AAAAgghr only 16bit

[alfspanners]

Hi Folks,

this is my first post on this forum though I used to be Mr Apocalypse on the original mackie forum many many moons ago. My workplace has a D8b V3(166mhz I think) 32mb ram with opt8 cards going via lightpipe to an Rme 52/96 in a carillon audio pc running cubase 6.

Last year I had occasion to look at cubase statistics for a file I had recorded with the d8b and noticed it said 16bit as the resolution(we were running cubase 4 at the time), i took some example files home and checked the statistics on a copy of sx2 which I had - they came up 24bit, so I thought panic over it's a bug with cubase. This year both myself and the studio upgraded to cubase 6 and i'm still getting 16 bit as the resolution of any files recorded with the d8b. I've analysed what's going out of the desk with stillwell audio's bitter bitscope, the top 16bits look normal, but the bottom 8 bits look like copies of a single bit though when i make a noise in the channel's mic this block of bits is increased and breaks down, but reaches the width of the bitscope which bitter's helpfile says is an indication of something being wrong. So technically there are 24bits being used cos the last 8bits are there, but there is something wrong/abnormal with them.
Now I know what you're thinking did I install the 16/24 patch that fixes the old vs 16bit output problem, well yes I did - years ago though as we were using an HDR24/96 i had no way of testing that it worked (our v3 is showing build 224 by the way). A horrible thought is that I've been recording in 16bit for the last 10 years.

So I'm still testing things - I've blown up the files in cubase and added 40db and compared them visually to known 24bit files - the mackie audio is definitely coming out 16bit. Of course cubase is writing them as 24bit files, so I tried reconverting the mackie files to 16bit proper and phase cancelling with the original _ a lot of noise is present in the cancelled out sum and distortion in the loud passages, noise is aprox -103db I have amplified it and blown it up this noise is 16bit stepped with the odd very miniscule 24bit sized ripple - this is the wierd blocking that i'm seeing in the lower 8 bits in bitter, but just viewd as a waveform

the remainder in 16bit sized blocks with 24bit sized ripples

remainder2.PNG (Array KiB) Viewed 2847 times

. I tried turning the mic down to minimum while watching bitter and could never get the 16th bit to switch off, even if I turned down the d8b digital trim to anything other than off. the only thing that will switch the 16th bit off is turning down the level to tape, interestingly you have to turn it down very far before cubase will acknowledge anything other that 16bit - I think I had to turn it down about 15bd's or so just to get a supposedly 20bit file (which makes me think that the old work around for the 16bit issue was an illusion as all it did was decrease the stepping in an already 16bit file). Also I've sent a known 24bit signal from cubase through the tape in 3 off the desk and routed to tape on channel one and It definitely comes back into cubase as a 24bit file. I've tried switching compression, eq gate on a off - no difference. The only time I can get the analogue going in to become 24bit going out is when I switch on the uv, bit it's only the uv that would appear to be operating at 24bit (or have some bits down there) and I don't want to be compressing uv afterwards . If this were the old fashioned analogue world I'd had said that the problem lies somewhere at the input (AD), but of course It would be that simple lol.

I need to fix this, or replace the desk. I realise that neither myself nor any other clients (including a number of successful musicians and producers) have noticed that there was a problem, but now that I know there is a problem it would be dishonest to pretend otherwise and professionally that is not an option. It's an old desk, true, but as it's not mine I'm not sure whether a replacement can be bought and how long that would take (the replacement certainly has to do less that the d8b started out doing - I only use it for ad da conversion and cue mixes - all my mixing in done inside cubase).

Does anyone else have/had the same problem

- is everyone else certain that they don't??????

Would upgrading to V5 fix this or is it possibly something else.

Help please!!!

thanks for reading

Mark

[Crash]

Do you have any UV22 dithering going on? You should have seen what the output of the console was with a HDR hanging off your d8b. It shows bit depth and clock speed. If you don't want to upgrade to Version 5.1, I would try a reinstall of the patch.

[alfspanners]

Hi,
thanks for your reply> No there is no Uv as there was no 22k spike in the noise profile when i examined it in my analyser, when i switched uv22 on - an obvious spike appeared. It always showed 24bit on my hdr and that's because the output is 24bit from the desk ie the wordlength is 24bit, it's just that when recoding from the mic ins, only the top 16bits contain the audio - the other eight seem to be broken - imitating a 16bit stepping pattern.

I've tried reinstalling the patch twice and it keeps telling me it's installed.

[Bilo2012]

Hi Alfspanners,

i think i know what your Problem is.
Did you checked your EPROMS on your Optical
I/O Cards ? If there V1 writes you have 16 Bit Translation.
You need V2 EPROMS from Mackie they are 24 Bits.

Hope this will the solution for you.

Best Regards

[alfspanners]

Thanks,

i haven't checked those (and I will tomorrow) , though I am able to send a 24bit file from cubase through the desk (via the opt 8 cards) and back in one piece and the end result is definitely 24bit so I'd be suprised if it is the eproms.
thanks again

[alfspanners]

Come to think of it - I thought it was only the DIO-8 cards that had the eprom issue anyway - ours are the opt-8 cards

[Crash]

Come to think of it - I thought it was only the DIO-8 cards that had the eprom issue anyway - ours are the opt-8 cards

That's was the way I remember it as well.

[Crash]

It always showed 24bit on my hdr and that's because the output is 24bit from the desk ie the wordlength is 24bit, it's just that when recoding from the mic ins, only the top 16bits contain the audio - the other eight seem to be broken - imitating a 16bit stepping pattern.

Maybe I am dense but I am not following this.... Are you not saying that you are getting 24 bits out of the desk right here? I am not trying to be a smartass but I am not sure I understand your "Mic ins" comment as it relates to 16 bit and then the statement that the desk output is 24 bit....

[Crash]

Ok, I think I am following what your saying now that I have read and re-read your posts. So you can send 24 bits digitally into the console and get 24 bits out, yes? But when you take an analog signal into the console, you are saying the output is 16 bit and not 24 bit, yes? Just trying to clarify here.

[alfspanners]

Hi crash, what I'm saying is as far as I can tell - the desks is outputting data in a 24bit format - there are 24 of those bits for each sample, however for some reason, the analogue ins (mic/line) are only sending a 16bit signal and the remaining 8 bits are being filled by sort of copies from the 16bit signal. When the signal is recorded in cubase the step size corresponds with 16bit rather that 24bit which is why Cubase 4 and 6's more intelligent statistic dialogue is saying that they are 16bit. Cubase sx2 must just have looked at the wordlength rather than the waveform. the waveform is clearly 16bit.
It is of course possible that the 8bit artefacts are the result of a 16bit sgnal going into the rme card, but the signal is 24bit for sure when I'm digitally thru-putting so it seems more likely the artefacts and truncation are from the mackie.