|
|
Post by donh50 on Sept 10, 2021 7:53:32 GMT -5
-20dbfs pertains to a digital signal level whereas 100mv pertains to a analog level. So they are two different things and used to measure in different ways. It all depends on context. Hope this helps. Lonnie Got it, thanks, it was the context I was sanity-checking. I wasn't sure if you really meant digital or were referencing some sort of analog "full scale" (e.g. at the output of a DAC). |
|
Lonnie
Emo Staff 
admin
Pay no attention to the man behind the curtain
Posts: 6,999
|
Post by Lonnie on Sept 10, 2021 7:57:00 GMT -5
If I could pass along one final thought on this subject.
I believe all of this could have been very positive if Amir had reached out to me when he was testing and said "Hey, I'm seeing this, is this normal?" I would have been happy to explain that it was the Dolby up mixer sitting in the background and if he lowered his test signal to -20dbfs it would all go away. If he was too busy to reach out which I totally get, we are all running a million miles an hour, at least put it into context. "I'm seeing these artifacts, don't know why but hey, they are -85db down so you will never know it." rather than simply stating this is a horrible machine and I can't recommend it.
Just my 0.02
Lonnie
|
|
|
|
Post by derwin on Sept 10, 2021 8:03:39 GMT -5
If I could pass along one final thought on this subject. I believe all of this could have been very positive if Amir had reached out to me when he was testing and said "Hey, I'm seeing this, is this normal?" I would have been happy to explain that it was the Dolby up mixer sitting in the background and if he lowered his test signal to -20dbfs it would all go away. If he was too busy to reach out which I totally get, we are all running a million miles an hour, at least put it into context. "I'm seeing these artifacts, don't know why but hey, they are -85db down so you will never know it." rather than simply stating this is a horrible machine and I can't recommend it. Just my 0.02 Lonnie ASR has done wonders for the audiophile consumer by being an objective and independent source of facts about the performance of gear (and helping us get away from absurdly subjective terms like air and sweeetness…). But I agree with this 100% - he seems like he has an internal set of logic he adheres to (so it’s not arbitrary), but this all could have gone so differently. On the plus side, I’ve learned a bunch listening to both you and Amir and the folks at ASR talk about what they may be seeing. But in summary, +1 - I don’t put this on him at all, he didn’t do this wrong, but I still wish he could have had the time to call you guys and avoid all this fun. |
|
|
|
Post by donh50 on Sept 10, 2021 8:25:51 GMT -5
I do want to say that I am impressed lonnie released the internal test reports. Shades of Emo of old and way more than I expected to see. Thanks Lonnie, that's transparency. On test levels: HT processors are not my field. Stand-alone data converters are typically tested at -1 dBFS, maybe 0 dBFS, because we control the input and output very tightly to avoid clipping. -1 dBFS provides about 10% margin, enough for a known and well-understood test signal, but not enough for a real-world signal which typically has 17 dB (music) to 20 dB (movies, may be higher for some) peak-to-average ratio. 17 dB is a power ratio of 50, and 20 dB is a power ratio of 100, meaning if you are listening your average power is 50~100 times lower than the peaks. Alternatively, you need 50~100 times the average power to prevent clipping the peaks. IME most people overestimate their average power needs and underestimate their peak needs. At any rate, I assume this is the basis for the -20 dBFS input. One last comment to avoid the appearance of incompetence: While technically dBFS refers to digital full scale in this context, it is also used in an analog way to indicate the analog input signal related to full-scale digital output from an ADC, and the analog output signal from a DAC relative to a full-scale digital input. That is why I asked Lonnie to clarify, and hopefully puts my question and his answer into context. Thanks, Don |
|
cawgijoe
Emo VIPs 
"When you come to a fork in the road, take it." - Yogi Berra
Posts: 5,033
|
Post by cawgijoe on Sept 10, 2021 8:26:41 GMT -5
If I could pass along one final thought on this subject. I believe all of this could have been very positive if Amir had reached out to me when he was testing and said "Hey, I'm seeing this, is this normal?" I would have been happy to explain that it was the Dolby up mixer sitting in the background and if he lowered his test signal to -20dbfs it would all go away. If he was too busy to reach out which I totally get, we are all running a million miles an hour, at least put it into context. "I'm seeing these artifacts, don't know why but hey, they are -85db down so you will never know it." rather than simply stating this is a horrible machine and I can't recommend it. Just my 0.02 Lonnie ASR has done wonders for the audiophile consumer by being an objective and independent source of facts about the performance of gear (and helping us get away from absurdly subjective terms like air and sweeetness…). But I agree with this 100% - he seems like he has an internal set of logic he adheres to (so it’s not arbitrary), but this all could have gone so differently. On the plus side, I’ve learned a bunch listening to both you and Amir and the folks at ASR talk about what they may be seeing. But in summary, +1 - I don’t put this on him at all, he didn’t do this wrong, but I still wish he could have had the time to call you guys and avoid all this fun. In my many years of reading reviews from all the major publications (Stereophile/Audio/Stereo Review/Home Theater/etc.), I have seen a handful of reviews where something was not quite right and the reviewer reaching out to the manufacturer for either another unit or an understanding of what the problem could be and it's mentioned in the article. IMO he should have caveated this issue with the stipulation that he would reach out to Emotiva for additional clarification and stay tuned. That avoids the "sky is falling scenario". Hindsight is 20/20, but if all three Emotiva units are based on the same architecture, you would think they would at least measure closely and if they didn't there was something wrong either with the unit, firmware, manufacturer testing, ASR testing, etc. Amir is a lot smarter than I am, but maybe slow down and double-check before publishing. |
|
Lonnie
Emo Staff
admin
Pay no attention to the man behind the curtain
Posts: 6,999
|
Post by Lonnie on Sept 10, 2021 8:42:40 GMT -5
I do want to say that I am impressed lonnie released the internal test reports. Shades of Emo of old and way more than I expected to see. Thanks Lonnie, that's transparency. On test levels: HT processors are not my field. Stand-alone data converters are typically tested at -1 dBFS, maybe 0 dBFS, because we control the input and output very tightly to avoid clipping. -1 dBFS provides about 10% margin, enough for a known and well-understood test signal, but not enough for a real-world signal which typically has 17 dB (music) to 20 dB (movies, may be higher for some) peak-to-average ratio. 17 dB is a power ratio of 50, and 20 dB is a power ratio of 100, meaning if you are listening your average power is 50~100 times lower than the peaks. Alternatively, you need 50~100 times the average power to prevent clipping the peaks. IME most people overestimate their average power needs and underestimate their peak needs. At any rate, I assume this is the basis for the -20 dBFS input. One last comment to avoid the appearance of incompetence: While technically dBFS refers to digital full scale in this context, it is also used in an analog way to indicate the analog input signal related to full-scale digital output from an ADC, and the analog output signal from a DAC relative to a full-scale digital input. That is why I asked Lonnie to clarify, and hopefully puts my question and his answer into context. Thanks, Don So just to help clarify things here. If we completely shut off all Dolby and DTS and literally route around them then you could send a 0dbfs test signal through the system and it would be clean as a whistle. You would see a ground floor noise level of roughly -130db down from full scale with minimal harmonics. The limitation we run into is for all practical purposes the heart of what a HT processor does and that is Dolby and DTS decoding and upmixing. While I give both companies props for what they do and how well they do it, it is unfortunately the limiting factor to our overall performance specs. I'm not complaining because if it weren't for them, we wouldn't have what we have. We would be an amp and DAC company.  . None-the-less, when you have all these black boxes sitting in the background doing their respective things, it does limit the performance and I fully understand why they want the system optimized for -20dbfs. It reduces the computational load significantly where the focus of the work can be put in rendering the channels. Please forgive me is I'm stating what you already know, but when I say -20dbfs, I am referring to the incoming digital signal. In a test configuration as in an AP test, this will hold at -20db, but in real life the signal is going vary all over the place and will reach peaks of 0db and silence of around -120db with the majority of the signals bouncing in-between. So the -20 optimization is simply a nominal point of reference for real world applications and not a hard ceiling. Hope this helps. Lonnie |
|
richb
Sensei
Oppo Beta Group - Audioholics Reviewer
Posts: 890
|
Post by richb on Sept 10, 2021 8:46:13 GMT -5
-20dbfs pertains to a digital signal level whereas 100mv pertains to a analog level. So they are two different things and used to measure in different ways. It all depends on context. Hope this helps. Lonnie Is is common/ubiquitous for digital music to be mastered close to or at 0 dBFS. Bass management and REQ requires digital headroom to apply boosts. Therefore, all processors must digitally attenuate when applying processing since the resulting signal CANNOT exceed 0 dBFS. There should be no need to digitally attenuate in Reference Stereo (or similarly named) mode where there is no processing. I have lobbied for ASR to measure a AVR/AVP center channel and to measure with simple bass management in place because this is a normal use case. This is a normal use case and it seems reasonable that this is a better indicator performance. Since ASR is primarily a music driven 2-channel site, then a representative signal is close to 0 dBFS. I expect sending -20 dB attenuated signals is not representative of compressed music. This is my first exposure to the concept that in processor THD+N (or SINAD) performance is reduced when the digital signal is at 0 dBFS (leaving out inter-sample overloads). Since digital attenuation does not attenuate noise, a 0 dBFS signal should produce an optimal result. I can't get my head around measuring a device with attenuated digital signals. Clearly, all sources must be handles so no matter its level, many times, the processor must digitally attenuate for bass management, PEQ, and REQ. - Rich |
|
Lsc
Emo VIPs
Posts: 3,434
|
Post by Lsc on Sept 10, 2021 8:47:52 GMT -5
ASR has done wonders for the audiophile consumer by being an objective and independent source of facts about the performance of gear (and helping us get away from absurdly subjective terms like air and sweeetness…). But I agree with this 100% - he seems like he has an internal set of logic he adheres to (so it’s not arbitrary), but this all could have gone so differently. On the plus side, I’ve learned a bunch listening to both you and Amir and the folks at ASR talk about what they may be seeing. But in summary, +1 - I don’t put this on him at all, he didn’t do this wrong, but I still wish he could have had the time to call you guys and avoid all this fun. In my many years of reading reviews from all the major publications (Stereophile/Audio/Stereo Review/Home Theater/etc.), I have seen a handful of reviews where something was not quite right and the reviewer reaching out to the manufacturer for either another unit or an understanding of what the problem could be and it's mentioned in the article. IMO he should have caveated this issue with the stipulation that he would reach out to Emotiva for additional clarification and stay tuned. That avoids the "sky is falling scenario". Hindsight is 20/20, but if all three Emotiva units are based on the same architecture, you would think they would at least measure closely and if they didn't there was something wrong either with the unit, firmware, manufacturer testing, ASR testing, etc. Amir is a lot smarter than I am, but maybe slow down and double-check before publishing. I think what may have happened is the Dolby upmixer Lonnie mentioned that runs in the background. This could be the cause of the noise that is making the XMC2 look worse than the original RMC1 test. Now, if you take the RMC1 with the current firmware I think it will yield the same results. So if the Dolby up mixer is the culprit, is there a way to remove that noise when playing 2 channel music (assuming the noise introduced by the upmixer adversely affects the sound)? I personally think it would be best to keep the signal for 2 channel as clean as possible but I’m not an engineer. Maybe in a future firmware release? EDIT: ok I was late again . Aren’t there filters that can be used to keep the noise out? Again, a non technical person asking. |
|
|
|
Post by donh50 on Sept 10, 2021 9:07:58 GMT -5
I do want to say that I am impressed lonnie released the internal test reports. Shades of Emo of old and way more than I expected to see. Thanks Lonnie, that's transparency. On test levels: HT processors are not my field. Stand-alone data converters are typically tested at -1 dBFS, maybe 0 dBFS, because we control the input and output very tightly to avoid clipping. -1 dBFS provides about 10% margin, enough for a known and well-understood test signal, but not enough for a real-world signal which typically has 17 dB (music) to 20 dB (movies, may be higher for some) peak-to-average ratio. 17 dB is a power ratio of 50, and 20 dB is a power ratio of 100, meaning if you are listening your average power is 50~100 times lower than the peaks. Alternatively, you need 50~100 times the average power to prevent clipping the peaks. IME most people overestimate their average power needs and underestimate their peak needs. At any rate, I assume this is the basis for the -20 dBFS input. One last comment to avoid the appearance of incompetence: While technically dBFS refers to digital full scale in this context, it is also used in an analog way to indicate the analog input signal related to full-scale digital output from an ADC, and the analog output signal from a DAC relative to a full-scale digital input. That is why I asked Lonnie to clarify, and hopefully puts my question and his answer into context. Thanks, Don So just to help clarify things here. If we completely shut off all Dolby and DTS and literally route around them then you could send a 0dbfs test signal through the system and it would be clean as a whistle. You would see a ground floor noise level of roughly -130db down from full scale with minimal harmonics. The limitation we run into is for all practical purposes the heart of what a HT processor does and that is Dolby and DTS decoding and upmixing. While I give both companies props for what they do and how well they do it, it is unfortunately the limiting factor to our overall performance specs. I'm not complaining because if it weren't for them, we wouldn't have what we have. We would be an amp and DAC company. . None-the-less, when you have all these black boxes sitting in the background doing their respective things, it does limit the performance and I fully understand why they want the system optimized for -20dbfs. It reduces the computational load significantly where the focus of the work can be put in rendering the channels. Please forgive me is I'm stating what you already know, but when I say -20dbfs, I am referring to the incoming digital signal. In a test configuration as in an AP test, this will hold at -20db, but in real life the signal is going vary all over the place and will reach peaks of 0db and silence of around -120db with the majority of the signals bouncing in-between. So the -20 optimization is simply a nominal point of reference for real world applications and not a hard ceiling. Hope this helps. Lonnie It does, thanks. I had not thought about the processing part, which is why I defer to HT types. Now, if only you could fix the HDMI spec madness... |
|
Lonnie
Emo Staff
admin
Pay no attention to the man behind the curtain
Posts: 6,999
|
Post by Lonnie on Sept 10, 2021 9:15:06 GMT -5
So just to help clarify things here. If we completely shut off all Dolby and DTS and literally route around them then you could send a 0dbfs test signal through the system and it would be clean as a whistle. You would see a ground floor noise level of roughly -130db down from full scale with minimal harmonics. The limitation we run into is for all practical purposes the heart of what a HT processor does and that is Dolby and DTS decoding and upmixing. While I give both companies props for what they do and how well they do it, it is unfortunately the limiting factor to our overall performance specs. I'm not complaining because if it weren't for them, we wouldn't have what we have. We would be an amp and DAC company. . None-the-less, when you have all these black boxes sitting in the background doing their respective things, it does limit the performance and I fully understand why they want the system optimized for -20dbfs. It reduces the computational load significantly where the focus of the work can be put in rendering the channels. Please forgive me is I'm stating what you already know, but when I say -20dbfs, I am referring to the incoming digital signal. In a test configuration as in an AP test, this will hold at -20db, but in real life the signal is going vary all over the place and will reach peaks of 0db and silence of around -120db with the majority of the signals bouncing in-between. So the -20 optimization is simply a nominal point of reference for real world applications and not a hard ceiling. Hope this helps. Lonnie It does, thanks. I had not thought about the processing part, which is why I defer to HT types. Now, if only you could fix the HDMI spec madness... OH, don't get me started on HDMI, I'll be on my soapbox all day. Trying to push 48gig of data down a line that was only designed to handle a few hundred meg, who thought that was a good idea? Lonnie |
|
cawgijoe
Emo VIPs
"When you come to a fork in the road, take it." - Yogi Berra
Posts: 5,033
|
Post by cawgijoe on Sept 10, 2021 10:11:46 GMT -5
We have so much to learn Lonnie. Maybe you can have a lunch end learn session you know….given all that free time you have 😬😂 Would love to. Seriously, what topics would everyone like to know more about? Nick and I were talking about doing some videos, but I would like to know what everyone would like to see. Lonnie Topics: 1) HDMI (current specs...future specs....issues....etc.) 2) Dolby Atmos (setting up...best types of speakers and locations for best sound....installation tips...etc.) 3) A video on Emotiva's measurement process for processors, amps, and preparation to test and send a product to a customer. I'm sure there are many others..... |
|
|
|
Post by geebo on Sept 10, 2021 10:58:32 GMT -5
I just looked at ASR and looked up reviews of AVRs and found 39 multichannel units. Only 7 got a "Recommended" from Amir which means a little over 82% are not. Some very expensive and highly regarded units receive a "No". Maybe he needs to look at AVR's a little differently.
|
|
|
|
Post by garbulky on Sept 10, 2021 11:08:20 GMT -5
I just looked at ASR and looked up reviews of AVRs and found 39 multichannel units. Only 7 got a "Recommended" from Amir which means a little over 82% are not. Some very expensive and highly regarded units receive a "No". Maybe he needs to look at AVR's a little differently. He does not do so. In general he doesn't care why they are limited. So for instance a tube amp may spec well for a tube amp and a multibit dac may spec well for a multibit dac. But it matters not to him. he is interested in low distortion. So for instance the -20db real world usage would likely not factor in to his determination, though I can't speak for him. |
|
|
|
Post by geebo on Sept 10, 2021 11:14:20 GMT -5
I just looked at ASR and looked up reviews of AVRs and found 39 multichannel units. Only 7 got a "Recommended" from Amir which means a little over 82% are not. Some very expensive and highly regarded units receive a "No". Maybe he needs to look at AVR's a little differently. He does not do so. In general he doesn't care why they are limited. So for instance a tube amp may spec well for a tube amp and a multibit dac may spec well for a multibit dac. But it matters not to him. he is interested in low distortion. So for instance the -20db real world usage would likely not factor in to his determination, though I can't speak for him. He's comparing apples and oranges. Would you compare ride quality of a motorcycle with that of a sedan? |
|
|
|
Post by AudioHTIT on Sept 10, 2021 11:16:06 GMT -5
If I could pass along one final thought on this subject. I believe all of this could have been very positive if Amir had reached out to me when he was testing and said "Hey, I'm seeing this, is this normal?" I would have been happy to explain that it was the Dolby up mixer sitting in the background and if he lowered his test signal to -20dbfs it would all go away. If he was too busy to reach out which I totally get, we are all running a million miles an hour, at least put it into context. "I'm seeing these artifacts, don't know why but hey, they are -85db down so you will never know it." rather than simply stating this is a horrible machine and I can't recommend it. Just my 0.02 Lonnie I’ve been reading through the corresponding thread on ASR, some there have also asked why he didn’t reach out before publishing, or why he doesn’t always do it. His response does have some logic, if a manufacturer sends him a product to test and he finds something unusual, he will hold his review and ask for comment. If however the product is submitted by a user / member, he does not (normally) ask the manufacturer for comment. While some of this discussion might have been avoided, this doesn’t seem like an unreasonable way to do business as he is busy too. You might think of it as an incentive to send products for testing, and also to make sure measurements are available so that someone can compare or check their results for ‘normalcy’. |
|
|
|
Post by AudioHTIT on Sept 10, 2021 11:28:36 GMT -5
Hi Lonnie. I'm trying to understand the -20db limitation. Why would sources output only a -20db and not a 0db or close to that digital signal? Are you counting lossless or music sources in this? I have numerous music sources that easily go past -20 db peaks digitally.… Please forgive me is I'm stating what you already know, but when I say -20dbfs, I am referring to the incoming digital signal. In a test configuration as in an AP test, this will hold at -20db, but in real life the signal is going vary all over the place and will reach peaks of 0db and silence of around -120db with the majority of the signals bouncing in-between. So the -20 optimization is simply a nominal point of reference for real world applications and not a hard ceiling. Hope this helps. Lonnie , hopefully you caught this from Lonnie, while we might want to see everyone test the same way, it may not give us the best real world idea of how the equipment performs. At this point I can’t say I'm sure which is right, so I’d probably say do both, but I can understand why each picks the level they do. |
|
stiehl11
Emo VIPs
Give me available light!
Posts: 7,269
|
Post by stiehl11 on Sept 10, 2021 12:10:03 GMT -5
He does not do so. In general he doesn't care why they are limited. So for instance a tube amp may spec well for a tube amp and a multibit dac may spec well for a multibit dac. But it matters not to him. he is interested in low distortion. So for instance the -20db real world usage would likely not factor in to his determination, though I can't speak for him. Would you compare ride quality of a motorcycle with that of a sedan? You would if you don't know the difference between a motorcycle and a sedan. A little knowledge is a dangerous thing. |
|
|
|
Post by garbulky on Sept 10, 2021 15:06:42 GMT -5
He does not do so. In general he doesn't care why they are limited. So for instance a tube amp may spec well for a tube amp and a multibit dac may spec well for a multibit dac. But it matters not to him. he is interested in low distortion. So for instance the -20db real world usage would likely not factor in to his determination, though I can't speak for him. He's comparing apples and oranges. Would you compare ride quality of a motorcycle with that of a sedan? If I bought the XMC-2, the priority would be for two channel PCM audio though it will still get used in HT. So in this case, yes, I would want it to perform at the distortion levels of DACs. I remember when the XMC-1 came out that I thought hey finally here's a purchase I could strive for that would measure as well as DACs and be fully balanced. Now, I'm not so sure. AudioHTITSadly, if I'm reading this right, it appears that the dolby filters cannot be turned off when it is not Dolby. Like in direct mode and stuff it looks like these filters are just allowed to be left in the loop. |
|
|
|
Post by geebo on Sept 10, 2021 15:38:44 GMT -5
He's comparing apples and oranges. Would you compare ride quality of a motorcycle with that of a sedan? If I bought the XMC-2, the priority would be for two channel PCM audio though it will still get used in HT. So in this case, yes, I would want it to perform at the distortion levels of DACs. I remember when the XMC-1 came out that I thought hey finally here's a purchase I could strive for that would measure as well as DACs and be fully balanced. Now, I'm not so sure. AudioHTITSadly, if I'm reading this right, it appears that the dolby filters cannot be turned off when it is not Dolby. Like in direct mode and stuff it looks like these filters are just allowed to be left in the loop. And if you bought a motorcycle your top priority would be the comfort and ride quality? Two channel DACS and multichannel processors are designed for very different things. It's also a reason why so many with HTP's of ANY brand also get a two channel DAC. |
|
|
|
Post by soupial on Sept 10, 2021 16:20:08 GMT -5
He's comparing apples and oranges. Would you compare ride quality of a motorcycle with that of a sedan? If I bought the XMC-2, the priority would be for two channel PCM audio though it will still get used in HT. So in this case, yes, I would want it to perform at the distortion levels of DACs. I remember when the XMC-1 came out that I thought hey finally here's a purchase I could strive for that would measure as well as DACs and be fully balanced. Now, I'm not so sure. AudioHTITSadly, if I'm reading this right, it appears that the dolby filters cannot be turned off when it is not Dolby. Like in direct mode and stuff it looks like these filters are just allowed to be left in the loop. Why would the dolby processing be used in a non-dolby format? If I'm playing 2 channel pcm without any up mixing then shouldn't there be a cleaner, less processed result? |
|
. None-the-less, when you have all these black boxes sitting in the background doing their respective things, it does limit the performance and I fully understand why they want the system optimized for -20dbfs. It reduces the computational load significantly where the focus of the work can be put in rendering the channels. Please forgive me is I'm stating what you already know, but when I say -20dbfs, I am referring to the incoming digital signal. In a test configuration as in an AP test, this will hold at -20db, but in real life the signal is going vary all over the place and will reach peaks of 0db and silence of around -120db with the majority of the signals bouncing in-between. So the -20 optimization is simply a nominal point of reference for real world applications and not a hard ceiling. 
