Topic: problem with SD 722 recording

[JasonSobel]

Hey folks,
I borrowed some gear for a couple of shows (big thanks to Pete for the mics and to Simon for the 722).
so I was running Neumann ak40 > lc3 > km100 > 18" canare star-quad XLR cables > 722

just to get all the basics out of the way, the 722 has firmware version 1.68 running.  it's not mine, so I didn't update the firmware.
on Thursday, I ran in the "normal" gain range for the opening band, and then switched to the "low" gain range for the main act.
on Friday, it was on the "normal" gain range the whole night.  I'm recording only to the internal HD, and I'm running 24/96.

I ran this setup two nights in a row.  the first night was my test run.  mics on-stage, ran for the opening band, and two full sets of the main act, a total of ~4 hours.  no problem, everything came out beautifully.  I was also running my normal rig MG 200 > V3 > HD-P2 on Thursday, because I was testing out far I could push the 722 battery.  and I wanted to compare.  so, everything is good.

Friday, exact same setup.  the only difference is that the Nuemann's aren't on-stage, but rather, in a hat
I'm sitting front row center, so I choose not to fool around with the gear too much.  I know the show is going to be in the 1.5-2 hour range, so I know I'll have plenty of battery power.   I start the recording before entering the venue, quickly check and adjust levels during the first song, and then close up the bag and don't look at it for the rest of the show.
after leaving the venue, I open my little bag to turn off the 722, and I see that the levels are getting pegged all the way to zero, a lot.  and it's pretty quiet, so I begin to suspect that something is up.

I get home, copy the files to my computer and check it out.  for the first 1.5 hours of the recording (I started the recording before going into the venue.  so even though it's about 95 minutes into the recording, it's actually only about an hour into the show), so, everything is great to start.  no problems, wonderful clear sound, not a lot of crowd noise.  and then, about an hour and half into the recording, there's the first click.  it's just one sample, and it's all the way at the top.  clearly out of place.  from there until the end of the show, these one sample blips just keep happening, and get way more frequent.  at first, it's just one every 10-20 seconds.  by the end of the show, these miss-placed sample are happened several times a second.  the recording is almost unlistenable.

I've attached some screen shots in WaveLab.  and also a 5 second sample (it's from the original 24/96 wav file, about 2.5 MB).

so, my question is, what went wrong?  why did the 722 mess up my recording so much?

as a "fix", in WaveLab, I've been able to zoom way in, highlight the one sample in each "click", and then use the waveform restorer to replace that sample with a linear interpolation between the samples on either side.  it seems to be working, and is unnoticable.  although by the end of the show, there are hundreds of these to fix, and I have't gotten to the really bad part.  I'm really hoping I can salvage this, even though I know it's going to take hours to do.

so, any thoughts?

http://www.turtleside.com/misc/sample.wav

[F.O.Bean]

both channels too huh? WEIRD

anyone?

[JasonSobel]

yup, both channels, and it seems pretty random when they occur.
I had the file splits set to 1 gig, because I prefer to work with smaller files.  all the files split seamless, as they should.
just thought I would mention that, because I left it out of the original post.

[F.O.Bean]

was the compression enabled?

[nickgregory]

that looks exactly like the waveforms I had when the A/D board was malfunctioning.  Do you know if that unit has had its A/D replaced?

[JasonSobel]

ok, I'm tracking out the d'Elf show from Thursday night, and I just noticed a few of these same errors during the second set.  so on Thursday, the 1 hour opening band was no problem.  the 2 hour first set was no problem, but in the 37 minute second set, there was only 6 of these problems.  much easier to fix than the following night's recording, but troublesome none the less.  on Thursday, the first one appeared ~32 minutes into the second set.

was the compression enabled?

nope, the limiter was not enabled either night.  and I wasn't close to 0dB anyway.

[JasonSobel]

that looks exactly like the waveforms I had when the A/D board was malfunctioning.  Do you know if that unit has had its A/D replaced?

I don't know for sure, but probably not.  Simon bought the 722 months ago, and has never used it.  I was borrowing it from him, and going to go through some settings with him.  is there a thread on here that talked about the A/D replacement?  perhaps with a range of serial numbers affected?  I didn't even know that there was a problem...

[nickgregory]

I know there was a thread on it somehwhere, but cant find it....here is the verbiage from SD support message board

Condition
Sound Devices has determined that the analog-to-digital converter circuit board in 722 and 744T units identified below has the potential to intermittently fail. The failure condition manifests as "popping" noise or digital hash when using analog inputs. This is not a problem which gets worse over time, and it happens rather infrequently and does not happen on all units. Typically, a power cycle will clear the condition. The converters are most prone to the failure at 96 kHz sampling rates. The problem can occur at other sampling rates with less frequency. Sound Devices has seen only a handful of failures, but we will exchange ALL suspect A/D boards as a preventative measure.

Affected Units:

722: Serial numbers 470405173000 and lower which do NOT have a "v14", "v15", or "v16" sticker directly below the serial number sticker on the back panel. The "v14", etc. stickers indicate the sub-revision of the unit.
744T: Serial numbers 460405164000 and lower which do NOT have a "v14", "v15", or "v16" sticker directly below the serial number sticker on the back panel.

There are two options to resolve this issue:

1. We can send you a new A/D board and replacement instructions if you are comfortable replacing it. It involves removing the top and bottom cover, snapping the new board on, then replacing the covers. WARNING: The covers can be very hard to get back together. If you are not comfortable doing this, PLEASE send the unit in.
2. You can send us your unit and we will do the board replacement here at the factory.

Please contact Sound Devices via phone or enter a case directly on our web site at www.sounddevices.com/contact_support.htm . Select "A/D Board Exchange" as the type. Sound Devices will contact you and send a replacement A/D board in the mail at no charge.

The new boards have identical frequency response, gain, etc. as the old ones. The only change is more robust digital communication from the A/D board to the DSP. The new, replacement boards are identical to what is being shipped in brand new units.

Thank you for purchasing Sound Devices products, and for your cooperation with this exchange program.

[JasonSobel]

Nick -
big thanks for the info.  what I find interesting is the description of the condition.  for my Friday night recording, it certainly did get worse over time.  and the power cycle did not clear the condition, because, as I just discovered, it happened towards the end of Set 2 on Thursday night, and again on Friday night.  anyway, I bet this is the problem, because the 722 is in the range of affected units.

[nickgregory]

Nick -
big thanks for the info.  what I find interesting is the description of the condition.  for my Friday night recording, it certainly did get worse over time.  and the power cycle did not clear the condition, because, as I just discovered, it happened towards the end of Set 2 on Thursday night, and again on Friday night.  anyway, I bet this is the problem, because the 722 is in the range of affected units.

when I had the problem, the power cycle didnt clear it either.  Call them tomorrow, as they (at least originally) sent out a replacement board with instructions on doing the repair.

[F.O.Bean]

sucks, hope this gets resolved quickly and thats the problem......

[CQBert]

When I ran a 722 I did this swap at home... it was very easy and took less than 30 mins total from open to close.

I never had this problem but was happy that SD had identified a potential problem out there and was correcting it free of charge and being pro-Active.  I got an email from them since I had registered and was effected....

Have no fear doing this work - very easy!!

Good Luck -

CQBert

[nickgregory]

When I ran a 722 I did this swap at home... it was very easy and took less than 30 mins total from open to close.

I never had this problem but was happy that SD had identified a potential problem out there and was correcting it free of charge and being pro-Active.  I got an email from them since I had registered and was effected....

Have no fear doing this work - very easy!!

Good Luck -

CQBert

definitely easy to do...and definitely props to SD.  I had the problem on a saturday and they had it troubleshot to the board and a replacement out to me within a few days.  The mass deployment came after that...

[MattD]

When I ran a 722 I did this swap at home... it was very easy and took less than 30 mins total from open to close.

definitely easy to do

It took about 2 minutes to do the swap, but putting the damn thing back together took about 45 minutes. The reassembly is definitely easier with another set of hands present.

[trajhip2000]

I had a very similar issue - but the problem was that the FW card in my PC had gotten unseated a little, and the spikes were happening randomly during the transfer to my PC. Until I realised what the problem was I spent hours making md5s of the files on the 744 HD (SLOW at the old FW speed), copying them to my HD and making md5s again, then comparing them. Sometimes it would take me 5 or 6 tries to get a clean transfer, other times I could do a whole batch of files with no problem. Unfortunately I had a couple shows where I deleted the originals before realising the copies were bad, and I haven't yet found the patience to go back and edit the spikes out.

Your problem is most likely the A/D board since the spikes are in the original files, so I'm just posting this as something else to investigate if you see spikes (especially on a unit with an upgraded A/D board).

Steve