Computer based Hifi - Page 22

i2k92 · 22nd December 2011, 10:10 PM

Quote:

Originally Posted by Javin5

boris81:

80-100 Watt should be enough. What is the fanless unit you found (brand, model)?
Aren´t there already existing Android apps for remote Windows-PC control?

Even 60w is enough, some reviews mentions that Fusion power consumption on max load is still below 40w. In case of music server, it is important to disable unnecessary peripheral and services so the consumption should fall below the number above.

Using remote desktop with a tablet is IMHO dreadful, due to small screen and touch interface. To do basic remote task (shutdown, manage services, etc) I uses OneID over the android, if I need to remote the server's desktop I use my laptop (which is rarely necessary).

qusp · 23rd December 2011, 01:16 PM

Quote:

Originally Posted by ChrisA

One thing that is VERY important are analog volume controls. Any digital control will reduce the dynamic range. It has to, it's something engineers can't work around.

yeah kinda like how engineers cant get past the fact that reducing the volume with ANY system results in lower dynamic range; its kinda what the whole thing is about

i think youve probably missed the last few years. when volume control is operating at 40bit or 64bit like quality implementations are now, you will not come anywhere near the same LACK of impact on the DNR with an analogue control

phofman · 06:58 PM

Quote:

Originally Posted by qusp

i think youve probably missed the last few years. when volume control is operating at 40bit or 64bit like quality implementations are now, you will not come anywhere near the same LACK of impact on the DNR with an analogue control

What matters are MSBs, not LSBs, since LSBs will be dropped before entering the maximum 24bit DACs with S/N resolution of 20bits max. Volume control in more than 24 bits brings no benefits (the first 24bits will be the same no matter how many further lower bits are used in the division, perhaps one more for dither calculation, but the 24th bit is below any real-world resolution anyway). Volume control in 32bits makes sense due to native numerical formats of CPUs. But the 64bit volume control is a nice selling point for commercial audio players developers :-)

DSP calculations are another story since they involve thousands of multiplications by often miniscule coefficients and additions on each sample where rounding errors at LSBs accumulate. Volume control is a single operation of dividing each sample by a static reasonably-sized number.

qusp · 06:07 AM

you misunderstand me, the dacs i'm talking about, like sabre 9018 (which is multichannel of course), the new AKM 32 bit parts, etc have an internal DSP/filter section which acts independently to the dac, also the DSP in my multichannel usb-i2s and also in completely software DSP. they all have integrated volume controls and i'm afraid the bits DO matter to the effectiveness and in some cases steepness of the filter (depending on the application), of course theres no free lunch, but compared to even the best analogue controls there is very little impact. channel matching is perfect all the way through the range, impedance is not effected, no extra parts are in the signal path, its far cheaper, it can be controlled in any number of ways including remotely, it doesnt wear. you can control as many channels as you want with one number etc etc.

analogue controls add noise of their own, the more you turn them down, but they attenuate the input noise, so DNR is impacted in 2 ways, digital the noise stays the same with respect to the signal, so when turned right down the ratio is of course higher, but given a noise floor that is low enough not to matter its a non-issue.

phofman · 24th December 2011, 12:18 PM

I am not saying digital volume control is inherently worse than the analog one. My argument is that more bits in volume control do not raise the headroom since the actual DAC circuitry does not transfer more than 20 most significant bits of resolution. No matter what is the number of bits below this level, some are hidden in the noise, the rest is dropped before entering the DAC.

DSP is performed before volume control, so that calculations are carried out on maximum information possible.

32bit DACs are built for convenience since 32 bit samples are natural language of CPUs, many transfer mechanisms (e.g. PCI controllers, Intel HDA) work at 32 bits only, and the samples had to be converted down to 24 bits before entering older 24-bit DACs. The actual resolution of 32bit DACs is the same as that of 24bit DACs, in fact I presume the actual DAC circuitry ignores the last 8 bits as it is way below physical resolution of the circuit and implementing them would be waste of energy and resources. And from outside of the chip only an electron microscope could tell

qusp · 01:37 PM

no, you really still dont get it. you have read some, but not all and understood some, but not all of what i said. lumped some bits in together, when i was talking specifically about something similar but different. i suggest you reread it, the 40 bit control i mentioned is NOT EVEN IN the DAC, neither is the 64bit, they are part of a multichannel usb-i2s convertors DSP and a piece of DSP software respectively. so i'm afraid they very much DO provide more bandwidth for the digital volume control, Crossover steepness and noise shaping filter. i'm just finishing up a multichannel digital XO, with dedicated individual i2s->dac->amp->driver path, the volume and filter are inseparable and calculated at the full bandwidth.

sabre also has one 256 step volume control, i spoke less of this, but there are all manner of processes running in it at higher bit depth and often at much higher bitrate if async is in use. it is a collection of parts in the one unit, the DSP and FIR filter section is actually operating at 48bit. i suggest you read this for some insight.

also the intel codec has little to do with 32bit support, 32bit float PCM has been in use on macs for years for audio mixing purposes

phofman · 24th December 2011, 01:54 PM

I do not argue about DSP. All I am saying is that doing volume control in significantly more bits than physical resolution of the DAC circuit gives no advantage since only as many most significant bits as is the resolution of the DAC will be used in the end and the rest of the bits will be dropped. If you disagree, please show us an example where this does not hold. I do not know of any since this is simple logic.

Merry Christmas :-)

24th December 2011, 05:56 AM	#214
qusp is choosing a less facetious title... diyAudio Member Join Date: Oct 2009 Location: Brisbane, Australia	you misunderstand me, the dacs i'm talking about, like sabre 9018 (which is multichannel of course), the new AKM 32 bit parts, etc have an internal DSP/filter section which acts independently to the dac, also the DSP in my multichannel usb-i2s and also in completely software DSP. they all have integrated volume controls and i'm afraid the bits DO matter to the effectiveness and in some cases steepness of the filter (depending on the application), of course theres no free lunch, but compared to even the best analogue controls there is very little impact. channel matching is perfect all the way through the range, impedance is not effected, no extra parts are in the signal path, its far cheaper, it can be controlled in any number of ways including remotely, it doesnt wear. you can control as many channels as you want with one number etc etc. analogue controls add noise of their own, the more you turn them down, but they attenuate the input noise, so DNR is impacted in 2 ways, digital the noise stays the same with respect to the signal, so when turned right down the ratio is of course higher, but given a noise floor that is low enough not to matter its a non-issue. Last edited by qusp; 24th December 2011 at 06:07 AM.

24th December 2011, 01:34 PM	#216
qusp is choosing a less facetious title... diyAudio Member Join Date: Oct 2009 Location: Brisbane, Australia	no, you really still dont get it. you have read some, but not all and understood some, but not all of what i said. lumped some bits in together, when i was talking specifically about something similar but different. i suggest you reread it, the 40 bit control i mentioned is NOT EVEN IN the DAC, neither is the 64bit, they are part of a multichannel usb-i2s convertors DSP and a piece of DSP software respectively. so i'm afraid they very much DO provide more bandwidth for the digital volume control, Crossover steepness and noise shaping filter. i'm just finishing up a multichannel digital XO, with dedicated individual i2s->dac->amp->driver path, the volume and filter are inseparable and calculated at the full bandwidth. sabre also has one 256 step volume control, i spoke less of this, but there are all manner of processes running in it at higher bit depth and often at much higher bitrate if async is in use. it is a collection of parts in the one unit, the DSP and FIR filter section is actually operating at 48bit. i suggest you read this for some insight. also the intel codec has little to do with 32bit support, 32bit float PCM has been in use on macs for years for audio mixing purposes Last edited by qusp; 24th December 2011 at 01:37 PM.

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search

New To Site?	Need Help?
Register to Participate Search Privacy Statement Contact Us	Frequently Asked Questions Did you forget your password? Mark Forums Read


	Please consider donating to help us continue to serve you. Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving

24th December 2011, 12:18 PM	#215
phofman diyAudio Member Join Date: Apr 2005 Location: Pilsen	I am not saying digital volume control is inherently worse than the analog one. My argument is that more bits in volume control do not raise the headroom since the actual DAC circuitry does not transfer more than 20 most significant bits of resolution. No matter what is the number of bits below this level, some are hidden in the noise, the rest is dropped before entering the DAC. DSP is performed before volume control, so that calculations are carried out on maximum information possible. 32bit DACs are built for convenience since 32 bit samples are natural language of CPUs, many transfer mechanisms (e.g. PCI controllers, Intel HDA) work at 32 bits only, and the samples had to be converted down to 24 bits before entering older 24-bit DACs. The actual resolution of 32bit DACs is the same as that of 24bit DACs, in fact I presume the actual DAC circuitry ignores the last 8 bits as it is way below physical resolution of the circuit and implementing them would be waste of energy and resources. And from outside of the chip only an electron microscope could tell

24th December 2011, 01:54 PM	#217
phofman diyAudio Member Join Date: Apr 2005 Location: Pilsen	I do not argue about DSP. All I am saying is that doing volume control in significantly more bits than physical resolution of the DAC circuit gives no advantage since only as many most significant bits as is the resolution of the DAC will be used in the end and the rest of the bits will be dropped. If you disagree, please show us an example where this does not hold. I do not know of any since this is simple logic. Merry Christmas :-)