วีดีโอ

I measured these with the miniDSP connected to the pre-out for the R channel of the processor I had at the time. I used L as the acoustic timing reference. The benefit is removing any impact of the LPF for the LFE channel which can't be changed on many devices. It's important to note that the L/R speakers need to be set to "Large" as well so the normal crossover on those channels doesn't impact things. I suggest this methodology (L timing and R temporary to subs; both set to "large") because it works on anything that has pre-outs for at least the LR speakers. I do things differently on processors with more advanced capabilities.

This is a little counterintuitive, but adding (enough) delay with identical polarity can look and sound like zero delay and polarity inverted on one sub. Inversion and delay are changing the *time* domain which changes the way the subs interact with one another at the measured listening position. If given the choice, when inverted with minimal delay looks better than "a lot" of delay, I'll take inverted with little delay every time. One thing to be super careful of - make sure you are using an acoustic timing reference with your measurements. Use the same speaker as the reference speaker for all measurements. When we're aligning subs, we're looking at phase interactions. In order to get accurate phase, we MUST use a timing reference. If the current output looks like the alignment tool, I'm guessing there's no timing reference.

You would run DLBC as 1 sub. So sub out from the AVR/Processor > miniDSP input > one sub on each miniDSP output. I used this in my own theater for a while and it provided the best of both worlds.

@ thanks. I appreciate the response

Yep. That's totally fine as long as you're not planning to use BEQ. I ran that way a long time myself.

It's been moved. Try this... make sure you have the ALL SPL tab selected. Then, click on the ACTIONS settings, not the CONTROLS settings.

It's exactly as @ryankramer mentioned - it's just been moved in the UI. In the beta release, you can go to ALL SPL, then right-click anywhere in the graph. Alignment Tool is at the bottom of the context menu that pops up. Just depends on which version you're running.

That's basic functionality. The only paid feature is "multi-input capture" for (I think) $100 USD. That option lets you capture input from more than one microphone at a time on a single measurement. If you don't see the other speakers, you have a config issue somewhere or perhaps a bad HDMI cable.

Thanks Jeff, I think it's because I normally use "pure direct" mode on my Marantz receiver, and use the right main out for measuring.

@@briansexton713 I think that's correct, but I'm rusty on the Marantz units. I think you'd need to set it to "Multi-Channel".

​@@jeffmeryso, I merely updated REW, and the options for other channels showed up. The warnings that the newer version of Rew not being able to open older files had my shy of updating.

Yep. I agree with everything you have above. In a perfect world, we actually want to be flat for a full octave above the crossover for subs and below the crossover for non-subs. We don't live in a perfect world though so we do the best we can with the equipment and room. My intent is to provide recommendations that work for the vast majority of folks though so flat to 120 Hz covers just about everything as you mentioned.

Correct. Delay directly into the processor.

Thanks for bringing this to my attention. I don't have control of what TH-cam runs, but can look to see if I have an option available on skipping ads.

Sorry for the delay and thanks for your patience. I was out of town most of the week. MSO runs only on Windows. It will run on anything with a Windows OS. If the laptop isn't powerful, the calculations will take longer to complete, but they will complete. I run a virtual machine on my MacBook for MSO when I'm calibrating client rooms.

@@jeffmeryHiya, thanks for the response. Do you still think it’s the best method to aim for a flat(ish) response and then let Dirac live do the room curve or to do the curve for the subs and pull the curtains in Dirac to not touch what you have done to the subs? One more thing (sorry) frustratingly the volume of both subs change (by the same amount). It’s frustrating because I increase their volumes and then after a while the volume returns to the earlier volume setting so I have to decrease the volume by the same amounts. Any idea what that could be? Cheers

@@jim586 It is absolutely best to optimize to flat in MSO and then add your house curve in Dirac. It's been a while but I believe if you did the curve in MSO and used curtains to restrict Dirac, it would limit Dirac's ability to integrate the non-sub speakers with the subs. This might be less of a concern if you're not using DLBC. I always recommend verifying results with REW measurements. Unfortunately I have no idea what the volume issue on the subs could be. Sorry!

Thanks for the answers, Jeff. Be good.

DLBC and MSO work to accomplish the same goals using slightly different methodologies. One or the other may provide better results in a given room. In my personal system, I got the best results using MSO and DLBC together. I optimized the subs in MSO, then presented a single sub to DLBC. The bass response was slightly improved over DLBC alone in my two outer listening positions, but I still got the advantage of good sub-to-speaker integration that DLBC also provides. Individual sub outputs is also a good consideration. I did not have that problem at the time, but it would be a blocker in my current setup. I change processors and no longer use DLBC in my own theater. I do calibrate quite a few systems with DLBC and sometimes DLBC with MSO if the situation warrants.

v2.0 was recently released. I haven't had time to do any update videos. There are significant changes based on feedback to the MSO creator and maintainer. He's done a ton of work in the last year or so. I don't really do side work on this. I have a day job and a side-gig doing full theater calibrations already! LOL! The best option is the MSO thread over on AVS Forum: www.avsforum.com/threads/optimizing-subwoofers-and-integration-with-mains-multi-sub-optimizer.2103074/. There are a couple of folks there that help if you post up the .msop file

MSO can certainly get more precise. However, that's not necessarily better. For a single position, I always use REW because it's much faster with equivalent results. For multiple positions it's highly room and equipment dependent. That being said, MSO works just as well for a single position as it does for multiple. In fact, it should work better for a single position than multiple since it doesn't have to balance the response at the MLP with the other seating positions. No wrong answer here.

Jeff thanks again I’m making great progress. Like you I’m using 2x4 HD. In your 9 minidsp export video you didn’t use all 10 PEQs available? I also thought I could allow 3 db boost vs 0 recommended by MSO Commends thanks to Carl

@@zeekbruno4869 It's not necessary to use every PEQ. I calibrated a theater manually this past week and only needed 2 PEQ filters to get a perfectly flat response. 3 dB is probably fine, but you need to know your equipment, its limits, and what you're trying to fix. No amount of boost will resolve a null, as an example. Keep in mind that a 3 dB increase requires 2x the power from the amplifier. That's why it's important to know your equipment and what you're trying to achieve.

Thanks Jeff

Jeff just a note to close off the effort I’m up and running. I was able to get beautiful sound from 14hz -130 hz Rythmic and Velodyne subs MSO was much better than I could do. On paper it reported ~0.09db rms spread. 7:42 I used all 10 PEQs and 3 I studied the performance metrics and notices that I had a 7db SPL weighted penalty ( using 75 db baseline setting ) so I think I should be able to split the different and put 3.5 db higher baseline ( 78.5) ? I’d i understand it Regards Db max boost

Yes it will do that, but probably not really reintroduce any issues. Dirac shouldn't make big corrections to something that's already been pre-corrected. Dirac doesn't EQ things that closely to force a super tight response. It might do that for a single point measurement, but for multi point calibrations with Dirac, the response is an average. There will be some lumps and bumps. It really should be a non-issue.

A little easier way to set the level is to just take a sweep that's loud enough to measure. Switch to the "Distortion" tab. At the bottom, you want to select the "Fundamental" and "Noise Floor" traces only. Look at the difference between these. In the video I said 40 dB, but anything 30 dB and up should be fine. Just look at the separation between the two lines and that will tell you if the sweep is loud enough.

There are two options here. The easy option is to increase the gain on ALL subs by the same amount. Just use the gain controls on the subs to do this. Technically, yes, differences can impact the calibration result, but if they're minor enough, they won't really be problematic for the result. Additionally, Audyssey will do some additional smoothing when it's run to further smooth the overall response. The other option is to try MSO v2.0 which was recently released. It has a new optimization method to maximize headroom and preserve as much output as possible. The magnitude response won't be quite as smooth. Again, Audyssey will further smooth the results. Preserving output helps quite a bit.

@@jeffmery Thanks Jeff! Appreciate the reply. I'll try this out.

Nope. No need to move the delay over.

There are many reasons that the result won't match the prediction. The most common are not using an acoustic timing reference, not resetting existing EQ and delays before measurement, and having an incorrect sample rate for the DSP device you're using. There's a full list of things to be aware of here: www.andyc.diy-audio-engineering.org/mso/html/reference-manual/faq.html Good luck if you decide to dig in again. Go slowly and have some patience while you work through things.

I'd love to do one but need to find the time. I have a list of videos I'd like to make. Life has just been in the way for quite a while now.

@@jeffmery Take your time..... waiting for your video

Thank you, and I appreciate the kind words! Life has just gotten in the way of content creation for me. I have more that I want to do, updates for new versions of software and show some other tips and tricks. I've just been buried on a few fronts. I haven't forgotten, I just need things to calm down to free up time.

Neither. I'm on Mac OS. My understanding is that Java has become the way to go on Windows. I have a Windows laptop around here somewhere. I really need to play with REW a bit more there. Sorry I'm not of more use!

You can absolutely just keep using it that way. The main reason to put it on the output PEQ filters is for BEQ users. BEQ implements all of its filters on the input side. If there's anything you need, BEQ will overwrite them. Not everyone uses BEQ though (I don't anymore myself).

I assume you mean the crossover between the subs and non-sub speakers? It's actually the exact same process and is commonly referred to as the "sub distance tweak". You just need to get the correct set of measurements. For home theater, subs are usually aligned with the C channel. 1. Set your crossover for C 2. Measure C from 20Hz-20kHz 3. Mute the subs, measure C from 20Hz-20kHz 4. Unmute the subs, temporarily disconnect the C speaker, measure C from 20Hz-20kHz You now have a "before" from step 2, a C-only measurement from step 3, and a sub-only measurement from step 4. In the alignment tool, you want to use the step 3 measurement in the top part of the alignment tool and the step 4 measurement in the bottom part. This will help to time align the subs to the C channel at the selected crossover frequency. Let me know if I missed the mark and you were actually referring to something else on the better crossover frequency.

Hi Jerry thanks for quick response I’ve built a fully active 3-way with sub. I’ve questioned whether I have selected the correct xover frequencies, ( sub to woofer, woofer to Mid, mid to tweeter). I’m thinking what I’ve learning from your video if I can start with woofer and align with mid ? Use the technique to see if I slid the freq up/ dwn if there is a sweet spot vs mfg xover freq?

@@zeekbruno4869 Ahh yes, I completely misunderstood. Yes, I think the alignment tool can be used for that use-case. But it doesn't have the ability to leave the timing fixed and just slide the crossover frequency around to see what works. I'm sure you know this, but crossover design gets complicated. It's about the frequency as well as the slopes on both sides, then gain, delay, inversion, and even AP filters can play a role. I honestly haven't done any driver-to-driver crossover work outside of implementing manufacturer supplied specifications for their active speakers. There are probably tricks in REW to make things go faster, but I just don't know them.

Jeff Thanks I’ll try it and keep in touch on my experiences Stay well and keep forging new Paths Bruno

You need to use something with a tweeter as the timing reference. I’d temporarily connect the miniDSP to the R channel output. Use L for timing and R to sweep subs. Hopefully you have RCA that can be used and not just XLR

There are different schools of thought on sub alignment to main speakers. Part of the challenge in using impulse responses is that they are biased to the arrival time of the higher frequencies. REW typically sweeps from low to high so a sub can show an odd impulse response making it difficult to align. This can some times lead to incorrect alignment and poor sound. This is what OCA is discussing when he talks about "cross correlation sometimes gets it wrong". The upside of this method is that typically the arrival time of peak energy is pretty well aligned. If this were me, and I only cared about a single seating position (as OCA does), I'd start with the alignment tool. I'd look at the combined subs and the L+R speaker together and see how I can get the best summation. I tend to favor as little delay as possible to get things aligned. Experiment with inverting the subs and delay to find the lowest amount of sub delay to give a good integration at the crossover.

I haven't found it to matter in my experience.

My subs were identical when this video was made so I gain matched them by setting their control to the same level. That works best when the subs are the same because we know they’re working equally hard. If the subs are different I’d start by level matching them at the MLP. There’s a really good method for mixed subs called “distortion matching”. There is a post by user “DaveBoswell” on the MSO thread at AVSForum that details the process. It’s more complicated than straight level matching. However, it helps to ensure that one sub isn’t completely overdriven if they have very different capabilities.

@@jeffmeryYes, my subs are identical, HSU vft2-mk5. Thanks,still in the early stages of how this whole process works, received my mini dsp/umik-1 mic 3 weeks ago and trying to set these up.

Just one more question if I could,after you have your aligned sub file(2-subs only)do you then export that file to the mini dsp?

@@kenestra123 Gain and delay from the alignment need to be input manually to the correct output channels on the miniDSP. If you took the next step and EQ'd the combined response, then yes, the EQ export needs to be imported and loaded on the output side of the miniDSP.

Yes, in my opinion it's better to use C or R instead of LFE. Set the C or L/R channels to "Large" in your AVR to bypass the crossover. Then you will get a true full-range measurement of the subs with no LPF in place.

@@jeffmery thanks for taking the time to reply 👍🏼 My old aventage receiver seems to have some sort of fixed filter

@@jeffmery Would you then have to connect the subwoofer RCA cable to the center channel pre-out and physically disconnect the center from the AVR?

@@chrislukowski1825 Correct! A bit of extra work, but worth it IMO.

When you have the ability to implement your house curve external to MSO, that is the way to go 100% of the time. If we try to implement a house curve in MSO, then every time we want to change it, we have to completely reoptimize in MSO. This can lead to inconsistent results. Manipulating a house curve in Dirac, by comparison, is significantly easier to do. In fact, I haven't run across a single room correction solution that is harder that MSO to implement a target curve. Additionally, if you implement the house curve in MSO, your room correction is going to come over the top of it and mash it back to flat or to whatever default target curve that system has. If you know your subs are flat you know the response they should have and it makes it much easier to return to a known good config.

I don't have that iMac any more so it was either direct or with an HDMI -> USB-C/Thunderbolt adapter. Make sure the one you purchase is at least HDMI 2.0b. Even though the audio is very low bandwidth, I had some issues with older HDMI 1.4 adapters I had laying around. I think the one I used most was from Anker and was maybe $15 on Amazon at the time. I now use an M1 MacBook Pro that has a native HDMI port (no dongle/adapter needed).

Thank you for your reply and advice. 👍

The context of that is trying to set an optimization target that only utilizes cuts and does not require us to add gain with any PEQ filters. Adding gain isn't a "bad" or "taboo" thing, it's just something that needs to be done carefully. A 3 dB increase in SPL requires 2x the power - yes double the power. It's easy to get into trouble and cook something. I just try to present things that won't get folks into trouble with their gear.

I appreciate the kinds words! I'd like to do an update with all the changes in MSO. I didn't have enough subs for a long time (went from 4 x 15" to 2 x 18" and now am back to 4 x 18"), and now things are just crazy busy. I have my fingers crossed to do something next month, but life has a way of interfering.

Have a look at this video to see results of EQ in REW. It's based on a different set of measurements, but the result is the same. th-cam.com/video/ARztXSmoQbE/w-d-xo.html

Yep. That is absolutely an option. I like starting out on the input side just because it allows for quick iteration. One small change affects all subs so it’s easy to try different settings very quickly.

MSO was written by AndyC56 (his handle on multiple forums like AV Nirvana and AVS Forum). He deserves 100% of the credit. I just created a tutorial video to help people get started.

Sorry for the delay! Those limits only apply to the filter parameters during the optimization. They don’t impact the actual filter limits on the miniDSP. It essentially lets you set different ranges/bounds on the inputs vs the outputs. In some cases that can be very useful. It’s not required that the be different and I believe one option is to use the limits for both the input and output filters. If you want to use an input filter for the house curve, there are a few ways to do that. In no particular order … * Don’t use input filters in the optimization. It’s possible to use 10 output filters and up to 8 crossover filters on each output for a total of 18 filters. You would need to enable this in the MSO options (don’t have it handy to check the specific menu options for you) and set the output filters to 8. When you export you get a regular and _xo file for import on the output PEQ and crossover for a given channel. * Use input filters, but leverage the “BEQ Export” option in MSO. This adds the input filters to any output filters leaving the input filters free. You are still subject to the limit of 18 filters as noted above and will potentially need to implement the same option in MSO for this to work. You may end up with one or two biquad files to implement on the miniDSP depending on how many filters you have and if you are using the XO biquads. I hope this helps!

@@jeffmery Thank you for the reply. I think I got some of it figured out. In the application options tab, when I enabled biquad file export, I set output biquad limit to 10 and input biquad limit to 8, and left crossover biquad limit at 8. When I ran the wizard, I selected 10 output filters and 8 input filters to be used. This generated 10 output filters for each sub and 8 shared filters. When I saved the optimization, the option I used was "Save miniDSP BEQ Biquad Text for this Configuration". It saved a crossover biquad file and an output biquad file for each sub. Due to the crossover biquads being generated from the shared filters, the two crossover biquad files are identical. It doesn't look like there is a handy upload file button in the crossover section, like there is in the PEQ section. But, I was able to copy the crossover biquad file into the crossover section of each sub output. Due to the crossover biquad files being identical I was able to copy it once, and paste it into each crossover section. I had the mini dsp input section free, so I set a low shelf filter with a 6db rise, I set the frequency to 47hz , and used a Q factor of 0.6. I haven't measured it with REW just yet, but the graph on the mini dsp looks like I will have a 6db house curve starting at around 100hz. I will turn the shelf filter back off to run Audyssey and then turn it back on again once the rest of my system is calibrated. It looks to me that it would be a good idea to use the MultEQ Editor App to turn off the Filter Frequency Range for the Subwoofer, so it doesn't mess with the subs optimization. I hope this isn't too confusing and hopefully I performed the process properly.

@@shaggs2richesYes that’s exactly correct. I have no idea why miniDSP didn’t put an “Import” button on the crossover biquads as well. I’m always paranoid that I’m going to miss a line doing the copy/paste 😂

@@jeffmery Thanks for your help. I am having a fun time messing around with different variables with the optimization. So far it seems that I get the best results optimizing from 12hz to 160hz. My subs (svs sb3000) are rated to 18hz, but I measure an output increase from 25hz to around 14hz. If I optimize down to only 15-20hz, the bump stays but ends up measuring as a 6b spike and 3hz wide. Optimized from 12 to 160hz, the line is very smooth-looking. Performance Metrics gives me a 0.51db seat-to-seat variation, 0.26db MLP target error, and a final error of 0.46db. REW sweeps measure the seats to look extremely similar to the MSO graph. I am quite impressed with how effective and relatively easy the program is to use once you get the hang of things. I'm sure there are a ton of things a more knowledgeable user can do with REW and MSO but at a beginner level, this seems to be really effective.

@@shaggs2riches Awesome! Welcome to the rabbit hole ;-). Those perf metric numbers are VERY good by the way.

I'm sorry for the late response! I completely missed the notification when you posted. It is possible to add a house curve with MSO. However, I don't advise that. The main reason is most people are using this for home theater and are going to run room correction after optimizing the subs. That room correction will almost certainly squash any house curve done in MSO. For that reason, I recommend optimizing flat in MSO, then running room correction. Use either your room correction software or the input filters of the miniDSP to implement a low-end house curve.

I do have plenty of output for that (tried to keep the video simple). However, it sounds like crap when crossed over that high. Things become too localizable and it really hurts the immersion. I don't remember exactly when I made this video. However, my solution is using intermediate bass management in Trinnov. I cross the Atmos speakers at 120 Hz over to the nearest listener level speaker, and then that gets crossed at 80 Hz to the subs. It works really well as those localizable frequencies are kept much closer (spatially) to the Atmos speaker in question so the immersion isn't broken. Only Trinnov and Storm have this capability natively to my knowledge. I use it quite often for clients and it works well. The real solution is to get Atmos speakers that don't suck. LOL! Also - you are correct regarding bass management filters for most processors. The LPF is 4th order on the subs. The HPF is 2nd order on the speaker. However, THX are assuming that the speaker is sealed and has a natural rolloff of 12 dB / octave (2nd order). When another 2nd order is applied to that natural roll off, the result is a 4th order HPF which matches the 4th order LPF on the sub side of the crossover. These are typically LR filters so the sum should be flat in magnitude and phase at Fc. It's worth validating the speaker in question's response to ensure reality matches THX's assumptions.

@@jeffmeryThanks, wow, Trinnov/Storm really do allow for some crazy bass management schemes - rolling off "to the nearest speaker" for speakers with high XOs sounds like an amazing thing. Any-who, just wanted to point out the 12/24 rolloffs usually done by the AVRs us mere mortals use :) THX-heritage, but seems to be the norm now, even for non-THX AVRs. I really wish selecting slope was available to us; many of us use LCR's getting down to the 40-60Hz-range, but we choose to still use a crossover at 80Hz, so many of us would likely be better off with 24/24 slopes (or 18/24 - assuming we all have sealed speakers rolling off at exactly 80Hz - in-room! - is a gutsy assumption).

@@DanielNilssonSeI agree that a selectable slope would be ideal.

Correct on your first statement. MSO allows you to set the min/max allowable values for any filter. This is helpful to get things dialed in or to stay within the capabilities of your equipment. It's been a while since I did the video so I don't recall why I chose 100 Hz for the video. It's best to understand the anechoic capabilities of the speakers in question, then compare that to the in-room response to determine a (potentially) good crossover frequency. It's a balance of not overdriving non-sub speakers, getting the most impact out of the bass frequencies, all while avoiding localization of sounds. I actually almost always start at 80 Hz and see where that gets me. It's pretty rare for me to deviate much from that by the time I get to the end results, but it does happen with exceptionally good (or bad) speakers. Once it's chosen, I simply double that value and use it for the upper bounds of the optimization range as well as the upper bounds for the center frequency (Fc) of any filters in use.

Correct. MSO's results are relative to the measurement conditions. So if MSO says you need 3 ms of delay and the original was 7.2 ms. then the new delay would be 10.2 ms. Does that make sense? I generally don't recommend doing subs+mains integrations in MSO when using some other automated room correction. It's not that it can't be done. It's that it's a more advanced topic and gets confusing quickly. Additionally, most processors don't have all the capabilities needed to properly implement things. I'm not trying to discourage anyone from doing that at all though. Learning is fun 🙂

I haven't worked with Tonewinner before. Two things I have seen that may help: ensure the processor is in a surround mode format or an all channel format and not forcing things to stereo, and try a different HDMI cable.

@@jeffmery Noted with thanks. I'll try a different HDMI and different adapter later. Tonewinner was setup in surround mode format so i dont think that the issue.

@@quantumacousticav If this is on a MacBook, I did have trouble with a few older HDMI adapters I had laying around. I picked up one from Anker on Amazon for about $15 that seemed to work well. Might be worth a shot. I upgraded to an M1 MacBook Pro a year or two ago that has HDMI built in so no longer need or use the adapters.

@@jeffmery Noted on the Anker adapter. Will go and buy one to test out shortly! Will update after testing! Thank you

Thanks for the feedback! With Dirac it depends on the specific AVR and what Dirac features it supports. Not all AVRs/Processors support all levels of Dirac so that's something you want to find out. There's Dirac Full-Range, DLBC (with single and multi-sub options), and ART. A miniDSP can help Dirac specifically with the "Full-Range" version. This is the entry level Dirac license that doesn't do anything special with subs or crossovers. In this case, the miniDSP would be used to pre-align and EQ the subs as one before running Dirac. Things get a bit more interesting with DLBC (Dirac Live Bass Control). DLBC does two things: automatically aligns and integrates multiple subs together, aligns the non-sub speakers with the subs at the crossover. DLBC can be purchased in a single-sub or multi-sub license. With multi-sub DLBC, there's really no reason to have a miniDSP in the signal chain. It does a great job on its own. With the single-sub license, a miniDSP can, and probably should, be used to present a single big sub to Dirac. In this case you're really getting just the benefit of integration at the crossovers with non-sub speakers (still very good). Depending on your level of comfort and how much you like to tinker, DLBC is the closest thing to a multi-sub "easy button" around right now. It's not perfect and takes some tinkering, but it's less complicated and less costly than doing that plus a miniDSP. ART is the top of the line room correction from Dirac. It's only available on Storm processors right now but others have announced that they'll add it in the future. With ART, you absolutely do not want a miniDSP between the subs and AVR. ART needs direct control of all subs and speakers to work properly. I hope this helps, but if not, keep asking questions and I'll help out 🙂

The video above is for 1.1.6, not 1.1.16 🙂. I'm not sure on the most recent version, but it should be 1.1.15 or newer. This is the link to download from the owner/maintainer's website? www.diy-audio-engineering.org/web_dl/apps/mso.zip

@@jeffmery Ohh. My mistake. Point 6 is a little bit lower than point 15. 😄 Thanks!

@@n3x3ff3ctHaha! No worries! Easy mistake to make (I've done it myself before).