KeithL
Administrator 
Posts: 10,273
|
Post by KeithL on May 3, 2022 10:14:24 GMT -5
Another MAJOR consideration with "passive preamps" is cable length and location. The output impedance of a "passive preamp" is MUCH higher than the output impedance of an active preamp... And it also varies considerably depending on where the volume control is set... The "passive preamp" puts a variable load on the source - which probably doesn't care much.
But it also offers a varying source impedance to the amp - which can make things act differently depending on the input impedance of the amp.
Therefore the cables between it and the power amp are going to be much more sensitive to picking up hum... And that sensitivity will vary depending on where the volume control is set...
And, if those cables are long, you may even experience some slight high frequency roll off, which may also vary depending on the volume setting... And both of those may be different with different amplifiers connected... (Especially tube amps - which tend to have a much higher input impedance than solid state amps.)
You will probably find that the cables between the preamp and the amp are FAR more likely to pick up hum with a tube amp connected. (So you'll need to be more careful about keeping those wires as short as possible and keeping them away from power lines.)
And you may also find that the output LEVEL "responds differently to where you set the knob"... (If you actually had balanced connections, and a balanced passive preamp, that combination would minimize the hum aspects of the situation.)
All of this will also depend on the exact value of potentiometers used in your "passive preamp" and how they're configured.... (10k is pretty common.)
So, depending on a lot of factors, a "passive preamp" may actually be a tiny bit better, but it could also be worse in other ways...
To preamp or not to preamp - that is the question... In my system, I need but two inputs, the DAC output from my music system and the analog output of my blu-ray player. I'm currently using the Emotiva BasX PT1 preamp as a switching device. I do, however, have a passive switching box that can put up to four inputs to a single analog out. My Emotiva Big Ego+ DAC has its own (digital) volume control built in. The disc player has fixed output. But I also have a remote-controlled passive volume pot that I could use either before the switching box (controlling volume on the disc player alone) or after the switching box (controlling volume on all sources). I'm curious to know if elimination of the active preamplifier will result in better sound. My best guess is that the active preamp will provide 95% the sound quality of the passive option. The PT1 is an exceptional preamplifier. In the past, however, my experience in removing an XPA-1, Gen. 2 from the system resulted in a HUGE improvement in soundstage. I'll post my results here... Boomzilla |
|
|
|
Post by Boomzilla on May 3, 2022 11:34:05 GMT -5
You seem to be asking about whether to use a crossover on the satellites… The short answer is that the right answer is going to depend on the actual response of the satellites... This means that regardless of the results I get, they won’t necessarily apply to anyone else’s system. Thanks for the info KeithL. I knew some of that, but not all. Another option I might try is to use the electronic crossover on the PT1 and ALSO engage the low-pass filter on the sub’s plate amplifier at the same frequency. This would roll off the sub at 24dB/ octave, reducing the chances of muddying the sound above the ℅ frequency. |
|
|
|
Post by leonski on May 3, 2022 20:58:33 GMT -5
I'd really love for an engineering type to chime in on the idea / measurementsof 2X crossovers of say.......12db / octave, IN SERIES....IF they are both low pass OR high pass.......
|
|
|
|
Post by Boomzilla on May 3, 2022 21:42:03 GMT -5
I've tried the two 12-dB filters in series (the built in 90-Hz low pass of the PT1 in series with the 90-Hz low pass of the subwoofer plate amplifier. Sounds much better than just the PT1 crossover alone. Rolling off the sub more steeply cleans up the crossover well.
But it's not all peaches at the ice cream parlor just yet... The 90 Hz crossover frequency is still too high for the transition to be seamless. Not the fault of the crossover - the bass can't be identified as a different source location by ear, but the speed difference between the satellite woofers and the sub cone can be clearly heard. My next attempt will be to run the satellites full-range, blending the sub in below the roll-off frequency of the satellites. This will mean that I can't use the PT1 crossover at all. Since the sub low-pass filter is "only" 12dB / octave, I'll need to buy and install an inline low-pass filter of roughly 50 Hz. Should be available from car stereo sources - otherwise I'll build my own.
My satellite speakers, being ported, will roll off naturally at 12dB / octave. If I roll off the sub at 50-65 Hz. at 24dB / octave, I should have an inaudible match.
Boomzilla
|
|
|
|
Post by leonski on May 4, 2022 0:00:11 GMT -5
My sub, a HSU is already 24db /octave. My main speakers are low-cut at about 50hz to 55hz at 12db / octave. If you are interested in an ACTIVE crossover solution? sound-au.com/project125.htmOthet solutions in the ESP universer are available....I saw a 2way with 18db / octave slope.... And yes, I tend to agree. a LOWEr crossover will benefit your end result. Less mush in the crossover region and less chance of phase issues mucking up what's LEFT. I would still like an engineer type to chime in on series crossovers.....do 2x@12db result in 24db or something higher? |
|
|
|
Post by Boomzilla on May 4, 2022 10:10:42 GMT -5
...I would still like an engineer type to chime in on series crossovers.....do 2x@12db result in 24db or something higher? ? My (non-audio-engineering) thoughts on this = It depends. If the two filters are directly and passively connected in series, then unintended effects may occur. However, if the two filters are separated by a buffer stage (op-amp, etc.) then the fixed input impedance of the buffer stage will isolate the two filters from each other. Virtually every component input has an independent buffer stage, so the likelihood of unintended effects should be low. For example, in my setup, the inline passive filter will be connected to the input of an Emotiva "Virtual Copper" wireless sub transmitter. The transmitter has its own input buffer stage. The receiver will be hooked directly to the subwoofer's input stage (that also has an input buffer). So in my setup, there is no possibility of the two filters interacting. Boomzilla |
|
KeithL
Administrator
Posts: 10,273
|
Post by KeithL on May 4, 2022 15:37:13 GMT -5
It's actually a little bit of both...
Putting passive filters in series tends to cause a LOT of interaction... Some suggest that the best way to minimize this is to feed a higher impedance filter from a much lower impedance filter. So, for example, use a 1k resistor and a 1 uF capacitor for the first one.... and a 10k resistor and 0.1 uF capacitor for the following one...
However, this causes other issues, so isn't a great solution... Putting a buffer between the two stages is the best way to avoid this...
The other issue, which remains, is that filters don't always calculate out the way you think.... For example: - a "12 dB/octave 80 Hz low pass filter" will be -3 dB at 80 Hz and roll off at 12 dB/octave far below that.... - a "24 dB/octave 80 Hz low pass filter" will be -3 dB at 80 Hz and roll off at 24 dB/octave far below that.... - BUT "two stacked 12 dB/octave 80 Hz low pass filters" would be -6 dB at 80 Hz
In addition to that there are lots of other factors involved with filters... - each type of filter has slightly different characteristics above the cutoff frequency - each type has different characteristics AROUND the cutoff frequency (you may get the flattest response by combining one filter with a "bump" and one with a "sag")
- and even the slopes near the cutoff frequency may actually be different (a real "12 dB / octave" filter does not have a nice round corner and then a ruler straight slope after that)
- and each also introduces different sorts of phase shift around the cutoff frequency
Phase shift is inevitable with filters - it's NOT a flaw, or something you can avoid by being careful, but is actually part of how a filter works (we're NOT talking about a few degrees either)
For example, one reason why the midrange is wired IN REVERSE in may three-way designs is because they use a two-stage passive crossover filter... And the 90 degree phase shift at the crossover point in each filter adds up to a total of 180 degrees of phase shift at that frequency...
So, in total, the signal is fully inverted at the crossover point, and so it ends up back in phase when you flip the wires.
The bottom line is that, when designing a filter with more stages, and a sharper cutoff... The designer may often use two different types of filters whose properties complement each other... And things like phase shift must need to be taken into account...
And you may get far more optimum performance by doing so that by simply "stacking two identical filters" (If you look at most professionally designed multi-stage filters they are NOT simply multiple identical stages stacked up.)
At a very minimum DO NOT assume that, if you do choose to stack two filters, their effects will "add cleanly"...
It's not that you can't do it, of can't get it to work well, but it is NOT fair to say that "the odds of at least some unintended effects will be low"... Don't count on it.
The short answer to Leonski's question is that, while it tends to end up that way well above and well below the cutoff frequency... The exact curve and slope near the cutoff frequency can end up being much more interesting.
...I would still like an engineer type to chime in on series crossovers.....do 2x@12db result in 24db or something higher? ? My (non-audio-engineering) thoughts on this = It depends. If the two filters are directly and passively connected in series, then unintended effects may occur. However, if the two filters are separated by a buffer stage (op-amp, etc.) then the fixed input impedance of the buffer stage will isolate the two filters from each other. Virtually every component input has an independent buffer stage, so the likelihood of unintended effects should be low. For example, in my setup, the inline passive filter will be connected to the input of an Emotiva "Virtual Copper" wireless sub transmitter. The transmitter has its own input buffer stage. The receiver will be hooked directly to the subwoofer's input stage (that also has an input buffer). So in my setup, there is no possibility of the two filters interacting. Boomzilla |
|
|
|
Post by Boomzilla on May 4, 2022 15:50:40 GMT -5
It's also fair to say that a LOT of crossover design issues can be detected with REW software. What the microphone hears tells you whether you have a dip, a bump, or a phase issue at the crossover point. Once the output of the filter combination is diagnosed using its actual acoustic output, correction can be applied (within reason) using parametric DSP. It should go without saying, but I'll say it anyway, moving the sub and then adjusting the phase on the subwoofer plate amp are the first steps in "fixing" crossover issues. In fact, the sequence for subwoofer troubleshooting should be:
1. Move the subwoofer to ensure that the crossover frequency issue you're measuring is NOT a room artifact.
2. Adjust the phase on the subwoofer plate amp to see if you can improve the behavior at the crossover frequency.
3. Once adjusted optimally, move the crossover frequency slightly up or down. Sometimes more or less overlap with the satellites can smooth out the crossover.
4. Once the crossover region is as flat as you can get it, then (and only then) apply DSP equalization to further smooth the crossover frequency (note: any more than 3dB of boost is not a good thing. Cuts of any amount are mostly fine.)
Boomzilla
|
|
|
|
Post by leonski on May 5, 2022 19:45:36 GMT -5
That's why IF I experimented with a MiniDSP, I'd 100% try to figure out FIR filters. NO phase shift thru the passband.....than adjust a few Miliseconds of time delay to 'time align' the drivers and bingo....
My Maggies use a 6db / octave high pass and a 12db / octave LOW for a crossover of 600hz....But the drivers are 90 degrees apart. So it is desired to have the late driver Nearer the listener by a few inches.....
Which now that I think about it, might be a DIRAC issue as well, depending on exact frequency which is measured......Or are they all 'sweeps'?
I got caught by Boom's #1, above. But I wasn't measuring, I was LISTENING. Sub was way off. And my den had a horrible one-note bass problem.
Moving the sub fixed everything and now it is quite musical without boom or other artifacts.
|
|
|
|
Post by Boomzilla on May 6, 2022 22:29:23 GMT -5
Now running my satellites full-range with inline, second-order, 50-Hz. low-pass filters on the sub. Since the speakers are rated for a -3dB of 45Hz. (and since the manufacturers all lie), this should work. Of course, women should follow me around like puppies after a pork chop, but no luck with that either… The bass just sounds so-so. I know the sub is in the right spot, so it may be time again to break out the old UMIK-1 again & fire up REW.
My wild & crazy audio amigo (who is right more often than not) advises me that his CD rips sound better through JRiver than through Exact Audio Copy. Anyone else compared the two? He also claims that his rips sound better when copied to a SD card than when ripped to an internal HDD, an external HDD, or a SSD.
Postscriptum:
My speakers sound FAR more open and image better when being driven by my PT1’s full-range output than when driven by the PT1’s onboard high-pass crossover. Am I just imagining this? Would the speakers sound better yet if I took the PT1 out completely and drove the power amps directly from the digital volume control of my Big-Ego+?
If I were to drive the amps directly, how could I create a blended sub signal without shorting the two channels into mono?
On a different note, before going to the hassle of REW measurements, I think I’ll try some Roon DSP. A small boost at the crossover frequency may restore how things should be sounding…
|
|
KeithL
Administrator
Posts: 10,273
|
Post by KeithL on May 8, 2022 3:21:22 GMT -5
If you (or your buddy) are hearing differences between RIPs then you should try doing an actual bit-compare on them. (I've found that the easiest way to do this is to run a checksum on one file... then "verify" the other with it. ) (This is assuming that, regardless of how they're ripped, they're ending up being actually played from the same media.)
Either they're identical - in which case the differences are placebo-related...
Or they're NOT identical - in which case something is doing the RIPs WRONG... (There is no possible excuse for two different programs to RIP the same disc and end up with different data - unless one of them isn't doing it correctly - or there are un-correctable errors on the disc itself.)
If you want to make sure you get proper RIPs you should use something to actually verify them - like AccurateRip. (Of course, if you have multiple copies of the CD, make sure you use the same exact one... even different "pressings" of the "same" disk are often slightly different.)
When you use the filters on the PT1 you are passing the signal through another stage of active electronics. There's also going to specifically be some phase shift around the crossover point - which is unavoidable with filters. (I've never listened to the outputs side by side but I wouldn't be surprised if they sound a tiny bit different.)
(Also remember that even a tiny mismatch in level, which may not be audible as such, will make things sound different when comparing them).
Now running my satellites full-range with inline, second-order, 50-Hz. low-pass filters on the sub. Since the speakers are rated for a -3dB of 45Hz. (and since the manufacturers all lie), this should work. Of course, women should follow me around like puppies after a pork chop, but no luck with that either… The bass just sounds so-so. I know the sub is in the right spot, so it may be time again to break out the old UMIK-1 again & fire up REW. My wild & crazy audio amigo (who is right more often than not) advises me that his CD rips sound better through JRiver than through Exact Audio Copy. Anyone else compared the two? He also claims that his rips sound better when copied to a SD card than when ripped to an internal HDD, an external HDD, or a SSD. Postscriptum: My speakers sound FAR more open and image better when being driven by my PT1’s full-range output than when driven by the PT1’s onboard high-pass crossover. Am I just imagining this? Would the speakers sound better yet if I took the PT1 out completely and drove the power amps directly from the digital volume control of my Big-Ego+? If I were to drive the amps directly, how could I create a blended sub signal without shorting the two channels into mono? On a different note, before going to the hassle of REW measurements, I think I’ll try some Roon DSP. A small boost at the crossover frequency may restore how things should be sounding… |
|
KeithL
Administrator
Posts: 10,273
|
Post by KeithL on May 8, 2022 3:44:24 GMT -5
Do bear in mind that, even with digital DSP-based filters, what they show you is a theoretical response... Even "digital filters" do not deliver an output that is "exactly perfectly what the picture shows"... And how close will depend on all sorts of complicated things you really don't want to get into (like how many taps they use and what level of precision)...
(They're usually far closer than makes any difference for audio - because our hearing isn't at all precise either.)
NOTE that, when you measure things with a MICROPHONE, in a ROOM, it's not at all the same as measuring an electronic SIGNAL. There are several options that determine exactly what test signal is used and how it is analyzed afterwards... You will find that REW offers you quite a few options in that regard... And, while Dirac doesn't offer you any options you can pick, they have their own method, which may not equate to ANY of the options in REW...
(It's not nearly as simple as "just set the window to exclude room reflections" or "pick a chirp instead of a click".) (And, of course, Dirac isn't even trying to exclude room reflections, since it's including them in its calculations....)
And, yes, it is generally a good idea to "time align" the drivers so that the arrival time at the crossover frequency matches. However note that, along with the electronic phase shift of the crossover, there is also a sort of "mechanical phase shift" associated with the physical driver itself. So don't assume that the "real phase shift" is going to match the "calculated phase shift" from the crossover alone. I have no idea how close this is on Magneplanars - but it can be off by hundreds of degrees with dynamic drivers.
It's much safer to play some sort of test signal that produces a transient at the crossover frequency...
Then stick a microphone at the listening position and look at the output on an oscilloscope... And "adjust things until you see one leading edge instead of separate ones from each driver" (that's kind of a simplification but the right idea).
And, unless you have a way to adjust the delay on each driver separately electronically, that's something that no room correction or EQ can exactly "fix in post". (Although components are often included in the passive crossover designed into the speaker to help electronically time align the drivers.)
That's why IF I experimented with a MiniDSP, I'd 100% try to figure out FIR filters. NO phase shift thru the passband.....than adjust a few Miliseconds of time delay to 'time align' the drivers and bingo.... My Maggies use a 6db / octave high pass and a 12db / octave LOW for a crossover of 600hz....But the drivers are 90 degrees apart. So it is desired to have the late driver Nearer the listener by a few inches..... Which now that I think about it, might be a DIRAC issue as well, depending on exact frequency which is measured......Or are they all 'sweeps'? I got caught by Boom's #1, above. But I wasn't measuring, I was LISTENING. Sub was way off. And my den had a horrible one-note bass problem. Moving the sub fixed everything and now it is quite musical without boom or other artifacts. |
|
|
|
Post by Boomzilla on May 8, 2022 13:33:31 GMT -5
I've heard the claim again and again - digital bits are bits. A copy is a copy and if the bits are identical, then the copy must sound the same as the original. I understand the error correction and verification algorithms just fine. And intellectually, I agree with the claim - an identical copy will always provide an identical sound. End of discussion.
But I'm no longer convinced that it's so. When I originally moved my music library from analog to digital, I use ALE format (Apple Lossless Encoding). My digital copies didn't sound as good as my analog sources. So I converted some files from ALE back to uncompressed WAV format, and heard a big improvement.
Next, I went back and re-ripped some of the source material directly to WAV. And the "native WAV rips" again sounded better than the ALE to WAV conversions.
Then, I used a variety of ripping programs including jRiver, Exact Audio Copy and others. I found that the different ripping programs sounded different from one another. I then used different ripping speeds. Sounded different. I then used different drives to rip from. Sounded different. Then I tried different destinations including WD drives, Seagate drives, SSD drives, internal drives, USB drives, and SD cards. Sounded different.
Is it credible that ALL these options are making less than bit-perfect copies of the source material? Really? And if the equipment IS consistently producing bit-perfect copies of the source material, what accounts for the differences in audible performance?
One theory I've heard is that different software and hardware introduces different amounts and amplitudes of jitter into the bitstream, and that whatever DAC you may be using may just "like" the output of one collection of hardware and software more than others. Of course, there is no way to quantify any of this without laboratory-grade test equipment (that I lack).
I'm becoming convinced, however, that there are still gremlins in digital audio reproduction that (while being only marginally measurable, if at all) definitely affect audible output. Equipment that I've heard that I consistently like the sound of:
Maxell CD drive (disc source)
Seagate Barracuda 7200 RPM destination drives (USB interface with lab-grade power supply)
Tascam SS-CDR250N recorder
jRiver ripping software @ 1x speed
Avid Pro Tools Studio editing software
SD card: SanDisk 128GB Extreme PRO UHS-II SDXC Memory Card
YMMV
|
|
|
|
Post by garbulky on May 8, 2022 14:26:43 GMT -5
I've heard the claim again and again - digital bits are bits. A copy is a copy and if the bits are identical, then the copy must sound the same as the original. I understand the error correction and verification algorithms just fine. And intellectually, I agree with the claim - an identical copy will always provide an identical sound. End of discussion. But I'm no longer convinced that it's so. When I originally moved my music library from analog to digital, I use ALE format (Apple Lossless Encoding). My digital copies didn't sound as good as my analog sources. So I converted some files from ALE back to uncompressed WAV format, and heard a big improvement. Next, I went back and re-ripped some of the source material directly to WAV. And the "native WAV rips" again sounded better than the ALE to WAV conversions. Then, I used a variety of ripping programs including jRiver, Exact Audio Copy and others. I found that the different ripping programs sounded different from one another. I then used different ripping speeds. Sounded different. I then used different drives to rip from. Sounded different. Then I tried different destinations including WD drives, Seagate drives, SSD drives, internal drives, USB drives, and SD cards. Sounded different. Is it credible that ALL these options are making less than bit-perfect copies of the source material? Really? And if the equipment IS consistently producing bit-perfect copies of the source material, what accounts for the differences in audible performance? One theory I've heard is that different software and hardware introduces different amounts and amplitudes of jitter into the bitstream, and that whatever DAC you may be using may just "like" the output of one collection of hardware and software more than others. Of course, there is no way to quantify any of this without laboratory-grade test equipment (that I lack). I'm becoming convinced, however, that there are still gremlins in digital audio reproduction that (while being only marginally measurable, if at all) definitely affect audible output. Equipment that I've heard that I consistently like the sound of: Maxell CD drive (disc source) Seagate Barracuda 7200 RPM destination drives (USB interface with lab-grade power supply) Tascam SS-CDR250N recorder jRiver ripping software @ 1x speed Avid Pro Tools Studio editing software SD card: SanDisk 128GB Extreme PRO UHS-II SDXC Memory Card YMMV So here's where you can verify. You mentioned you converted the ALE to Wav. And then also ripped natively to wav. So you have TWO files that should theoretically be exactly the same. To find out if they really are: You can use this tool: steinberg.help/wavelab_pro/v10/en/wavelab/topics/audio_analysis/audio_files_comparing_t.htmlIt will compare the two files and generate a new audio file with ONLY the difference present in the audio files. So basically if you can hear anything in this file, then you know the two files are NOT identical. You can use this online selector to compare the files and see if there is a difference blue2digital.com/apps/compare-audios.html |
|
|
|
Post by leonski on May 8, 2022 16:35:03 GMT -5
NOTE that, when you measure things with a MICROPHONE, in a ROOM, it's not at all the same as measuring an electronic SIGNAL.
There are several options that determine exactly what test signal is used and how it is analyzed afterwards...
You will find that REW offers you quite a few options in that regard...
And, while Dirac doesn't offer you any options you can pick, they have their own method, which may not equate to ANY of the options in REW...
(It's not nearly as simple as "just set the window to exclude room reflections" or "pick a chirp instead of a click".)
(And, of course, Dirac isn't even trying to exclude room reflections, since it's including them in its calculations....)
Keith, I measured stuff for a living for a number of years. That's how I got in minor trouble on the DIRAC thread. Imagine! Trying for repeatability! The Shame!
I just saw that Mini has a 2nd, more expensive microphone choice. Some claims are made which may, to the extent they are correct, help in some measurement situaitons.
|
|
KeithL
Administrator
Posts: 10,273
|
Post by KeithL on May 9, 2022 11:24:03 GMT -5
At this point you're doing a bit of "dancing around the issue"... The point is that, while you "understand" the error correction algorithms, you apparently haven't actually tested the files.
(And that's the obvious way to find out if they're really the same or not.)
The short answer is that, yes, it is quite "credible" that different ripping programs may end up with different results...
And, if you have a disc that is scratched or damaged, that possibility rises from "credible" to "relatively likely"... And, yes, occasionally discs even have errors in the optical master, which may result in read errors when you rip them... One time I had several copies of the same pressing of the same album... and all had different errors... all centered around the same physical location. (Clearly there was a flaw in the optical master that was producing discs that were "weak" to varying degrees at that spot on their surfaces.)
It's also quite possible that, even with lossless files, a certain conversion program may simply be making mistakes when it converts them. Or that it may deliberately make changes because it is invoking some option like like like "replay gain"...
(Sometimes, just like other software, audio programs may simply have errors in them.)
Options like AccurateRIP actually do a checksum of the file and compare it to checksums from other RIPs of the same file. The results are incredibly informative... (if you got the same checksum as several other people then you probably have a good RIP)
(And, for example, you may discover that, on two different pressings, all tracks are identical but one, which has different versions on each.)
Your ideas about jitter are well taken - and quite valid.
HOWEVER, while jitter can affect a digital audio STREAM, it CANNOT affect the actual data STORED in a file. Jitter is a variation in the data clock... When you RIP a CD, the data is stored, while the clock is discarded... Then, when you play that data back, a new clock is created... Therefore any variations in the original clock don't count (unless they produce errors in the data - which would change the data). (However it is possible that HOW the file is stored MIGHT affect the quality of that new clock that is added when it is read back.)
There are all sorts of ways in which different data paths can affect jitter...
(Although most well designed modern DACs do a pretty good job at minimizing the effects of jitter altogether...) (Asynchronous USB connections are especially good at eliminating jitter... but this is much less true for Optical and Coax connections.) For example, a certain computer, or a certain program, could have more jitter when playing from an HDD than from an SSD. And there could major differences between an internal HDD and a USB HDD... There could even be differences depending on where the file resides on the surface of the drive... If the file is fragmented in storage then the computer will be "doing more work to read and reassemble the pieces". (In general using an internal SSD will minimize this.) Likewise, even though the data is the same, some lossless file formats may take more effort to decode than others... And this may vary with different programs... or different computers... (Even the FLAC format lets you select a minor tradeoff between "better compression" and "easier encoding and decoding".)
However, with all of these variables, the point at which the data is RIPped puts a SOLID WALL between the two sides of the process. At that point, either you have the correct DATA stored in the file, or you do not.
Someone else mentioned a Steinberg "audio compare" program. I'm not familiar with it... but I'm not sure it would be a good choice. (I believe it is intended to detect "audible differences" rather than "bit differences"... )
If you want to see if two FILES are identical there are more certain ways of comparing them directly.
I don't usually use it, but there is even a command line option: FC /b filename1.ext filename2.ext Unless the result is "no differences encountered" then your two files are NOT the same.
(Note that it is possible for RIPped files to contain trivial differences that may not be audible... but that's another story... for another time.)
I've heard the claim again and again - digital bits are bits. A copy is a copy and if the bits are identical, then the copy must sound the same as the original. I understand the error correction and verification algorithms just fine. And intellectually, I agree with the claim - an identical copy will always provide an identical sound. End of discussion. But I'm no longer convinced that it's so. When I originally moved my music library from analog to digital, I use ALE format (Apple Lossless Encoding). My digital copies didn't sound as good as my analog sources. So I converted some files from ALE back to uncompressed WAV format, and heard a big improvement. Next, I went back and re-ripped some of the source material directly to WAV. And the "native WAV rips" again sounded better than the ALE to WAV conversions. Then, I used a variety of ripping programs including jRiver, Exact Audio Copy and others. I found that the different ripping programs sounded different from one another. I then used different ripping speeds. Sounded different. I then used different drives to rip from. Sounded different. Then I tried different destinations including WD drives, Seagate drives, SSD drives, internal drives, USB drives, and SD cards. Sounded different. Is it credible that ALL these options are making less than bit-perfect copies of the source material? Really? And if the equipment IS consistently producing bit-perfect copies of the source material, what accounts for the differences in audible performance? One theory I've heard is that different software and hardware introduces different amounts and amplitudes of jitter into the bitstream, and that whatever DAC you may be using may just "like" the output of one collection of hardware and software more than others. Of course, there is no way to quantify any of this without laboratory-grade test equipment (that I lack). I'm becoming convinced, however, that there are still gremlins in digital audio reproduction that (while being only marginally measurable, if at all) definitely affect audible output. Equipment that I've heard that I consistently like the sound of: Maxell CD drive (disc source) Seagate Barracuda 7200 RPM destination drives (USB interface with lab-grade power supply) Tascam SS-CDR250N recorder jRiver ripping software @ 1x speed Avid Pro Tools Studio editing software SD card: SanDisk 128GB Extreme PRO UHS-II SDXC Memory Card YMMV |
|
KeithL
Administrator
Posts: 10,273
|
Post by KeithL on May 10, 2022 14:53:41 GMT -5
The real issue is that, with room correction, you are NOT making anechoic measurements, and you are NOT making purely directional measurements either. Therefore, since there's no way to get real precision, there's bound to be a lot of "art" in figuring out what really matters and to what degree. That's what I really like about digital audio... At least at the point where you have a digital audio FILE you CAN have absolute precision and certainty. There are no margins for error... and no approximations... either a number is right or it's not... and either a file is exactly identical or it isn't.
And it isn't that difficult to make sure the numbers really are exactly right...
(It's one of the few places on Earth where that's actually true...)
For room correction....
You can make the equivalent of an anechoic measurement... By holding the microphone at 90 degrees to the speaker... Using the 90 degree calibration file... And using a short enough window to pass direct sounds but exclude reflections... Except that, even to do that, you will need to set a rather precise window. What if your front speakers are ten feet from the microphone...
And two feet out from the wall behind them... And your rear speakers are 22 feet away...
In order to get proper "pseudo-anechoic" measurements for both you would need to set different sample windows. And, if those rears are mounted on the wall, there's no way to exclude all boundary interactions with that wall. But, with room correction, NOW we need information about the room. So now we have one microphone, with one calibration curve, picking up reflected sound from all directions.
And, even if we also had a zero-degree curve, there's no place to enter it, and we certainly have no way to use the mic's polar response pattern.) Therefore, for that part of the test, there are going to be a LOT of approximations, and a LOT of assumptions about "typical rooms".
Therefore, while we can get a pretty accurate "pseudo-anechoic" response for each speaker. We don't have nearly enough data to accurately model the room. We could do a lot better with a 3D model of the room... With acoustic absorption data for each area of surface filled in... And, of course, we could use full directional response curves for each of the individual speakers too.
But, of course, we don't have anywhere near that much data. However, since it's pretty likely that every microphone is going to have a different polar response pattern... It's pretty obvious that the non-90-degree data from different ones will be different... So, even with the same program, I would expect somewhat different results with different microphones... Even if they both have equivalent and extremely accurate 90 degree calibration curves... And, once you factor in the fact that different programs are bound to be using different approximations and assumptions... (And add in the factor that we don't have unlimited resources to calculate with a huge number of variables.)
I would expect different results with different combinations.
But, to address your comment about available microphones... I would suggest that the best results would be obtained when using the microphone that the software "is expecting"... For example... - if the software expects the microphone to have a high frequency response that falls off rapidly off-axis - but you use a microphone that has really flat off-axis response - then the assumptions the software is making in its calculations won't match - which could result in a less accurate result
And, as for direct repeatability..... At 10 kHz a 1/2 inch shift in microphone position will end up causing a 180 degree error.... Or could move you from a node to a null or vice versa...
(And you aren't going to get that accuracy with a plain old microphone stand.)
NOTE that, when you measure things with a MICROPHONE, in a ROOM, it's not at all the same as measuring an electronic SIGNAL. There are several options that determine exactly what test signal is used and how it is analyzed afterwards... You will find that REW offers you quite a few options in that regard... And, while Dirac doesn't offer you any options you can pick, they have their own method, which may not equate to ANY of the options in REW... (It's not nearly as simple as "just set the window to exclude room reflections" or "pick a chirp instead of a click".) (And, of course, Dirac isn't even trying to exclude room reflections, since it's including them in its calculations....) Keith, I measured stuff for a living for a number of years. That's how I got in minor trouble on the DIRAC thread. Imagine! Trying for repeatability! The Shame! I just saw that Mini has a 2nd, more expensive microphone choice. Some claims are made which may, to the extent they are correct, help in some measurement situaitons. |
|
|
|
Post by leonski on May 10, 2022 22:05:13 GMT -5
The 'art' is to keep the microphone in the SAME locations for each set of measurments.
The issue isn't 'the polar response pattern' of any microphone....it is doing it the same WITH YOUR MIC. I don't care what mic you
are using or its response pattern. Even the program doen't matter much IF you use the same program (and maybe revision?) each time.
A possible exception would be to compare results from REW with DIRAC. Someone with experience with both systems may have some good input.
As for mic location? Yes a small movement makes larger differences as frequency RISES.
As it turns out? there is a way to help this mess out. You could take the same measurement a number of times. They will NOT all be identical unless the
software is playing with the data too much. this is one thing that helps. YOu'll get an idea of measurment variability. What I did with MY metrology gear was
to get 3 poeple (I'm not one of 'em) to measure 10 samples, 3x each in random order. The machine required ONLY the operator to install the sample. I had marked an area
to be measured in the form of a 12mm (about 1/2") diameter circle on a 6" wafer. Across such a small area I KNEW the thickness would only vary a max of <1%.
Crunching the numbers (I had a fill-in-the-blanks form) resulted in sort of a quality measure of the SYSTEM.
I could ascribe some results to Operator Variability. Other differences were pure machine.
For a single point? once the mic is placed, that's IT.
If you want to check out OTHER variability, run ONE test per day from the 'same point'. I would expect MORE variability here based in part of Keith's points.
Put the mic away between uses and RESET to where you think it goes. Reinstall the next day for your One Point measure. If everythng were 'perfect' the graphs would be no worse
than the repeat 10x without touching anything. But that's NOT going to happen....
If you want to REALLY toss in a wildcard? Use a different mic from a given single point setup.......That's gonna cover another of Keith's points......
|
|
|
|
Post by Boomzilla on May 19, 2022 6:28:28 GMT -5
Sigh of relief - After two-weeks plus of intensive work, I've finished the illustrations for my book. I was afraid that I'd have to spend $$$ paying a commercial artist to do these, but using (mostly) Adobe Photoshop Elements for Mac, I've got useful vector drawings at 600 dpi resolution that will satisfy my publisher. Took between a half and a full day of work for each, and they aren't quite as professional as a commercial artist might have done, but since they're diagrams rather than photo-style illustrations, they will suffice. And now to move on! |
|
KeithL
Administrator
Posts: 10,273
|
Post by KeithL on May 19, 2022 10:04:53 GMT -5
Those drawings look just fine... and quite clear... which is the point.
I'm a bit confused because those obviously are vector pictures... but I'm guessing your publisher wanted them in 600 DPI raster versions...
And I'm guessing that you drew them in vector format and then imported them into Photoshop Elements to rasterize them...
(I'm thinking that it might have been easier to export them to a raster format from the vector drawing program you used.)
Sigh of relief - After two-weeks plus of intensive work, I've finished the illustrations for my book. I was afraid that I'd have to spend $$$ paying a commercial artist to do these, but using (mostly) Adobe Photoshop Elements for Mac, I've got useful vector drawings at 600 dpi resolution that will satisfy my publisher. Took between a half and a full day of work for each, and they aren't quite as professional as a commercial artist might have done, but since they're diagrams rather than photo-style illustrations, they will suffice. And now to move on! |
|
