Review: Dual sub EQ with miniDSP + Audyssey XT

[rojo]

I got my miniDSP a few days ago. w00p w00p!

Since I've got two subs but my receiver's flavor of Audyssey doesn't have the ability to correct multiple subs, I figured a miniDSP would be a suitable, perhaps superior, substitute. The miniDSP, a UMIK-1 and a plugin cost me around $200 -- a worthwhile investment to make sure my subs perform to their full potential, as well as for a shiny new hacky gadget for me to play with.

However, I still want to leave my receiver's Audyssey bass correction enabled. I enjoy Audyssey's Dynamic EQ feature which boosts loudness at lower volumes while flattening response when approaching reference volume.

I opted for the 4way Advanced plugin. I don't really get why recommendations and tutorials often point to the 2way Advanced plugin for equalizing dual subs. 2way Adv requires the LFE signal to be split into both inputs, and only two curves can be applied. As far as I can tell, the 4way Adv plugin is better suited for the task by allowing a 3rd or 4th sub to be added later, and only requires a single signal without a splitter. 2way Adv would be better suited for stereo signal corrections. (Edit: fuzz explained what I was missing below.)

Get to the point already!

Anyway, Google is littered with threads of users asking how to implement a miniDSP to correct subs, but it isn't very revealing for people like me who also appreciate Audyssey's other general corrections, who wish to use both at the same time. Maybe this thread will relieve that shortcoming a little.

The process

After I ordered my miniDSP and UMIK-1 kit, I downloaded and installed the latest beta of REW.

The first step, before taking any measurements, is to calibrate the sound card. It is generally recommended to use a USB sound card, as a PC's onboard sound is often of dubious quality. Since I just happened to have an old Creative Xmod on hand, I tried that. I failed. It turns out that the Creative Xmod is not a very good sound card for running REW. After a couple dozen attempts at various permutations of output and input levels, loopback calibration always resulted warnings about unreliable results or clipping; and the graphs all looked like tall blades of grass. Many of the sound card driver's volume control options are grayed out or missing in Win 7, so I was unable to defeat the apparent feedback. Maybe the Xmod was better with XP drivers, but it's useless for my setup.

My motherboard's integrated sound card worked just fine, though -- an NVidia high definition audio device as part of an older nForce4 chipset.

With my sound card calibrated, my progress halted while I waited for Hong Kong Speedpost. In the meantime, I read a good chunk of the REW help file. I'm sorry to say that by about page 75, I admitted defeat, unable to absorb any more theory until I could witness REW in action. To be honest, between my having read that far and a Google search or two when I got stuck, I acquired enough basic understanding that I never needed to resume the slog through the documentation any farther.

After only a few days, my miniDSP kit arrived, and I could begin in earnest.

Using the 4way Advanced plugin, I muted all sub channels but one. In REW, I ran a 4-sweep series with the mic at each of my listening positions from 10Hz - 160Hz, measuring each sub individually, and averaging the results for each sub on the All SPL tab. With all the measurements made, I saved all to a measurement file for each sub for a later step. For this series of measurements, I let REW choose the appropriate target level in EQ -> Target Settings.

After EQ -> Filter Tasks -> "Match Response to Target" and "Send Filter Settings to Equaliser" and loading the filter for each sub into 4way Advanced, I unmuted all sub channels (except the one where nothing is plugged), set a 48dB/octave high pass filter at 15Hz to provide a little protection for my ported subs, and re-measured. The response was pretty flat. Satisfied with the measurement, I moved my miniDSP from my sound card to my receiver's LFE channel -- out of development and into production, if you'll pardon the geeky metaphor.

Next, I ran Audyssey setup to let my receiver properly appreciate the curiously flat response of my subs, with the intention of flattening any previously stored Audyssey sub corrections.

After tweaking my receiver's crossover settings, etc, I plugged the miniDSP back into my PC and reloaded my saved measurements into REW -- this time with a house curve added to REW's preferences. My house curve is as follows (logarithmic interpolation enabled):

20 1.5
30 6.0
80 0.0

(Note: I initially tried setting the peak at 20Hz. It sounded pretty rumbly, even for passages that shouldn't have any rumble. The revised curve above sounds more natural. My room is about 2300 cubic feet, or a little over 65 cubic meters.)

This time I lowered the target level as needed to keep my desired curve underneath the actual uncorrected response of my subs. I set the house curve after running Audyssey because I didn't want Audyssey flattening / correcting the slope I intentionally introduced.

The results

These measurements were taken with 2.1 (stereo mains + LFE), simply because I was too lazy to unplug my mains from my receiver. The mic was in my primary listening postition.

Was it worth it?

Well, the subs with miniDSP + correction sound excellent. But honestly, Audyssey did a great job with its auto calibration as well. To me, the biggest benefit of miniDSP over Audyssey alone is that with miniDSP, I can manually configure the delay / distance from each sub to the primary listening position; whereas with Audyssey alone, both subs are doomed to play at the same time regardless of distance from listener.

Is this a problem? I'm guessing it could be a problem for difficult rooms -- rooms with modes / dips in inconvenient frequencies, rooms with a brittle reverb, or subs in non-optimal but non-negotiable placements that create standing waves or cancel each other. My room is apparently pretty tame, though. To be honest, I wasn't unhappy with my sound before. I can't really decide whether I'm any more or less happy with the sound with the miniDSP in place.

So, if the addition of the miniDSP wasn't a profound improvement, was it worth the 200 bucks and the hours of studying help documentation, Googling, measuring, tweaking and poking?

For the knowledge I gained, I'd have to say, yes. I've had a lot of fun playing with my new toy. Besides, I have a problem trusting automation. I don't let Windows decide which of my files to back up. I archive old emails manually. My car has a manual transmission. And now I don't let Audyssey decide my sub EQ. But most importantly, I've got the timing adjusted on my subs so the sound from both hits my ears at the same time, and it sounds like one sub. Well, no, actually, it sounds more like my bookshelf speakers are defying physics and producing all the ground rumbling bass, and the subs aren't really there.

 

[BoredSysAdmin]

Well, no, actually, it sounds more like my bookshelf speakers are defying physics and producing all the ground rumbling bass, and the subs aren't really there.

Great write-up and great another confirmation that Audyssey could do pretty decent job fixing room issue as your measurements confirm

Also best subs are ones you don't notice

 

[fuzz092888]

I opted for the 4way Advanced plugin. I don't really get why recommendations and tutorials often point to the 2way Advanced plugin for equalizing dual subs.

I have the 2 way advanced and as far as I can tell from the diagram for the 4 way crossover, it is theoretically superior to the 4 way advanced plugin for dual subs.

2 way adv requires the LFE signal to be split into both inputs, and only two curves can be applied.

I'm not sure where you're getting this information, but the second half of it is incorrect. First of all, the 4 way advanced plugin (according to the diagram anyways) only gives you 6 bands of PEQ for both inputs, and then 6 bands/output. The 2 way advanced plugin gives you 6 bands per input and then 6 bands/output giving you a total of 12 bands per sub for dual sub vs 6 bands that affect both subs and then 6 bands that affect each individual sub for dual subs with the 4 way advanced.

As far as I can tell, the 4way Adv plugin is better suited for the task by allowing a 3rd or 4th sub to be added later, and only requires a single signal without a splitter. 2way Adv would be better suited for stereo signal corrections.

A second miniDSP is your best option for a 3rd and 4th sub.

EDIT: On the bright side it's only $10 to switch over if you ever feel curious enough. I'm not saying you'll get a huge difference, but every little bit can help. Also, I'm not sure if you played with the settings in REW for what you wanted your target to be, but if not, learning how to adjust those can help yield better results as well.

 

[rojo]

Ah. The 4way Advanced plugin is mono input. The first block lets the user select either left or right to use as the input. Since it's mono, you're right that the input peq is applied to all outputs indiscriminately. I keep mine flat. But as long as I'm fine with 6 or fewer peq corrections per output, I can have up to 4 subs. Indeed, I am using the third channel for my bass shakers. My jbl sub is only using 3 peqs and my paradigm, 4. It sounds good to me. *shrug*

When you talk about the target eq, are you talking about the house curve? Or within the eq section of rew? I did the former, but not much of the latter -- only what's described in the write-up above. If I hear anything that bugs me (or if I just get bored and feel like tinkering one day), I'll take your suggestion and dig deeper. Thanks!

Sent from my LG-VS980 using Tapatalk

 

[fuzz092888]

Ah. The 4way Advanced plugin is mono input. The first block lets the user select either left or right to use as the input. Since it's mono, you're right that the input peq is applied to all outputs indiscriminately. I keep mine flat. But as long as I'm fine with 6 or fewer peq corrections per output, I can have up to 4 subs. Indeed, I am using the third channel for my bass shakers. My jbl sub is only using 3 peqs and my paradigm, 4. It sounds good to me. *shrug*

For my dual opposed each driver is getting 12 bands of EQ and I easily use all 12, but then again I think my EQ tolerances are set tighter than you have yours.

When you talk about the target eq, are you talking about the house curve? Or within the eq section of rew? I did the former, but not much of the latter -- only what's described in the write-up above. If I hear anything that bugs me (or if I just get bored and feel like tinkering one day), I'll take your suggestion and dig deeper. Thanks!

Sent from my LG-VS980 using Tapatalk

Target EQ in the REW EQ section. All those parameters that you can specify before you let it auto EQ. Plus you can hit "optimize" this or that. Between the REW software and the miniDSP software, it's all pretty customizable.

 

[rojo]

Further chronicling my journey, I decided it'd be nice to have REW detect the distance of each of my subs from the mic -- not physical distance, necessarily, but the time between signal initiation and the first significant impulse, which could conceivably be a reflection. The difference of this value between each sub will be set in my miniDSP, and will hopefully be more accurate than my guesswork.

The measurement I need requires "Use Loopback As Timing Reference" to be checked on the Analysis tab of REW's Preferences, as well as, of course, a loopback from an unused output to an unused input. Since I'm using a USB mic and HDMI signal, my implementation of a loopback isn't exactly going to be traditional.

I've seen a thread or two describing the use of ASIO4ALL as an HDMI sound interface driver within REW to redirect unused channels as a loopback. As far as I can tell, ASIO is basically a software audio device router. It's apparently not for activating multiple interfaces simultaneously, though.

Anyway, at first, my Geforce GT610's HDMI output wasn't available to ASIO4ALL. It took me a while to figure out that the NVidia HDMI device needed to be configured as a 5.1 channel output in Windows Sound Playback Devices. It took me a while longer to discover that my receiver had to be powered on and in true, not simulated, multichannel mode before Windows would make any configurations available other than 2.0 channel.

Doing all this and changing the sound card driver in REW's Preferences window to ASIO reveals 8 channels for the outputs, one of which I'm sure is my subwoofer channel. I'll trial and error to find out. Curiously, my UMIK-1 has two selectable inputs. I'm not sure yet how I'm going to twist that into a loopback. I haven't experimented beyond this yet. More later.

Of course, advice is welcome to save me from re-inventing the wheel.

Edit: I'm stumped at the moment. Obviously, the UMIK-1 can't be used in a loopback. I tried plugging a 3.5mm-3.5mm cable between my analog speaker and line in jacks on my PC which worked fine to calibrate my sound card under REW's Java driver, but it seems ASIO is expecting my analog output to be multichannel. If I try to configure that jack as having multichannel output in Windows' Playback Devices, Windows lets me get to the end of the wizard and says, "Format not supported by the device." If I ignore all this and just leave the jack configured as stereo, REW + ASIO still gives me "HD Audio Speaker 1" through 8 from which to choose my Timing Reference Output. But each selection results in an unmoving Ref In meter. I'd thought about piping the audio back from my receiver to my PC's optical or digital coax jack, but my receiver has no digital out other than HDMI for zone 2. I'd also considered piping audio from my PC to my receiver, to my television and then back to the PC via optical, but my TV has only one optical out which already serves another purpose. Suggestions?

I think I'll read as much of this thread as I can stand and see whether I find any clues there. I also found AustinJerry's guide which looks quite promising.

 

[andy_c]

Edit: I'm stumped at the moment. Obviously, the UMIK-1 can't be used in a loopback.

(...)

I think I'll read as much of this thread as I can stand and see whether I find any clues there. I also found AustinJerry's guide which looks quite promising.

I don't think you'll find much about loopback in the thread you're referring to. It's a great thread, but few if any of its participants are using loopback AFAIK.

What I've been able to find regarding loopback with USB mics is a couple of threads at HTS. In post 3 of this thread, John Mulcahy says this:

Before you could combine measurements in that way you would need to make sure they were properly time aligned. Time alignment needs a timing reference for REW, which is provided using a loopback connection - but that is not possible with a USB mic.

In a post 5 of this thread, he talks about trying to find a workaround for this, but it is only in the conceptual stage.

Wouldn't really help, there remains an unknown and variable latency between REW asking the audio system to start generating the test signal and something actually appearing at the audio output - that can be substantial (tens or even hundreds of ms). There is another trick that Holm uses, its 'keep audio active' option, which basically keeps the audio interface running (sending it audio data, even if silence, and capturing audio data from it) for as long as the program is running, in which case a running sample count can be kept to reference subsequent measurements to the first measurement taken. That might be another option, though it only allows measurements to be referenced (timing wise) to the first measurement made in any given session so it wouldn't be possible to compare measurements made in different sessions.

The HolmImpulse feature he's talking about is called "time locking" and is discussed a bit in this thread in the Parts Express forum.

Even when not using a USB mic, getting loopback to work with HDMI is tricky. I figured out a way to do it while doing some measurements on the bass management of the Emotiva UMC-200. My post describing that is here. The relevant text is below. The Tascam device being referred to is the US144MkII.

To do all these tests, it's necessary to apply the test signals using HDMI. But also one needs a stable loopback timing reference to be able to measure changes in group delay. The first thing I tried was energizing only the left channel of the UMC-200 using HDMI. The loopback timing reference consisted of a cable from Tascam channel 2 analog out back to channel 2 analog in, and specifying these two signals as the REW timing reference output and input respectively. The results from those measurements made no sense, and my guess of the cause is a lack of timing synchronization between the Tascam analog out and the HDMI out of my laptop. Using two different sound "cards" in this way is possible because I was using ASIO4ALL, and all sound devices enabled in the ASIO4ALL configuration appear as "ASIO4ALL v2 device" or something similar in REW.

So the next thing I tried was using the HDMI right channel output as the timing reference output for REW, in addition to the left-channel HDMI test channel signal. This means that the loopback timing reference now goes through the UMC-200 itself, rather than just being confined to the test equipment. This required taking the UMC-200 right-channel preamp output and bringing it back into the channel 2 analog input of the Tascam for the timing reference input. The measurement input was the Tascam channel 1 analog input and was connected to the preamp output of the channel under test of the UMC-200. To keep the timing reference consistent, the distance of the right channel of the UMC-200 (now part of the timing reference loop) was kept fixed at its minimum-delay value of 32.8 ft (maximum distance). This way, the test signal path (HDMI out, analog in) was symmetrical with the timing loopback signal path (also HDMI out, analog in). This test setup was quite fussy, but with a lot of trial and error I was able to get good data.

 

[rojo]

Interesting. I'd come to believe that, even using ASIO4All to perform audio routing, although it's possible to use a separate device for capture than the one selected for playback, it's still impossible to have two playback devices or two capture devices active at the same time. I'd briefly considered getting an HDMI video game capture device and selecting that as the capture device for the loopback, but that wouldn't have solved the issue of having the UMIK-1 perform the primary capture without having to rely on a different device for loopback capture. It looks like a loopback is impossible with a USB mic.

Wonder whether there's a USB mic that'll record the left channel from the air and the right from an rca plug? That'd be the easiest solution it seems. Otherwise, your method of plugging this XLR interface mic into this Tascam mixer would probably be pretty much the only worthwhile solution. I'll have to read and Google some more, but I'm not really ready to spend more money on this yet.

At this point, I've already used Audyssey to detect and reveal the aurally calculated distance of each of my subs, and I dialed the difference into my miniDSP without relying on REW for the measurement. I guess it worked as intended. *shrug* My subs are so close to my seating area I'm not sure it matters much, to be honest. But they do sound good both in my primary listening position and across the room, for all that matters.

Sent from my LG-VS980 using Tapatalk

 

[andy_c]

I'm using an EMM-6 (but with the Parts Express calibration) into a Tascam US144 MkII USB interface, which has the mic preamp with phantom power built in. I'm very happy with it. The cheaper US122 MkII is almost the same, but does not have S/PDIF output.

Or if you want to use just a USB mic, there's always HolmImpulse. It won't give correct absolute delay numbers, but if you do one measurement, then enable time locking, all subsequent measurements in that session will have the correct delay relative to the first measurement.

 

[rojo]

LOL I just had a complete Frankenstein idea. I could plug the UMIK-1 into my little Linux appliance computer and have it pass through audio untouched into only the left channel of a 3.5mm - RCA splitter. For the right channel, I'd plug a preamp out from my receiver. The splitter's 3.5mm plug end would then go into the line in jack of my HTPC.

Ah, then I'd have to calibrate my Linux computer's sound card, as well as merge that calibration with my HTPC's sound card. Brain hurt. The universe would probably implode as soon as I hit "Begin sweep" anyway.

I guess I'm done for now.

 

[rojo]

I think I might've figured it out, for Windows at least. Mac and Linux users might be able to accomplish the same thing using JACK, though it won't be as easy.

1. Install the free / donationware VB-Cable. (Download link is currently in the middle of the page, in the center column as of the writing of this post.)

2. In Control Panel --> Sound --> Playback devices, there'll be a new Windows sound device. Configure it to the same number of channels as you have your HDMI interface configured. (Not sure whether that's necessary, but it can't hurt.) Then go into its properties and un-mute it.

3. In Recording devices, there'll be a new Windows sound device there as well. Go into its properties and set it to the same sample rate as you have REW.

4. Open REW. Go into Preferences. Open ASIO's control panel. Make sure your new virtual device is enabled, as well as your HDMI interface and your microphone interface.

5. Close and re-open REW, and go back to Preferences. Set your Output to your HDMI device, to whatever channel you want to measure (1 for left front, 2 for right front, 3 for center, 4 for subwoofer, etc). Set your Input to UMIK-1. Set your timing reference output and timing reference input to the same name. Hit Check Levels and make sure that both the microphone meter and the timing reference meters are doing something useful.

I haven't run any measurements with this setup yet, but I think it should work.

 

[andy_c]

This looks promising. I'd be interested in your results when you try it. Maybe do measurements before and after a distance setting change in the channel under test to see if the impulse response shifts by the expected amount?

 

[rojo]

If I get the opportunity, I will. I was kind of hoping someone with more experience in such matters might do some testing for me though. When I'm home, I'm using wrangling a toddler or catching up on quality time with my wife, so I don't get much time to play with measurements. The time I do get, the HVAC is usually running and the toddler is generally raising my noise floor.

 

[fuzz092888]

I haven't put any time or effort into this because it's one if the things Audyssey does really well. Plus you don't need to have Audyssey engaged to get those benefits.

 

[rojo]

I may not know what I'm doing, but I don't think this is going to work. If I measure, then perform a second measurement, the 100% peak impulse occurs at successively higher times with each measurement. It'll start at 30-something ms, then go to 40-something, then 50-something... To get close to a plausible value, I have to close all measurements before beginning a new one. Weird.

But even if I start with a clean slate before each measurement, the 100% peak impulse measurement is unpredictable. As follows are some measured values, all taken playing the same sub with the mic in the same position:

 miniDSP         measured
out delay      peak impulse
---------      ------------
   0ms              35ms
   4ms              32ms
  7.5ms             36ms
   0ms              26ms
   4ms              41ms

Oh well. I guess there really is no substitute for putting the loopback on the same i/o devices as are performing the measuring. I really hope someone else will verify my test results, though.

 

[andy_c]

But even if I start with a clean slate before each measurement, the 100% peak impulse measurement is unpredictable. As follows are some measured values, all taken playing the same sub with the mic in the same position:

 miniDSP         measured
out delay      peak impulse
---------      ------------
   0ms              35ms
   4ms              32ms
  7.5ms             36ms
   0ms              26ms
   4ms              41ms

Oh well. I guess there really is no substitute for putting the loopback on the same i/o devices as are performing the measuring. I really hope someone else will verify my test results, though.

That looks like the same kind of thing I was seeing when I first tried loopback using an HDMI measurement. I got random, nonsensical delay measurements. What I found was that the measurement and loopback outputs had to be from the same device, and the measurement and loopback inputs also had to be from the same device, but the output device could be different from the input device.

Cool pic! Haha.

 

[andy_c]

Or if you want to use just a USB mic, there's always HolmImpulse. It won't give correct absolute delay numbers, but if you do one measurement, then enable time locking, all subsequent measurements in that session will have the correct delay relative to the first measurement.

In doing more research, it appears my statement above is wrong. I found a thread in the Parts Express forum in which the OP had a problem with HolmImpulse "time locking" not working. Bill Waslo (author of the OmniMic measurement software) made the interesting observation that if the A/D and D/A clocks of the sound device(s) are not locked to one another, that Holm time locking won't work.

In the case of a USB mic, the A/D clock is in the mic and the D/A clock is in the sound device, so these can't possibly be locked to one another. This seems to indicate that users of USB mics are SOL as far as trying to obtain a reliable timing reference in their measurements, regardless of the measurement software used.