Any interest in an ASR community speaker project?

[thewas]

Low order filters, which help blend directivity characteristics, are much easier to implement due to delay.

Low order filters enhance though vertical directivity problems (lobes).

 

[dwkdnvr]

I'll add a bit more depth, but as a quick reply I'll register my interest. This will compete for time with my own project, but I'd be willing to see how I can help out. DSP stuff and a CNC router at least for prototype baffles if we want to try something interesting, plus maybe measuring if I get my butt in gear, since I have to do that for my project as well.

My gut reaction is that an ASR1 really should have a waveguide, but I guess we'll see how that falls out. 2 alternate drivers to think about. SB10PGC21 is a 3" mid for $16, and according to the datasheet measures fantastically well through the midrange. ND25FW (as used in the C-note) or vifa XT19+visaton WR-146 as inexpensive waveguide tweeter choices. (the wr-146 is a bit big, but see here http://www.5een.co.uk/Waves.htm )

 

[briskly]

My understanding is that the Visaton and Monacor waveguides are similar in profile, but the throat transitions are different. Does anyone know what that difference is?

ScanSpeak Discovery sort of level.

Any particular reason to pick up the Disco models over the original Peerless versions?
Also, in the hypothetical SB 3-way, I think the 23NRX is a better choice. TSPs aside, the material choice is not going to be a major factor with such a low expected x-over point, and NRX is cheaper than the NAC/NBAC.

 

[617]

MY understanding is that the Visaton and Monacor waveguides are similar in profile, but the throat transitions are different. Does anyone know what that difference is?


Any particular reason to pick up the Disco models over the original Peerless versions?
Also, in the hypothetical SB 3-way, I think the 23NRX is a better choice. TSPs aside, the material choice is not going to be a major factor with such a low expected x-over point, and NRX is cheaper than the NAC/NBAC.

You make good points. I think driver choice is a lot easier with a fully active dsp system, and many woofers would work. I would like to use something which works well in a sealed enclosure as the small size is very appealing to me, but obviously going vented is up to the builder's discretion.

I don't know or can't remember the difference between the monacor and visaton waveguides, I believe heissman acoustics has that info on his website.

Regarding scan discovery, I know most of the tweeters are old peerless/vifa designs, but I don't believe the midranges were? The 15m fiberglass unit with foam surround for example is unique to the discovery line if I am not mistaken.

 

[617]

I have used these Sure Electronics or WONDOM brand amplifiers in many projects. Parts Express has them re-branded with red PCBs and makes nicer user manuals for them. I bought mine before Parts Express started selling these but I would prefer to source the boards from them since it is the same price and comes from their US warehouse rather than China direct. I have also used their Bluetooth equipped amp boards but I would not recommend them due to nasty ground loop noises and DC offset when using the supplied volume pot.
View attachment 66711
View attachment 66712
I turned this old General Electric speaker into a battery powered bluetooth speaker using the 1x100w amp with DSP (equivalent to the DSPB-100) and the 1x100w amp without DSP (WONDOM AA-AB31184). As you can see they are mounted on the KAB-AB L-type Aluminum Bracket using the
PC Board M3 Standoff Kit. I have used the 1x100w and 2x50w amps without DSP for a few different bluetooth speaker projects and have been quite satisfied. I would bet that their measured performance is mediocre but they sound perfectly adequate to me. Thankfully there was no ground loop issue when powering both amplifier boards from the same power supply. The gain of these two amps is different but once I had the volume properly set, there didn't seem to be any audible difference between the amps while I was deciding which to use for woofer and which for tweeter. Noise levels weren't super low but they were about on par with my LSR305's, although I was using the not so efficient Peerless DX20BF00-04. After this project I realized it would have made more sense to just use the 2x50w amp with DSP for a 2-way speaker.
View attachment 66713
View attachment 66714
View attachment 66715
This is an old XAM 4E speaker that I am turning into a self power speaker with built in phono preamp. I'm using the 2x50w amp with DSP for this one with the same Peerless tweeter and the Dayton Audio DC160-4.

All of these DSP units from Sure Electronics use the ADAU1701 chip which is the same part used in the standard miniDSP 2x4. I wouldn't call the miniDSP 2x4 simple to use, but the Sure Electronics units are even less user friendly imo. First of all they require the Dayton Audio DSPB-ICP1 for programming, which is an added cost. But more annoyingly they use the SigmaStudio software for programming which is not intuitive. Now, once you have learned to use the software and have created the perfect crossover, it is easy to save that configuration and share it with others. However just getting the program to recognize the DSP board and getting the configuration properly synced can be a pita. I wouldn't consider this a deal breaker but someone will have to create a detailed guide if this project is meant to be shared with a larger audience.

Also the Dayton Audio DSPB-K is interesting but when used with the Dayton Audio DSPB-KE i/o board it only gives you 3 outputs. I'm not sure if the DSPB-K board is inherently limited to 3 outputs but it's a shame the connection board designed for it is limited. When you consider the price of these two boards together along with the price of the DSPB-ICP1 programming board, and keep in mind that it only has 3 outputs, I would prefer to use the miniDSP 2x4 Kit. And I would not be surprised if the miniDSP has better measured performance when you look at the supporting circuitry around the ADAU1701. The DSPB-250 is definitely the sweet spot when it comes to price vs features seeing as you're getting a 2x50w amplifier along with two more outputs from a 3.5mm jack. Pairing that with the 1x100w amp without DSP woud make a lot of sense for a 3-way speaker. And then there's still an extra output for a subwoofer if you want it, although I think it would make more sense to use a preamplifier or processor with subwoofer outputs before the speakers.

Wow, thanks for all the images. It didn't occur to me that the boards could be stacked. Love seeing your projects, very much in the DIY spirit. I especially like the little 3d printed escutcheon you made for that tweeter.

I agree that the minidsp stuff is a bit easier to program, I looked at sigmastudio and it doesn't look very inviting. However, I'm fairly confident I can figure it out.

I sent off the DSPB 250 and KAB 100 boards to be tested. I am particularly interested in if it the kab100 has any sort of hum or noise like you mentioned.

 

[somebodyelse]

I agree that the minidsp stuff is a bit easier to program, I looked at sigmastudio and it doesn't look very inviting. However, I'm fairly confident I can figure it out.

Hifiberry have a fair bit of stuff about using SigmaStudio for the BeoCreate and DAC+ DSP, and FreeDSP may also be helpful. Porting some of their software may be viable too - HifiBerry's via a Pi, or Auverdion's web gui on an esp32 maybe. That's going somewhat beyond an initial prototype though.

 

[dwkdnvr]

I think there is some benefit in keeping the 'design concept' pieces separate from the 'dsp implementation' pieces, even if "we" are working through them together.

the 'design concept' would be the driver lineup, specific cabinet design/drawings, and a set of biquads/gains that implement the xover transfer functions. This design concept could in principle be implemented in a variety of ways.

Then, we have the 'canonical implementation' of the design using a particular dsp/amp setup - the Dayton DSP amps potentially seem to be a good choice, but maybe something else will crop up.

Obviously there is some potential overlap - i.e. does the cabinet design include provision for embedding the dsp/amps to facilitate an 'active monitor' form factor? But keeping a bit of separation IMHO does a couple things:
- makes collaboration easier since it decouples the 2 aspects
- broadens the potential appeal, since you aren't necessarily locked into specific hardware.

Digression:
One of the big barriers to DIY DSP systems is that creating a truly universal system that satisfies everyone's particular requirements is a bit difficult. Some folks are 2-channel PC audio only, in which case JRiver for example works great. Others need HT integration, which means some form of analog input is necessary. Others recoil completely from the idea of an extra AD/DA cycle. Some need remote control, some don't, ditto multiple sources. Some get hung up on avoiding sample-rate-conversion and need a system that adapts, while others are OK with SRC. etc.

So, I see a great benefit to decoupling the 2 aspects since it balances giving folks an 'all in one' recipe with flexibility to go your own way if you want. It also avoids a potential pitfall with device availability - if the Dayton DSP amps suddenly become unavailable, the project can retool pretty easily. Plus, if/when we finally find a good cheap AVR, something like a Raspberry Pi 4 outputting HDMI to an AVR would be a fantastic general purpose DSP environment available to basically anybody (as long as you are music-only and can live with limitations of input source)
/Digression

In terms of design concept, I think a full 3-way DSP is probably the right idea. It's probably worth at least considering the idea of a hybrid passive/active design though. A passive M/T crossover coupled with an active dsp xover to the woofers would be realizable with 4 channels of DSP/amp rather than 6, opening up things like the Minidsp 2x4HD or even SHD. The passive M/T xover could be 'assisted' by dsp (e.g. for baffle step if needed) and so wouldn't be a tremendously difficult xover if decent drivers were selected. But, it addresses the single biggest challenges of 3-ways noted earlier - sensitivity matching of the woofer to the M/T and the large/expensive xover components needed for low-ish woofer xovers.

Procedural ideas:
- VituixCad for the speaker design (pretty much a no-brainer as far as I'm concerned)
- github project to host/share artifacts and documentation etc.

 

[dwkdnvr]

Getting back to driver choices, I think the biggest points of agreement that have to be arrived at are
- budget
- acceptable cabinet size
- (maybe) topology - sealed, ported, PR?

My concern about a sealed alignment is that output is ultimately limited and it needs power, at least if you're trying to stay small. If sealed is really desired, the RSS210HO probably deserves a look - not too much more expensive than the SB23's, but more xmax and does ok in about 0.5 cu ft. PR alignments can work well for small boxes, but are trickier to get right. The Dayton DSA175 prs are pretty cheap though, so I may putz around and see whether I can model up something that might be an interesting alternative.

 

[dwkdnvr]

Ok, throwing a potential woofer option out there. A single Anarchy woofer ($66 from diysoundgroup) with 2 Dayton DSA215 PRs ($20 ea) in 19L will be flat-ish to 30Hz and produce 100dB with 80W input before the PR excursion becomes a problem. 150g added weight to each PR, but I'm not actually sure they will support that much weight. 17L and 100g each looks good to the mid-30's and gets you 104dB @100W (modeled with WinISD, which I believe is half-space)

this is the 8-ohm Anarchy. Normally I'd go for the 4-ohm to take better advantage of amp power, but full T/S params for the 4-ohm aren't available.

 

[GDDGEE]

@617 First off, I really like this idea! It appears coming up on the design constraints will be the hardest part to get this off the ground, which is likely to be expected. It sounds as if you have a vision in mind, so perhaps state your goals for a design, get going, and see who wants to join in and call it the "ASR-1" as @Thomas savage suggested. Perhaps the way to go about it is to consider the speaker an 'open source' design where everything you're doing is 'published' for comment/update. I think a clean way to do this would be:

1. Go active and 'publish' the complete setup of filters/settings - give the Dayton boards a try. If we find out the Dayton amps are low quality, it seems straightforward-ish to copy the settings to a MiniDSP/ICEPOWER/Hypex setup.

2. Use a translaminated box design (see attached as an example). Once the shape is determined, anyone could get the CAD files and remake it exactly like yours and it this way it doesn't have to be a plain box.

3. Use common, readily available drivers (as you've already stated).

I would try to contribute however I can if you pick something and go with it.

 

[617]

Getting back to driver choices, I think the biggest points of agreement that have to be arrived at are
- budget
- acceptable cabinet size
- (maybe) topology - sealed, ported, PR?

My concern about a sealed alignment is that output is ultimately limited and it needs power, at least if you're trying to stay small. If sealed is really desired, the RSS210HO probably deserves a look - not too much more expensive than the SB23's, but more xmax and does ok in about 0.5 cu ft. PR alignments can work well for small boxes, but are trickier to get right. The Dayton DSA175 prs are pretty cheap though, so I may putz around and see whether I can model up something that might be an interesting alternative.

A few thoughts. Dsp enables designs which would be impossible to realize with passive components. However, you can keep almost all the advantages of an active design by going to a hybrid design where W/M is an active crossover, and M/T is passive. The latter passive network doesn't need to be too fancy, since you can apply PEQ to that channel, so really it's just a matter of getting the crossover integration correct.

Going in this direction opens up some possibilities for amplification, as well, since it is a bit easier to find a stereo amp to stick in a speaker than it is to find 3 channels of amplification. Having said that, if the tests on the 50x50x100 module come back looking good, I think that is a good direction, since the entirely active implementation is a clean, easy thing to build, compared to a system having both an active and passive crossover.

Now, regarding woofers. This is an interesting topic. My thinking is that I need to do some simulation on basic crossover points and driver sizes to see what works best, but I am thinking that if we work our way from the top, with a 26mm tweeter, the biggest mid you would want to use for really smooth off axis is probably around 120mm (or less). Now, going from the mid down to the woofer, you don't want to get into a situation where you have a huge woofer with big xmax and resonances all over the midrange, since you're using it up relatively high (say we want drama free bandwidth to 500hz or above.) This doesn't invalidate the use of a subwoofer, but it's something we need to look into.

Now, regarding budget. Part of the appeal of DSP is that you can, to a certain extent, use ornery drivers with good harmonic distortion characteristics, but poor linear distortion - rising responses, low breakup and so on. As a result of this, and the fact that level matching is less of a priority, driver cost can be somewhat low. A lot of what you pay for in expensive drivers is ease of use with passive crossovers, but when you have the ability to use an 8th order filter on your woofer and unlimited EQ on every driver, the demands for the drivers are reduced somewhat. What becomes important is low harmonic distortion, usable bandwidth and consistency. I think for the sake of fun, we should be looking at a budget of $160 per speaker. I think this is the sweet spot.

@hardisj can you chime in and give your advice on woofer selection? We're designing trying to design an active/DSP speaker, and we're thinking of using a sealed 8-10" woofer for the low end. Requirements are smooth response to 500hz, reasonable extension in a smallish cabinet, under $120 or so, and high output. I know you used to measure these things so your intuition is valuable here.

 

[hardisj]

@hardisj can you chime in and give your advice on woofer selection? We're designing trying to design an active/DSP speaker, and we're thinking of using a sealed 8-10" woofer for the low end. Requirements are smooth response to 500hz, reasonable extension in a smallish cabinet, under $120 or so, and high output. I know you used to measure these things so your intuition is valuable here.

$120/pair or $120/driver?
Up to 500hz implies midbass passband to me (i.e., 70hz-500hz). Is that correct? IOW, this isn't a subwoofer, right?

I always default to Scanspeak and SB Acoustics. Scanspeak's Discovery line is what I used in my car. 10 inch midbass covering 80-300hz. Around the $120/driver mark.

But, I'll wait to get the answers before I throw out any suggestions.

 

[617]

Ok, throwing a potential woofer option out there. A single Anarchy woofer ($66 from diysoundgroup) with 2 Dayton DSA215 PRs ($20 ea) in 19L will be flat-ish to 30Hz and produce 100dB with 80W input before the PR excursion becomes a problem. 150g added weight to each PR, but I'm not actually sure they will support that much weight. 17L and 100g each looks good to the mid-30's and gets you 104dB @100W (modeled with WinISD, which I believe is half-space)

this is the 8-ohm Anarchy. Normally I'd go for the 4-ohm to take better advantage of amp power, but full T/S params for the 4-ohm aren't available.

I like the Anarchy for obvious reasons, it is a beast. I also like it because it is small. However, I think one of the objectives here should be to keep cabinet size down, so designing for exclusive use with PRs I don't like. However, giving builders the option to build a bigger cabinet, using a port, or a PR (or transmission line) is a very desirable option. One problem I see with the Anarchy of course is that availability is spotty; only through DIYSG, and they're sold out of some of the other models. I would much prefer to get a more available driver from PE/Madisound.

Can you simulate the 708 in WinISD or whatever?

 

[617]

$120/pair or $120/driver?
Up to 500hz implies midbass passband to me (i.e., 70hz-500hz). Is that correct? IOW, this isn't a subwoofer, right?

I always default to Scanspeak and SB Acoustics. Scanspeak's Discovery line is what I used in my car. 10 inch midbass covering 80-300hz. Around the $120/driver mark.

But, I'll wait to get the answers before I throw out any suggestions.

$120 a driver (or less). I like the Dayton RS270p, for example, but if we can get an 8" with enough xmax to beat it that might be better.

 

[hex168]

@dwkdnvr I like where you are going, particularly the decoupling of the active option. I'd like to read more about your Raspberry Pi/ AVR concept when you have a chance. Can you take a look at the Dayton DSA175? It has lower distortion and higher efficiency than the Anarchy (although I would think Erich H or Matt Grant could supply the 4 ohm TS parameters). I'm also curious about the Faital Pro 6FE100 since it could easily acheive high outout on an AVR, but I have not found any independent tests.

 

[hardisj]

$120 a driver (or less). I like the Dayton RS270p, for example, but if we can get an 8" with enough xmax to beat it that might be better.

In general "There's no replacement for displacement". I'd go with a 10 inch midbass; you can sacrifice a little xmax in return for higher sensitivity. Plus, it doesn't beam until about 700hz so you're good to there.

I'll list a few that I would look at.
https://www.madisoundspeakerstore.c...very-26w/4534g-10-aluminum-cone-woofer-4-ohm/
(or the 8 ohm version)

I see PE has this as a buy-out (they don't normally carry Scanspeak) with 13mm linear excursion. Not super high sensitivity but 8 ohm and small enclosure space.
https://www.jdoqocy.com/click-7732025-13715689?url=http://www.parts-express.com/scan-speak--830772-10-paper-cone-subwoofer-8-ohm--299-729&cjsku=299-729



I've used this and preferred it over the SS Disco 10" above:
https://usspeaker.com/ciare-hw251n-1.htm


And, of course, Dayton drivers are always an option, like you said. You'll just have to find something that works with the planned enclosure you need.


No SB acoustic drivers in this price range in a 10 inch, unfortunately. But a word of caution with SB/Satori specs: their "linear xmax" values are usually close to half of what I have measured with my Klippel in the past.

 

[Mashcky]

I'm a member of a maker's space that just reopened on limited terms, so I now have access to a very nice CNC machine again. I will volunteer to help out using this resource however I can. While I am a bit of a CNC novice, cutting a simple baffle for testing would be very easy. I even have access to other woodworking supplies for making complete cabinets and have experience making decent cabinets for my er18dxt.

I won't add any new ideas to this thread because there seem to be enough already. I will say that I'm excited by OP's goals of using this project as a source of fun, distraction, and ASR community pride. I also like the idea of an inexpensive, small speaker that can be an office/shop/kitchen set. I don't suspect anything we come up with will displace the primary setup of hardcore ASR members, so why try? A low barrier speaker could also mean greater development participation, wider adoption, and faster iteration. For the same reason I'm also partial to high sensitivity, passive speakers, as both of these factors allows for cheaper, fewer amps. As for actives, I see no reason prototypes couldn't be sent to different parties to work on both an active version and a passive version in parallel but this (like all factors) needs to be decided by members who will actually commit to contribution.

 

[617]

Some data. I simulated most woofers and subs in the 8-10" size range from about 50-$120, also including some 7" models such as the Anarchy 708 and Dayton Esoteric 180ti. I graphed their maximum SPL/Frequency when put in a closed box of optimum size. I think selected a sort of median driver, and turned off everything that had less bass output (SPL at 40hz) than that, leaving 12 drivers. The ones which made the cut are somewhat interesting.

Looking at the data, the highest midband SPL (200hz) comes in at 110db, lowest is 105. Highest SPL at 30hz is 98db, lowest in this sample is 92db. Things are somewhat more complicated by power handling abilities of course, but this is a good sample to select a woofer from. I will compile some more data shortly.

 

[dwkdnvr]

A few thoughts. Dsp enables designs which would be impossible to realize with passive components. However, you can keep almost all the advantages of an active design by going to a hybrid design where W/M is an active crossover, and M/T is passive. The latter passive network doesn't need to be too fancy, since you can apply PEQ to that channel, so really it's just a matter of getting the crossover integration correct.

Going in this direction opens up some possibilities for amplification, as well, since it is a bit easier to find a stereo amp to stick in a speaker than it is to find 3 channels of amplification.

Yeah, the more I think about it the more I like the hybrid idea with passive M/T. Now, a lot of that is probably that I have the SHD and so my main project is exactly that type of design. I just think it opens up more flexibility overall - more DSP options, easy to do with a pair of stereo amps etc.

And, since we can assume that baffle effects can mostly be handled in the DSP, it means we can use 'infinite baffle' crossover designs which are easy to simulate and iterate in VituixCad using either datasheets or HiFiCompass measurements.

Plus, it makes it a bit easier to conceive of this as a modular design - the same M/T unit could be easily grafted onto a variety of woofer choices depending on target use - small sealed version for desktop use, larger ported/PR version for use in-room.

Having said that, if the tests on the 50x50x100 module come back looking good, I think that is a good direction, since the entirely active implementation is a clean, easy thing to build, compared to a system having both an active and passive crossover.

Did you send them to Amir, or somewhere else? Based on the datasheet, the amp itself is 'ok', although doesn't actually hit the power targets (a lot of these chips are rated at 10% distortion). If it uses the built-in converters in the ADAU dsp, it should be similar-ish to the original non-HD minidsp units - so 'ok, not great'.

Now, regarding woofers. This is an interesting topic. My thinking is that I need to do some simulation on basic crossover points and driver sizes to see what works best, but I am thinking that if we work our way from the top, with a 26mm tweeter, the biggest mid you would want to use for really smooth off axis is probably around 120mm (or less). Now, going from the mid down to the woofer, you don't want to get into a situation where you have a huge woofer with big xmax and resonances all over the midrange, since you're using it up relatively high (say we want drama free bandwidth to 500hz or above.) This doesn't invalidate the use of a subwoofer, but it's something we need to look into.

I'd target a 300Hz W/M xover. Low enough to keep a lot of flexibility in woofer choice, but high enough that even 10cm series drivers like the Scan 10f and SB10PGC can remain in contention. I like the idea of a 3" mid if only because it leaves more room in our hypothetical 0.5-0.6 cu ft cabinet for the woofer.
As for 'subwoofers', you need to be careful. The Dayton RSS drivers - particularly the HF line - are really more like inefficient woofers than typical subwoofers. Their cost and box size requirements probably eliminate them from this project (although I think I could hit 160/side using the RSS210HF if I had to), but there are very highly regarded designs out there that use the RSS210HF up to 800Hz and the 12" RSS315HF up to ~500.

I think the only other design constraint that needs to be agreed on is exactly how much bass extension and output we need, at least for the "reference" design. If we assume 0.6 cu ft net as our upper limit, that's probably a reasonable constraint. Extension and output become 'as much as possible under the limits'.

I think for the sake of fun, we should be looking at a budget of $160 per speaker. I think this is the sweet spot.

Okay, challenge accepted The problem won't be coming up with ideas, it'll be limiting it down to a couple serious suggestions.

 

[dwkdnvr]

Some data. I simulated most woofers and subs in the 8-10" size range from about 50-$120, also including some 7" models such as the Anarchy 708 and Dayton Esoteric 180ti. I graphed their maximum SPL/Frequency when put in a closed box of optimum size. I think selected a sort of median driver, and turned off everything that had less bass output (SPL at 40hz) than that, leaving 12 drivers. The ones which made the cut are somewhat interesting.
.

Interesting - nice work. What is the range of box sizes that results? If the Peerless 830668 can make it up to ~300Hz, it might be a great choice - I tend to forget about this driver since the ported box requirement is a bit big, but sealed with an LT might work very well.