Then why not just fix the gain structure by individual L-pads for each channel?
UGH!
how about a properly designed digital/analog output in the the first place?
no need for L pads if the system si done right and what now i got to gain aroun 8 lpads??
Id be better of designing and building my own 8 channel preamp around the cs3318
which I don't have time for..
guess i'll stick with cascaded behringers for now.
how about a properly designed digital/analog output in the the first place?
no need for L pads if the system si done right and what now i got to gain aroun 8 lpads??
Id be better of designing and building my own 8 channel preamp around the cs3318
which I don't have time for..
guess i'll stick with cascaded behringers for now.
It would be enough to make one -23 db L-pad for each channel, the 9 db difference to the most attenuated channel is so small so it wouldn't impact SQ.
mhh, the analog outputs of the Behringer DCX is +22dBu FS = 9.75Vrms which is more compatible to pro gear than hifi equipment. Thus with respect to propper gain structure with hifi amps the miniDSPs are more compatible
How does a properly designed digital/analog output look like, having in mind that you have to interface with a wide variety of systems (amps with different gains and particularly different speakers ranging from <80dB to >110dB) and not to forget the really affordable price? It's always easy to ask for a product that does fit ones own individual needs and doesn't cost too much
mine are 5 v modded unit..
and the stk behringer has -15 db of cut
a proper digital volume is done as said before in the cpu post processing. just like the jriver or foobar media players do..
analog is easy after you get the signal form the dac.. making 8 channel track close enough is tough..
and the stk behringer has -15 db of cut
a proper digital volume is done as said before in the cpu post processing. just like the jriver or foobar media players do..
analog is easy after you get the signal form the dac.. making 8 channel track close enough is tough..
I don't see how it would matter where the calculations are done, the only thing that matters is that it is done and with what precision the calculations are done in. It makes no difference if the attenuation is done in the media player, or in the OS volume controller, or in the MiniDSP or in the built in digital volume controller in the DAC itself.
REREAD THE ESS SABER DOCS..
WHERE the volume contorl is done in the digital domain makes a HUGE difference..
you wouldn't want the dac making the volume changes to a 16 bit signal post processing .. you could end up with 8 bit s or less of signal resolution.
my point is with 32/64 bit FP available there is no reason not to do it correctly.
WHERE the volume contorl is done in the digital domain makes a HUGE difference..
you wouldn't want the dac making the volume changes to a 16 bit signal post processing .. you could end up with 8 bit s or less of signal resolution.
my point is with 32/64 bit FP available there is no reason not to do it correctly.
I don't think you understand how digital attenuation actually works : /
You do digital attenuation like this: You have a full signal in 24 bit words. You want to attenuate the signal by ~ 48 db, so you divide the 24 bit words by 256 since 2^8 = 256. You might also want to add dither before but lets ignore that for now. The important thing is that the output of the attenuation is the original signal / X, where X is how much you want to attenuate. And now for the even more important fact: You could do this calculation with 24 bits of information in a 99999 bit word length and the output would be exactly the same except for lots of unused zeroes. Sure if your DAC can input 32/64 bit floating point input words this would be another story but I have yet to see one.
And if you have to do volume changes to a 16 bit signal which would be a 24 bit signal that already has ~ 48 db of attenuation then your gain structure is messed up. If your gain structure is not messed up that means full signal volume in to the DAC means max volume in your actual system. And that means that cutting bits from this full signal equals attenuation.
If your system has such a gain structure and a 24 bit DAC then it doesn't matter where you do the attenuation because any noise that is introduced due to the resolution loss cannot be heard by humans. You don't need 32 bit DACs, or floatnig point DACs because nice and normal 24 bit DACs are good enough.
Last edited:
OllBoll, I generally agree but want to add that it can make a difference if you do attenuate (significantly) before the DSP doing a lot of math for filter and EQ calculations vs. after the DSP. Thus from the technical point of view (not marketing or subjectivism ) it should be done after the DSP, equal in the DSP itself or in the DAC.
The ESS Paper seems to be quite old since they are talking about 16Bit DACs vs their 32Bit implementation. Or maybe they just want to emphasize how good they are Do the same math with a 24Bit system which is actually standard and you will not see significant influence. Finally as long as one really believes digital volume control is evil, he will not be satisfied with such a system
) it should be done after the DSP, equal in the DSP itself or in the DAC.The ESS Paper seems to be quite old since they are talking about 16Bit DACs vs their 32Bit implementation. Or maybe they just want to emphasize how good they are
Do the same math with a 24Bit system which is actually standard and you will not see significant influence. Finally as long as one really believes digital volume control is evil, he will not be satisfied with such a system
Last edited:
OllBoll, I generally agree but want to add that it can make a difference if you do attenuate (significantly) before the DSP doing a lot of math for filter and EQ calculations vs. after the DSP. Thus from the technical point of view (not marketing or subjectivism
Of course, but in practice with this level of attenuation the signal level will be so low so you shouln't hear it as long as said before the gain structure is ok. A prime example of this is that most of all music today is stored in 16 bit format since you don't need more and thus you'd have to attenuate more than 48 db to even start to reduce SQ.
DONE EVER SAY IT CANT BE HEARD.. I can hear a 40khz ultrasonic welder running!..makes my damn teeth hurt when I walk into the welding room.IF any one welder is running..I'm not getting into audible bs on audio..
I got a 12 yr old with better hearing an musical pitch than anybody on this forum. He can hear the difference when I attenuate digital levels in the dcx below about 10 db
and I know how to do digital math..that was my whole point they are using a 28 bit processor.. so how do they do the processing and volume control?
MY laptop does its signal processing in 32 or 64 bit depending on which laptop I use and send the digital out 24 96 read about jriver or foobar
I'm saying if you are using 32 bit processing 28 bit or whatever.. you have to add 0's to then end of the bit chain!!!!! this is he PROPER way to do it before processing.. then you process then.. you can just truncate the bits n not use some funky algorithm . you will use the algorithm to convert SAMPLE RATES not word length!!
use foobar/jriver/ soundeasy and cut the volume in the digital domain using a 64 bit processor. even if you cut it 60 db!!!
you still get the max S/N ratio of the dac.. do that to a 16 bit signal and u get ****!
all I wanted to know was the mini engineered properly?? and how did it handle the digital volume..
I got a 12 yr old with better hearing an musical pitch than anybody on this forum. He can hear the difference when I attenuate digital levels in the dcx below about 10 db
and I know how to do digital math..that was my whole point they are using a 28 bit processor.. so how do they do the processing and volume control?
MY laptop does its signal processing in 32 or 64 bit depending on which laptop I use and send the digital out 24 96 read about jriver or foobar
I'm saying if you are using 32 bit processing 28 bit or whatever.. you have to add 0's to then end of the bit chain!!!!! this is he PROPER way to do it before processing.. then you process then.. you can just truncate the bits n not use some funky algorithm . you will use the algorithm to convert SAMPLE RATES not word length!!
use foobar/jriver/ soundeasy and cut the volume in the digital domain using a 64 bit processor. even if you cut it 60 db!!!
you still get the max S/N ratio of the dac.. do that to a 16 bit signal and u get ****!
all I wanted to know was the mini engineered properly?? and how did it handle the digital volume..
You sir, should contact the Guinness Book World Records!
I agree =)
And IMSTOOPID, there is so much wrong in your post I don't know where to start.
You've obviously not understood how sample values are stored and transmitted from the source to the DAC. It doesn't matter how many bits you use when calculating the attenuated audio samples. What matters is the minimum bit resolution choke point in the chain.
In this case the choke point is that the audio has to be transmitted as 24 bit samples @ your chosen sampling rate per second. This means that full usage of those 24 bits equals full volume. Less than full usage of the bits equals less than full volume => attenuation. You can't magically remove this physical limitation by doing the calculations in a 64 bit processor at the source.
If you, however, contrary to my understanding indeed have realized a way to do this then I'd suggest you write a paper about it because then you would get famous, very very famous (and probably also very rich).
you add bits before processing all zeros to end of signal chain
whats so hard to understand about that.. it raises the s/n ration by adding 0's at the end of the chain whether you add dither or not is up to whoever is doing the processing.
then you do the processing.. duh! then you set you max ( digital 0 or something a bit less preferably) and truncate ( or dither) the rest of the bits to 24.. simple enough!!
uh that's how it done on most pro studio software!!!
troll the pro audio boards
as far as the 40khz welders if you ever been around one you;d understand..if not you wont and cant explain it. like i said make my teeth hurt!
whats so hard to understand about that.. it raises the s/n ration by adding 0's at the end of the chain whether you add dither or not is up to whoever is doing the processing.
then you do the processing.. duh! then you set you max ( digital 0 or something a bit less preferably) and truncate ( or dither) the rest of the bits to 24.. simple enough!!
uh that's how it done on most pro studio software!!!
troll the pro audio boards
as far as the 40khz welders if you ever been around one you;d understand..if not you wont and cant explain it. like i said make my teeth hurt!
What you are proposing would be similarly impossible to this:
Say we have a road from point A to point B. In between A and B there is a single file bridge, so only 1 car can pass at any time.
Building a 4711 file highway from A to this bridge will not magically remove the limitation that a single car can only pass the bridge. The maximum flow of cars from A to B at any given time is still 1. Sure you could increase the speed of the cars to improve throughput but that isn't an option in our case since the protocols don't work that way.
In the same way doing calculations with extra zeroes at the source will not negate the limitation that they will be sent as 24 bit samples.
Say we have a road from point A to point B. In between A and B there is a single file bridge, so only 1 car can pass at any time.
Building a 4711 file highway from A to this bridge will not magically remove the limitation that a single car can only pass the bridge. The maximum flow of cars from A to B at any given time is still 1. Sure you could increase the speed of the cars to improve throughput but that isn't an option in our case since the protocols don't work that way.
In the same way doing calculations with extra zeroes at the source will not negate the limitation that they will be sent as 24 bit samples.
That is definitely not 40kHz that you are hearing, but the effects of changes in the load on the ultrasonic transducer, during the welding cycle, that have harmonics within the audio band.
update
So on a different note......my miniDSP and UMIK-1 mic would be arriving any day now but the DevTeam has delayed shipments of the mic until correction have been made to the sensitivity in the calibration file. They say it should only be a few days so I'll keep my fingers crossed.
From what I've read, once the sensitivity issue is sorted out the calibration file for the mic should be spot on. Since this is an instrument that requires precision, I'd say that's a good thing.
So on a different note......my miniDSP and UMIK-1 mic would be arriving any day now but the DevTeam has delayed shipments of the mic until correction have been made to the sensitivity in the calibration file. They say it should only be a few days so I'll keep my fingers crossed.
From what I've read, once the sensitivity issue is sorted out the calibration file for the mic should be spot on. Since this is an instrument that requires precision, I'd say that's a good thing.
So on a different note......my miniDSP and UMIK-1 mic would be arriving any day now but the DevTeam has delayed shipments of the mic until correction have been made to the sensitivity in the calibration file. They say it should only be a few days so I'll keep my fingers crossed.
From what I've read, once the sensitivity issue is sorted out the calibration file for the mic should be spot on. Since this is an instrument that requires precision, I'd say that's a good thing.
Fingers crossed too. Shame, have everything but one final ingredient for the recipe to be complete. Have the means to take/make measurements, but no crossover ie this calibration debacle is holding up the key part of the order. Received confirmation this morning of the delay. Ordered Feb 22nd
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.