Enginerding with Weeb Labs

[somebodyelse]

Note that despite appearances, the lower driver is not a BMR; it is simply a flat coned midwoofer selected to match the appearance of the BMR above it.

Are you sure it's not a second BMR, but low passed to prop up the bottom end a bit? I guess that would be a 1.5 way.

 

[voodooless]

Are you sure it's not a second BMR, but low passed to prop up the bottom end a bit? I guess that would be a 1.5 way.

You can clearly see the breakup of the second woofer in the individual plots.

 

[Rednaxela]

I will take these outside tomorrow and capture some anechoic measurements, which I anticipate will be rather flat.

Most interesting. Looking forward to these.

 

[Weeb Labs]

On the other hand here ins Europe you can get often a pair of active JBL 305 for 200 € which probably beat them in most objective measurements.

Definitely. You probably wouldn’t be looking at these unless you specifically want something small or with very wide directivity.

 

[Weeb Labs]

I have just finished capturing all of the requisite data for a CEA-2034A along with other measurements and will be publishing a full review shortly. Most likely this evening or tomorrow.

Here's a preview. There are a few things which complicate measurement for this type of driver.

 

[Rednaxela]

Nice!

They could use a little EQ but DI doesn’t look too bad.

 

[Weeb Labs]

Nice!

They could use a little EQ but DI doesn’t look too bad.

The 10KHz on-axis dip and reduced DI at that point is primarily being caused by the grille.

 

[Multicore]

Congratulations! Very well done. I think this should be promoted to home page, @amirm

I would suggest high pass on the voice. It sounds like you've got quite a proximity effect going on. I use this free plugin called Nova for spoken voice processing. It can do a lot more than correct proximity effect and I didn't notice anything else to comment on in your audio so you could use something simpler.

 

[Weeb Labs]

Here's a crazy idea. Take one of these radar occupancy sensors with support for 30 zones and use it in conjunction with an ADAU1701, CamillaDSP or similar to trigger various indexed PEQs or IRs depending upon one's position in the room. The FP2 is very responsive.

The result is listener tracking room correction. If there is more than one pair of speakers in the room (desk and couch for example), we could even switch between them.

 

[somebodyelse]

One step beyond having it follow you from room to room - I assume that's already possible given various home automation servers supporting at least LMS and presumablyother multi-room audio systems.

 

[MCH]

Here's a crazy idea. Take one of these radar occupancy sensors with support for 30 zones and use it in conjunction with an ADAU1701, CamillaDSP or similar to trigger various indexed PEQs or IRs depending upon one's position in the room. The FP2 is very responsive.

The result is listener tracking room correction. If there is more than one pair of speakers in the room (desk and couch for example), we could even switch between them.

I would imagine that the change would have to be gradual, otherwise it would be very noticeable and I bet weird sounding. Can the DSP tools you are considering do gradual EQ changes?
And what about audiophiles that don't live alone? Do we have to move around the house like a herd or the rest of the household has to deal with strange sound effects?

 

[Weeb Labs]

I would imagine that the change would have to be gradual, otherwise it would be very noticeable and I bet weird sounding. Can the DSP tools you are considering do gradual EQ changes?
And what about audiophiles that don't live alone? Do we have to move around the house like a herd or the rest of the household has to deal with strange sound effects?

The idea is not to implement something that would follow the listener around and gradually adjust the correction but rather to demark specific positions around the room which the listener frequently occupies and apply the relevant filters when they are entered.

This would not be particularly useful for multiple listeners but as the radar sensor is capable of tracking multiple listeners, we could either switch filters or disable the correction altogether when more than one listener is detected.

 

[MCH]

Hey @Weeb Labs I am thinking on doing something with the SRC4192, or maybe the 93. The objective is to have several sources at different sample rates and switch between them and have a SRC4192 to output a constant sample rate to a rpi (I will be using camilladsp).
How did your board work for you? Anything to complain about? Did you ever do source switching or sample rate changes on the fly? Does it react well? Did you use tdm with it? Any feedback would be appreciated, thanks!

 

[Weeb Labs]

Hey @Weeb Labs I am thinking on doing something with the SRC4192, or maybe the 93. The objective is to have several sources at different sample rates and switch between them and have a SRC4192 to output a constant sample rate to a rpi (I will be using camilladsp).
How did your board work for you? Anything to complain about? Did you ever do source switching or sample rate changes on the fly? Does it react well? Did you use tdm with it? Any feedback would be appreciated, thanks!

Not sure how I missed this post but I hope you managed to accomplish your goal. The SRC4192 did work well for me and responded gracefully to sample rate changes. I did not test it with TDM.

My current project is a new board called Prism, which takes an 8-channel 48KHz ADAT input (from an audio interface or otherwise) and outputs four (4x2ch) SPDIFs. Here's some very early stage schematic capture.

It can either perform clock recovery from the ADAT input (relatively high jitter as always) or take a word clock input from the master device. The ADAT decoder is a Coolaudio V1402 and the SPDIF TX are DIT4096s. BOM will be about €18 at small (<10) volume.

I have been experimenting with some uncommon approaches to DSP (write-ups/videos soon) and wanted the ability to make use of the two ADAT outputs on my audio interface but was disappointed to find that ADAT to 4xSPDIF is not a product that exists on the market. So, I am designing this first and foremost for my own purposes but will be open sourcing it.

 

[MCH]

Not sure how I missed this post but I hope you managed to accomplish your goal. The SRC4192 did work well for me and responded gracefully to sample rate changes. I did not test it with TDM.

My current project is a new board called Prism, which takes an 8-channel 48KHz ADAT input (from an audio interface or otherwise) and outputs four (4x2ch) SPDIFs. Here's some very early stage schematic capture.

View attachment 453321

It can either perform clock recovery from the ADAT input (relatively high jitter as always) or take a word clock input from the master device. The ADAT decoder is a Coolaudio V1402 and the SPDIF TX are DIT4096s. BOM will be about €18 at small (<10) volume.

I have been experimenting with some uncommon approaches to DSP (write-ups/videos soon) and wanted the ability to make use of the two ADAT outputs on my audio interface but was disappointed to find that ADAT to 4xSPDIF is not a product that exists on the market. So, I am designing this first and foremost for my own purposes but will be open sourcing it.

Cool project!
I never finished the SRC4192 project because I couldn't get the ICs cheap. Now I have seen them at 4-5 euros, but x4 is still quite some money for what it is. I do have one stereo board that I never find time to test.

Where do you get the v1402 cheap? I see 16$ in their website and the AL1402 from Ali are ca. 8 eur min.
You probably have seen this but in case:

https://www.xmos.com/download/AN01004:-ADAT---Replacing-Wavefront-AL1401-402-with-xCORE(1.0.1rc2).pdf/

 

[Weeb Labs]

Cool project!
I never finished the SRC4192 project because I couldn't get the ICs cheap. Now I have seen them at 4-5 euros, but x4 is still quite some money for what it is. I do have one stereo board that I never find time to test.

Where do you get the v1402 cheap? I see 16$ in their website and the AL1402 from Ali are ca. 8 eur min.
You probably have seen this but in case:

https://www.xmos.com/download/AN01004:-ADAT---Replacing-Wavefront-AL1401-402-with-xCORE(1.0.1rc2).pdf/

The price on Coolaudio's website is $7.65 (€6.50) for me. I did see the XMOS alternative but decided that I wanted to use something less fussy, with hardware control.

 

[Weeb Labs]

Further progress on the design. I decided to give this board a properly buffered word clock loop out, as the additional expense is minimal and I will no doubt find it useful in the event that I end up with one of these boards on each ADAT output. Nothing particularly fancy for the SPDIF outputs; just AC coupled with a divider to bring the 5V Vio TTL down to 0.5Vpp. At 5mA drive, it's well under the 30mA source capacity of the DIR4096. Routing soon!

 

[MCH]

I am using lmk1c110x in a project I am doing. It multiplexes the clocks x times.

 

[Weeb Labs]

I am finally ready to call the schematic complete. I had considered the LMK1C110X but decided that I wanted to stick with parts readily available from LCSC to streamline the workflow somewhat. Only the V1402 needs to be acquired independently (and I already have a stock of those).

This was originally intended to be bus powered but I ultimately decided to use a standard 12V input.

Time to route.

 

[mikeylemur]

I picked up a pair of Cambridge Audio Minx MIN 22s this evening and they have been a rather pleasant surprise. Much like the MIN 12s, these are 2.25" BMR drivers with a claimed 180 degree directivity.

Note that despite appearances, the lower driver is not a BMR; it is simply a flat coned midwoofer selected to match the appearance of the BMR above it.


Here is a quick in-room measurement on my desk.

We have reflections around 600Hz and modes below that, both of which can be ignored. I paid EUR 90 for this pair and this is rather impressive performance. I will take these outside tomorrow and capture some anechoic measurements, which I anticipate will be rather flat.

Let's take a look at directivity. These measurements are quite low resolution but directivity looks rather good. It is smooth and off-axis response doesn't drop below -8dB until we are a whopping 90 degrees (cyan line) off axis! Not unexpected for a BMR driver.

Breakup doesn't set in until just above 16KHz, which encompasses the vast majority of the audible band.


Then we have distortion at 76dB/1M. First as a percentage.

And then relative to signal.

This is really quite impressive. We are reaching down to -60dB at 8KHz.

Finally, some nearfield driver measurements. Green is the midwoofer and orange is the BMR.

Capturing meaningful nearfield measurements of these drivers is quite tricky, as the BMR's directivity is extremely wide and it is in very close proximity to the midwoofer.

On the whole, I find these little speakers to be rather impressive. Paired with a cheap subwoofer and given appropriate correction, they would make a lovely little desktop monitor.

Do you still have the graphs for these measurements, the images seem to be down for me