miniDSP Tide16 - Holy Grail with 16 Channel Atmos/DTS:X, high SINAD

[peng]

Anything using DSP is a suspect.
The ones we know for sure are impeccable are RME and Fiio (I think, if these are the ones who use DSP dedicated chip).
MiniDSP also applied a partial fix after the thread (and because of it)

We've seen similar hits at Topping's implementation, or Bluesound for example.
There's a nice paper at the thread I linked about filter stacking and precision.

Thanks, were the lower numbers mainly due to increased noise or distortions? Also, if the lower performance is in proportion across different platforms, lowering from 118 dB SINAD is still going to be lowering from 100 dB. (100-30=70, 118-30=88 (still quite good).

Regardless, for SINAD chasers who want to use the same HT system for pure direct audio enjoyment, the Tide 16 might be the only game for the time being, which other AVP can do better than 110 dB or even 107 dB SINAD in pure direct mode, other than miniDSP that anyone is aware of?

 

[Sokel]

Regardless, for SINAD chasers who want to use the same HT system for pure direct audio enjoyment, the Tide 16 might be the only game for the time being, which other AVP can do better than 110 dB or even 107 dB SINAD in pure direct mode, other than miniDSP that anyone is aware of?

Let's hope they did it this time.
It's noise that give the penalties, people who measure at the miniDSP forums moaned for years about it.

Best bet is a dedicated DSP modern chip and the software to make it shine.

 

[Bjorn]

It's a miniDSP so I am quite sure (99%?) that manual will work except the FIR filter part as they might still be disabled/lockout if Dirac license is installed just not used.
According the miniDSP's website info, the rephase FIR tool can be used:

What you will need​

1. The rePhase program. rePhase can be downloaded from the official rephase page.
2. A convolver/filtering engine. For phase linearization, one of:
   • A miniDSP Flex Eight without Dirac Live
   • An OpenDRC-DI and the OpenDRC 2x2 plugin
   • For a linear-phase crossover, one of:
   • A miniDSP 2x4 HD
   • A miniDSP Flex without Dirac Live

So, unless I misunderstood something there, without Dirac Live would mean you did not purchase and install the license, once you have done that, the hardware has Dira Live "baked in". I also don't understand fully what that rePhase program is, but I assume it is the tool for users who want to do their own manual FIR filters. I am further assuming the same that applied to the Flex series also would apply to the Tide series, lots of assumption, sorry.

But these products don't have enough processing power to run both Dirac and FIR. The Tide 16 has plenty, so we may see both combined here. Crossing my fingers.

 

[nothingman]

Small channel count and Dirac ART seems to be 2 ideas at odds. This is where a Tide8 would lose out to an AVR that can do ART. An 11-channel 3800 with ART will run around $2k. Can a Tide8 hope to go below that, include ART, and still make a profit?

I think there are enough 5.1, 5.1.2, 5.2, 7.1, PC systems out there only using a 2x4 HD that could be make a Tide 8 viable assuming they add HDMI 2.1 and maybe streaming/roon capability.

I understand the value argument, but it can be exhausting to see 5.x setups or 7.x setups treated as budget and unworthy of flagship performance. Some people just can’t expand their setup with more spakers, and it’s appreciated to have options for high SINAD performance without buying something with way more channels than I need (and the price to match).

A Tide 8 would also be a good back up (easier to do than using the HTx) for my 7.1.4 HT system. To me, if a Tide 8 is price down to about $2.5 K, it would be very attractive, as that would be the cost of a stupid Cinema 40 here in Canada. Obviously if not for curing OCD/audiophile things, the X3800H is still the value king that a Tide 8 cannot compete with…[cont.]

I appreciate these attempts to will a “Tide8” into existence, but I think over-buying channels is just the nature of the game. Yes, it feels like overkill and over-paying, but the economics of a smaller channel device aren’t great. It won’t save manufacturers that much in hardware and it will rule out a large portion of their potential customer base.

At least miniDSP is packaging 16 channels in a unit much smaller than anything from Denon/Marantz. More than the cost and unecessary channels, using one of their aircraft carrier sized AVRs just as a processor would be extremely annoying.

Hope I’m wrong! The most I can manage in my system would be LCR plus four small subs, so seven channels total, but I have no expectation that I’m going to get ART in that sort of package, at least not for several years when it’s thoroughly dispersed downmarket.

 

[peng]

But these products don't have enough processing power to run both Dirac and FIR. The Tide 16 has plenty, so we may see both combined here. Crossing my fingers.

That's entirely possible, albeit on the optimistic side because with the seemingly much more powerful DSP, the channel count increased to 16, plus ART capability, plus the immersive surround sound codecs etc.. so there are major offsets. Anyway, I have already asked them (support) a few questions and hopefully they would respond.

 

[voodooless]

Also, the paper at the end of the thread describes very closely these penalties, regardless of the hardware they use.

Edit: here it is, very old but...

https://firmware-developments.com/WEB/DOC/REF/ZauchaIIR_AES.pdf

The paper is about fixed-point math DSP operations. So on a float platform (like the SHARC, XMOS will probably run fixed-point math for DSP), there will be slightly different trade-offs to be made, but for sure, you'll incur penalties when doing math with finite resolution. I'm pretty sure the implementations of these systems can be carefully tweaked to have much improved performance. The general problem solver will probably be to add some dither to remove quantization errors where needed.

 

[Sokel]

The paper is about fixed-point math DSP operations. So on a float platform (like the SHARC, XMOS will probably run fixed-point math for DSP), there will be slightly different trade-offs to be made, but for sure, you'll incur penalties when doing math with finite resolution. I'm pretty sure the implementations of these systems can be carefully tweaked to have much improved performance. The general problem solver will probably be to add some dither to remove quantization errors where needed.

I'll take your word for it, I have understood the hard way that filter design, stacking, etc is way above my pay grade, let alone their hardware implementation.

Judging by the rarity of formally educated people who know EXACTLY what they do at forums (and rightfully so, they sell themselves like hot cakes, no reason to do it for free) , I suspect that a few around know how to play the game exactly as it should.

 

[Ultrasonic]

Other miniDSP's for example can be used either with DIRAC or manual but not both (correct me here if that's not true these days) .

You've got this wrong. Other miniDSPs with Dirac Functionality can still apply manual EQ as well (although there's no reason to from my testing), as well as all crossover, delay, signal routing functionality that is very useful. What is traded for Dirac Functionality is the ability to apply manual FIR filters.

 

[Sokel]

You've got this wrong. Other miniDSPs with Dirac Functionality can still apply manual EQ as well (although there's no reason to from my testing), as well as all crossover, delay, signal routing functionality that is very useful. What is traded for Dirac Functionality is the ability to apply manual FIR filters.

FIR is the whole game here, that's what we're asking for ages.
That's why I specifically wrote about full potential.

 

[Ultrasonic]

I would really like to know if the Tide uses dedicated volume control ICs such as most if not all higher end AVPs do and even avrs such as D+M’s. I asked them that question when I bought the HTx, but they wouldn’t tell me, answered all my other questions though.

If anyone knows the answer, please kindly share.

I don't know specifically for the TIDE16 but it is standard across other miniDSP devices that the volume control is performed on the DSP chip so I'd be very surprised if the same wasn't true for this latest model. Not sure this is any sort of problem though.

 

[Ultrasonic]

is the Tide8 on their roadmap? What’s the estimated release date?

Not that has been announced, and miniDSP's approach is never to say what they're planning but rather just announce products when release is imminent, like with the Tide16 here.

 

[Ultrasonic]

FIR is the whole game here, that's what we're asking for ages.
That's why I specifically wrote about full potential.

Well, no, it's not. You may want it but that doesn't make everything else irrelevant. Not even close.

 

[Sokel]

Well, no, it's not. You may want it but that doesn't make everything else irrelevant. Not even close.

I guess that since premium is not up there, you're probably right about it.
Niche of a niche is hard to get.

 

[peng]

I don't know specifically for the TIDE16 but it is standard across other miniDSP devices that the volume control is perform on the DSP chip so I'd be very surprised if the same wasn't true for this latest model. Not sure this is any sort of problem though.

Not a problem as such, but psychologically, I don't feel good when using modest gain power amps such as those with 25-26 dB gain when using Dirac Live that grab about 10 dB for reserve, resulting in the Flex (even for the balanced output) volume having to crank up to near 0 for moderately loud level from 8-11 feet with speakers that have sensitivity below 90 dB. With dedicated volume chip such as Marantz, Anthem AVPs you get higher maximum output voltages than the mini's 4-4.2 V. The trade off, I guess is, by skipping the volume IC, miniDSP can achieve higher SINAD, along with their quad differential/balance DAC implementation, the Tide is expected to achieve a whopping 118 dB SINAD at 4 V, that's 12 dB better than the AV10 and AVM90 that uses reference class ESS DAC IC, that' a real case of no reference class ESS DAC IC plus better implementation, trumps the use of flagship DAC ICs. It's rare (as most talks about how implementation is what counts are typically bs/excuses), but we can actually see the results in the miniDSP's cases.

 

[Oddball]

…or any other pre/pro on the market for that matter. Hey, it’s a fantastic box and a great value, but it has the same problem that every other pre/pro has.

The problem is two fold:
• It doesn’t have linear phase crossovers (or support for them)
• Its DSP only supports processing at 48hHz
(if there is a unit that has all this, it has to run around $30k or more)

Well, we all have our preferences. I value simplicity and reliability but try to get best out of gear that gives me that.

Just a note that latest Dirac ART software that is now available on gear such as Tide16, HTP-1, or D&M gear (used to be only Storm) can be in below $4K range (or $2K for entry D&M AVRs). ART does not have a traditional crossover concept and blends the speakers and subs through collaboration and mutual support. Does other things like reducing decay and has ability to fill in the dips of one speaker by assistance of other supporting speakers. Nothing is perfect so there are still some time alignment issues between sub 150hz range and lack of EQ support below 20hz.

Supposed to be available as PC version later this year. So perhaps one less thing to worry.

Have always been happy with 48kHz DSP processing so not aware of the landscape above that.

 

[boXem]

Thankfully, you are here to enlighten us...

See below

You aren't exempt from that bias.

Reason why I try to STFU when I didn't actually practice the subject I am talking about.

 

[mdsimon2]

It makes me wonder if such deterioration would occur and more less equally, similarly to all AVPs, AVRs, ie. not just to miniDSP's.

I think it is a very common issue that is not measured at ASR. You see hints of it when filters are inadvertently engaged.

For example, see Emotive RMC 1+ results below, with speakers set to Small a HPF is engaged and you get a low frequency noise penalty, with speakers set to Large the HPF goes away and so does the low frequency noise (https://www.audiosciencereview.com/forum/index.php?threads/emotiva-rmc-1-av-processor-review.66077/).

I've long advocated for some standardized tests in this area, but I don't think they are coming. Better to have your own measurement capability if you are curious.

I think I would do something pretty simple and hopefully representative of how most people use crossovers.

1) 30 Hz -1 dBFS tone while applying a 80 Hz 4th order LPF (mimicking a sub LPF x-over)
2) 100 Hz -1 dBFS tone while applying a 80 Hz 4th order HPF (mimicking a main HPF x-over)

I would also love the tests below with no filters applied to see how the DAC handles intersample over clipping.

3) 11.025 kHz +3 dBFS tone with DAC volume control at 0 dB
4) 11.025 kHz +3 dBFS tone with DAC volume control at -3 dB

Michael

Michael

 

[voodooless]

Reason why I try to STFU when I didn't actually practice the subject I am talking about.

Clearly, you (think you) know enough to see that people are wrong. The polite thing would be to point that out, rather than to use an ad hominem. Because otherwise: what's the point?

 

[Ultrasonic]

Not a problem as such, but psychologically, I don't feel good when using modest gain power amps such as those with 25-26 dB gain when using Dirac Live that grab about 10 dB for reserve, resulting in the Flex (even for the balanced output) volume having to crank up to near 0 for moderately loud level from 8-11 feet with speakers that have sensitivity below 90 dB. With dedicated volume chip such as Marantz, Anthem AVPs you get higher maximum output voltages than the mini's 4-4.2 V. The trade off, I guess is, by skipping the volume IC, miniDSP can achieve higher SINAD, along with their quad differential/balance DAC implementation, the Tide is expected to achieve a whopping 118 dB SINAD at 4 V, that's 12 dB better than the AV10 and AVM90 that uses reference class ESS DAC IC, that' a real case of no reference class ESS DAC IC plus better implementation, trumps the use of flagship DAC ICs. It's rare (as most talks about how implementation is what counts are typically bs/excuses), but we can actually see the results in the miniDSP's cases.

You make a good point. My AVR is capable of higher output levels than my SHD and I'm not sufficiently familiar with general home theatre setups to know if this is something that may be necessary in a lot of cases.

One thing that would help is if digital headroom is automatically applied internally on the Tide16, which I'd hope it might be with all the advanced Dirac functionality.

 

[jeffaegrim]

No, my first statement is not absolute. Linear phase filters are undesirable because they add latency. In a video system, that will cause lip sync problems. The solution is to delay the video. In order to do that, you would need a lot of RAM depending on how much latency you need to achieve.

I think that JRiver has a lip sync feature, but I do not know what the maximum delay is. The problem is that not videos can be passed through JRiver, for e.g. if you want to watch a Youtube video or stream from Netflix. I have never tried to find a solution for video, so this is something I am largely ignorant about. If I had to do it, I would simply use minimum-phase. If you (or anybody else) knows of a way to delay ALL video to achieve lip sync, I would love to learn about it.

I don't think that pre-ring is much of a problem. It can be managed so that it is inaudible. AFAIK the only downside of linear-phase that can not be overcome by most of us is its latency.

Thanks for the response Keith. And no - I sadly have no knowledge of how latency can by compensated for system-wide. I guess I will just stick with minimum phase for now.