LDC capsule self-noise?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

tynkerd

Member
Joined
Oct 17, 2018
Messages
5
So, looking again at those AliExpress cheap LDC capsules, I noticed they state an equivalent noise of 15dBA.

Does anyone know what kinds of self noise a capsule itself introduces? I can't believe these capsules can be that noisy...

For reference, the entire Rode NT1-A mic ( praised for it's low nose floor) is only 5dBA...capsule and circuitry combined!

Is this just standard Chinese sellers "label everything, even if it's wrong" mentality?
 
A capsule, being a pure reactance (capacitance) has no noise at all. If the capsule also has a built in impedance converter, like most electrets, then that amplifier will have noise, but I don't think that's common for large diaphragm capsules.

It might be a typical value when used with a certain FET, bias resistor, and polarizing voltage. It could also just be some number picked out of thin air to make the buyer happier and the product seem more "high tech" and well constructed. I would not pay much attention to it personally. And, 15dBA isn't all that noisy anyway - a KM 84 is around 22dBA and you can make fine recordings with it. There aren't many (any?) recording studios these days that even get close to the 20s, so don't get too obsessed.
 
Monte McGuire said:
A capsule, being a pure reactance (capacitance) has no noise at all.
Indeed, in itself, it has no electrical noise. However, that is not all.
Due to the acoustic radiation impedance, the capsule picks up noise. Another way of looking at it is that the air molecules, agitated by brownian movement, hit the diaphragm. The resulting noise level depends on the diaphragm area (and other physical conditions, such as temperature).
Because this noise is erratic, it increases according to the square root of the area, when the signal increases linearly with the area, resulting in an increase of 3dB S/N with each doubling of the area. So LDC's have intrinsically a better S/N than SDC's.
Now there is another very significant factor, the diaphragm damping. Damping is necessary to linearize the frequency response, which would be a huge midrange peak without it. This damping reduces the sensitivity.
Another technique has been developed, which consists in applying less damping and compensating with electronic EQ. This allows reducing the intrinsic noise to sub-zero values.
 
As pointed out above, 15dB(A) isn't very loud and is a pretty typical range for a microphone capsule to be spec'ed at.  For example, a standard 1/2" measurement microphone from B&K would also have around a 15dB(A) noise spec (which is the capsule itself and doesn't include the electronics that follow which would bump the total spec up to ~18-19dBA). 

The noise spec of the Rode NT1a is certainly impressive at first glance.  However, I'm not sure how they achieve that.  My best guess would be that they actually attenuate the output of the whole microphone to bring down the overall level (including the noise).  This is just a guess.  That mic's other specs (if they are correct as listed) imply this is the case though.  The max output level is rated at 137 dBA but they don't say what the distortion level is at that SPL.  1-3% would be typical for a max SPL spec but their sensitivity and max voltage output spec indicate that the max output for 1% THD actually happens in the neighborhood of 88dB SPL at the mic which is less impressive and pretty low.  Maybe the listed specs aren't all correct though (wouldn't be the first time microphone specs were labeled/listed wrong). 
 
spica said:
My best guess would be that they actually attenuate the output of the whole microphone to bring down the overall level (including the noise).  This is just a guess. 
That is a wrong guess. Equivalent noise is another way of expressing nominal S/N ratio. Decreasing sensitivity decreases both Signal and noise.
The schemo does not indicate any tricks in the electronics.
 
Let's compare NT1a to current NT1. Main topology is truly similar between both, just with slightly modfications in new version.

Sensitivity:
-31.9dB re 1V/Pa (25mV/Pa) vs. -29 dB re 1V/Pa (35mV/Pa)
Equivalent Noise:
5dBA SPL  vs. 4.5 dBA SPL (as per IEC651)
Dynamic Range:
>132dB vs. 128 dB SPL
Maximum SPL:
137dB SPL vs. 132 dB SPL
Signal / Noise:
88dB SPL vs. 90 dBA SPL (as per IEC651)
Maximum Output:
+13.7dBu vs. +8 dBu (1kHz, 1% THD into 1K Ω load)
Polarisation voltage:
75V vs. 65V
 
abbey road d enfer said:
That is a wrong guess. Equivalent noise is another way of expressing nominal S/N ratio. Decreasing sensitivity decreases both Signal and noise.
The schemo does not indicate any tricks in the electronics.

Probably shouldn't have guessed.  Admittedly (and verified as) a bad guess.  It was largely based on the Max output spec I saw on the Rode website which may have been intended to be listed in volts and not millivolts.  http://www.rode.com/microphones/nt1-a

The listed spec is 13.7mV (or ~88 dB SPL based on a sensitivity of 25mV/Pa)... 13.7 Volts would make more sense as it would equal about 129dB. 

Note: just noticing ln76d's post noting a max output of 13.7dBu.  That seems more likey.  Seems to be a type-o either way. 


 
ln76d said:
Let's compare NT1a to current NT1. Main topology is truly similar between both, just with slightly modfications in new version.

Sensitivity:
-31.9dB re 1V/Pa (25mV/Pa) vs. -29 dB re 1V/Pa (35mV/Pa)
Equivalent Noise:
5dBA SPL  vs. 4.5 dBA SPL (as per IEC651)
Dynamic Range:
>132dB vs. 128 dB SPL
Maximum SPL:
137dB SPL vs. 132 dB SPL
Signal / Noise:
88dB SPL vs. 90 dBA SPL (as per IEC651)
Maximum Output:
+13.7dBu vs. +8 dBu (1kHz, 1% THD into 1K Ω load)
Polarisation voltage:
75V vs. 65V
The basic specs are pretty similar to a Neumann TLM49 or U47 FET, which publish an intrinsic noise that's about 10+dB higher. I think it can't be explained by a single factor, but rather by a combination of several:
-Higher polarization voltage, but that would account for about 1dB max
-Lower diaphragm damping, but that would show in the frequency response (the head amp is essentially linear)
-Acoustic enhancement, but that would show in the polar diagram
-Non-standard measurement procedure
- Data doctoring by mktg dept
-?
 
abbey road d enfer said:
Is that right? Do you have a tested and true schematic attesting to that?

Rode don't share schematics, i reworked lot of NT1a without it, tracing circuit. There was somewhere here on gdiy unofficial schematic, which practically fits 90% real circuit (if i remember correctly there was few marginal mistakes with components values).
Main difference (but not only) is in capacitors value at the output. NT1a use 47uF which cause fixed hpf, NT1use 100uF which gives almost flat response. Frequency responses in datasheets of course are somoothed and not 100% accurate, especially that k67 version of nt1a have even bigger roloff and hf freq bump due to transducer character, but even there are visible differences on graphs.  Polarisation voltage difference which i showed in previous post also took from my own measurements.
 
ln76d said:
Rode don't share schematics, i reworked lot of NT1a without it, tracing circuit. There was somewhere here on gdiy unofficial schematic, which practically fits 90% real circuit (if i remember correctly there was few marginal mistakes with components values).
Main difference (but not only) is in capacitors value at the output. NT1a use 47uF which cause fixed hpf, NT1use 100uF which gives almost flat response. Frequency responses in datasheets of course are somoothed and not 100% accurate, especially that k67 version of nt1a have even bigger roloff and hf freq bump due to transducer character, but even there are visible differences on graphs.
According to schematics that are flying around the web, both NT1 and NT1a have rather flat response.
What I'm talking about is large midrange dip that would compensate for an undamped diaphragm. Typically, that technique of reducing self-noise involves about 20dB midrange cut. I don't see anything in the schematics that would produce such a large EQ.
 
abbey road d enfer said:
According to schematics that are flying around the web, both NT1 and NT1a have rather flat response.
What I'm talking about is large midrange dip that would compensate for an undamped diaphragm. Typically, that technique of reducing self-noise involves about 20dB midrange cut. I don't see anything in the schematics that would produce such a large EQ.

I'm not writing about first NT1 at all, which was schoeps circuit adaptation by Jim Williams only about NT1a and NT1 (current version, also named NT1, black body finish). For the current revision i doubt there's even unofficial schematic over web, if there's any please share the link.
Electric measurement of NT1a circuit didn't showed me any midrange dip, but clearly showed low end rolloff. 
 
I've just done detailed noise measurements of a DIY variable pattern mike using a Chinese 'CK12' clone from Oceano99's 2011 group buy.  The circuit uses 2 FETs and the Schoeps output circuit rather than switching Vp to the capsule.

This is a very quiet mike; about 10dBA is my best estimate.

I can clearly see the effect of switching between omni, fig-8 & cardioid and the noise of the resistors this puts in the signal path. 

Also the '1G & capsule capacitance' noise and the effect of using 5G.

No effort was made to optimise the circuit for noise and I don't think the actual capsule acoustic damping noise is significant here.  capsule is somewhat underdamped

So it is certainly possible, with attention to detail, to make a mike with this capsule and a modified Schoeps circuit have sub 10dBA noise.

Eric Benjamin of the Yahoo MicBuilders group has detailed measurements of Rode NT1A including noise at 5.5dBA

My measurements of the DIY mike are on MicBuilders under Files/Mic Measurements/Zephyr.doc  You have to join.
 
capsule is somewhat underdamped
abbey road d enfer said:
Shouldn'it show some kind of midrange lift, then?
Yes.  Omni & Cardioid show a broad 2-3dB lift 2k5-16k Hz.  Probably sounds good

But the main fault in this capsule is Fig-8 which has 6-7dB peak at 5kHz and generally sloping up to HF.  This shows up in the 180 cardioid response too.

I think most of the 'clones' try (and generally manage) to get good cardioid.  Usually omni works out good too.  Where most of them fall down is getting good Fig-8

This little exercise has got me itchy to start designing capsules again.  8)

BTW, no one seems to have posted definitive responses of the U87 EQ.  My investigations circa jul 2011 suggest this would 'correct' the omni & cardioid responses of this capsule.  ChinaMod+U87hybrid.doc in my Yahoo MicBuilders Files.  You have to join.

Now I'm a pretend LTspice guru, maybe I should re-visit this  8)
 
ricardo said:
BTW, no one seems to have posted definitive responses of the U87 EQ.  My investigations circa jul 2011 suggest this would 'correct' the omni & cardioid responses of this capsule.  ChinaMod+U87hybrid.doc in my Yahoo MicBuilders Files.  You have to join.
I have simulated the U87 circuit with LTspice. As soon as I get back home I'll post the results. There is nothing magic in the response, ordinary 1st-order shelves. Could be easily done with any mixer's EQ.
 
ricardo said:
Yes.  Omni & Cardioid show a broad 2-3dB lift 2k5-16k Hz.
Wouldn't it rather be the result of diffraction? I would think the fundamental resonant frequency of the diaphragm to be much lower...
Anyway, 2-3 dB doesn't explain the very low self noise, that is about 5 dB below the best contenders.
 
abbey road d enfer said:
I have simulated the U87 circuit with LTspice. As soon as I get back home I'll post the results. There is nothing magic in the response, ordinary 1st-order shelves. Could be easily done with any mixer's EQ.
Yes please Abbey.  If you could post your LTspice model too, that would be much appreciated.

BTW, if you are just injecting your signal to the 'Test Point' doesn't give the 'real' EQ.  You need your source floating with the capsule.  My DOS Circuit Analysis programme can't do this conveniently.

Yes.  Omni & Cardioid show a broad 2-3dB lift 2k5-16k Hz.
Wouldn't it rather be the result of diffraction? I would think the fundamental resonant frequency of the diaphragm to be much lower...
Anyway, 2-3 dB doesn't explain the very low self noise, that is about 5 dB below the best contenders.
There's several things affecting 'HF' response.  The 'diffraction' includes cavity effects of the mounting too and we could consider them 'resonant'.  Diaphragm damping is effective for these.

My measurements of Zephyr's CK12 clone mike tells me you can get below 10dBA spl without invoking the 'underdamped diaphragm' trick.

The original Senheisser paper where this trick is used to the fullest advantage also shows a Fig-8 with lopsided 'diaphragm resonance' so they are using a lot of cavity 'resonance' in their mikes to extend HF

A dual diaphragm LDC, like CK12 and clones, isn't a strict 'slack' fig-8 diaphragm or a 'tight' omni diaphragm.  Gotta be careful.  I've never designed a dual diaphragm capsule so might be pontificating out of the wrong orifice  ;D
 
Back
Top