Like a lot of people on this forum I want a quality measurement microphone but can't really justify a full professional set up.
B&K (or clone) external polarisation capsule + proprietary preamp body + external power source is the laboratory reference option but is expensive, inflexible and inconvenient (not easily portable).
But the cheap Electret Condenser Mics tend to distort and lack stability over time.
So what are the options in the middle?
1. Use a B&K style capsule connected to a 48V Phantom powered body to provide the 200 V polarisation and the preamp (actually an impedance buffer)
Gefell and Josephson make such bodies but they are expensive and would be somewhat difficult to DIY.
[Edit] So it's a nice option if you have the money to buy ready-made "top of the line" hardware.
2. Use a B&K style pre-polarised capsule connected to a P48 V body that only has to do impedance buffer duties.
These capsules are fairly common and almost a commodity item, made by B&K, DPA, G.R.A.S., Gefell, Norsonic, ACO Pacific and PCB Piezotronics, so there's a reasonable chance to find one at a reasonable price.
This is a simpler system and naturally there are more commercial options.
ACO Pacific, and PCB Piezotronics do "measurement mics" and DPA market their's as an omnidirectional, flat response "studio mic".
MicW is the studio mic division of BSWA (like DPA to B&K) and claims to have a decent measurement mic.
These are less expensive but still not cheap ~$500 for the body.
I have had a look at some of the JFET buffer circuits from Borbely and Bob Cordell and it seems it shouldn't be too expensive.
[Edit] So a DIY kit could be almost "top of the line" and more than adequate but far cheaper.
Since option 2 is my choice - has anybody seen a DIY "preamp" body kit or have ideas for the circuit?
I am tempted to try a fully complementary circuit but a kit makes sense, if only to simplify the mechanics, so I don't have to cut odd sized capsule threads and find Teflon insulators and similar obscure parts.
David
B&K (or clone) external polarisation capsule + proprietary preamp body + external power source is the laboratory reference option but is expensive, inflexible and inconvenient (not easily portable).
But the cheap Electret Condenser Mics tend to distort and lack stability over time.
So what are the options in the middle?
1. Use a B&K style capsule connected to a 48V Phantom powered body to provide the 200 V polarisation and the preamp (actually an impedance buffer)
Gefell and Josephson make such bodies but they are expensive and would be somewhat difficult to DIY.
[Edit] So it's a nice option if you have the money to buy ready-made "top of the line" hardware.
2. Use a B&K style pre-polarised capsule connected to a P48 V body that only has to do impedance buffer duties.
These capsules are fairly common and almost a commodity item, made by B&K, DPA, G.R.A.S., Gefell, Norsonic, ACO Pacific and PCB Piezotronics, so there's a reasonable chance to find one at a reasonable price.
This is a simpler system and naturally there are more commercial options.
ACO Pacific, and PCB Piezotronics do "measurement mics" and DPA market their's as an omnidirectional, flat response "studio mic".
MicW is the studio mic division of BSWA (like DPA to B&K) and claims to have a decent measurement mic.
These are less expensive but still not cheap ~$500 for the body.
I have had a look at some of the JFET buffer circuits from Borbely and Bob Cordell and it seems it shouldn't be too expensive.
[Edit] So a DIY kit could be almost "top of the line" and more than adequate but far cheaper.
Since option 2 is my choice - has anybody seen a DIY "preamp" body kit or have ideas for the circuit?
I am tempted to try a fully complementary circuit but a kit makes sense, if only to simplify the mechanics, so I don't have to cut odd sized capsule threads and find Teflon insulators and similar obscure parts.
David
Last edited:
Why would you opt for a 200V system? What are the mean pressure levels you want to measure? The 200V polarized capsules are perfectly suited for small (acoustic) signal and/or high signal to noise. You'd need that when doing transmission tests in an acoustics lab. In a lot of situations there is no need for that though. Simply because the background levels are so high that the measurement doesn't justify the S/N ratio of the system.
My 2ct: unless you have an acoustics lab, opt for a traditional 48V setup. Back electrets have proven to be stable in time, more rugged (don't ever drop a 200V capsule) and comply to the measurement praxis.
[Edit]Distortion hardly is an issue either (to me at least). I missed BSWA in your list of manufacturers. No personal experience with them, but they seem a viable budget option. Don't know if you have access to them though.
[Edit2]added a product sheet
My 2ct: unless you have an acoustics lab, opt for a traditional 48V setup. Back electrets have proven to be stable in time, more rugged (don't ever drop a 200V capsule) and comply to the measurement praxis.
[Edit]Distortion hardly is an issue either (to me at least). I missed BSWA in your list of manufacturers. No personal experience with them, but they seem a viable budget option. Don't know if you have access to them though.
[Edit2]added a product sheet
Last edited:
I don't intend to, my choice is option 2, the P48 system, I included the 200 V system for context as the reference standard.
Also to introduce the P48 to 200 V option, I only just learned of this and it seems to be little known, so it seemed worthwhile to mention it as a possibility for those who need it.
I will edit the post to make it clearer.
I have only just become aware of them and I don't know their product line well but I couldn't find a suitable P48 body.
[Edit] None of the ones on your sheet seem to be P48.
Not sure about their availability in Australia, certainly less than B&K!
Thanks for the reminder
Best wishes
David
Last edited:
Dave Z, I'm not sure about your particular needs of measurement.
B&K and GRAS are my prefered mic's to use in measurement systems. The older B&K measuring amplifiers are flexible for 1/2" & 1" sizes in either 0V, 28V or 200V. You can expand this way and I still use the B&K 2636 / B&K 1617 setup for 1/3 octave measurements. The meter scales are interchangeable for different mic sensitivities, including direct input.
B&K and GRAS are my prefered mic's to use in measurement systems. The older B&K measuring amplifiers are flexible for 1/2" & 1" sizes in either 0V, 28V or 200V. You can expand this way and I still use the B&K 2636 / B&K 1617 setup for 1/3 octave measurements. The meter scales are interchangeable for different mic sensitivities, including direct input.
Hi BKnut (is that short for Brüel & Kjær nut?)
My measurement requirements are for speaker development and the usual room optimisation.
I would like to play around with some driver fabrication ideas so distortion as well as the usual frequency response needed if one just puts commercial drivers in a cabinet.
Nice to meet an Australian B&K enthusiast, a relative did postdoc work there so I've had an interest for many decades.
What are your own measurements for? speaker stuff is usually not 1/3 octave any more.
Do you happen to have a schematic of a B&K buffer/preamp by any chance?
I don't want to violate their IP but I would be curious to see their approach.
I have seen a dismantled body and it looked fairly simple.
Presumably the later TEDS bodies would be more complex.
Best wishes
David
Last edited:
Now I won’t stop you from building your own preamp, but I’m inclined to think that a pretty straightforward and cheap usb audio interface with decent preamps and an entry level class 2 microphone (with correction sheet if you like) would do. Possibly an investment in an Earthworks or NTI class 1 mike would be justified but you’d still be under or around the cost of a B&K setup without the interface and the ease of use. And with pretty decent results.
There is a problem with a USB mic for my application.
My plan is to build a Klippel style Near Field Scanner that does two concentric scans around the speaker.
This makes it possible to separate out the outward sound waves and inward reflections and thus calculate anechoic response even in a normal room.
To achieve this nice trick requires absolute phase and delay but the inbuilt clock of the USB interface creates synchronisation problems.
This requirement is also why I want better stability than is usually needed for simple DIY measurements.
The Earthworks or NTI are a fallback option but I have found that BSWA make (made?) a P48 powered prepolarised measurement mic that is not quite reference standard but still pretty respectable.
And the price is not bad, save a lot of work on a preamp project.
I am very tempted. I think you have talked me into it😉
Best wishes
David
There was a pretty extensive thread in DIYaudio that you can find read.
My contribution was mainly the maths.
The OP was unable to continue so I put it on the back-burner but I haven't dropped the idea.
There is now also some excellent work over at AudioScience review by "NTK".
So the maths and software are fairly well understood.
Now I need a mike and have to build the damn structure.
I will start a thread on the construction when (if) I have some results to show.
Best wishes
David
OK, that wasn't clear to me. If you're into building the software anyway, you might have considered the UMIK-X arrays from MiniDSP, have you? It's not my forte, but my guess is that with those arrays a practical variant of the Klippel method would be far easier to accomplish.
Furthermore, since traditional gated measurements do fine from 200-300Hz upwards, you'd need the NF scanner primarily for LF. I can't help thinking phase/timing issues in cheap sound cards aren't very problematic in that frequency range, especially since you could work with a reference (two channel measurement) and physical errors (varying mike to source distance) are more likely to screw results up. But maybe I'm wrong there.
Furthermore, since traditional gated measurements do fine from 200-300Hz upwards, you'd need the NF scanner primarily for LF. I can't help thinking phase/timing issues in cheap sound cards aren't very problematic in that frequency range, especially since you could work with a reference (two channel measurement) and physical errors (varying mike to source distance) are more likely to screw results up. But maybe I'm wrong there.
I can probably provide more details on air condenser measurement Mike's than most here. I have 7 different B&K' s and others along with different preamp and docs. If you need long term stability you need the metal diaphragm. PM me for scans of docs.
Otherwise the little electrets are great up to around 30 KHz and way cheaper. And way easier to implement. You may be able to make an array of them and not need any mechanical stuff to move the mikes by scanning electronically. Or go nuts with a 16 channel recording interface collecting it all in one pass.
Otherwise the little electrets are great up to around 30 KHz and way cheaper. And way easier to implement. You may be able to make an array of them and not need any mechanical stuff to move the mikes by scanning electronically. Or go nuts with a 16 channel recording interface collecting it all in one pass.
Hi Demian, I had hoped you would show up, which ones do you have?
My interest is in a 1/2" free field for speakers and eventually a 1/4" pressure mic for plane wave tube use.
The 1/2" preamp can be adapted to fit a 1/4" capsule but I suspect the authentic B&K adaptor will cost about as much as a whole measurement mic from a less exclusive company, and the threads sizes are a bit unusual so not easy to find, any hints?
I am not sure if I really need the stability of a metal condensor mic. but it would reassure me.
A Klippel style scan is typically ~1000 measurements, each of a few seconds, so around an hour.
So, plenty of time for variation - temperature and humidity can be controlled indoors but pressure may vary a little. Outdoor measurements obviously more variable, probably required in the early phase while I work on room reflection removal.
Anyway, I don't mind an excuse to justify the acquisition of a nicer microphone.😉
My PMs are limited but my Email address is in my profile.
Could I ask for a copy of a preamp schematic?
I initially planned to use the famous Panasonic capsule but they became unavailable and the quality control of the replacements is a bit uncertain.
I believe the Primos are decent but I have more or decided on a BSWA capsule, it's calibrated and has the stability of a metal membrane.
We need ~1000 measurements, I don't think an array of that size is very practical.
A 16 channel (or better still 18/19 for 10° increments over a half circle) would mean a lot of ADCs and a multichannel recorder.
And it would still have to be scanned around one axis.
There's an outfit who do this commercially, where time is money and all, but it's a lot of extra trouble for a hobbyist who does the occasional test.
Best wishes and thanks for the offer
David
Last edited:
As I indicated in response to Demian, I don't see this would be practical but if you have an idea I would be happy to look at it.
Yes, the Kilppel does this, actually couple of octaves above 300 Hz but same principle.
I plan to do the same.
Two channels would, of course, mean a extra mic and preamp and complexity.
REW does offer this option, precisely because they found USB sound-cards problematic, but the Vituix developer says it's just a pain and won't support it.
I want to avoid the whole issue if I can, and stick with one mic, just like Klippel.
I don't think their work is above criticism but I am inclined to think they have this bit correct.
Best wishes
David
Last edited:
The mic capsule I use is a Primo EM258. It has pins and is meant to mount to a PCB. The maximum thickness of the PCB should be under .062”.
https://primomic.com/pdf/EM258.pdf
Primo prefers to sell direct only. The standard package is 500 pieces. They make them to order so lead time is a few weeks or so.
FYI last I knew Bose used a curved holder with I believe 18 microphones mounted on it. So they just have to step the loudspeaker around. Audioxpress used to sell a turntable for such measurements. Typical spacing is of course 5 degrees. So with 18 microphones and 36 steps you should be able to get a standard polar plot. Of course you should be able to do at least 28 different frequency bands with great accuracy and a bit less for the other 4.
If you really go with multiple microphones, 648 would be enough. I suspect the standard CMOS 4051 analog selector ICs would allow multiplexing of the capsules. All you need is space and a budget of around $2,000.00.
https://primomic.com/pdf/EM258.pdf
Primo prefers to sell direct only. The standard package is 500 pieces. They make them to order so lead time is a few weeks or so.
FYI last I knew Bose used a curved holder with I believe 18 microphones mounted on it. So they just have to step the loudspeaker around. Audioxpress used to sell a turntable for such measurements. Typical spacing is of course 5 degrees. So with 18 microphones and 36 steps you should be able to get a standard polar plot. Of course you should be able to do at least 28 different frequency bands with great accuracy and a bit less for the other 4.
If you really go with multiple microphones, 648 would be enough. I suspect the standard CMOS 4051 analog selector ICs would allow multiplexing of the capsules. All you need is space and a budget of around $2,000.00.
Last edited:
I don't quite understand yet why an USB audio interface would present timing issues. In traditional 2ch measurement you use one channel for reference at the speaker input. Thus an absolute timebase reference is possible. That trick would work also with more elaborate multiple input audio interfaces. Be it your sampling rate presents you with a limit, but with 192k that issue also would be out of the way.
RE microphones.
Per request I have:
B&K
4136 1/4"
4135 1/4"
4134 1/2"
4133 1/2"
4165 1/2"
4145 (2X) 1"
probe microphone
multiple preamps
multiple power supplies
Multiple measuring amps.
Also HP 1/2" and 1" + preamps
And GR preamp.
If you want one of these microphones I suggest shopping for a GR 1560-P42 which has an internal bias generator for the capsule and a 10X preamp internal and interfaces through an XLR connector (non-standard pinout). its the easiest to deploy. All it needs is a simple 20V power supply.
RE USB- USB as an interface if all the audio passes through a single interface its all fine. However if part (e.g. source) does not then you have latency issues since the timing over the USB link is not predictable. When measuring phase you need the samples to be simultaneous. Some early soundcards uses a single ADC for both channels and there was always a phase shift between channels. That has passed fortunately.
You can get up to 30 channels in some of the pro interfaces. You may need to make a number of preamps to support that.
For the Klippel type device what angle step for each measurement? Were you planning to build the software in Matlab or Octave? You can get support in Linux for these multichannel interfaces and a really clean path for the signals.
Outline a little more of what is needed in acquisition to support the analysis.
Per request I have:
B&K
4136 1/4"
4135 1/4"
4134 1/2"
4133 1/2"
4165 1/2"
4145 (2X) 1"
probe microphone
multiple preamps
multiple power supplies
Multiple measuring amps.
Also HP 1/2" and 1" + preamps
And GR preamp.
If you want one of these microphones I suggest shopping for a GR 1560-P42 which has an internal bias generator for the capsule and a 10X preamp internal and interfaces through an XLR connector (non-standard pinout). its the easiest to deploy. All it needs is a simple 20V power supply.
RE USB- USB as an interface if all the audio passes through a single interface its all fine. However if part (e.g. source) does not then you have latency issues since the timing over the USB link is not predictable. When measuring phase you need the samples to be simultaneous. Some early soundcards uses a single ADC for both channels and there was always a phase shift between channels. That has passed fortunately.
You can get up to 30 channels in some of the pro interfaces. You may need to make a number of preamps to support that.
For the Klippel type device what angle step for each measurement? Were you planning to build the software in Matlab or Octave? You can get support in Linux for these multichannel interfaces and a really clean path for the signals.
Outline a little more of what is needed in acquisition to support the analysis.
BSWA (not a really good name for a tech company). I looked at the web side and realized I had used some of their microphones in a headphone measurement system. Not bad. However they have a "revolutionary" speaker with a Rubidium magnet CA918 Free-field Sound Source_BSWA Technology Co., Ltd. Having never heard of this I looked and cannot find any reference to Rubidium for magnets.
I just checked and someone is selling a handful of the GR preamps on eBay cheap. I would get another but just no need: GR GenRad General Radio 1560-P42 Preamplifier | eBay
Bill Waslo found these first and worked up an improvised connector for the interface on the back.
Bill Waslo found these first and worked up an improvised connector for the interface on the back.
I only learned of them recently, they seem to be about 75% of the quality of B&K - but they are only about 25% of the price so the cost/benefit is pretty decent.
That reference looks totally spurious., the physics of a Rubidium magnet makes no sense.
I could, of course, be mistaken - because magnetism is a remarkably difficult quantum effect with no obvious intuitive explanation.
But I'd bet money it's a translation error or salesman's nonsense, or some such.
More extended response to your other comments soon as I have time for a considered reply.
Best wishes
David
I was a bit surprised by this myself.
As I understand, it's not the USB audio interface but rather when the mic is a USB mic with it's own clock and latency, not synchronised with the computer audio interface.
That is consistent with Demian's comment, but I use an internal sound card so I have no personal experience yet.
I just want to avoid even to have to think about it.
Nice collection, no pre-polarised I notice.
I was unaware of that GenRad product, they are rare in Australia, nice to learn about and looks excellent
At that price I would be inclined to buy it just on principle but the one you linked won't ship to here.
So I will probably just buy a BSWA and phantom power pre-amp.
That is actually a tricky question, there's issues of optimisation of the accuracy where it's needed, on axis, versus an even spread of points.
And steps near the "pole" are smaller than steps near the "pole" for the same number of radians or °.
So we should work in steradians but most people have no mental model for that.
But think in the order of 10°, or ~1000 points in 2 concentric spheres.
Yes, Octave
NOT interested, for reasons already mentioned.😉
Basically, first step is to measure the response of the speaker at many points in a full sphere.
This allows calculation of the CTA (AKA CEA) 2034 frequency response polars and so-called "spinorama" metrics.
If the measurements are repeated on a second, concentric sphere then the outward sounds and inward reflections can be separated and we can achieve anechoic chamber results (or better) in a normal room.
Plan is a 2 axis pivoted scanner with one measurement microphone.
Phase one is manual steps to create CTA 2034 data.
Next phase is replace pivots with stepper motors, automate the process.
The data can be time windowed and done outdoors to minimise the effects of reflections
Final phase is option to move the microphone to a different radius, repeat the scan and separate the sound fields.
No 648 channels required, no ~$2,000 and boxes of 500 microphones😉
Just simple and incremental.
Best wishes
David
Last edited: