Strange phenomenon!

[Razmo]

Hi!

Just thought I had to share this with you guys:

I'm as everyone here surely knows a gearslut, and have woved my life to analog mixing etc. (those of you who don't know this do not spend enough time here as I do ).

now I've been toying with the idear of recording my analog mixes directly into 24bit 192Khz... until now I've not really done it (using 44.1Khz), but after trying it recently, I stumbled on a weird phenomenon... When I recorded my analog mix faster than 44.1 Khz, there seemed to be more noise introduced, and at 192Khz it simply went insane! the noisefloor raised from about -75db to a staggering -44db!!! ... I found that this noisefloor raised proportionaly with the samplerate, so 96Khz gave a raise to the noisefloor also, though not as much.

My brain stopped trying to figure out the reason, and I wrote a mail to E-Mu who make the 1212M soundcard I'm using for recording... and after three or four mails, I finaly found the reason for the phenomenon:

Using an analog mixer, you also have "higher than human ear" frequencies (way above 44.1Khz), and when you start recording in 192Khz, you let a lot of supersonic frequency material through the recording process, and sometimes this noise can be quite loud!

I started on a journey to find WHAT in my studio that is the culprit... turning off all gear, I had a beutiful noisefloor at about -70dB at 192Khz... and it stayed there with all the rest of the equipment exept for ONE! ... the Elektron Machinedrum! ... that bloody culprit is oozing highfrequency material! ... guess I'll have to stick with 44.1Khz, or find myself a lowpassfilter for the Machinedrum

Just wanted to share the experience... maybe some of you could end up in a like situation... just don't forget that supersonic frequency! (this has no relevance for plugin sluts I guess).

[PsychNerD]

Same soundcard here so im curious,

you could run loops on each piece of gear (that can) using RMAA, and then you'll know what happens and where...

http://www.rightmark.org/

[Razmo]

PsychNerD:

Yes... I think I'll do just that one day, just for the fun of it... as written above, I've found the culprit (machinedrum). When recording in 192Khz, 24bit with the mixers input turned down for the Machinedrum, the noisefloor lies below -70db (as per the E-Mu PatchMix control software's meters). I'm perfectly satisfied with anything below -66db which is the borderline for what is considdered "silence"... in the past I was VERY noise-fobic, but not anymore... as long as it stay below those -66db that is

It's just incredible that a piece of gear, that is this new and modern (and darn expensive!) has such a huge amount of supersonic frequency noise.... I wonder why (have written Elektron to find out)... I'm beginning to wonder if it's a hardware fault in the machine

[PsychNerD]

I bet it's related to internal clock or other digital signal leaking into analog... (based on the frequency). Like you said, you can simply use a lowpass (capacitor) at the output at a very high frequency or insert an expander/gate that will cut the signal when unused.

[Razmo]

PsychNerD:

Yes I could do that, but I got my reply from Elektron today, and they say the reason lies with the DAC... when the Elektron was developed 192khz sampling was not used. They actualy plan on putting new DACs in future SPS-1s.. now as if that will help me!

They suggested a device called "mutator" to tame those high frequencies, and make other fun things with... got to have a look at that device.... heards the name, but cannot remember what it really is... or even worse; how expensive it was

anyways, it's not really that big a deal... I'll manage... I just thought it was a not so obvious reason to the noisefloor prob. that you do not really consider when trying to find a reason for this problem.

I'm questioning the DAC issue though... I've got digital gear from all the way back to the mid 80's... they do not "noise" like that!... I bet they have this capacitor filtering on their output, so it's simply bad hardware planning on Elektrons side I would believe! ... and it does not surprise me... they are a relatively new hardware company, and their first product, the SID Station showed imense signs on this (crashes and malfunction + bad build)... now the Machinedrum has gone a long step from that SID Station, but obviously they still have their small quirky auras

[PsychNerD]

I would be curious to see the spectrum of that noise (plugged directly into the 1212) to see if it's a narrow/peak or a wide spectrum noise...

I think i will start to do some RMAA tests on my side, synths noise spectrum, freq response of console, recoder, compressors, etc... maybe I will have some surprises on my side, but i'll stay in 44.1k anyway... respecting the garbage in garbage out principle (or rather quality in, garbage out)... no need for 192 here... i have a 32 channel LPF called a MX3282A in the path anyway ...

Mutator ?

http://www.mutronics.co.uk/html/mutator.htm ? It seems to cost around a grand... to fix a problem... what a shame Elektron... I would've suggested a passive lowpass with a relatively high frequency cutoff before paying that much for a dedicated patch to the problem... it can even be fixed internally...



with a 500Ohms and 0.1µF passive filter you have about a 32kHz corner freq with 1 pole slope...

anyway, you probably know this stuff already, was thinking out loud...

Hail hardware!

[Razmo]

PychNerD:

No, actually I'm just an electronics novice, so I know shit about this, but understand your point... I'm not so dumb though, that I'll buy a device at that price to fix the problem, and funny enough, I thought of exactly something like what you just illustrated... I just do not know how to do this I guess.

About the noise, I can surely record a 192khz 24bit wav file for you, if you are that interrested in analyzing it

Actually I'm not much of a noise hunter... I mean, I suspect that a cablemess mixed with several powercords and wallwarts on the floor is subject to more noise also... not to speak that the most of the connections to the mixer is unballanced (most synths have unballanced jacks)... so I simply go with the -66db principle... but of course it would be nice to be able to find out if anything can be minimized... but for now, my mount doom eye of focus is set on the Machinedrum!

I was thinking; can this passive filter you wrote about maybe even be build directly into a cable? ... should just find a jack with enough room... will not attempt to fix anything inside the Machinedrum... mostly because I do not want to damage anything, but also because Elektron use some really weird screws for which I do not have the right tool.


Now that 44.1 vs 192 debate... I'm not so sure what to think myself, but to be honest, I've read somewhere that the post processing DSP can gain extra benefits from a 192 file compared to a 44.1 one . I'm thinking about the mastering process here, doing multiband compression, EQ and levelling. I'm not even sure I'd be able to hear the difference either... and in the end we end with a degrading MP3 conversion also... I just keep thinking: "Will the end MP3 benefit from having mastered at 192khz? ... until now all this has been thoughts only, and nothing has been proved in practice for me.

Also what about downsampling? ... if you record in 44.1 there will be no downsampling... but with 192 there will... and when at it, I believe it should be better to go higher than 44.1 in multiples, so actually 176khz should be better because this would be a fraction-less downsampling DSP operation... ARGH! without any audible proof, it's impossible to believe what's true/best maybe this is why I'm tempted to go with the "believe your ears" approach now...


Your last line: you're absolutely right!
no, seriously I can see many tempting aspects in the plugs world... also many woes etc. but for me it all boils down to the fact, that when you have listened to an analog synth you KNOW you got to have this sound! ... that's how it works for me. I've dissed plugs and PC music production totally simply to avoid PC instabillity and to constrain myself... work best with limits... I have a tendency to want the best of the best, and when I'm presented with the sheer number of options in the software department, I short curcuit my brain! ... it would be impossible for me to get anything done because I'd want everything

erm.... anything else you'd like to know?

[PsychNerD]

About the noise, I can surely record a 192khz 24bit wav file for you

Only if you have some spare time, could be interesting...

Actually I'm not much of a noise hunter...

When you are working with hardware, even more when you have old gear, you must forget about that noiseless sanitized sound

can this passive filter you wrote about maybe even be build directly into a cable

Shame I did'nt think of it... of course... a resistor and a cap is not that big... that would be perfect. My two concerns would be final gain loss and final impedance.... I would personally not be afraid to give it a go at these values...

processing DSP can gain extra benefits from a 192 file

For processing that's a fact, just figure yourself working with a 300dpi picture instead of a 72 one... more manipulations means lesser loss at higher resolution, same for audio...
also, you get more resolution if you record real (acoustic) instruments or real analog gear (that generates harmonics over 22k)... other than that, digital in the path screw it all anyway creating aliasing and the rest... I personally don't care and go for 16b/44.1k even if I could at least go to 48, everything often needs rate conversions and I hate these manipulations even more. Can't hear much of a difference anyway...

Also what about downsampling?

Depends on what does the conversion... but it is in fact processing. But keep in mind that the EMu1212 probably does internal conversion when running at 44.1k anyway, dont know it's 'native' rate (or if it is truly switching it's clock). Cheaper chipsets ofter works internally at 48k and working at 44.1 makes a 44.1/48/44.1 conversion path... anyway...

no, seriously I can see many tempting aspects in the plugs world

<flamebait>plugs have their problems too... I am obviously hardware oriented so for me vsti are like 'virtual sex'... better go for the real thing... </flamebait>

when you have listened to an analog synth you KNOW you got to have this sound! ... that's how it works for me

same here... anyway, they often are too easy to obtain, you don't get any fun out of this. it's like activating cheats on a videogame... you play and quickly get bored.

constrain myself... work best with limits

it's the 303 story, not easy, does'nt sound well (by the specs) but god... I can get it out of my head... (no ELO pun here)

anything else you'd like to know?

Oh well... not really... but that was an interesting thread...

[Razmo]

PsychNerD: Why work at 16 bits when you can work at 24!? ... I don't quite get why you'd do that?

Also, the 1212 IS switching it's clockrate internally... that was their biggest mistake with their Audio Production Studio card earlier. There is no rate conversions going on in that card, and it's possible to see this cause if you choose 98 or 192 khz, nothing that play in another rate will sound at all.... exept that it seems like if something IS going on... the rates seem to be gangeed in pairs... 44/48, 88/96, 176/192... any playback within the pairs will work (48 at 44.1 and vice versa) so maybe there IS some convertion going on ... I cannot tell... but the specs state that no conversion is going on, so maybe they are right.

I have the same "virtual sex" opinion like you... though there are some things that I miss from the plug world, like automation... everything is saved with the project. Saving changes made in hardware for a project is a F****** nightmare! ... second miss are the "clean studio look"... it takes much less space. though, virtual have all odds against it with respect to hardware... to me that is.

and hey! that last comment I made was not meant in a way that I did not like your response... it was aimed at meself because I easily get into a rant ... just ask anyone else

[PsychNerD]

I work at 16b only because my recorder doesnt support 24... I dont record directly on the 1212, I have a 32x8 console, the 8 subgroups are connected to a DR8 that I use for recording (at 16/44.1) then when i'm done, i transfer the 8 tracks via the 1212 but using lightpipe/adat... I prefer having the 8 track flexibility, everything is grouped in a way I can do corrections after recording. Like you said, it's a nightmare to do reworks once it's recorded... having it multi-tracked helps a bit...

I just benchmarked my recording setup and it does'nt seems to be that bad for all the low-end gear i have... here is the spectrum from :

1212m out->console input->bus routing->clx-440->dr8 in->dr8 out->1212m in

that's my standard path (except for the DR8 analog out that I replace with lightpipe...)



-86.5 noise floor mostly falling below -100 when the expanders kicks in... seems that I have a bit of 60hz hum I did'nt realized I could fix... of course it's only for 1 channel... when everything is on, add all the synths noise.



And btw, i did'nt took your last comment like a rant or anything like it...

For the gear mess, looks like we have a setup about the same size... but I do have room space, that helps a lot. My setup is not messy at all.

[Razmo]

PsychNerD:

I have a little question on the picture you posted of the lowpass diagram; I understand it like this:

the "IN" connection is where the Machinedrum's output (possitive lead) should go, and the "OUT" connection is where the positive lead is connected to the cable right? (even though this is happening IN the cable)... also, the "GROUND" after the capacitor is the connection to the negative lead in the Mchinedrum Output... right?

What if the cornerfreq was to be at approx. 44.1 or 48 khz? do you know what the specs of the two components should be then?

EDIT: I've located the WIKI where you found that diagram, and the calculator, so far so good... but when I insert those values you wrote, it does not relate to 32khz... it fits with 50 Ohms, was that more like what you meant?

I also think that page explained how to wire it all up.. simple and easy really... I'll try it out later and see what it does for me Also, I believe that the better the tollerance of the resistor, the more precise...

But thank you very much for the idear... very much appreciated!

[Analog-X64]

Hey Razmo!!!

Buy this one for me... let me know and I'll give you my home addy.

http://cgi.ebay.ca/ws/eBayISAPI.dll?Vie ... 0173415397

[Razmo]

ANX: he he!... sure you'd like that! ... but that one is surely too expensive for what it does ... besides, I'm not the biggest of Roland filter fan... Roland's filters are a bit too warm and soft for me, and those who spend mega bucks on a Jupiter 8, is definitely NOT me!