Roland (Cakewalk) UA-1G
I focussed on use to transcribe from vinyl to digital.
There is a thorough review about recording from guitar input at
http://the-gadgeteer.com/2009/07/08/cakewalk-ua-1g-usb-audio-interface-review/
Manufacturers Specification:
2 channel (stereo in/out at analog line level and optical digital, plus headphone out)
1 channel microphone and guitar in
SAMPLING FREQUENCY
Advanced (ASIO) driver On: 24 bit 32/44.1/48/96 kHz
Advanced Driver Off: 16 bit 32/44.1/48 kHz
LINE IN
Twin RCA jacks, nominal input level -10 to +4 dBu
GUITAR IN (switchable to MIC IN)
1/4” phone jack, nominal input level -30 to -16 dBu
MIC IN (monaural dynamic type)
1/4” phone jack, nominal input level -40 to -26 dBu
MIC IN (plug-in powered monaural miniature condenser type)
1/8” powered phone jack; can be used simultaneously with the ¼” guitar/mic in jack
DIGITAL IN (shared jack with powered mic in)
Optical mini type conforms to IEC60958
LINE OUT
Twin RCA jacks
Nominal output level -10 dBu
PHONES OUT
1/8" TRS stereo jack
adjustable volume knob
DIGITAL OUT (shared jack with phones out)
Optical mini type conforms to IEC60958
USB 1.1 type A
POWER SUPPLY
USB connection 5V 200 mA
CONTROLLERS
Input level dial (for all analog inputs)
Phones volume dial
Switch for guitar/mic level from shared jack
Switches for sample rate, play or record at 96 kHZ, input monitor (below 96 kHz), analog or digital input, advanced driver on/off.
INDICATORS
USB (red LED for power from the USB port)
IN (green LED flashes above a moderate analog signal level to ADC)
OUT (green LED flashes above a moderate analog signal level from DAC)
PEAK (red LED flashes “if audio input signal exceeds the allowable level”??)
Performance tested using:
HP mini 5102 (Intel Atom N450 1.66GHz, X-25M G2 SSD, 2GB RAM)
USB 2.0 port
Windows 7 pro 32 bit
Audacity 1.3.12 beta
Noise levels (tested by recording with no input connections at 24 bit in Sonar LE then examining zoomed waveforms in Audacity) were around -78 dBFS with the UA-1G input level knob set at 100% (5 o’clock), -86 dBFS at 2 o’clock, and -89 dBFS at 0% (7 o’clock).
Set-up per manufacturer instructions (with Advanced Driver installed and switched on) gives Default Format = 2 channel 16 bit 44,100 Hz (CD Quality) under Windows Sound > Recording Devices > Microphone Properties > Advanced. Cakewalk Technical Support advised that this setting applies only to use in shared mode (simultaneous use by multiple applications). I changed it to 24 bit in any case.
The owner’s manual states for each analog input to “Adjust the input level knob until the level is as high as you can get it without causing the PEAK indicator to light”. I tried this with various input signal strengths. With music at nominal 350 mV (-9 dbV) from my Rotel RA-1312 consumer pre-amp it needed the input level dial at about 2 o’clock. With higher input signals (up to nominal 1.5 V from an FM receiver) it required correspondingly lower settings on the dial. But in every case, recording this way (with software input level set at 100%) in either Audacity or Sonar LE always gave the peaks of the recorded waveform at -10 dBFS. I emailed Cakewalk Technical Support several times to try to establish exactly what are the triggers for the peak indicator. I asked explicitly whether it is intended to indicate clipping at the analog input stage or clipping at the digital output stage or some defined (-10 dBFS) headroom. I explained that I was trying to establish whether it was sensible to ignore this peak indicator and set levels based on the recorded waveform when users did not want to trade S/N ratio for 10 dBFS of headroom (for example when transcribing to digital from finished vinyl records). I did not get an answer about the exact triggers, but the advice was
“This is a good question. The peak indicator on the UA-1G is more of a warning signal than a traditional peak light.
Basically, the UA-1G is geared toward beginners; the average consumer. It was made to not peak or clip, which is why there is a maximum output of -10 dBu on the device. The only way your signal would sound distorted when coming through the UA-1G is if the signal coming into the device was already super hot and over driven.
So yes, you can sensibly ignore the peak indicator on the UA-1G in this case. For more accurate dBu levels, we recommend devices like the UA-25EX which has better preamps and a higher output level than the
UA-1G.”
That is confusing to me, as the peak indicator has nothing to do with the output line level (-10 dBu). So I gave up on the emails (no offense, compared to some other companies Cakewalk were great to reply at all and the replies were relevant if slightly vague; but it is a noreply system so you have to jump through hoops again on the Cakewalk web site every time you would like clarification). Instead I ran a test for clipping when recording the same track (from the Thelma Houston Sheffield Lab LP via my Rotel RA-1312 pre-amp line out) with:
- The UA-1G peak indicator just off (which required the UA-1G input level knob at 2 o’clock and gave recorded peaks near -10 dBFS)
- The recorded waveform peaks around -1 dBFS (which required the UA-1G input level knob at full scale and illuminated the UA-1G peak indicator much of the time).
In both cases, the Audacity = Windows input level slider was left at 100%. From the zoomed waveforms (note the different vertical zoom levels), I think we can conclude that:
(i) With this kind of (consumer line out) signal level the UA-1G peak indicator can indeed be ignored (it seems to be set for -10 dBFS digital output).
(ii) The input level knob can be set to full scale without analog clipping.
(iii) The recorded waveform can be used to confirm that there has been no digital clipping.
(iv) Edit: Noting that the noise floor increases by almost the same amount as the recorded peaks (8-9 dBFS) over this range (2-5 o’clock on the input level knob) the S/N ratio is not going to be improved much (sorry, I got this wrong in the initial post).
The recorded tracks sounded fine to me. Given that plans are afoot to implement 24-foot recording in PartAudio / Audacity for Windows: if you have equipment that can benefit from the lower noise floor, or if you believe in other magical benefits of 24-bit recording (and you don’t pay too much attention to that PEAK LED), the UA-1G may be worth the extra cost. Unfortunately, you may still then need to compile a private version of Audacity with ASIO to get the 24-bit recording capability (or use an ASIO-capable program like Sonar LE which is supplied with the UA-1G device). If you want those extra ins and outs the UA-1G is an obvious choice.
