Topic: UA-5 > 661 tests need some analysis

[Massive Dynamic]

I've been playing around with some new gear and made some test recordings with an Oade T+modded UA-5 running coax into a Marantz PMD 661. The Marantz is really particular about accepting a digital signal. You have to set bit-depth and sample rate in the preferences menu before recording, and the source has to match those settings.

For the first test, I set the 661 to record at 24/44.1. I set the UA-5 output accordingly and made a short test recording. After hitting stop on the 661, but while the UA-5 and 661 were still powered up, I switched the UA-5's sample rate knob to 48 and tried to record the next clip. The 661 recorded just fine, but the file was still 24/44.1. The UA-5 has to be powered down before changes in output will be effective. After powering down and changing the UA-5 output to 24/48, I tried to record but the 661, still set to 24/44.1, would not. This was a consistent result in all the permutations I tried, and all test files were checked for bit-depth and sample rate in Wave Editor.

The other curious thing happened by accident. The 661 had been previously set to record at 16/48. I forgot, and set the UA-5 to output 24/48. The 661 took the signal, and the resulting file was indeed a 16/48 file. I don't know why the 661 will accept a 24/48 input signal while the preferences are set to 16/48 when it won't accept mis-matched signals in my other tests. Is the UA-5 output 24-bit, 16-bit, or both at the same time? Or is this something unique to the 661? Is there another test I should try? I am aware of the previous discussions about UA-5 output so these results were a surprise.

[runonce]

I've been playing around with some new gear and made some test recordings with an Oade T+modded UA-5 running coax into a Marantz PMD 661. The Marantz is really particular about accepting a digital signal. You have to set bit-depth and sample rate in the preferences menu before recording, and the source has to match those settings.

For the first test, I set the 661 to record at 24/44.1. I set the UA-5 output accordingly and made a short test recording. After hitting stop on the 661, but while the UA-5 and 661 were still powered up, I switched the UA-5's sample rate knob to 48 and tried to record the next clip. The 661 recorded just fine, but the file was still 24/44.1. The UA-5 has to be powered down before changes in output will be effective. After powering down and changing the UA-5 output to 24/48, I tried to record but the 661, still set to 24/44.1, would not. This was a consistent result in all the permutations I tried, and all test files were checked for bit-depth and sample rate in Wave Editor.

The other curious thing happened by accident. The 661 had been previously set to record at 16/48. I forgot, and set the UA-5 to output 24/48. The 661 took the signal, and the resulting file was indeed a 16/48 file. I don't know why the 661 will accept a 24/48 input signal while the preferences are set to 16/48 when it won't accept mis-matched signals in my other tests. Is the UA-5 output 24-bit, 16-bit, or both at the same time? Or is this something unique to the 661? Is there another test I should try? I am aware of the previous discussions about UA-5 output so these results were a surprise.

The second issue could be by design. The ability to truncate bits and still record might be seen as a "feature." It accepted the signal, but chopped off 8 bits without dither. Not sure why anyone would intentionally do this. But perhaps allowing it was better than generating an error. Not unique, most 16 bit recorders will do the same when you throw them a 24 bit signal. And certainly nothing to do with the UA5

[SmokinJoe]

Your results make perfect sense in the view of previous tribal knowledge.

I seem to remember someone running some tests with a UA-5 (and probably an MT), and we concluded that the UA-5 always puts out 24 bit signals, and 16bit bitbuckets (typically JB3s and H120s at the time) would truncate.  I'm not saying we had iron-clad proof of that, but it made sense in the view of what we saw.  It's probably in "Team UA-5" posts from a couple of years ago.

The other thing is that Grace and Apogee and other vendors make a huge deal out of the super dithering technology they have, and why not?, because it's a non-trivial engineering effort.  It's much simpler to just to spit out the same 24 bit signal.  Especially when you consider that the UA-5 was primarily designed for USB recording, and the digi-out was kind of a nice to have feature for compatibility purposes.

Having to power down the UA-5 when you make changes sounds familiar.  Again, I think they originally figured you would be clicking on your computer screen, and the driver would give it a soft reset or something.  Without the computer, you have to do a hard reset.

My limited experiences with some other devices (PMD670) is that "if they are required to lock on, the bitrate settings have to match", which only makes sense.  Some devices (like Rockbox H120) will try to guess what it is and lock on.  When it guesses wrong you get the "funny sounding file syndrome".  My R4 resamples the digi in, so I can be fed a 96k signal and record to 44.1k if I want, it doesn't care.

[Massive Dynamic]

I ran one more test. I played a 16/44.1 DAT from my R300 into the 661 set to receive a 24/44.1 signal. Recorded fine. When I tried to play a 16/48 DAT, the 661 would not accept the input. Seems to be more finicky about sample rate than bit rate. My assumption is that the 661 is truncating the signal rather than dithering, but I don't know how to confirm that.

I don't have any other digital output devices to test, and I am not sure these tests tell us any more about the UA-5 or the 661 than we already knew, but I thought I'd post them just the same.

The other thing is that Grace and Apogee and other vendors make a huge deal out of the super dithering technology they have, and why not?, because it's a non-trivial engineering effort.

That's true, but it is still amazing to me that no one could tell by listening that the files had been truncated instead of dithered until you and others did some testing.

[runonce]

I ran one more test. I played a 16/44.1 DAT from my R300 into the 661 set to receive a 24/44.1 signal. Recorded fine. When I tried to play a 16/48 DAT, the 661 would not accept the input. Seems to be more finicky about sample rate than bit rate. My assumption is that the 661 is truncating the signal rather than dithering, but I don't know how to confirm that.

I don't have any other digital output devices to test, and I am not sure these tests tell us any more about the UA-5 or the 661 than we already knew, but I thought I'd post them just the same.

The other thing is that Grace and Apogee and other vendors make a huge deal out of the super dithering technology they have, and why not?, because it's a non-trivial engineering effort.

That's true, but it is still amazing to me that no one could tell by listening that the files had been truncated instead of dithered until you and others did some testing.

Neither truncation or dither is in play in that last experiment...If the sample rate doesn't match...it wont lock.

Truncation involves reducing bit depth...that experiment is increasing the bit depth...(sort of)