Topic: UA5>optical>h120 weirdness

[scarletfire1]

Had an interesting experience last night. Not sure whether it's the UA5-digi, the opti cable or the h120. One thread mentioned reformatting after each show which I haven't done but will start doing. Another thread mentioned reseating the optical cable, which I did between sets (there are also no instances of the tell-tale static associated with sketchy digital transfers). That thread also raised some concerns about optical as a robust/reliable signal path. Between sets, my bag got jostled and I had some power flakiness for a minute with the DVD batt cable and lost power but upon reset, power remained continuously on during the music (Impact DVD batt at 50% after 3 hrs).

The weirdness was that the h-120 was trying to record at a sampling frequency of 96 khz despite being set to Record Source = digital, Frequency = same as source. The UA5 was clearly set at 16/44.1. I checked it and rechecked it - many times. The track-length timer on the h-120 was running at half-speed and the wav file headers came in at 96 khz (I didn't even realize it would record at 96). Opening the file in my editor sounded like chipmunks till I changed sample rate to 44.1 then it sounded fine.

I recorded 4 bands last night night. The headliner and two openers recorded fine once I switched sampling freq.. One didn't save at all despite my having monitored the levels and the clock (even at half speed) throughout the set and stop at the end seemed uneventful (hard drive light flashing as if writing away). The one that didn't save was giving me an alpha-numeric error message at the top of the screen while recording but I can't recall the message. I power-cycled the h-120 after that set, Rockbox reboots to record screen, headliner recorded and saved but at 96kHz wav-header. The H120 battery was near full. Hard drive >60% free.

I've run maybe 12 shows in 5 months on this rig and until last night, it's performed reliably and exactly as expected. Some operator learning experiences, of course, but the hardware's run fine.

I'm posting here because I don't know if it was Rockbox that went goofy, an h-120 issue, opti cable, or the UA5 signal. Please feel free to move wherever the questions best fits, but any advice is appreciated.

[sunjan]

The weirdness was that the h-120 was trying to record at a sampling frequency of 96 khz despite being set to Record Source = digital, Frequency = same as source. The UA5 was clearly set at 16/44.1. I checked it and rechecked it - many times. The track-length timer on the h-120 was running at half-speed and the wav file headers came in at 96 khz (I didn't even realize it would record at 96). Opening the file in my editor sounded like chipmunks till I changed sample rate to 44.1 then it sounded fine.

Did you update Rockbox recently?

Search the forum for chipmunks, it's an issue that the Hxxx series suffer from. You're supposed to switch on the UA5 and connect the optical cable before switching on the H120, to make sure that the recorder knows what frequency it's being fed. Even when doing so, it went flakey on me a few times.

To be safe, update the Rockbox to the latest build, and run a chkdsk /f /r. That usually helps...

[run_run_run]

I know with mine, I have to make sure the UA-5 is powered up for a few secs before I turn on my Iriver to lock on to the signal

[scarletfire1]

Ahh, so if power to the UA-5 was interrupted while idling between sets while everything was connected/powered on/paused and I just started a new recording wave file when the UA-5 came back up rather than restarting the h120 to synch, that might cause sample freq confusion in the h120?

[run_run_run]

Ahh, so if power to the UA-5 was interrupted while idling between sets while everything was connected/powered on/paused and I just started a new recording wave file when the UA-5 came back up rather than restarting the h120 to synch, that might cause sample freq confusion in the h120?

yep

[SmokinJoe]

Ahh, so if power to the UA-5 was interrupted while idling between sets while everything was connected/powered on/paused and I just started a new recording wave file when the UA-5 came back up rather than restarting the h120 to synch, that might cause sample freq confusion in the h120?

yep

Seconded.

Look here for how to fix the chipmunk header. http://taperssection.com/index.php/topic,80888.0.html

By the way... I don't reformat the disk, ever... I think that's mostly on JB3s and R4s.  If you format, you will have to copy your Rockbox directory back.  When it gets full I delete the stuff and do a defrag.

[rsimms3]

That's weird.  The only way a H120 will record in 96khz is when fed a digital signal that is 96khz, it isn't even an option in analog.  A H120 will record optical in 44.1/48/88.1/96 if fed a digital signal at those rates. 

[SmokinJoe]

That's weird.  The only way a H120 will record in 96khz is when fed a digital signal that is 96khz, it isn't even an option in analog.  A H120 will record optical in 44.1/48/88.1/96 if fed a digital signal at those rates. 

I suspect when the UA-5 is powered up the digi out spews some bizarre bitstream... which the H120 sees as "lock in to 96 khz".  It's not supposed to work like that, but then again, it was designed to have a computer attached.

[scarletfire1]

That's weird.  The only way a H120 will record in 96khz is when fed a digital signal that is 96khz, it isn't even an option in analog.  A H120 will record optical in 44.1/48/88.1/96 if fed a digital signal at those rates. 

That's why while watching the track timer crawl along, I kept checking to make sure the UA-5 was set to send 16/44.1 and not one of the ADV settings. I have to believe the UA-5 was sending a 96kHz sample rate signal because in the spectral analysis, the files actually contain data all the way to 48,000 Hz. I expected the header to have been goofy but for the data to cut off at 22,050 hz but these were actual 96k files.

[SmokinJoe]

That's weird.  The only way a H120 will record in 96khz is when fed a digital signal that is 96khz, it isn't even an option in analog.  A H120 will record optical in 44.1/48/88.1/96 if fed a digital signal at those rates. 

That's why while watching the track timer crawl along, I kept checking to make sure the UA-5 was set to send 16/44.1 and not one of the ADV settings. I have to believe the UA-5 was sending a 96kHz sample rate signal because in the spectral analysis, the files actually contain data all the way to 48,000 Hz. I expected the header to have been goofy but for the data to cut off at 22,050 hz but these were actual 96k files.

Correction: The only way a H120 will record in 96khz is when it thinks it's being fed a digital signal that is 96khz

They probably are 44100 files, not 96k files.  This type of mis-sync happens to almost everybody once, and the key is to powering up the UA-5 (wait a few seconds until that red light comes on), and then power up the H120, otherwise the H120 is guessing what the sync rate is.  I haven't looked inside the Rockbox source code, but I think it probably goes something like this: look for a 44100 signal, not seeing anything... look for a 48000 signal, not seeing anything... look for 88200 signal, not seeing anything.... look for 96000 signal, NOW I'm seeing something so I'll assume I'm getting a 96000 signal.

The above explanation is a known issue that happens to everyone.  I don't think the UA-5 was sending 96k data, I don't think the UA-5 would make that mistake.  Fix the header in the wav file, and you should be fine.

A 16 bit wave file looks like this:
- header (contains info on what the the machine thinks the file is, including sample rate)
- raw data, which is just 16bit samples one after the other, regardless of the bit rate.

The H120 creates an erroneous header file, and copies in the raw data from the UA-5 as it comes.  Because it thinks it is getting 96000 samples/second, and it's only getting 44100 samples/second, the clock runs slow.  It's not paying attention to an internal clock, it's just adding 1/96000 second per sample, where it should be adding 1/44100 second per sample.  On playback it tries to playback these samples at 96000, when the really are at 44100, so it's playing back too fast, hence the chipmunk sound.

Until you fix the wave header, spectral analysis doesn't mean anything, because it's not looking at the data correctly.  The data is just data... without any clock reference.

[run_run_run]

Do this mean if a UA-5 was really sending a 24/96 signal, the iriver could take it as 16/96?

[SmokinJoe]

Do this mean if a UA-5 was really sending a 24/96 signal, the iriver could take it as 16/96?

It would try, but it wouldn't be able to do it. It can't process it and write it to disk fast enough.  I tried this once, as a test.  The H120 says 96k on the display, and while you are recording you think it's working fine, but the file isn't usable.  I don't remember exactly what the problem was, but I think it had lots of skips.  88.2k had less skips, but still way too many.  44.1k and 48k are usable.

[rsimms3]

Do this mean if a UA-5 was really sending a 24/96 signal, the iriver could take it as 16/96?

It would try, but it wouldn't be able to do it. It can't process it and write it to disk fast enough.  I tried this once, as a test.  The H120 says 96k on the display, and while you are recording you think it's working fine, but the file isn't usable.  I don't remember exactly what the problem was, but I think it had lots of skips.  88.2k had less skips, but still way too many.  44.1k and 48k are usable.

I tried this too a while back and I think it worked okay.  This has peaked my curiosity and I am going to revisit it when I get home from work.  Sorry for any misinformation about the H120 only recording when fed a signal, learned something new about it guessing on the sample rate.

[cgrooves]

Do this mean if a UA-5 was really sending a 24/96 signal, the iriver could take it as 16/96?

It would try, but it wouldn't be able to do it. It can't process it and write it to disk fast enough.  I tried this once, as a test.  The H120 says 96k on the display, and while you are recording you think it's working fine, but the file isn't usable.  I don't remember exactly what the problem was, but I think it had lots of skips.  88.2k had less skips, but still way too many.  44.1k and 48k are usable.

I tried this too a while back and I think it worked okay.  This has peaked my curiosity and I am going to revisit it when I get home from work.  Sorry for any misinformation about the H120 only recording when fed a signal, learned something new about it guessing on the sample rate.

I think I have recorded a 24/96 signal successfully in the past as well using the UA5 / H1xx combo.

Also, hasn't it been established that all digi mod UA5's always send out a 24 bit signal (i.e. the ADV switch setting only applies to changing the sample rates on digi mod UA5's).  Your bit bucket just truncates that 24 bit signal to 16 bit if it can't use the extra 8 bits of info. 
See here:  http://taperssection.com/index.php/topic,14389.msg1197143.html#msg1197143


EDIT:  As everyone else has said....Always turn on the H1xx after a signal (red light) is being sent through the optical cable if you don't want to have messed up header files.  This is a well known problem.