since I was using the AD2K as a reference level meter i ran the same +11.5 dBU signal into the AD2K at 24/44 to see what the noise floor looked like there. Highest sensitivit yof the AD2K is 14dBU=0dBFS, so we see the peaks bout 2dBFS lower:
AD2K, 24/44 line in (+11.5 dBU), peak -2.5, -2.4 dBFS (L/R)
Left Channel Right Channel
Maximum sample value (dB) -97.567 -96.890
RMS level (dB) -111.239 -110.870
Integrated Loudness (LUFS) -Inf.
Loudness Range (LU) 0.00
Maximum True Peak Loudness (dBTP) -96.26
Maximum Short-Term Loudness (LUFS) -Inf.
Maximum Momentary Loudness (LUFS) -Inf.
of course the AD2K beats the A10 handily on noise level, by about 15 dB, but its a good reference by which to measure the line-in noise of the A10
-
and of course, the comp. just a simple 2 files this time
one of these is the ICD-UX560 with source signal peaking around -16 dBU
one of these is the PCM-A10 with source signal peaking around 0 dBU with input level set to minimum (01)
peaks on both recorders were at a conservative -12dBFS, plenty of room for headroom.
both files were then normalized. can you tell them apart? if so, which do you prefer?
https://drive.google.com/drive/folders/1fIlaMWwA2NWWKqdZpj9YMJEFk22ERCFj?usp=sharing
is it even a fair fight? assuming youd probably be adding a bit of noise when boosting gain 16dB to feed the A10 at its optimum, in most field situations (nothings free!)
-
my bad on that shared link,i had the settings wrong. is fully shareable now
-
Maybe the A sample has a tiny bit more weight in the bass. But really, comparing either of these to a version playing from Google Play Music, there's not much in it. My ears are old and shot but I can't picture someone taking exception to either in terms of overall sound quality.
-
I could hardly tell them apart. But I did notice a clicking sound in the hey19-B recording on the left channel at about 0:49. Any idea what could have caused this?
-
dropped sample, good catch!
https://i.imgur.com/dOtMFUD.png
further testing is in order
the samples are identified in the hidden text below, roll over to see
A = A10
B = ICD-UX560
------------------------------------------
-
i was thinking of unloading the 570 as i like the 560 better for field recording due to peak margin and the slightly slimmer feel
however, the three extra nav buttons on the 570 add a lot of playback functionality when listening to long form stuff like audiobooks, podcasts, etc