Beocreate as a standalone crossover (needs power)?

[MeloManiac]

 

I came across the Beocreate 4-channel  amplifier.

https://www.raspberrypi.org/magpi/beocreate-4-channel-amplifier-review/

https://www.hifiberry.com/shop/beocreate/beocreate-4-channel-amplifier/

https://www.analog.com/en/design-center/evaluation-hardware-and-software/software/ss_sigst_02.html#software-overview

 

As far as I understand,  together with a Raspberry Pi, it can be used as a full blown digital amplifier. But what makes it even more interesting, imho, is that it can also be used as a fully programmable, digital crossover. It was originally conceived  for B&O speakers, but according to the review, it will work with any passive speaker: "(...) it supports the creation of DSP ‘profiles’ using the provided SigmaStudio software. You can play endlessly with crossover frequencies and filters, create a setup you like, and then store it on the Beocreate so anything playing on the Pi uses it. This is a level of control that can only have been imagined just a few years ago."

The review concludes: "An audiophile’s dream, the Beocreate is the missing link between the worlds of classic high-end speakers and modern streaming technology."

 

So I'm curious: could this tiny thing be used to upgrade classic Klipsch speakers too? I'd love to hear what crossover specialist think of this.

 

 

[Chris A]

2 hours ago, LeftEyeShooter said:

So I'm curious: could this tiny thing be used to upgrade classic Klipsch speakers too? I'd love to hear what crossover specialist think of this.

 

Looks like this is a project environment for Raspberry Pi and/or SigmaStudio enthusiasts.  Even on diyAudio, the subject still seems to be a bit esoteric.

 

I'm not sure that many would go this route, for instance, a miniDSP 2x4 HD costs $205 with a box...so what you're really getting for the money is two 30w amplifier channels and two 60w amplifier channels, all class D, on one board--without a box or connectors, etc.  The Beocreate 4-channel class D amplifier provides a direct plug-in interface for a Raspberry Pi, and some sort of application library to start off, but I don't see finished solutions for hi-fi audio use.  I do see a some ideas presented, including "internet of things" type of applications, but the DSP crossover portion of the puzzle seems to be not well documented when I searched.

 

One might look at this as a "toy computer + 4 channel amplifier"...which looks like a DIY audio-only AVR with four channels for approximately $300+...once you get all of the stuff you need together to do it.  If that interests you and you have a penchant for home brew DSP processing and development, it looks like an interesting project.  However, I have too many of those right now...and the concept doesn't solve or address any audio problems that I currently see. 

 

I'd be more inclined to invest in a miniDSP 2x4 HD and a pair of 50ASX2 class D board amplifiers to bi-amp a pair of two-way loudspeakers, or 2 miniDSP 2x4 HDs and three 50ASX2 ICEpower modules to tri-amp two three-way loudspeakers.  The miniDSPs come with pretty good DSP applications and there is a relative wealth of information at the miniDSP web site--including using freeware FIR filter applications and REW, etc.--so that portion of the problem space is pretty much under control. 

 

The only remaining question is the resulting hi-fi sound quality, which has been an issue with miniDSP products in the past. I'm not sure that Raspberry Pi is up to the task--as the miniDSPs are making full use of SHARC processors to do what they do.

 

Chris

[MeloManiac]

18 minutes ago, Chris A said:

 

The only remaining question is the resulting hi-fi sound quality, which has been an issue with miniDSP products in the past. I'm not sure that Raspberry Pi is up to the task--as the miniDSPs are making full use of SHARC processors to do what they do.

 

Chris

 

Once the eprom on the Beocrate is flashed, it can operate without the raspberry pi (but it will need power). I'm 99% sure the Beocrate also uses the SHARC processor. The hifi sound quality of the beocrate add on should be far superior to the sound quality of the raspberry pi itself.

 

This is the more elaborate video, step by step. It seems pretty do-able. The SD-card that is used here, I guess, is configured with presets for this specific speaker. For Klipsch speakers, they have a 'safe' standard setting that then can be fine-tuned by the user in SigmaStudio.

 

 

[Edgar]

16 minutes ago, Chris A said:

The only remaining question is the resulting hi-fi sound quality, which has been an issue with miniDSP products in the past. I'm not sure that Raspberry Pi is up to the task--as the miniDSPs are making full use of SHARC processors to do what they do.

 

5 minutes ago, LeftEyeShooter said:

The hifi sound quality of the beocrate add on should be far superior to the sound quality of the raspberry pi itself.

 

Except for esoteric issues of floating-point vs. fixed-point, word size and numerical precision, limit-cycles, etc., that are of importance to programmers, assuming that the arithmetic is done properly it makes no difference whether the digital signal processing is performed in an ARM (Raspberry Pi), a Sharc (Beocrate), or by counting on one's fingers. If there are differences in sound quality between various processors, it is either because the analog stages are different, or it is because the programmer didn't implement the arithmetic correctly. The former can be avoided by digital transfers; the latter by employing competent programmers.

 

Also, the Raspberry Pi 3 has a 1.2 GHz quad-core ARM processor with a NEON floating-point coprocessor. That's a lot of processing power. I did a project with the Pi3 a while back, and ultimately gave up because Linux ALSA is not ready for prime time (IMHO), and because there are problems with the Pi3 access to more than two of the eight HDMI audio channels.

[Chris A]

I should have said "Raspberry Pi development environment".  Even miniDSP has sound quality issues, probably due to insufficient analog stages and perhaps the level of DSP processing to achieve high quality sound out the back end.  All the criticisms of miniDSPs currently seem to be:

1. lack of sufficient number of PEQs (biquads) available per channel
2. lack of sufficient gain or attenuation per channel
3. fidelity of the resulting DSP algorithms to produce high fidelity--a comment that was made by the reviewer looking at the output waveforms vs. a higher performance commercial-grade DSP crossover
4. a high DSP-induced noise floor in the output analog channels
5. not enough FIR taps to be useful to lower frequencies at higher sampling rates (i.e., 96 kHz/channel).

 

All of these issues are unknown for the B&O DSP/amplifier board mentioned above, and each line item is a potential deal-killer for my uses.  For the present, it seems prudent to stick with something that has proven performance, if home hi-fi is the goal.  If that isn't the goal, rather if you are just tinkering, then by all means--buy one and try it out.  Then let us know if you were ultimately successful.

 

Chris

[Edgar]

5 minutes ago, Chris A said:

I should have said "Raspberry Pi development environment".

 

Yes, for an end-user there simply isn't one. The user would have to know how to program from scratch. In my case I used C++ and PortAudio under Linux.

 

Quote

lack of sufficient number of PEQs (biquads) available per channel

 

Designer's choice. No intrinsic limitation, until your number of biquads gets huge, at which point you also start to run into numerical precision problems.

 

Quote

lack of sufficient gain or attenuation per channel

 

That's best done in the analog domain.

 

Quote

fidelity of the resulting DSP algorithms to produce high fidelity (a comment that was made by the reviewer looking at the output waveforms vs. a higher performance commercial-grade DSP crossover

 

Now that's a programming problem. The Sharc is a well-established, well-respected processor for audio. There should be no problem with sound quality.

 

Quote

a high DSP-induced noise floor in the output analog channels

 

Ditto the above.

 

Quote

not enough FIR taps to be useful to lower frequencies

 

Same answer as number of PEQs -- designer's choice.

 

I inquired long ago whether miniDSP would release a SDK (software development kit). They refused, saying that they wanted to keep complete control over the firmware. A pity, because there are some of us out here who know what they're doing. I would probably enjoy designing custom crossovers and EQ for Klipsch products.

 

[Chris A]

21 minutes ago, Edgar said:

I inquired long ago whether miniDSP would release a SDK (software development kit). They refused, saying that they wanted to keep complete control over the firmware. A pity, because there are some of us out here who know what they're doing.

Yes--and I sense that is also the case over at diyAudio where a greater concentration of digital guys seem to reside. Since I classify myself as a user presently, I make comments only from that standpoint.  It seems to me that Xilica is pretty much meeting the standard for commercial loudspeaker processors at the "right price" by offering only the features that the user needs at the lowest price for commercial processors. When you add FIR filtering and a AES/EBU interface, the price doubles (XP series --> XD series).  That seems like a wise marketing choice to me in today's hardware environment.

 

As far as the quality/price of loudspeaker processors (DSP crossovers), I see there is an overall tendency of home hi-fi users (mostly DIY types) to gravitate to the lowest priced products, when, if you look at the prices of esoteric amplifiers and preamplifiers, cables, sources/DACs, etc.--this all seems to me to be curiously bad economics in action. (This isn't the first time that I've said this in this forum.) In my experience, it's wise to put your money on the DSP crossover(s) considering the state of the present marketplace, then if you have to skimp, do that on the amplifiers and preamps since you can find extremely high quality components for very little money--especially when considering other Class D amplifier board offerings from B&O. 

 

Also, DSP crossovers can be used on any loudspeaker or collection of drivers/horns that the owner chooses, to be reused from setup and speakers to other setups and speakers for the life of the user.  This is much different than passive crossovers--where even if you replace a blown or aged diaphragm in one driver, you've basically invalidated that network unless you physically modify the network.  I think that a lot of audio enthusiasts really haven't realized this.

 

Chris

[Edgar]

1 hour ago, Chris A said:

As far as the quality/price of loudspeaker processors (DSP crossovers), I see there is an overall tendency of home hi-fi users (mostly DIY types) to gravitate to the lowest priced products, when, if you look at the prices of esoteric amplifiers and preamplifiers, cables, sources/DACs, etc.--this all seems to me to be curiously bad economics in action.

 

Yes! Yes! Yes! Years ago I made the comment on this very forum that I couldn't understand why people would be willing to spend thousands of dollars on a preamp, but only $100 on a crossover, whose job is much harder. Back then most crossovers were analog, but the idea still holds true for digital.

[MeloManiac]

19 minutes ago, Edgar said:

 

Yes! Yes! Yes! Years ago I made the comment on this very forum that I couldn't understand why people would be willing to spend thousands of dollars on a preamp, but only $100 on a crossover, whose job is much harder. Back then most crossovers were analog, but the idea still holds true for digital.

 

A while ago I discovered Siegfried Linkwitz and his speakers. Both in the US and in Europe, you can buy his DIY kits. The LX521.4 is the flag ship of the line-up. The components are around $2,100 and then you must add $3,000 for the 'powerbox' (an advanced DSP). Mr Linkwitz agrees with you here...

According to some,  this speaker set is one of the best in the world. The price doesn't include the cabinet (a blueprint is provided though), but does include a license to build the speaker and a step by step manual, and the speaker specific code for the 'powerbox'.

 

https://www.magiclx521.com/epages/17940394.sf/en_GB/?ObjectPath=/Shops/17940394/Products/"LX521 kit"

 

PowerBox 6pro NCore with Hypex DLCP digital signal processor. 

• All you need between your music source (CD, vinyl, BluetoothAP, SOONOS;......) and your speakers
• PowerBox = Preamp & ADC & DSP & 6 DACs & 6 Hypex power amplifiers
• Tailored performance: 2x100W, 2x125W, 2x250W, all 2Ohms capable!
• Slim & quiet, fan-less design. Massive 3mm full aluminum case with stainless steel front plate. Dimensions ca 44 x 36 x 5,5 cm
• Professional high current SpeakON output: Two SpeakONs replace 12 (!) “banana” plugs
• High End Hypex DSP with carefully designed clock-, DAC- and analogue output stages.
• Balanced signal from input to output, remote volume control.
• LXspeaker owners get preloaded filter sets (LXmini, LXmini+2, LXstudio). LX521.4 needs two of these PowerBoxes, a MASTER  and a FOLLOWER unit.
• IN: analog balanced XLR, digital RCA, optical, AES XLR & RJ45, USB for individual filter, gain, delay and inv. programming!
• Free Hypex Filter Design software for any speaker project.

 

[Edgar]

7 minutes ago, LeftEyeShooter said:

 

A while ago I discovered Siegfried Linkwitz and his speakers.

 

 

My very few dealings with Siegfried Linkwitz were just barely cordial -- he and I didn't agree on the importance of minimum phase response. But otherwise he was very creative and prolific, and your example shows that he was largely on the right track. R.I.P.

[Chris A]

I was going to say that I've got a viewpoint about Linkwitz' loudspeaker designs (which I personally don't consider to be the best approaches).  I think that he approached the subject from a direct radiating loudspeaker background and converged in his later loudspeaker designs on some familiar revelations of what makes loudspeakers sound accurate, clear, and subjectively engaging--that I'll highlight below:

1. the importance of somewhat controlled directivity obtained via dipole radiation to control horizontal and vertical directivity
2. he increasingly saw the importance of controlled phase behavior toward the end of his life, and even distanced himself from the crossover network topology that was named for him (and was originally invented by Russ Riley--as it has been reported).  I think that as you read through his website on the more "advanced topics" areas, you begin to see that he realized that phase is at least as important as flat amplitude (SPL) response after something is done about directivity issues in the the full range frequencies.

In other words, I think that he was converging on the same "truths" of loudspeaker performance as that which I personally have recently seen to be true.  If he had lived another 20 years, perhaps he would have arrived at the point where a few others here may have arrived:

• full-range controlled directivity via horns to minimize in-room nearfield reflections vs. frequency
• low modulation distortion (which Linkwitz missed--like so many others) for realistic full dynamic range production and clarity of presentation, and
• minimum phase behavior of the loudspeakers, i.e., without significant phase growth at least considering the electrical crossover networks

I think Danley found these three topics --starting at very high powered acoustic wave levitation and progressing to very high powered direct radiating subwoofers, then the MEH (multiple entry horn) concept and the resulting the Synergy™ horn . 

 

But there are other ways to achieve at least two out of three of these factors to get at least good results. 
 

Chris

[Edgar]

40 minutes ago, Chris A said:

he increasingly saw the importance of controlled phase behavior toward the end of his life, and even distanced himself from the crossover network topology that was named for him (and was originally invented by Russ Riley--as it has been reported). 

 

Interesting ... the fact that a Linkwitz-Riley crossover creates an allpass filter, which is nonminimum phase, was the main source of disagreement between him and me at the time. (This was many years ago.) I argued that nonminimum phase was a problem; he argued that it didn' t matter.

 

As for Riley; the story that I heard was that Riley noticed that the difference between a 2^nd-order highpass transfer function and a 2^nd-order lowpass function was arithmetically equal to a 1^st-order allpass function. Linkwitz extended it to higher-order transfer functions and applied it to crossovers. At least that's what I read somewhere. Could be wrong.

 

[Chris A]

It's nice to know who did what. 

 

The good news is that a few of the Klipsch designs (the ones with full-range directivity) can also be made to have nearly minimum phase response through careful choice of crossover filters using a DSP crossover.  I've recently found that the shortened-down Jubs have really outstanding phase and group delay response as measured in and around the listening positions in my (treated) listening room.  I'm talking performance that rivals the SH-50 performance in these areas.  The Jubs of course have better directivity control to a lower frequency (about 100 Hz) than the SH-50 (about 250-300 Hz before the -6 dB point begins to interact with the side walls if placed near a corner).

 

 

 

Chris

[Chris A]

I suppose I should post the same data for the SH-50 to prove my boast above:

 

 

 

Chris

[Edgar]

Chris, is the gray line the minimum-phase response for the same magnitude response?

[Chris A]

For each loudspeaker, the gray line is minimum phase/excess group delay.  It looks as if the Jubilee SPL is 86 dB and the SH-50 is 91 dB, both at 1 m on-axis.