I created these while investigating what type of environment I might be putting the 18H+ into.
I wasn't super stritch on the comparison, I just wanted a ball park idea, and to see if further investigation was needed.
The graphs are; My MLTL, a ~300 liter BR, a ~338L Sealed box, and a 380-400ish liter TH....Getting real specific, I know. I had to trick HR into displaying a figure for sealed, my closed box is just my MLTL box with the smaller CSA length changed to the smallest figure HR will allow. From there these are the Horn throat simulations at 1 watt. I think all of these are probably fine, I was just really interested, and like I thought, the "sealed" box has the highest pressures generated.
Attachments
Last edited:
Fluid, you were advising me about Linear crossover....I would like to entertain that option, I've found that I can very desirable results crossing the 18 to 15 with minimum phase...but the crossover to the Axi is going to require a high order crossover, and linear phase is the way to do that so...
Does anyone have any products in mind? I could still use the amps DSP to do voicing, but a module before the amps doing the FIR filtering sounds like how this go....
Does anyone have any products in mind? I could still use the amps DSP to do voicing, but a module before the amps doing the FIR filtering sounds like how this go....
There are a few options depending on the source you will be using. If the audio is coming from a computer then convolution is a simple way to apply the slopes if separate outputs are available. If not then rephase can generate phase only corrections that can be applied to the stereo signal before any minimum phase crossover in your amp.
Another computer can be used as a realtime convolution or DSP engine to input audio process it and output the filtered responses to the individual drivers. Using a VST host or linux command line software works for that.
Then there is hardware like MiniDSP or professional Lake type processors with FIR capabilities. These will be more tap limited than the software solutions. There are some cheap chinese DSP boards based on Analog Devices chips with FIR capabilities that could use Sigma Studio to generate the filters. Not the easiest program to learn though.
Another computer can be used as a realtime convolution or DSP engine to input audio process it and output the filtered responses to the individual drivers. Using a VST host or linux command line software works for that.
Then there is hardware like MiniDSP or professional Lake type processors with FIR capabilities. These will be more tap limited than the software solutions. There are some cheap chinese DSP boards based on Analog Devices chips with FIR capabilities that could use Sigma Studio to generate the filters. Not the easiest program to learn though.
- thats leaving out a lot of details, to say the least. The less taps the less latency, I think is what you showed me earlier in the thread. Lake lm26 and Dolby Lake hardware comes up on my radar, but I don't know to judge them.
A computer having multiple outputs is a special thing, what sound card is that? Where I can apply individual filters to each channel....all channels are balanced signal to the amp...ebay?
Looks like MiniDsp and Lake are the only options.
Last edited:
Yes it is
If you are serious about implementing it then I will happily explain it in as much detail as you want. What I have found through answering similar questions in other threads is that people have very specific requirements relating to what sources they want to use, what their pre conceived notions are regarding computers in audio etc. A lot of time is wasted explaining things in great detail only to have them say they can't do that because of insert random problem here
In a basic sense yes, if the number of taps is lower the latency will be lower but so will the frequency resolution of the filter which starts to become an issue at the 200Hz frequency you are looking at.
Lake stuff is good but expensive. Mark100 uses a different system that I can't remember the name of but is also not cheap.
Using a computer with multiple outputs is an easy enough way to make an active crossover, the software to use depends on your input sources to some degree as to which will work better. For your situation I would be seriously looking at a professional soundcard like maybe a MOTU Ultralite AVB, something from RME, Focusrite or Steinberg depending on how much you have to spend and any other features they have that might be useful to you in other ways.
Jriver is quite flexible as a media player and DSP engine, multichannel convolution, VST effects delay EQ, channel matrixing. It also has an ASIO line in which you could use with external sources plugged into the soundcard.
One reason I like a computer based playback sysytem if it works is that there is no need for sample rate converter chips, which you will get in all the hardware based systems to allow them to function at their set rate. Jriver can change rates on the fly if you have it set correctly.
You need a software layer to apply the filtering or other DSP functions.
That is where Jriver media centre (or similar) software comes in, that has a DSP engine where you can run convolution for FIR filters or VST plugins that can generate FIR filters and apply it to stored music on playback or incoming audio via the ASIO line input and send the separately filtered parts to individual soundcard outputs.
It is also possible to use any other media playback software or DAW and link the output of that to either Jriver or a VST Host software via ASIO loopback software.
That is where Jriver media centre (or similar) software comes in, that has a DSP engine where you can run convolution for FIR filters or VST plugins that can generate FIR filters and apply it to stored music on playback or incoming audio via the ASIO line input and send the separately filtered parts to individual soundcard outputs.
It is also possible to use any other media playback software or DAW and link the output of that to either Jriver or a VST Host software via ASIO loopback software.
Like Fluid says, to use FIR filters, you need either software or hardware implementation.
On stereo 4-ways:
I've used JRiver with the VST plugins, and a 8-channel DAC.
Works great for FIR convolution, but still needs further software brought into the equation for normal speaker management tasks...IIR xovers, EQ's, delays, etc.
This may be easy to do, but I simply don't trust my high powered rigs with PC (Win) processing very much, and wouldn't continue on with attemting to fully implement PC speaker management. If the VST plugins worked with JRiver running linux, I might have continued on...
Also used miniDSP. An openDRC to start, then 4 openDRC DIs. This was a nice setup, all the taps reasonably needed, and enough IIR capability to negate the need for any further processing.
Only pains were that a PC can only connect to one miniDSP box at a time (for loading FIR files or making adjustments; and finding a way to insert their digital I/O between source and multi-channel DAC.
Now, I'm using a QSC Core-110f, a standalone linux box,running QSYS Designer compiled schematics.
I ended up with this by accident really. I bought some used qsc network amps off ebay at really great prices, because I read you could use them in standalone mode like regular amps.
Well, you can, but only after they are loaded with a design you built on a QSYS Core....dummy me..
Anyway, turned out to be one of the best goofs I've ever made.
Whenever I had looked at qsys designer previously, I was completely turned off by the apparent complexity. But with the amps I bought, I was forced to buy a used Core (about $1250 ebay) and give it a go. Took the online training course for qsys.
It's easy !! and you can do anything with it !! Highly recommended!!! Selling all my miniDSP.
Symetrix is another open architecture processor, but doesn't seem to have the following QSC has.
I do believe these open architecture processors are the future, here today.
As far as Lake goes, their price and waay overly complicated software have kept me away. If you do want to go that route, be aware they have limited custom FIR capability.
And credible rumor has it a new round of Lakes is about to be released. That could make existing units drop alot and maybe turn them into a good deal. Guy's like Clair Bros may dump a ton of them..
Bottom line with Lake imo, is be cautious....
Another way to implement FIR is with leading edge pro-amps like Crown I-Tech HD's or certain Powersoft or Linea Research models. This would be a pretty high $$ route to take though...
On stereo 4-ways:
I've used JRiver with the VST plugins, and a 8-channel DAC.
Works great for FIR convolution, but still needs further software brought into the equation for normal speaker management tasks...IIR xovers, EQ's, delays, etc.
This may be easy to do, but I simply don't trust my high powered rigs with PC (Win) processing very much, and wouldn't continue on with attemting to fully implement PC speaker management. If the VST plugins worked with JRiver running linux, I might have continued on...
Also used miniDSP. An openDRC to start, then 4 openDRC DIs. This was a nice setup, all the taps reasonably needed, and enough IIR capability to negate the need for any further processing.
Only pains were that a PC can only connect to one miniDSP box at a time (for loading FIR files or making adjustments; and finding a way to insert their digital I/O between source and multi-channel DAC.
Now, I'm using a QSC Core-110f, a standalone linux box,running QSYS Designer compiled schematics.
I ended up with this by accident really. I bought some used qsc network amps off ebay at really great prices, because I read you could use them in standalone mode like regular amps.
Well, you can, but only after they are loaded with a design you built on a QSYS Core....dummy me..
Anyway, turned out to be one of the best goofs I've ever made.
Whenever I had looked at qsys designer previously, I was completely turned off by the apparent complexity. But with the amps I bought, I was forced to buy a used Core (about $1250 ebay) and give it a go. Took the online training course for qsys.
It's easy !! and you can do anything with it !! Highly recommended!!! Selling all my miniDSP.
Symetrix is another open architecture processor, but doesn't seem to have the following QSC has.
I do believe these open architecture processors are the future, here today.
As far as Lake goes, their price and waay overly complicated software have kept me away. If you do want to go that route, be aware they have limited custom FIR capability.
And credible rumor has it a new round of Lakes is about to be released. That could make existing units drop alot and maybe turn them into a good deal. Guy's like Clair Bros may dump a ton of them..
Bottom line with Lake imo, is be cautious....
Another way to implement FIR is with leading edge pro-amps like Crown I-Tech HD's or certain Powersoft or Linea Research models. This would be a pretty high $$ route to take though...
No need for VST's for all of those features, Jriver has built in parametric EQ, delay and some IIR crossover ability, convolution does not have to have any more than a single sample delay associated with it either. All of that is available on the linux version as far as I know, I use a ton of VST plugins with my set up as I like the graphical tweaking rather than a list of figures Jriver uses and I have had no stability issues on Windows.
Totally understand why in a live sound or extremely high spl environment a hardware system gives greater peace of mind.
Camplo if you have linux skills then all of this can be done in linux if you don't the learning curve is somewhat vertical. Basic Jriver on linux is much easier even for someone who has never done it before.
I put Bassbox Pro back on my computer today, and out of curiosity ran some quick simulations of the AE drivers as they were in the database.
I used an EBS vented alignment and sealed in the 18 for comparison. The vent needed for the 18 is not at all practical, the 15 is getting pretty long too.
The 18H+ does not seem to be designed as a subwoofer, the TD15H goes lower in frequency but gives up output to the bigger driver. Depending on how much you are targeting very low frequency response, sealed could be a good choice.
I used an EBS vented alignment and sealed in the 18 for comparison. The vent needed for the 18 is not at all practical, the 15 is getting pretty long too.
The 18H+ does not seem to be designed as a subwoofer, the TD15H goes lower in frequency but gives up output to the bigger driver. Depending on how much you are targeting very low frequency response, sealed could be a good choice.
Attachments
I guess JRiver's DSP Studio has evolved since I tried it...probably back at MC22.
At that time, the type and order of available IIR xovers was very limited, and timing increments too coarse for HF/VHF phase alignments.
For FIR, I used the Config File component in the free ConvolverVST, which was Win only. Had to convert my JRiver NUC from linux to Win.
No doubt there are plenty more options today.
Is most of your JRiver use with full range, or full range with a sub?
I could see how JRiver would have worked fine for that, even when I tried. I'm thinking this might be a case similar to DRC...that works better for full rangers, than for developing multi-ways? Just spitball wondering
JRiver has a built in Convolver too in case that was not mentioned: Convolution - JRiverWiki
So does Roon: Roon Knowledge Base - DSP Engine: Convolution and a number of other software media players.
For commercial DSP software, I can recommend Acourate and Audiolense that have digital XO capability, with virtually every type of crossover and slope adjustment you would want. Linear phase, min phase XO's plus minimum, linear, mixed phase targets. In addition to driver time alignment, room correction and excess phase correction. It's the full suite, hence the commercial software price.
PS. Those Purifi drivers are amazing!
Good luck camplo!
So does Roon: Roon Knowledge Base - DSP Engine: Convolution and a number of other software media players.
For commercial DSP software, I can recommend Acourate and Audiolense that have digital XO capability, with virtually every type of crossover and slope adjustment you would want. Linear phase, min phase XO's plus minimum, linear, mixed phase targets. In addition to driver time alignment, room correction and excess phase correction. It's the full suite, hence the commercial software price.
PS. Those Purifi drivers are amazing!
Good luck camplo!
Good stuff Mitchba, I see JRiver has evolved indeed.
I have another old JR ID running linux I'm going to recommission.
I can add FirDesigner to the list of recommended commercial FIR generators. Super capable and easy to use. Takes less than 5 minutes once measurement is in hand, to crank out a FIR file.
But you know, the more I've learned about FIR, min phase, lin phase, and alignment in general .....well, other than the time consuming pain of manual EQs, ....good ole, entirely manual, rePhase, works
Windows based JRiver has had this for years. I've been using it since version 17.
Plus FIR processing allows you to do all the crossover you'd ever want quite easily. Even without using vst plugins, the two PEQ banks that have been there for ages etc. It is very versatile and has been for years, really. At least the Windows release.
Same goes for DRC. But it's up to the user to do it, not automated but certainly not impossible.
Plus FIR processing allows you to do all the crossover you'd ever want quite easily. Even without using vst plugins, the two PEQ banks that have been there for ages etc. It is very versatile and has been for years, really. At least the Windows release.
Same goes for DRC. But it's up to the user to do it, not automated but certainly not impossible.
Maybe I'm missing something. The two PEQ banks seem very limited (MC24).
Are there some settings, maybe number of channels or something else, that it takes to open up multi-way crossovers, butterworth, LR, bessel, odd and even orders etc. All pass, etc. Without embedding it all into FIR?
Hope so, that would rock!
Not really the thread for it but the crossovers in JRiver are even-order butterworth filters (6 db, 12db, 24db, 36db, 48db).
From: Parametric Equalizer - JRiverWiki It should also explain the many other options available in the PEQ banks.
The PEQ banks have never shown me any limits as to how much PEQ's fit in there (i've tried at least over 80+ PEQ's per bank without any trouble), for all channels I use (depending on the output device used, not on the program itself) so I can't call them limited. And there are two of those banks to fill. Add in another VST EQ plugin if that isn't not enough for you.
Together with the fixed crossovers one could use the PEQ tools and shelve functions and you can make the driver follow just about any curve you could ever desire. Not enough? Add a VST plugin that does that job better. Or simply use one of the FIR tools to achieve those goals, as that makes it super easy. The convolver is on board, just stick the filters in there. Multi channel and auto swapping between sample rates if you want.
I see it as a really open program/platform that can house anything I can wish for, there's always a solution to be found, be it in a VST or the added stuff that keeps on coming in every following build.
I'd really wish for a connection with REW and RePhase for JRiver, being able to load and save setting between the three. That would be a sweet deal for me, personally. However RePhase and REW can work together right now, giving me every bit of freedom I'd ever need to do pré EQ, crossover, post EQ by simply stacking it in RePhase's own available banks and making a FIR filter out of that. Directly usable in JRiver's convolver.
There are lots of ways to get anything you'd want. But you don't need JRiver to get there, there are lots of other solutions as well. It's just one heck of a useful tool to house everything and anything that I will ever need. Especially due to the vst options, where I can add a plugin wrapper/chainer that can load it's own vst's and even has tools of its own for even more fun to play with. So I can get as fine delay for each channel if JRiver's own setting are too coarse for me. simply by routing and adding upon a stable foundation that JRiver offers me. The above has been available since well, forever it seems. At least since 2012 when I stumbled over it and decided to give it a try.
It feels limitless to me... and I've tried many ways and many ideas using this platform. Get a little creative with what you've got and it will prove itself in being very versatile. That doesn't stop with JRiver. There's always a way.
With all the things I do in my chain I couldn't think of a single piece of hardware able to do the same. Sure I needed a couple of VST's. But that's part of the beauty of it. Most other fixed solutions would feel limited for me. In here I can choose what I want and how to get there. The limit being my fantasy, really.
Cascade them to get Linkwitz Riley crossovers.
From: Parametric Equalizer - JRiverWiki It should also explain the many other options available in the PEQ banks.
The PEQ banks have never shown me any limits as to how much PEQ's fit in there (i've tried at least over 80+ PEQ's per bank without any trouble), for all channels I use (depending on the output device used, not on the program itself) so I can't call them limited. And there are two of those banks to fill. Add in another VST EQ plugin if that isn't not enough for you.
Together with the fixed crossovers one could use the PEQ tools and shelve functions and you can make the driver follow just about any curve you could ever desire. Not enough? Add a VST plugin that does that job better. Or simply use one of the FIR tools to achieve those goals, as that makes it super easy. The convolver is on board, just stick the filters in there. Multi channel and auto swapping between sample rates if you want.
I see it as a really open program/platform that can house anything I can wish for, there's always a solution to be found, be it in a VST or the added stuff that keeps on coming in every following build.
I'd really wish for a connection with REW and RePhase for JRiver, being able to load and save setting between the three. That would be a sweet deal for me, personally. However RePhase and REW can work together right now, giving me every bit of freedom I'd ever need to do pré EQ, crossover, post EQ by simply stacking it in RePhase's own available banks and making a FIR filter out of that. Directly usable in JRiver's convolver.
There are lots of ways to get anything you'd want. But you don't need JRiver to get there, there are lots of other solutions as well. It's just one heck of a useful tool to house everything and anything that I will ever need. Especially due to the vst options, where I can add a plugin wrapper/chainer that can load it's own vst's and even has tools of its own for even more fun to play with. So I can get as fine delay for each channel if JRiver's own setting are too coarse for me. simply by routing and adding upon a stable foundation that JRiver offers me. The above has been available since well, forever it seems. At least since 2012 when I stumbled over it and decided to give it a try.
It feels limitless to me... and I've tried many ways and many ideas using this platform. Get a little creative with what you've got and it will prove itself in being very versatile. That doesn't stop with JRiver.
There's always a way.With all the things I do in my chain I couldn't think of a single piece of hardware able to do the same. Sure I needed a couple of VST's. But that's part of the beauty of it. Most other fixed solutions would feel limited for me. In here I can choose what I want and how to get there. The limit being my fantasy, really.
Last edited:
I guess JRiver's DSP Studio has evolved since I tried it...probably back at MC22.
At that time, the type and order of available IIR xovers was very limited, and timing increments too coarse for HF/VHF phase alignments.
For FIR, I used the Config File component in the free ConvolverVST, which was Win only. Had to convert my JRiver NUC from linux to Win.
No doubt there are plenty more options today.
Is most of your JRiver use with full range, or full range with a sub?
I could see how JRiver would have worked fine for that, even when I tried. I'm thinking this might be a case similar to DRC...that works better for full rangers, than for developing multi-ways? Just spitball wondering
I have recently upgraded to version 26 from 22, but I was not really using those features myself so I can't say if they are better. I can set 0.01ms as the shortest delay in version 26 today, any less and it goes to 0. Crossovers are fixed shape not sure if that is LR or Butterworth, 6dB to 48dB which can always be stacked at the same frequency for higher slopes. If you want anything specific convolution is way to get it, with proper config files the convolver can be run separately across the channels.
For now I run full range arrays no sub, but there are two Dayton HF Reference 15" sub drivers sitting in boxes waiting patiently for me to enclose them in cabinets. Wesayso has shown that adding subs is no issue with Jriver, I was in contact with someone who recreated the LX521 crossover in Jriver so I don't think it is a limiting factor.
DRC is designed for overall stereo correction but you can make it do much more than that if you know how to ask
Edit crossed over with wesasyo again, Butterworth filters, which are the building blocks for LR, no Bessel, but odd order Butterworth is possible. I've never been able to get an exact answer with Jriver Linux but I imagine it might be able to use LADSPA plugins as they are native to Linux.
Last edited:
I already there with you =)
The fs of the 18h+ is 24hz.
I was just about to make a new thread on designing a Aperiodic vent in HornResp
Does anyone know how to design an Aperiodic Vent?
Last edited:
After much debate, I have decided to invest into the development of an "Active Radiator Sub", using the 18H+ plus an active radiator, which allows me to retain the attributes of the sealed, while gaining the added output of the ported. Unlike passive radiating, "Active Radiator Technology" allows unmatched "port" performance, achieving the next level of SQ, taking off were passive radiator, falls short!
Last edited: