I attempted to record at the weekend on my WD my passport hard drive. I have used it many times. This was the first time I had tried since the most recent firmware update. I formatted the drive and when I armed it, it showed 2 errors. The error message functionality was not available before the recent update of course.
I tried my other WD my passport HD which I have also used before and it showed 2 error messages also. I didn’t have time to play around with it so I just hit record and left it.
Ironically the recording worked fine.
So………how reliable is the error message and have the errors always happened but I had no idea before this feature was added??
Errors might be equivalent to the error count on a linux NIC - when data stays in the buffer for any time - so it’s an early warning system.
I had three show up in the first few minutes of a carol service on Sunday, none for the next couple of hours. (Using an HDD as someone else had my USB stick)
It would be interesting to have an official word on what the errors might be…
Yes I’m thinking that having the ability to see the errors detected could cause unnecessary concern if the errors shown do not seem to actually effect the recording.It would be nice to know what they actually meant. Bear in mind, the same amount of errors showed up on 2 different hard drives - both of which have been used many times for recording on the QU16 and have both been re-formatted as per A&H instructions.
I’ll try again this evening if I get time…
Hi all,
The error system indicates a buffer overflow. When this happens, the content of the buffer is lost, this could be up to 2 seconds of audio across all 18 tracks.
While testing USB storage devices we noticed (especially with magnetic HDs) that one or several errors might get reported at the very start of a recording due to the mechanics of the drive getting up to speed. If no other error is logged afterwards, this should not be a cause of concern.
So it is actually lost data.
Is there any chance of “warnings” when the buffer gets past some level?
Is there any benefit? I don’t know how much buffer there is (although it sounds like there is only a 2 second buffer (21848*24=41k, so I assume 64kb??))
Would we expect to see that buffer generally running empty?
one second recording of 18 channels at 48kHz with 24 bit is about 2.5MBytes, just to mention (18480003)…
a warning potentially could be displayed at 50% fill level, so the operator has one second to recognize, understand and act…
Whoops - I forgot the “k”
It’s more that an operator would get a feel for devices which are often tripping the warning, and for those which never even get that full (i.e. trustworthy devices)
That what’s Nicola said: If there are few errors (=Buffer Overruns) during startup this could be accepted if the error count does not further increase during recording. This is for HDD recording. For recording on SSD/Flash I guess only 0 errors are acceptable, since they neither have to spin up nor vary in access times due to seeking around.
And just to repeat some hints for recording multitrack on HDD:
- always start recording on a freshly formatted device (no fragmentation at all)
- do not copy additional files from a PC to that drive (recording need to seek around that data)
- do not erase files/sessions (this also requires additional seeking)
- do not defragment a multitrack recording (interleaved tracks will get sorted, resulting in additional seek times)
Thanks for explaining why this happens Nicola.
And thanks to everyone for your help…
I’ve been seeing reports of errors during recording to my 500 GB Western Digital Elements drive (not right at the start of recording, but during recording). It tends to run into 10’s of errors over an hour or so, though it has run for up to four hours with no errors, so it is intermittent. This is even after a fresh format on the A&H.
These have been incidental recordings so far so I’ve not been too worried (yet), but I’m starting to wonder if this drive is really suitable, even though it is on the recommended list.
The drive is new (well, bought in July, but predominantly only used for this purpose).
Is there anything I can do ‘offline’ to determine if this is a drive error or something else? - would the normal read/write and smart tests be of any indication of what’s going on? The only thing that comes to mind is that I have to position the mixer next to my drums so it’s probably subjected to some level of vibration!
The other thing I’ve noticed - this is a USB 3 drive, but copying wav files produced by the QU to my PC it runs at about 5 MB/s! As an experiment I tried dumping another file on the same drive (same formatting etc.) and it was as expected - about 80 MB/s. So it something to do with the files themselves that causes really low transfer speeds.
Checking on the “improved” version with USB3, you’ll notice that this is a 5400 rpm model whereas the older USB2 version was 7200 rpm. My guess is that the slower spin rate is a likely suspect. I was told “back when” that 7200 was the desired spin rate for recording as opposed to simple mass storage devices…which it would seem yours may be.
That’s my best guess.
Interesting find Dick.
Yes, I remember those days! If I recall correctly though, in isolation of other factors, 5400 rpm should provide plenty of throughput for recording multiple channels of LPCM - 16 channels of 24 bit 48 kHz should certainly be well below the bottleneck imposed by the spin speed. SOS had a pretty good article on this a while back…
However, I guess it’s not inconceivable that something else has changed (as part of the move to lower rpm) on the latest version of this drive to cause issues. In which case, the approved drive list really needs updating (this is what I used to help decide my purchase in the first place) and unfortunately make and model doesn’t necessarily guarantee you’ll be ok!
Does anyone else have issues with transfer speeds copying their recordings to PC using a USB3 drive (i.e. getting only about 5 MB/s).
I can always relegate this drive to other purposes, but if I’m going to buy another I want to make sure the make/model and particular variant will be error free (and preferably allow me to copy my tracks at USB 3 speeds…)
Without wanting to hijack this read - can anyone recommend a drive they have bought recently that reports no errors and transfers A&H recorded files at a more tolerable rate?
EDIT TO ABOVE:
The drive on the tested list IS the USB 3 variant, so I guess it ought to be removed.
In fairness the list is caveated that a particular drive may not work, but, is that because the QU is rather too sensitive, or are the HD manufacturers sometimes employing techniques that mess up real-time transfer? Either way, it doesn’t help us users much. It’s all a bit trial and error.
Now, if these are genuine errors, surely we need A&H to improve their drive validation process during formatting (or provide a PC utility if the mixer isn’t powerful enough to do it)?
Regarding your concern about read speed, remember how multitrack recordings are written to the disk. Each file is subdivided into fixed size units called clusters. When multiple files are written in parallel, consecutive clusters of each file will be interleaved with the clusters of all other streams. This reduces the requirement of head movements during recording (and playback) dramatically.
When these (fragmented) files are copied back to the PC, many read requests need to be processed by the device, since the files are not consecutive. For each read request the head must wait for the data coming from the rotating surface, this time is shorter for faster rotating drives.
When you defragment the drive or re-check with a file copied from PC, clusters are properly ordered and data can be retrieved with less read requests, raising copy speed dramatically (as observed).
On the other hand, playback from a defragmented QuDrive (or project copied back to the disk from PC) will probably not work for the same reason, since heads need to be repositioned for each cluster.
Anyway, regarding observed errors during recording you should remember that HDDs are mechanical drives with a head flying in fractions of a mm above the magnetic surface. Many drives contain some kind of shock sensors and quickly park the head in a save position if some mechanical movement is detected. Drives without such sensors may simply destroy head and/or surface.
So: If you record live, its probably better to record on flash sticks or SSD.
The forum recommendation is the Sandisk extreme usb3 stick. I have the 64gb version
Andreas - thanks. Re. cluster interleaving, that makes perfect sense. Wrt errors - well, in my case my mixer is usually setup next to my drums! The same thoughts crossed my mind, but I have observed errors being reported when not playing. A read/write test of every sector does not throw any issues, so I don’t think there’s been sustained mechanical damage (i.e. it doesn’t explain what’s happening under stable conditions).
The suggestion to switch to a non-mechanical drive makes sense. I guess the only issue is memory wear, but I gather most drives/sticks achieve upwards of a million P/E cycles these days?
Thanks for the tip Bob - I wasn’t sure about using a stick as most seem to lack the raw throughput, but reading around the SanDisk does appear to do the job. Worth a punt for 35 quid! Is this the exact model? (Getting paranoid now!)
Looks like it. I also have a short USB extension cable with a right angle at the desk end, so the whole thing is low profile at the desk.
Maybe checking the reallocation sector count from the drive’s S.M.A.R.T. info would help. Regular read-write tests don’t tell much, since the drive internally performs CRC checks and reallocates logical sectors (as accessed from the host) to spare physical sectors in case of excess errors.
Maybe your drive received some damage during transport and observed write errors may indicate new sector reallocations.
…or that particular 5400 drive is simply too slow regarding seek times…
Hi Andreas, in your post above, you said:
"And just to repeat some hints for recording multitrack on HDD:
– always start recording on a freshly formatted device (no fragmentation at all)
UNDERSTOOD TO MEAN FORMATTED BY THE QU DESK.
– do not copy additional files from a PC to that drive (recording need to seek around that data)
DO YOU MEAN LIKE WAV FILES FOR 2 TK PLAYBACK FOR PRESHOW ETC. OR SHOW, OR PATCH INFORMATION?
– do not erase files/sessions (this also requires additional seeking)
I TAKE THIS TO MEAN THAT AFTER TRANSFERRING MULTI TRACK SESSIONS TO THE COMPUTER THAT ONE CAN NOT JUST ERASE THE SESSION. RATHER, THE DRIVE NEEDS TO BE REFORMATTED BEFORE THE NEXT SESSION. IS THAT CORRECT?
– do not defragment a multitrack recording (interleaved tracks will get sorted, resulting in additional seek times)
I’m using WD Passport 500 gig and have yet to get consistent multitrack recordings from my QU24 desk. One 2hr show will record just fine and the next won’t. I think what your saying is that the drive should be reformatted before every session and that if I wanted to use another feature like 2tk playback, or virtual sound check, that it should be on a separate drive. is that correct?
Thanks,
David