Do some speakers play better with DSP? (Recommendations?)

[Pariah Zero]

I'm looking at updating my listening setup with better speakers (budget: ~$500 USD). My question is if there are speaker styles (like KEF's coaxial) that take to DSP better, vs being able to bi-amp and use the DSP as an active crossover (like the Klipsch RP-600m II's).

At present, my system looks like:

Listening room: 6 x 4 meter (20x12 foot), speakers set along the long axis
Amp: 2x Fosi v3
miniDSP Flex (2-in, 4-out)
Speakers: Dayton T-652-AIR
Sub: 10" ported, Dayton. Nothing special, but I understand we're relatively deaf to distortion at sub frequencies.

So: my thinking is to update the speakers. I can bi-amp and use the miniDSP as an active crossover.

I'm an electrical engineer, so I'm at least passingly familiar with how to read & apply measurements, but I really don't know much about how speaker styles (things like coaxials vs TM, MTM, 2.5-way, 3-way, etc.) effect how well I can use EQ and/or DSP.

Ideally, I'd want to get a speaker that will give me the most flexibility to EQ or otherwise improve its in-room response.

I gave the KEF and Klipsch as examples as I know KEF's are coaxial (but not bi-ampable), and I'd largely be limited to EQ. Conversely, there's the Klipsch, which is bi-ampable and I'd be able to do at least some active crossover...

And to throw in a third option, I enjoy DIY, and the Mechano23 looks pretty promising, and it can be a pure bi-amped active crossover speaker.

My question is which path seems most likely to let me make the most of what the DSP & offers me?

(Edit: spelling fix)

 

[DVDdoug]

Crossover shouldn't be a problem but when you start boosting bass you can push the speaker (or amplifier) into distortion. It's mostly a problem in the bass where you often need several dB of boost/correction. No matter how much you boost, you aren't going to get bass you can feel from a 5-inch woofer.

Ported speakers drop-off more quickly than sealed speakers and if you boost at lower frequencies you can end-up with the woofer flopping around without generating much sound (except maybe adding to the distortion). A lot of "small" sealed active subs have built-in EQ and tons of amplifier power to overcome the low-frequency drop-off.

This isn't what you asked but this post shows the generalized difference between sealed and ported speakers. The ported design goes lower but then drops off more steeply. St some point the curves cross and the sealed speaker is putting-out more.

Pro subs are often large and ported. Usually they are tuned to down to around 40Hz and they are very efficient and go very loud to fill a large venue with bass you can feel in your body.

I enjoy DIY, and the Mechano23 looks pretty promising, and it can be a pure bi-amped active crossover speaker.

IMO - The main advantage of an active speaker is ability to build-in corrective EQ (analog or DSP). And once you've got a power supply and electronics inside it's a pretty-good economic trade-off to make it bi-amplified with an active crossover.

 

[Keith_W]

If you enjoy tinkering with DSP, the best speaker is one where each driver has its own DAC and amplifier channel. You can create such a speaker by DIY or by modifying a commercial design. Bypass the passive crossover, and solder each driver cable to a binding post.

Bear in mind that when you DSP, there are two steps. The first step is to DSP the loudspeaker, and the second step is to integrate the loudspeaker with subwoofer(s) and the room.

When you DSP the loudspeaker, you will need to obtain an anechoic measurement of the loudspeaker or individual drivers (i.e. free of room reflections). This is exceedingly difficult with some loudspeakers, ESPECIALLY very large loudspeakers, panel speakers, ported speakers, or any speaker with woofers or drivers close to the floor. For obvious reasons, it is easier to obtain a quasi-anechoic measurement of individual drivers than an entire loudspeaker. If you can not obtain a clean measurement, you can not DSP the loudspeaker, period. Otherwise you will be applying meaningless corrections for problems that are not there.

There are definitely advantages to replacing a passive XO with a DSP XO. How much of an advantage depends on how compromised the passive XO was and how much correction your speaker needs. The ability for designers to correct speaker problems via passive XO's is limited. More components = higher cost and more insertion loss. It is possible to transform a poorly designed XO into a perfect one with DSP.

Room correction is a different beast. This time, you want to capture the loudspeaker with the room response. And you can only do room correction up to a certain frequency (usually between Schroder to 4x Schroder). Above this, as Toole says, you can only apply "broad, low Q, tone-control" DSP. This is because every tiny wiggle you see at high freqs may not be real, there is a strong possibility that it is an artefact of microphone positioning.

So the question comes back to you - are you prepared to DIY or modify a commercial design? If not, then the role of DSP is a bit more limited. You will be stuck trying to DSP a whole loudspeaker (which isn't easy) or limiting yourself to room correction only.

 

[kemmler3D]

Pretty much what @Keith_W and @DVDdoug said.

I guess my question for you is what you hope to accomplish with DSP in particular. Is it to maximize the improvement to the speaker by using DSP? IMO it's mostly good for 1) more complex crossovers than is practical with passive, 2) phase correction, 3) room correction.

I'd want to get a speaker that will give me the most flexibility to EQ or otherwise improve its in-room response.

DIY definitely affords some good opportunities to leverage DSP, but DSP is just a gateway to the entire can of worms that is DIY. In the end, DSP is the easy part, and measurement is the difficult / time-consuming part.

I think your problem is that when DSP is deployed for the crossover, when it comes to commercial speakers, big improvements are really only possible if you start with a genuinely bad speaker.

Coaxials will be more forgiving if you try to EQ beyond room correction due to more even dispersion, but in the case of KEF speakers they usually don't need a lot of help there.

In the case of 1, 2, 3-way, MTM or whatever, DSP is basically just an easier and more powerful way to correct for the shortcomings / challenges of the design. I don't think there's a special type of speaker where DSP magically gets you way more improvement.

Redoing the crossover of an existing speaker does seem like it would yield a big improvement, but unless it was poorly done in the first place you're not going to get a night-and-day better speaker at the end. And bi-amping isn't enough, you would need to bypass the existing crossover.

I guess where DSP might shine the most is when you want to push the limits of drivers in a DIY context in a way that requires very steep xover slopes. This is something you can do with FIR filters that isn't really possible with passive crossovers, and done right you can sort of cheat the limitations of drivers by cutting off their response very quickly, so you can squeeze higher / lower crossovers out of drivers, dodge distortion, etc. This is kind of advanced DIY work, you can easily end up with a weird sounding speaker, but it's one approach that is vastly harder without DSP.

 

[Pariah Zero]

So the question comes back to you - are you prepared to DIY or modify a commercial design? If not, then the role of DSP is a bit more limited. You will be stuck trying to DSP a whole loudspeaker (which isn't easy) or limiting yourself to room correction only.

I'm prepared to do either, and perhaps both:

The Dayton T652-AIR speakers I have are (in)famous for having a single capacitor on the tweeter for the "crossover." (You can reference ASR's B652-AIR speaker review... just be careful because there's a loose panther head rolling around somewhere).

I already have replacement binding posts so I can re-wire and give a DSP channel for the tweeter, and one for the woofer(s) (i.e. Bi-amp and use the DSP for the crossover). It'll take a bit of routing the back panel to fit the new rear panel assembly (ie plastic bit + binding posts). I'm kind of planning on doing that modding (and the quasi anechoic measurements) already. I have a UMIK-1 measurement microphone, stand, and lazy Susans are easy to get to place the speaker.

That said, "Black Friday" sales are now, and I'm thinking of having "better" speakers waiting for later. (Even if those Black Friday sales are for, example, a kit speaker or drivers for the Mechano23.)

I guess my question for you is what you hope to accomplish with DSP in particular.

#1: I have the DSP, and I'd like to tinker and learn how to use it - I realize it's a gateway drug to DIY, but I love to DIY and tinker. I've made a number of microphones as well - ribbon, condenser (traditional hi-z and RF), and dynamic. I don't want to chase the dragon endlessly, but I would like to be able to build the occasional speaker set and then use DSP to make the most of what it can do.

 

[Torin Krell]

DSP correction usually works best on speakers with smooth, reasonably even dispersion / directivity characteristics. Once you have chosen quality speakers that are also reasonably flat (or Harman curve like), low distortion (as well as other important parameters listed elsewhere on this forum), optimizing speaker position, listening position and room treatment should ideally come next. After these are optimized, DSP will have less to correct for, ("and need to work less hard") which will translate into better final results.

 

[Kal Rubinson]

I gave the KEF and Klipsch as examples as I know KEF's are coaxial (but not bi-ampable), and I'd largely be limited to EQ.

Only the 2way KEFs are not capable of being bi-amped. The 3-way KEFs can.

 

[Chrispy]

I'm prepared to do either, and perhaps both:

The Dayton T652-AIR speakers I have are (in)famous for having a single capacitor on the tweeter for the "crossover." (You can reference ASR's B652-AIR speaker review... just be careful because there's a loose panther head rolling around somewhere).

I already have replacement binding posts so I can re-wire and give a DSP channel for the tweeter, and one for the woofer(s) (i.e. Bi-amp and use the DSP for the crossover). It'll take a bit of routing the back panel to fit the new rear panel assembly (ie plastic bit + binding posts). I'm kind of planning on doing that modding (and the quasi anechoic measurements) already. I have a UMIK-1 measurement microphone, stand, and lazy Susans are easy to get to place the speaker.

That said, "Black Friday" sales are now, and I'm thinking of having "better" speakers waiting for later. (Even if those Black Friday sales are for, example, a kit speaker or drivers for the Mechano23.)


#1: I have the DSP, and I'd like to tinker and learn how to use it - I realize it's a gateway drug to DIY, but I love to DIY and tinker. I've made a number of microphones as well - ribbon, condenser (traditional hi-z and RF), and dynamic. I don't want to chase the dragon endlessly, but I would like to be able to build the occasional speaker set and then use DSP to make the most of what it can do.

I'd just move to better speakers.

 

[Pariah Zero]

I'd just move to better speakers.

Hence my question if there's a particular kind of speaker that is more flexible if I want to tinker with the DSP for EQ and/or room equalization.

It sounds like there really isn't, per se, which is good to know.

 

[Pariah Zero]

Only the 2way KEFs are not capable of being bi-amped. The 3-way KEFs can.

I'm not familiar with all of KEF's line, but I don't recall seeing currently-manufactured models which are under $500 and bi-amped. Still, good to know for the future.

 

[Chrispy]

Hence my question if there's a particular kind of speaker that is more flexible if I want to tinker with the DSP for EQ and/or room equalization.

It sounds like there really isn't, per se, which is good to know.

Yes, generally a more capable speaker can handle more variance via eq. Not sure what you mean by "kind" particularly otoh.

 

[Keith_W]

I'm prepared to do either, and perhaps both

Good! In that case I have some suggestions that will make your life easier:

- Avoid any speaker with ports. This is because ports need to be measured separately from the woofer with a loopback timing reference so your UMIK-1 is not suitable. Also, ports can not be corrected independent of the woofer - this makes filter design a bit more difficult.
- Avoid any speaker with drivers close to the floor. It makes it more difficult to take a measurement without early reflections from the floor. One trick is to turn the speaker upside down, but that requires the speaker to have a flat top.
- Avoid any dipole, any omni, and any panel speaker. It's not that they are bad speakers, but it's because taking a good measurement is much more challenging and will require special precautions.
- Flat baffles are easier to model than curved baffles. For example, when I took measurements of my woofer, I placed my mic 10cm from the woofer to obtain a quasi-anechoic response. I looked for the reflection in the ETC and windowed it out. But measuring so close does not account for the baffle step, so I simulated the baffle step and convolved it with the measurement.

None of these are absolutes, and of course you can go with whatever loudspeaker you want. I am only pointing out that some loudspeakers are easier to measure than others, and these types of loudspeakers are more suitable for us hobbyists with limited resources.

 

[Pariah Zero]

Good! In that case I have some suggestions that will make your life easier:

- Avoid any speaker with ports. This is because ports need to be measured separately from the woofer with a loopback timing reference so your UMIK-1 is not suitable. Also, ports can not be corrected independent of the woofer - this makes filter design a bit more difficult.

As you might expect, as I have made multiple microphones - I have a few multichannel USB audio interfaces as well - would they be able to do the loopback timing reference? (obviously with a different calibration mic, but, still...)

 

[Keith_W]

As you might expect, as I have made multiple microphones - I have a few multichannel USB audio interfaces as well - would they be able to do the loopback timing reference? (obviously with a different calibration mic, but, still...)

To do a loopback timing reference, you need an interface or a UMIK-2. Some interfaces (e.g. Focusrite) make you use an actual loopback cable which means you lose one analog output and one analog input. Others (e.g. RME) lets you use a software loopback.

 

[Duke]

Hence my question if there's a particular kind of speaker that is more flexible if I want to tinker with the DSP for EQ and/or room equalization.

Well I'm not sure about "more flexible", but this comes to mind:

You might look to speakers which already do well in areas that matter to you which DSP cannot fix. For example,DSP cannot increase a speaker's efficiency or thermal power handling, nor can it fix problems with the radiation pattern.

On the other hand if a speaker has a lot of untapped headroom, DSP can help you take advantage of that.

Several years ago the manufacturer of a computer-based front end used a pair of my stand-mount speakers along with someone else's subwoofers at CES in Las Vegas. The subwoofers crapped out on the first day of the show. Well my stand-mount speakers had 12" prosound midwoofers with a lot of linear excursion and 400 watts RMS thermal power handling. He was able to take advantage of those attributes with the EQ capability in his computer-based system.