• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

DSP in Separates System

Joined
Dec 19, 2020
Messages
73
Likes
21
Location
Pacific Northwest, U.S.A.
Hi All,

I have a friend who has a system that goes like this:

Streamer - > Schiit Modius (DAC) > Schiit Freya+ (PreAmp) > Mono-block Schiit Tyr (Amps) > MartinLogan 60XTi

He wants to add a subwoofer. I told him that if he wants to do that, he probably needs to implement some kind of room correction. I had heard about people using MiniDSP for room correction with separates, but while I assume the MiniDSP goes in between the streamer and DAC in the chain, I don't know where the subs would connect so that they ALSO are part of the room correction on a separate channel. Also, I'm not sure which exact MiniDSP model we would go for.

To me, the closest "application" I can find on their site is this page:


But then it makes me wonder if the SHD replaces his DAC altogether. Would the sub just be a third channel out?


Thanks
 
Check out the miniDSP Flex.



It could replace both his Modius and Freya+, and provide room correction and subwoofer integration.
 
The easiest way to add a sub, which would require the least amount of component swapping, would be to get a studio sub with XLR Line in and Line out, as well as a built-in crossover. Something like the Kali WS-6.2

Such a sub could simply be inserted between the Freya and Tyr, and your friend could keep all his Schiit gear and continue using the Freya to control volume.

If instead one would want to use a miniDSP with a digital crossover and potentially room correction, then your friend would ideally remove the Modius and Freya from the signal chain, since he would no longer be able to control volume downstream of the DSP unit, as that would mess up the balance between mains and sub.
 
Last edited:
Room correction is different from a crossover, which splits the low frequencies and everything else.

The miniDSP can do both.

AVRs have "bass management" which works as a crossover and mixes-in the "point one" LFE channel in movies.

Some subwoofers have line-level pass-through with a full crossover.

Or you can buy an active crossover which is easier to set-up but not as "flexible" as the miniDSP (and no room correction)).
 
When you add a sub to a system, you need to:

1. Volume match the sub to the mains. Failure to do this will result in thin or overwhelming bass.
2. Smooth out the frequency peaks and dips in the room because ALL bass does this, not just subs. Failure to do this will result in lumpy sounding bass.
3. Low pass the sub and high pass the mains. Failure to do this will result in misaligned phase over the band where freqs overlap, producing cancellation at many points in the FR.
4. Align the phase at the XO region. Failure to do this may produce cancellation at the XO point.
5. Time align the sub to the mains. Failure to do this will result in "flabby" sounding bass, where the upper freqs are heard first, and the bass freqs 10-20ms later. It will sound disjointed.

And all this is assuming that the subwoofers are properly placed in the first place. These are your friend's options, from worst to best:

1. Do not use any type of crossover. Use the mains full range, low pass the sub only.
This is what the majority of people seem to do (the subjective side of audio anyway). All they do is adjust the controls on the SW plate amp until it sounds right - i.e. rough volume and phase matching and maybe some PEQ. Most people do this subjectively, but it is extremely difficult to do without measurements, and it is impossible to achieve a satisfactory result with any degree of precision. This is REL's recommended approach. I think it is utterly stupid, it is really disappointing to see a subwoofer manufacturer mislead their customers this way. REL surely knows better, yet they spread misinformation.

2. High pass the mains, low pass the sub with an external analog crossover, or built-in XO in the sub.
This is what the remaining majority of the subjectivists do, usually without measurements. This is better, because the overlap region is smaller, and therefore less potential for phase cancellation. But - still no time alignment nor phase alignment, and no DSP for the natural peaks and dips in the FR. Furthermore, if the subs have DSP but the mains don't (e.g. as seen in Velodyne subwoofers), the DSP itself introduces additional latency, up to 30ms. This is HUGE and it will definitely worsen the time misalignment and produce flabby sounding bass. This DSP is usually low resolution and IIR because of the requirement to keep latency as low as possible whilst running on hardware with minimal processing power, so it will not correct bass freq peaks/dips in a rough fashion.

3. Use a hardware based DSP to high pass the mains, and low pass the subs
Examples of hardware based DSP: MiniDSP, built-in bass management of AVR's, Genelec GLM, DLBC/Dirac Art, etc. This is the most common recommendation you will see on ASR but it is still not the ultimate. This is better, because it ticks nearly all the boxes of the requirements laid out above. Nearly all hardware based DSP use automated software algorithms that removes decision from the user and these algorithms usually get it wrong, usually with no way to over-ride the algorithm. For e.g. some AVR's calculate the delay by directing you to enter the distance of the sub from the listener. This is wrong, the delay should be measured with a microphone with a timing impulse (we can discuss why). Also, processing is done on low powered hardware, and the low latency requirement again means IIR filters or mixed phase filters and USB microphones (or even worse - cheap uncalibrated microphones). So while there is some bass management, it is not the best - better precision, but not great precision. It may correct the bass up to 80-90% of what needs to be done if the algorithm gets it right. If you use this approach, you need to be aware of the shortcomings and do what you can to mitigate it.

4. Use software based DSP to high pass the mains, and low pass the subs
Examples: Acourate, Audiolense, Focus Fidelity, REW/RePhase, Matlab, Octave. These software packages vary in the degree of automation (i.e. use of a software algorithm), but most let you over-ride the defaults and give you more control over what you need to do. Some (like Acourate, REW/RePhase) are completely manual and relies on you to interpret graphs and make decisions. Octave and Matlab are even more manual in that you need to enter mathematical equations. It is my belief that manual correction is superior to "one button DSP" if you know what you are doing and most people don't, even ASR readers. The disadvantage of this approach is the steep learning curve, requirement for a certain type of hardware configuration (i.e. some kind of computer must always be in the signal chain), lack of convenience, and difficulty of use.

At the end of this, you need to explain to your friend that better correction comes at the cost of greater difficulty. He needs to choose how serious he is about bass correction, and how much inconvenience he is prepared to put up with.
 
Thank you everyone for your help. This is a lot of information. @staticV3 That was really helpful to explain that the DSP would not allow for matched volume control downstream. @Keith_W Do you do this for a living!? I like option 3. I have, in my 'system', just an AVR with Audyssey XT. It works pretty well for my limited application (My bottleneck is the room, not my system). Option 3 would basically be something like Audyssey, right?
 
Correct me if I am wrong but as much as I can see those expensive Schiit monoblocks have both unbalanced RCA input and balanced XLR inputs and switch gain controller in between?!
Mini DSP's like Flex HT, Flex and SHD (with a streamer) do all by hand except equal loudness normalisation meaning they are multichannel DAC and preamp PEQ's and FIR, they have limited Dirac (not full range or multi sub) support which you pay extra for. Choice would he go balanced or unbalanced means one or the other to the end (including sub's and DSP). Recommendation is multiple sub's and in 2.2 setup cut high.

On the other hand those speakers with 2x 8" woffer's don't really need either so strong monoblocks (94 dB claimed sensitivity per W into 8 Ohms) nor subwoofer (4x 8"=12~13" sub). He would most definitely benefit from deacent room correction (DSP).

It's not mine to judge nor I know the goal he wants to achieve or how big and long the space (room) is and listening at which distance. Is it mind blowing chest trumping physical bass feel at 120 dB SPL or nice mid 70 dB SPL comfortable listening (or less) with bass loudness correction.
 
First, all the above is true. This is just my 2cents and experience, i had the 2x4HD, Flex and Flex HT.

Streamer - > Schiit Modius (DAC) > Schiit Freya+ (PreAmp) > Mono-block Schiit Tyr (Amps) > MartinLogan 60XTi
I would replace the Schiit Modius (DAC) and Schiit Freya+ (PreAmp) with a MiniDSP Flex. You also need a measurement microphone (MiniDSP UMIK1) and a laptop that can run REW (Windows, Linux, Mac).

Here is the application if you have one subwoofer:

The MiniDSP 2x4HD or Flex have several inputs and 4 RCA outputs. You would use out1 for the left channel from 80Hz up, out2 for right channel from 80Hz up and out3 for the sum of left+right and below 80Hz, which goes to the subwoofer. out4 would be unused, but maybe your friend wants to add another subwoofer later.

The 2x4HD is a bit awkward in optics and user interface, so i would go for the Flex. It has a nice large display, all connections on the back, far more inputs and a nice volume knob. In terms of fidelity the 2x4HD is fine unless you listen in the near field (it has some noise, some can hear a faint hiss). The Flex is better, no noise, even in the near field. Sadly the Flex costs more.
 
@Keith_W nailed it with his answer. I've been through every version of his list and can attest that the best results are having a small computer in the chain along with a multichannel DAC. Yes, there is a steep learning curve learning the software and understanding the results but in the end it is worth it. Depending on how far down the rabbit hole you want to go miniDSP is a decent starting place to start learning about it all but ultimately lacks the processing power.
 
Thank you everyone for your help. This is a lot of information. @staticV3 That was really helpful to explain that the DSP would not allow for matched volume control downstream. @Keith_W Do you do this for a living!? I like option 3. I have, in my 'system', just an AVR with Audyssey XT. It works pretty well for my limited application (My bottleneck is the room, not my system). Option 3 would basically be something like Audyssey, right?

I don't do this for a living. I am a hobbyist like most people on ASR. I just take my hobbies seriously ;)

Everybody is bottlenecked by the room and the system. And for those of us who DSP, we are also bottlenecked by our knowledge and skill.

Yes, option 3 would be something like Audyssey. Don't ask me anything about Audyssey, I know very little about it.
 
When you add a sub to a system, you need to:

1. Volume match the sub to the mains. Failure to do this will result in thin or overwhelming bass.
2. Smooth out the frequency peaks and dips in the room because ALL bass does this, not just subs. Failure to do this will result in lumpy sounding bass.
3. Low pass the sub and high pass the mains. Failure to do this will result in misaligned phase over the band where freqs overlap, producing cancellation at many points in the FR.
4. Align the phase at the XO region. Failure to do this may produce cancellation at the XO point.
5. Time align the sub to the mains. Failure to do this will result in "flabby" sounding bass, where the upper freqs are heard first, and the bass freqs 10-20ms later. It will sound disjointed.

And all this is assuming that the subwoofers are properly placed in the first place. These are your friend's options, from worst to best:

1. Do not use any type of crossover. Use the mains full range, low pass the sub only.
This is what the majority of people seem to do (the subjective side of audio anyway). All they do is adjust the controls on the SW plate amp until it sounds right - i.e. rough volume and phase matching and maybe some PEQ. Most people do this subjectively, but it is extremely difficult to do without measurements, and it is impossible to achieve a satisfactory result with any degree of precision. This is REL's recommended approach. I think it is utterly stupid, it is really disappointing to see a subwoofer manufacturer mislead their customers this way. REL surely knows better, yet they spread misinformation.

2. High pass the mains, low pass the sub with an external analog crossover, or built-in XO in the sub.
This is what the remaining majority of the subjectivists do, usually without measurements. This is better, because the overlap region is smaller, and therefore less potential for phase cancellation. But - still no time alignment nor phase alignment, and no DSP for the natural peaks and dips in the FR. Furthermore, if the subs have DSP but the mains don't (e.g. as seen in Velodyne subwoofers), the DSP itself introduces additional latency, up to 30ms. This is HUGE and it will definitely worsen the time misalignment and produce flabby sounding bass. This DSP is usually low resolution and IIR because of the requirement to keep latency as low as possible whilst running on hardware with minimal processing power, so it will not correct bass freq peaks/dips in a rough fashion.

3. Use a hardware based DSP to high pass the mains, and low pass the subs
Examples of hardware based DSP: MiniDSP, built-in bass management of AVR's, Genelec GLM, DLBC/Dirac Art, etc. This is the most common recommendation you will see on ASR but it is still not the ultimate. This is better, because it ticks nearly all the boxes of the requirements laid out above. Nearly all hardware based DSP use automated software algorithms that removes decision from the user and these algorithms usually get it wrong, usually with no way to over-ride the algorithm. For e.g. some AVR's calculate the delay by directing you to enter the distance of the sub from the listener. This is wrong, the delay should be measured with a microphone with a timing impulse (we can discuss why). Also, processing is done on low powered hardware, and the low latency requirement again means IIR filters or mixed phase filters and USB microphones (or even worse - cheap uncalibrated microphones). So while there is some bass management, it is not the best - better precision, but not great precision. It may correct the bass up to 80-90% of what needs to be done if the algorithm gets it right. If you use this approach, you need to be aware of the shortcomings and do what you can to mitigate it.

4. Use software based DSP to high pass the mains, and low pass the subs
Examples: Acourate, Audiolense, Focus Fidelity, REW/RePhase, Matlab, Octave. These software packages vary in the degree of automation (i.e. use of a software algorithm), but most let you over-ride the defaults and give you more control over what you need to do. Some (like Acourate, REW/RePhase) are completely manual and relies on you to interpret graphs and make decisions. Octave and Matlab are even more manual in that you need to enter mathematical equations. It is my belief that manual correction is superior to "one button DSP" if you know what you are doing and most people don't, even ASR readers. The disadvantage of this approach is the steep learning curve, requirement for a certain type of hardware configuration (i.e. some kind of computer must always be in the signal chain), lack of convenience, and difficulty of use.

At the end of this, you need to explain to your friend that better correction comes at the cost of greater difficulty. He needs to choose how serious he is about bass correction, and how much inconvenience he is prepared to put up with.
Hi Keith:

How do I align the phase in XO region? I have a MiniDSP FLEX and REW. If it’s RePhase then I am stuck due to tap limitations. Other options?

Thanks.
 
Hi Keith:

How do I align the phase in XO region? I have a MiniDSP FLEX and REW. If it’s RePhase then I am stuck due to tap limitations. Other options?

Thanks.

This is why I don't use MiniDSP. Your only other option is to move to software based DSP. If you are really OCD, use one of the manual options - it gives you more control. Examples: REW + RePhase + DRC-FIR, or Acourate. With all manual options, the authors assume you know what you are doing. The former combo is surprisingly powerful and it's free, but it lacks a lot of niceties. All you need to do is look at the instructions for DRC-FIR and RePhase and you will see that DRC-FIR is programmed by using a config batch file which you have to modify with constant reference to the manual. RePhase has a confusing interface. REW is great, but you use it to do measurements / obtain information and needs the other two to make filters. Acourate does it all in one, and it costs money, but it's still cheap.

The downsides of going down this route is having to acquire all the bits you need to replicate MiniDSP's functionality and the learning curve. The solution also isn't as robust as a MiniDSP in that one bad Windows update can break your audio system, whereas a MiniDSP is "set and forget". Going down the software route gives you more processing power, more control, and ultimately measurably better results. Different products for different needs.
 
This is why I don't use MiniDSP. Your only other option is to move to software based DSP. If you are really OCD, use one of the manual options - it gives you more control. Examples: REW + RePhase + DRC-FIR, or Acourate. With all manual options, the authors assume you know what you are doing. The former combo is surprisingly powerful and it's free, but it lacks a lot of niceties. All you need to do is look at the instructions for DRC-FIR and RePhase and you will see that DRC-FIR is programmed by using a config batch file which you have to modify with constant reference to the manual. RePhase has a confusing interface. REW is great, but you use it to do measurements / obtain information and needs the other two to make filters. Acourate does it all in one, and it costs money, but it's still cheap.

The downsides of going down this route is having to acquire all the bits you need to replicate MiniDSP's functionality and the learning curve. The solution also isn't as robust as a MiniDSP in that one bad Windows update can break your audio system, whereas a MiniDSP is "set and forget". Going down the software route gives you more processing power, more control, and ultimately measurably better results. Different products for different needs.
Thanks for the quick and thorough response. I will need to do a bit of research and then decide if it’s a step that is within my realm of capability and patience :)
 
Yeah I think my friend will be moving to bigger speakers haha. This is just way too much.
 
Back
Top Bottom