Hi,
First of all, thanks Mabat for a heroic effort and a fantastic tool and thread!
I'm really interested in your PWT efforts and i have a bunch of microphones that i can donate if you are interested. I have 6 BSWA MPA416 microphones, they are 1/4" ICP units ( BSWA Technology Co., Ltd.-Productions ). They are by no means perfect but should be quite usable
.
BR,
Anders
First of all, thanks Mabat for a heroic effort and a fantastic tool and thread!
I'm really interested in your PWT efforts and i have a bunch of microphones that i can donate if you are interested. I have 6 BSWA MPA416 microphones, they are 1/4" ICP units ( BSWA Technology Co., Ltd.-Productions ). They are by no means perfect but should be quite usable
.BR,
Anders
Hi Anders,
what do you mean by donating them? This is still pretty expensive stuff
At the moment I have six Panasonics WM-61A that I bought years ago when I though it would be a good idea to have them around. I should have bought much more, they were dirt cheap. Of course I would be more than happy to use better mics, especially if noise could be reduced as this will be probably much more critical than I anticipaded.
I really consider bying a lot more of these (cheap) 6 mm electrets and make a dense array along the tube by simply fixing them into the wall, starting right next to the driver. But let's see first how far I can get with the setup I have...
what do you mean by donating them? This is still pretty expensive stuff
At the moment I have six Panasonics WM-61A that I bought years ago when I though it would be a good idea to have them around. I should have bought much more, they were dirt cheap. Of course I would be more than happy to use better mics, especially if noise could be reduced as this will be probably much more critical than I anticipaded.
I really consider bying a lot more of these (cheap) 6 mm electrets and make a dense array along the tube by simply fixing them into the wall, starting right next to the driver. But let's see first how far I can get with the setup I have...
These looked nice... staring to get the itch... but the sensitivity? And what size driver and kind?
Einhorn | Martion
//
but the sensitivity? And what size driver and kind?Einhorn | Martion
//
I don't know if there would be any real benefit, but you might consider using MEMS microphones. Compared to typical 6mm capsules, MEMS mics are smaller (while having similar self noise levels), have a higher overload point, and are much more consistent unit-to-unit in frequency response and sensitivity. The Knowles SPM0687LR5H-1 or the TDK InvenSense ICS-40730 would probably work well.
I used ATH 4 to create a waveguide. Once the software is set up it's extremely easy to create the waveguides. ABEC software was a little difficult to find, however. I wonder why they stopped distributing it.
I only have two problems now:
1) I don't understand anything about waveguides or how to properly design them.
2) I haven't figured out how to work with STL files so I can make templates for tools to DIY the waveguides.
I wish there was a DIY waveguide/horn sticky like the crossover sticky Allen B made. He explained the basic issues of crossovers and how to design and measure them.
Anyway, the software itself is very easy to use.
I only have two problems now:
1) I don't understand anything about waveguides or how to properly design them.
2) I haven't figured out how to work with STL files so I can make templates for tools to DIY the waveguides.
I wish there was a DIY waveguide/horn sticky like the crossover sticky Allen B made. He explained the basic issues of crossovers and how to design and measure them.
Anyway, the software itself is very easy to use.
Attachments
All true, except you left out one thing. They have very high internal noise levels. (who cares for cell phones and the like.) Trust me, I know because I was Director of Research at Knowes when the first MEMs mics were being developed. In microphones there is no way around smaller size = higher noise.
some things to consider:
1) the angle of the waveguide walls largely determine the coverage angle. IE, if the walls are 90 x 90, you'll get a wavefront that's (mostly) 90 x 90.
2) The diameter of the waveguide determines how low it will work. For instance, 1khz is 13.5" long. So if you have a waveguide that's 13.5" in diameter, it will control the wavefront down to about 1khz.
It's surprisingly simple! But like all things, The Devil is in the Details.
The claimed self noise for the Knowles SPM0687LR5H-1 (datasheet here) is 24dB(A) and the TDK InvenSense ICS-40730 (datasheet here) is 20dB(A). For comparison, the 6mm Panasonic WM-61A has an SNR of "More than 62dB" according to the datasheet (ref 1Pa, A-weighted), so the self noise is less than 32dB(A) (typical is not given). The lowest noise 6mm capsule at I could find at digikey (datasheet here) has a self noise spec of 20dB(A). The Earthworks M series measurement mics (1/4in) have a self noise spec of 20dB(A).
its just shot noise. air particles are just hitting the membrane causing impulse noise. if you have a big diaphram, this averages out more, leaving you with less impulses (noise) and more signal (average pressure) same logic in reverse for a small diaphram.
Thanks. That's a big help. I try to read your posts on this topic and Earl Geddes'. The two things you just clarified are basics I didn't understand.
I was just reading this thread today where Earl Geddes shared information on wave guides.
It seems like one of the least easy to make devices in DIY home audio.
Here's a couple of silly anecdotes you might enjoy:
1) In 2005, I bought a plane ticket to Denver, so that I could listen to Earl's speakers. While I was at the RMAF audio show, I bought a copy of his book. I dug into the book that night, but couldn't find what I was looking for: the formula for an OS waveguide. The next day, I sheepishly mentioned that I couldn't find the formula. Turns out it's not in the book. Earl wrote it down for me on a post it note, which I still have in my copy in his book.
2) A big part of my day job is like that movie Office Space. Basically there's a really smart guy who understands something, and then there's a customer who pays for stuff. I often have to interface between the smart person and the customer. This may be why I'm semi decent at translating technical documents into something that's a little easier to digest for a broad audience.
1) In 2005, I bought a plane ticket to Denver, so that I could listen to Earl's speakers. While I was at the RMAF audio show, I bought a copy of his book. I dug into the book that night, but couldn't find what I was looking for: the formula for an OS waveguide. The next day, I sheepishly mentioned that I couldn't find the formula. Turns out it's not in the book. Earl wrote it down for me on a post it note, which I still have in my copy in his book.
2) A big part of my day job is like that movie Office Space. Basically there's a really smart guy who understands something, and then there's a customer who pays for stuff. I often have to interface between the smart person and the customer. This may be why I'm semi decent at translating technical documents into something that's a little easier to digest for a broad audience.
In regards to the difficulty of waveguide design:
A waveguide is basically a cone.
We have an entrance at one end (the throat) and an exit at another (the mouth.)
If it was perfectly conical (like a cone) that wouldn't be ideal, because there would be a mismatch at the entrance and the exit. That's where the oblate pheroidal stuff comes in; we have to match the curve of the waveguide, and the device that's driving it. That's why the exit angle of the compression driver is so important, and why the wavefront shape matters a lot.
Once all that makes sense, you'll start seeing it apply to all kinds of strange things in the world. For instance, I saw the same ideas pop up in an article that I read about freeway offramps. Basically the freeway engineers figured out that there's exactly *one* shape that's optimal for matching the offramp of a freeway with the road that it's attached to.
A waveguide is basically a cone.
We have an entrance at one end (the throat) and an exit at another (the mouth.)
If it was perfectly conical (like a cone) that wouldn't be ideal, because there would be a mismatch at the entrance and the exit. That's where the oblate pheroidal stuff comes in; we have to match the curve of the waveguide, and the device that's driving it. That's why the exit angle of the compression driver is so important, and why the wavefront shape matters a lot.
Once all that makes sense, you'll start seeing it apply to all kinds of strange things in the world. For instance, I saw the same ideas pop up in an article that I read about freeway offramps. Basically the freeway engineers figured out that there's exactly *one* shape that's optimal for matching the offramp of a freeway with the road that it's attached to.
It depends on the math stability. Noise makes the math less stable.But as you say this can be improved with time averaging.
The noise is thermal noise (see Johnson–Nyquist noise - Wikipedia - Thermal noise on capacitors) As the condenser mic gets small the capacitance gets smaller and the noise goes up. It has always been fundamental physics that this has to be true.
As I said above, the limitations are physical laws. How and if the numbers you quote are true then I have no idea how they do it. When I was working at Knowles the best small hearing aid mics were 24 dB(A) which were best-in-class. The MEMs mics were about 40 dB(a), which matched theory almost exactly.
Using several MEMs mics on a single substrate is a big advantage (which I have heard they do,) but I really have to suspect the TDK spec of 20 dB(A) as it matches a very high quality Earthworks mic with a much larger diaphragm. Either Earthworks is incompetent (but those are very good numbers) or someone is fudging the data.
Last edited:
Hopefully the manufacturers aren't lying. How long ago was it that you were working at Knowles? It seems to me that there have been significant improvements in MEMS mics in just the last few years. I don't know much about the details, but I did find a relatively recent (2018) article (PDF) that talks a bit about SNR optimization (pages 9-12).
I know a bit about MEMS microphones having spent ~a dozen years working in MEMS and I would say that the 20 db(A) is definately attainable. It's just a question of cost. We were able to get to 20 db(A) several years ago, it was just more costly than the customers (mostly mobile phone guys) wanted to pay for.
Thanks for the tips. I ordered 25 of those capsules. I like the ease of mounting - with MEMS it would be more complicated I guess.
Hopefully it works out. The acoustic overload point isn't terribly high for those capsules (claimed 110dBSPL <3%THD), but maybe it doesn't matter.
Yeah, the bottom port MEMS mics would basically require PCBs for this application. You can manually solder wires to them (I've done this for a measurement mic), but mounting would be difficult.
Top port mics might be easier to work with, but they tend to have lower performance (especially at low frequencies, but I guess that probably doesn't matter for this application). The TDK InvenSense ICS-40619 might be workable. Its self noise isn't as low (27dB(A)) as the 6mm capsule, but it actually has significantly better dynamic range due to its much higher acoustic overload point (127dBSPL for 3%THD and 132dBSPL for 10%THD).
Yeah, the bottom port MEMS mics would basically require PCBs for this application. You can manually solder wires to them (I've done this for a measurement mic), but mounting would be difficult.
Top port mics might be easier to work with, but they tend to have lower performance (especially at low frequencies, but I guess that probably doesn't matter for this application). The TDK InvenSense ICS-40619 might be workable. Its self noise isn't as low (27dB(A)) as the 6mm capsule, but it actually has significantly better dynamic range due to its much higher acoustic overload point (127dBSPL for 3%THD and 132dBSPL for 10%THD).