Where is the Flaw in most Multichannel Preamps/ AVRs?

[mdsimon2]

AVRs typically have digitally controlled analog volume control so using SINAD / DR measurements at full output isn't indicative of residual noise at attenuated output levels. At lower output levels an analog volume control will provide lower noise than a digital volume control given the same full level dynamic range. This is why many use AVRs and don't report hiss issues. In addition, as mentioned by others, AVR SINAD is often dominated by distortion which is most less audible than noise.

Standalone DACs typically use digital volume control so need higher dynamic range to avoid hiss.

Using the analog output of an AVR into a miniDSP HTx would clearly result in more noise than just using the AVR itself, however I think such a system should still have acceptable noise performance. The noise at the output of the miniDSP will be a combination of noise from the AVR + ADC + DAC, therefore I think you will achieve the best noise performance by running the AVR at max volume into the ADC and using the miniDSP for volume control. As you lower the volume in the miniDSP, it will effectively attenuate the noise from the AVR and ADC.

A bit of quick modeling shows that if you are starting with an AVR that has 4V DR of 108 dB, after about 20 dB of attenuation in the miniDSP, the noise level will be very close to the noise level of the miniDSP DAC itself.

Of course, if you go the digital route into the miniDSP HTx with something like the Arvus, the only noise contribution will be from the miniDSP DAC.

Michael

 

[Fredygump]

SINAD may or may not correlate with residual noise. SINAD is THD+Noise. So a SINAD of -90 db could be very, very low noise of -120 db with some harmonic distortion at -90 db. Or it could be distortion at -120 db with noise which is at -90 db. You don't know just using that one number.

The measurement in Amir's reviews you want to look at is dynamic range. It will give you the noise level over 0hz to 20 khz vs max output level. From that you could get an idea of the microvolts of noise and with the gain of the rest of your gear get an idea if noise will be audible.

For instance the Monoprice HTP-1 has a dynamic range of 112 db relative to 4 volts output.

I went back and looked at the THD + N charts that Amir published. If I ignore the results and just look at the graphs themselves, I think I'm seeing the trend. It appears that the SINAD is being brought down by distortion. Some of them seem to have very low noise, but the units with lower noise fail to reduce the distortion. So the data for the quieter ones looks really bad, since the distortion peaks are high above the noise floor.

I don't know why distortion would not be improved, but I guess it doesn't really matter. I know I can't hear distortion that is even -80dB.

But for noise, if the noise floor is -120dB or below, then I don't think I have anything to worry about. So I'll just ignore the SINAD and look at the noise part of the graph...

 

[Blumlein 88]

I went back and looked at the THD + N charts that Amir published. If I ignore the results and just look at the graphs themselves, I think I'm seeing the trend. It appears that the SINAD is being brought down by distortion. Some of them seem to have very low noise, but the units with lower noise fail to reduce the distortion. So the data for the quieter ones looks really bad, since the distortion peaks are high above the noise floor.

I don't know why distortion would not be improved, but I guess it doesn't really matter. I know I can't hear distortion that is even -80dB.

But for noise, if the noise floor is -120dB or below, then I don't think I have anything to worry about. So I'll just ignore the SINAD and look at the noise part of the graph...

Don't take offense if you already know, but do you understand what you are seeing in FFT graphs? If not they can be confusing or deceptive on noise floors. The single spec you want is the dynamic range number. It is the level of the noise vs max level and does not include distortion in any way. So you get a clean apples to apples comparison between how quiet two pieces of gear are.

I can explain the FFTs in fairly simple fashion if don't already know about them.

 

[Fredygump]

Don't take offense if you already know, but do you understand what you are seeing in FFT graphs? If not they can be confusing or deceptive on noise floors. The single spec you want is the dynamic range number. It is the level of the noise vs max level and does not include distortion in any way. So you get a clean apples to apples comparison between how quiet two pieces of gear are.

I can explain the FFTs in fairly simple fashion if don't already know about them.

I don't know what you think I'm missing?

I'm looking at a Denon plot at the moment, and the FFT with a 1khz tone chart seems to correlate exactly to the multitone chart, with the dynamic range being the difference fron the signal to the highest distortion peak...approx 107dB. But the single tone test seems to show that in the absence of a tone, the noise level is -135dB. But in the multitone test, the noise floor is obscured by the distortion.

For the purposes of this inquiry, I'm just hoping to not bet burned by a noisy device that reverses the progress I've made with my system! Simple, right? My old processor had a rated noise floor of -87dB, which is something like 35µV. That number is with no signal present, which is the spec I'm looking for. I'm sure my speakers have more distortion than the measurements below, so I'm not really worried about distortion.

And yeah, I get my dB units mixed up, so I'm just using "dB". I'm not always sure if it is dBu, dBa, dBfs, dBrA, or whatever the correct unit is for various measurements. But I'm not an electrical engineer.

 

[Blumlein 88]

Okay, imagine you had a voltmeter that would measure evenly up to 20 khz. Imagine you had white noise which is an even amount of energy at every frequency. Now implement a pair of filters. One goes from 0 to 10 khz, and the other goes from 10 khz to 20 khz. Measure the same thing with your voltmeter and each half of the band will measure -3 db lower. Implement more filters say from 0 to 5 khz, 5 khz to 10 khz, 10 khz to 15 khz and 15 khz to 20 khz. Now each one will measure the noise another -3 db lower or -6 db from the full band measurement. This is essentially what an FFT does. So a 32k FFT will split the band into 16,000 equal bands. The noise level in each band will drop because of this. A graph from an FFT showing a floor around -135 db on the graph doesn't mean that is the noise level. Over the whole band it will be higher. A 32K FFT which is what Amir uses will lower the noise floor on the graph by 39 db. So that -135 db would be maybe only -96 db for the whole band. Now in reality noise is usually pink in nature with more low frequency than high so that complicates the picture a little bit.

So the two things you'd really like to know are Signal to noise ratio which is max signal vs no signal. Or dynamic range which is a -60 db tone notched out to leave only noise. Distortion at that low level will nearly always be below the noise floor. Amir uses dynamic range on DACs because some DACs with no signal turn the output to all zeroes giving a false picture of the noise floor. A low level signal keeps the DAC turned on for a truer picture. The multitone will often be close to a picture of the nosie floor level, but not always.

A good example is the recently reviewed Bricasti. It has a dynamic range of 115-116 db. Yet the graph of the 1 khz tone the noise levels are well below distortion showing -150 db on the graph. Yet that is an effect of the FFT graphing. So dynamic range is close enough to measuring noise to tell you what it is you wish to know. More so than the multitone graph.

https://www.audiosciencereview.com/forum/index.php?attachments/bricasti-m1se-high-end-balanced-stereo-dac-rca-measurement-png.354410/

 

[Mikig]

from my experience: the other day I connected the Theta Casanova to use it as a 2-channel preamplifier.
System untouched, so set exactly as always. When I put my ears close to the speaker, I noticed the classic noise of the system being turned on with music turned off, that slight hiss, which, however, by raising the volume you can hear even when moving away from the speakers. With my passive preamp (passive attenuator) this hiss cannot be heard: not even at 0db which obviously corresponds to the maximum volume. Particular to think that in a "silent" system the preamp can make so much difference to the background noise.

 

[Mikig]

speaking of AVR: do you know of a basic AVR, preferably with balanced outputs, to use with my TV but which has a small box? the HT preamps I own are entirely too big, so I can't put them in the cabinet. I would like to at least be able to mount the center channel. thanks!

 

[Blumlein 88]

speaking of AVR: do you know of a basic AVR, preferably with balanced outputs, to use with my TV but which has a small box? the HT preamps I own are entirely too big, so I can't put them in the cabinet. I would like to at least be able to mount the center channel. thanks!

The smallest I know of is the Emotiva pre/pro. It has only RCA outputs. It is something like the normal 17 inch width though pretty thin top to bottom.

 

[Mikig]

The smallest I know of is the Emotiva pre/pro. It has only RCA outputs. It is something like the normal 17 inch width though pretty thin top to bottom.

Thank you! I have a Casanova and a Casablanca II left but they are too deep and large. My lady has forbidden me, in the living room, anything that has visible wires or that is not contained in the space of the furniture!! So I can't use them….

 

[Blumlein 88]

Thank you! I have a Casanova and a Casablanca II left but they are too deep and large. My lady has forbidden me, in the living room, anything that has visible wires or that is not contained in the space of the furniture!! So I can't use them….

Well their smallest one has grown a little, but still pretty thin. Grown a little in price as well.

BasX MC1 13.2 Channel Dolby Atmos® & DTS:X™ Cinema Processor

The BasX MC1 is a high performance cinema surround sound processor that supports 4k UHD video, including HDR and Dolby Vision, enhanced ARC (eARC), and the latest Dolby Atmos and DTS:X immersive surround sound formats.

emotiva.com

 

[Mikig]

Well their smallest one has grown a little, but still pretty thin. Grown a little in price as well.

BasX MC1 13.2 Channel Dolby Atmos® & DTS:X™ Cinema Processor

The BasX MC1 is a high performance cinema surround sound processor that supports 4k UHD video, including HDR and Dolby Vision, enhanced ARC (eARC), and the latest Dolby Atmos and DTS:X immersive surround sound formats.

emotiva.com

I was also thinking about a device like the Topping DM7 8 channels? but then how do I do it? my source for tv is the cable box. We should first put a processor in it that can send the already corrected and separated signal.

 

[Blumlein 88]

I was also thinking about a device like the Topping DM7 8 channels? but then how do I do it? my source for tv is the cable box. We should first put a processor in it that can send the already corrected and separated signal.

There is no good way to get the Dolby decoding for that. I wish there were. They aren't going to let you at the decoded signals in digital form. A shame, they lock this down and other than very expensive gear (Trinnov or Lyngdorf or similar) most of it has mediocre performance.

 

[GXAlan]

I am surprised no one has mentioned that the "flaw" is that you simply can't do this, at least not without difficulty.
The industry doesn't want bit-perfect copies, so inexpensive AVR's don't output a digital signal after they decode.

+1000

You basically have the JBLs with Dante. The reliability of Dante on JBL seems spotty based upon AVSForum.

You have the Trinnov Altitude 32 and Storm Audio.

The original Trinnov’s were about 90 dB SINAD, but they had low noise and moderate distortion. Moderate distortion meaning that Harmonics still 100 dB below the fundamental which tells you that as long as the noise is low enough, that level of distortion is inaudible when listening to movies where 0 dBFS should hit 105 dB.

 

[Peewee12]

The smallest I know of is the Emotiva pre/pro. It has only RCA outputs. It is something like the normal 17 inch width though pretty thin top to bottom.

Is this well-regarded around here? I was under the impression that it wasn't great. I've been tempted by this for a long time. It would be perfect for me if it had balanced outputs but other than that, it's perfect.

 

[Blumlein 88]

Is this well-regarded around here? I was under the impression that it wasn't great. I've been tempted by this for a long time. It would be perfect for me if it had balanced outputs but other than that, it's perfect.

I'd say not well regarded around here. I posted my own measurements of a previous version which weren't good. Amir has tested their XMC-1 which was probably fine, but disappointing.

 

[Fredygump]

Okay, imagine you had a voltmeter that would measure evenly up to 20 khz. Imagine you had white noise which is an even amount of energy at every frequency. Now implement a pair of filters. One goes from 0 to 10 khz, and the other goes from 10 khz to 20 khz. Measure the same thing with your voltmeter and each half of the band will measure -3 db lower. Implement more filters say from 0 to 5 khz, 5 khz to 10 khz, 10 khz to 15 khz and 15 khz to 20 khz. Now each one will measure the noise another -3 db lower or -6 db from the full band measurement. This is essentially what an FFT does. So a 32k FFT will split the band into 16,000 equal bands. The noise level in each band will drop because of this. A graph from an FFT showing a floor around -135 db on the graph doesn't mean that is the noise level. Over the whole band it will be higher. A 32K FFT which is what Amir uses will lower the noise floor on the graph by 39 db. So that -135 db would be maybe only -96 db for the whole band. Now in reality noise is usually pink in nature with more low frequency than high so that complicates the picture a little bit.

So the two things you'd really like to know are Signal to noise ratio which is max signal vs no signal. Or dynamic range which is a -60 db tone notched out to leave only noise. Distortion at that low level will nearly always be below the noise floor. Amir uses dynamic range on DACs because some DACs with no signal turn the output to all zeroes giving a false picture of the noise floor. A low level signal keeps the DAC turned on for a truer picture. The multitone will often be close to a picture of the nosie floor level, but not always.

A good example is the recently reviewed Bricasti. It has a dynamic range of 115-116 db. Yet the graph of the 1 khz tone the noise levels are well below distortion showing -150 db on the graph. Yet that is an effect of the FFT graphing. So dynamic range is close enough to measuring noise to tell you what it is you wish to know. More so than the multitone graph.

https://www.audiosciencereview.com/forum/index.php?attachments/bricasti-m1se-high-end-balanced-stereo-dac-rca-measurement-png.354410/

If I am reading your response correctly, the trace that looks like noise on the FFT is influenced by a lot of factors, so it means different things in different situations? That makes sense.

In an attempt to make progress toward the original question, is there something inherent to the design of pre/pros that makes them under-perform when compared to 2 channel DACs? Or do manufacturers think "audiophile" performance doesn't matter on these units, so they intentionally design multichannel processors to a lower standard than stereo DACs, prioritizing other features instead? They do love adding features!

The comments have not developed into a pattern that I can discern. One person said that using better quality DACs won't make a difference, and then another says that if the device could output a decoded digital signal, then there wouldn't be any noise associated with the digital signal. Another said that a specific processor used a specific DAC chip that is well known, but people didn't like it.

One reason for asking is I'm wondering if the performance will get better in the future, or if manufacturers will keep performance at around the same level, while adding features to try to convince us to upgrade?

 

[Fredygump]

I was also thinking about a device like the Topping DM7 8 channels? but then how do I do it? my source for tv is the cable box. We should first put a processor in it that can send the already corrected and separated signal.

My reading about the MiniDSP HTx is that an AppleTV 4k will pass decoded multichannel audio through to the HTx. But it is unclear what other devices work like that. And I expect the Topping is similar to the MIniDSP, so you will probably get the same results.

 

[Blumlein 88]

If I am reading your response correctly, the trace that looks like noise on the FFT is influenced by a lot of factors, so it means different things in different situations? That makes sense.

In an attempt to make progress toward the original question, is there something inherent to the design of pre/pros that makes them under-perform when compared to 2 channel DACs? Or do manufacturers think "audiophile" performance doesn't matter on these units, so they intentionally design multichannel processors to a lower standard than stereo DACs, prioritizing other features instead? They do love adding features!

The comments have not developed into a pattern that I can discern. One person said that using better quality DACs won't make a difference, and then another says that if the device could output a decoded digital signal, then there wouldn't be any noise associated with the digital signal. Another said that a specific processor used a specific DAC chip that is well known, but people didn't like it.

One reason for asking is I'm wondering if the performance will get better in the future, or if manufacturers will keep performance at around the same level, while adding features to try to convince us to upgrade?

Dolby won't let people do much as they have a monopoly so good enough is that and no more. Very nearly all AVRs and Pre/pro of any brand are made by one company. Same situation, good enough is good enough and no incentive to do better. Nothing in their bottom line. Whether more is possible or the processing needed for video formats using the approved chips puts a limit on it I don't know. A few devices are better than average, but far from SOTA.

The FFT's don't really mean different things in different situations you just have to understand what they are showing to make sense of them.

If you use digital output into better DACs it would be better. Digital output signals don't add noise. However, you cannot do that, as licensing agreements aren't going to allow that. So it is a moot question unless that changes.

I don't expect much improvement in the future on the video end of things. Though many only have distortion at the -85 db level, that is not going to be audible. If they have dynamic range or SNR around 100 db which some do that also is unlikely to be a problem. So good enough is probably good enough just no headroom for error. I don't like it, but until someone breaks Dolby's grip on the video industry don't expect much change.

 

[jayapple]

I searched through all of Amir's reviews of multichannel preamps, and I made a list of all the ones that have balanced outputs. The rest of my system has balanced interconnects, and I want to keep the entire analog signal chain balanced if possible.

So I made a list, and the recurring theme is that these devices do not seem to perform very well compared to modern DACs. My list:

View attachment 362127

Has anyone tried to dig any deeper to figure out where the performance issue comes from? Is it from processing the digital signal, or is it the DACs themselves?

I imagine with 16+ DACs in a single unit, a company would be tempted to choose something cheaper. But if the DACs were better, would the audio quality be better?

I'm asking because I recently bought a MiniDSP HTx, which rates top of the chart for SINAD...but it doesn't decode multichannel audio. And I don't have a multichannel preamp, though I'm curious about going that route.

What I'd like to know is, what happens if I get a relatively inexpensive AVR/ preamp that can decode the audio, and then send a digital signal to my HTx?

Would I end up with the high quality audio the HTx's review indicates it is capable of? Or would I still get the noisier/ lower quality audio described by the preamp reviews?

The professional decoder circuit design is simple, as ESS or AKM highly integrated chips can handle all the problems and achieve high separation stereo. However, AVR is a multi-channel processor that requires signal processing by DSP before being transferred to DAC, making multi-channel volume management relatively complex. While high-performance chips like ESS may not match the DSP chip and multi-channel volume management, it's worth noting that there is currently limited demand for multi-channel music and watching movies doesn't require such high parameters. Therefore, the performance of AVR is more than sufficient for use.​

 

[er|κzvio1in]

My reading about the MiniDSP HTx is that an AppleTV 4k will pass decoded multichannel audio through to the HTx. But it is unclear what other devices work like that. And I expect the Topping is similar to the MIniDSP, so you will probably get the same results.

I investigated this topic the past week and my conclusion is that the Apple 4k route is the only major application for the flex HT when using the HDMI eARC input. Other devices don't convert surround bitstream signals to LPCM. I've even consulted LG about this to confirm.

So if you don't want to use HDMI, you're better off with another MiniDSP product. If you do want the HDMI input but without Apple 4K TV, the signal is limited to stereo. A flex HT with decoder built in would be much more versatile and popular, but more expensive because the additional electronics and license costs.