Any simple mods for lowering the noise floor of microphones?

[jp008]

So I bought a pile of cheap microphones to have some fun with.. a couple of dynamics, a condenser and some type of crazy karaoke mic. Upon plugging them into my preamp, I jacked up the gain and you know the rest. Mmmmm.. delicious white noise.

Then I got to thinking again (dangerous in my case, believe it), WHAT IF instead of expecting a bunch of white noise when cranking up the gain on my preamp, I just solder a bunch of stuff between the mic capsule and the XLR output to see if I can lower the noise floor BEFORE plugging it into the preamp. Eureka !

Then I did and internet search and.. nothing. Amazing ! I think I found a diamond in the rough thinking of microphone designers throughout the ages.

That said, please post any possible simple 'noise floor lowering' microphone modifications ideas you might have, so I can keep myself busy. Thank you.

Again, it doesn't have to be elaborate, just some simple stuff to try with 3 or 4 components. It's not like I could fit a big fancy circuit board into a microphone anyways. Duh?

[Tj Shredder]

The reason why they are cheap...
The only way is software, spectral noise reduction. Most likely the capsule itself is noisy, whatever you place between would rather create additional noise...
But aside, check impedances, if they are not matched between mic and preamp you get more noise...

[DJ Warmonger]

I just solder a bunch of stuff between the mic capsule and the XLR output to see if I can lower the noise floor BEFORE plugging it into the preamp. Eureka !

This is against basic laws of physics ¯\_(ツ)_/¯ https://www.everythingrf.com/community/ ... ise-factor

[whyterabbyt]

So I bought a pile of cheap microphones to have some fun with.. a couple of dynamics, a condenser and some type of crazy karaoke mic. Upon plugging them into my preamp, I jacked up the gain and you know the rest. Mmmmm.. delicious white noise.

Then I got to thinking again (dangerous in my case, believe it), WHAT IF instead of expecting a bunch of white noise when cranking up the gain on my preamp, I just solder a bunch of stuff between the mic capsule and the XLR output to see if I can lower the noise floor BEFORE plugging it into the preamp. Eureka !

Then I did and internet search and.. nothing. Amazing ! I think I found a diamond in the rough thinking of microphone designers throughout the ages.

That said, please post any possible simple 'noise floor lowering' microphone modifications ideas you might have, so I can keep myself busy. Thank you.

Again, it doesn't have to be elaborate, just some simple stuff to try with 3 or 4 components. It's not like I could fit a big fancy circuit board into a microphone anyways. Duh?

If 3 or 4 components could make any significant difference to lowering the noise floor of a cheapo microphone, those 3 or 4 components would be built into cheapo microphones as standard.

[aciddose]

There are actually several ways you can easily beat the preamp's noise performance by using buffers or amplifiers at the source rather than at the destination. This is why active mics already have such a buffer/amp built in.

One method that works quite well is if you're able to trade distortion for noise... for example you might introduce harmonics like 2nd -30, 3rd -20 or similar. This is a fairly nasty level of distortion, but by using a non-linear amplifier with much higher gain ("soft clipping") you can dramatically increase your headroom and reduce overall noise by amplifying the signal from the mic before the noise is introduced.

There are a number of different buffering strategies that work best depending upon your exact aims... but generally speaking if you're using a good active mic it's already using the best possible method to minimize noise and maximize linearity (minimize THD.) So if that's your aim (the most common aim), you're best to use quality equipment designed for that purpose.

The only time you're going to do any better is as I mentioned in a situation where you're trading noise vs. distortion vs. bandwidth, and only because you're making some sacrifice that the designers aiming to satisfy the common demand were unable or unwilling to make.

Sometimes a mic with a "poor" THD is absolutely great.

Try follower -> common base amplifier -> follower (3 transistors, 3 resistors, 2 capacitors). This will be highly non-linear but provide great gain with minimal noise. If you don't want non-linear you need either a more complicated amp or you're stuck with a simple single stage buffer. If bandwidth and input impedance are an issue, the solution is generally a NJFET buffer rather than a BJT transistor... but then you have the issue of VGS variation.

[jp008]

Try follower -> common base amplifier -> follower (3 transistors, 3 resistors, 2 capacitors). This will be highly non-linear but provide great gain with minimal noise. If you don't want non-linear you need either a more complicated amp or you're stuck with a simple single stage buffer. If bandwidth and input impedance are an issue, the solution is generally a NJFET buffer rather than a BJT transistor... but then you have the issue of VGS variation.

Thanks for the suggestion, but I'm afraid something like that would be far too complicated unless I had a serious schematic showing what goes where and in what order. Actually, what I'd really be interested in to start with is something much more simple like a 100hz HPF circuit such as those small barrel adapters have, maybe even a LPF in the neighborhood of 10-12k. What does inside of those little adapters look like? I know there's a manufacturer named PSC that makes the HPF at around 140hz and I was thinking of buying one just to open it up and see how it works. I mean, it can't be that complicated with puny 2" x 1/2" dimensions. A lot of microphones even already have a HPF installed, such as the SM7b, which makes me wonder what that part of circuit looks like. I owned one years ago, but never bothered to have a better look at the inside..

[BertKoor]

What's inside... intriguing indeed!

Seeing and thus knowing what's inside is something entirely different than understanding what's going on in there.

I have a basic understanding of what a resistor, capacitor & transistor do. These are the basic building blocks of everything electronic. I could explain their basic behaviour. But a decent preamp like aciddose described, even if my life depended on it I could not come up with the design from scratch.

It's intriguing what stuff can be fitted into a plug these days. For instance I have a cable that converts VGA to composite video. This takes more than "a few" transistors. So I'm sure it's "just" a chip inside. Really nothing to see there.

A basic hi-pass filter can be made with just a capacitor and a resistor. You do have to know though the exact impedances of what's before and after it. A few weeks ago JCJR demonstrated his skills in computing the design. It only works with high impedance capsules and low impedance preamps. Otherwise you need a buffer (amp stage) inbetween.

Why would you hi-pass? Destroy hum?
Why would you low-pass? Destroy noise?

These are the things the EQ section of a mixer's channel strip can do. Better buy a small Behringer Xenyx mixer and/or graphical EQ. Far more versatile than a XLR adapter with one filter.

But there are limits to what analog electronics can do. Especially keeping noise levels down is a serious hard problem to solve. Reducing noise without destroying the signal in some other aspect is impossible.

DSP however is very smart. Record some noise of your mic, let the plugin analyse it, and it will deduct the noise fingerprint from real signal. Excellent job, impossible to do analog.

Or buy actual good mics. There are limits to how far you can polish a turd

[jp008]

....

[jp008]

....

[jp008]

Why would you hi-pass? Destroy hum?
Why would you low-pass? Destroy noise?

Without getting too technical, let's just say I have a small sound box that has L/R analog mic inputs on it (not USB) and I tried like ten different dynamic mics going direct into the box, but for some bizarre reason, only 1 of the mics had the strange ability to produce an enormous amount of clean gain. It's almost as if the mic has a built-in preamp because all of the other mics were not even audible. The mic is just a basic dynamic made in Korea, but I've never heard a microphone do that before. It basically eliminated the need for an external preamp. Very strange. Only problem is.. wait for it.. the low end on the mic needs something. Take a guess what.

Hint: I want to put it inside the microphone. Care to share your basic resistor/cap combo ? I want to try something basic that might roll everything off below like 250hz and then maybe a slight 10-12k bump or shelf, so not really an LPF in the highs, just HPF in the lows with something to maybe nudge the top up a bit, but not essential. Hell, I should just buy the damn little HPF adapter made by PSC and take it apart..

Please don't say "mixer" or "Behringer" in the same sentence again.

[BertKoor]

We need to get technical: what exactly is that little box of yours with two mic inputs? Does it have switchable phantom power?
Do you have any details about that Korean mic?

Like said, you need to know the exact impedances of both in order to compute which capacitor to put in that forms the ideal HPF for this specific case.

I'll say it again: Behringer makes devent mixers for the price you pay. The X32 is quite attractive.

[jp008]

We need to get technical: what exactly is that little box of yours with two mic inputs? Does it have switchable phantom power?
Do you have any details about that Korean mic?

Like said, you need to know the exact impedances of both in order to compute which capacitor to put in that forms the ideal HPF for this specific case.

I'll say it again: Behringer makes devent mixers for the price you pay. The X32 is quite attractive.

No offense if you use them, but I don't use mixers anymore and I'm never going back to them, ever. I only use channel strips or rack mounts these days, so there's no need to keep trying to convince me to use one. Thanks for the suggestion though.

As for the sound box and mic I'm using, the box is just a generic Chinese thing, which actually works better for vocals than my both my RME ADI and my TC Electronic Gold converters, since they're both hyped in the higher frequencies (after A/D is achieved), which just destroys things with weird artifacts, almost always in the 2000hz nightmare area. Cursed frequency! But when using the $30 Chinese box recording vocals, I honestly heard a big difference in naturalness, so as a matter of fact, in my case, the little box sounds better than $1500 worth of elaborate technology when recording vocals. No, I'm not using phantom power either, since as I previously said, ALL the mics I tested are Dynamics, so there's no need..

As for the microphone's impedance, I have no idea because it's too old and there's no information online. Maybe you could tell me how to test it with a voltage tester?

[BertKoor]

M'kay, interesting to hear a RME converter lets someone down.

Here's the thread I meant, really good explanation there :
viewtopic.php?f=102&t=534236&start=20

[aciddose]

Try follower -> common base amplifier -> follower (3 transistors, 3 resistors, 2 capacitors). This will be highly non-linear but provide great gain with minimal noise. If you don't want non-linear you need either a more complicated amp or you're stuck with a simple single stage buffer. If bandwidth and input impedance are an issue, the solution is generally a NJFET buffer rather than a BJT transistor... but then you have the issue of VGS variation.

Thanks for the suggestion, but I'm afraid something like that would be far too complicated unless I had a serious schematic showing what goes where and in what order.

Well, the issue is the specifics of the design depend completely upon what sort of mic you're using, what the output is, what input impedance you're trying to achieve and so on.

Here's a rough ltspice schematic, save this as a text file named buffered_common_base_amp.asc or similar and you can load it.

Code: Select all

Version 4
SHEET 1 880 680
WIRE -224 80 -224 64
WIRE -144 80 -224 80
WIRE -16 80 -144 80
WIRE 64 80 -16 80
WIRE 176 80 64 80
WIRE -224 96 -224 80
WIRE -16 160 -16 80
WIRE 64 160 64 80
WIRE -144 176 -144 80
WIRE -320 224 -336 224
WIRE -304 224 -320 224
WIRE -224 224 -224 176
WIRE -224 224 -240 224
WIRE -208 224 -224 224
WIRE 176 240 176 80
WIRE -144 288 -144 272
WIRE -96 288 -144 288
WIRE -64 288 -96 288
WIRE 64 288 64 240
WIRE 64 288 32 288
WIRE 96 288 64 288
WIRE 112 288 96 288
WIRE -320 336 -320 224
WIRE -16 352 -16 240
WIRE 176 352 176 336
WIRE 208 352 176 352
WIRE 304 352 272 352
WIRE 336 352 304 352
WIRE -16 368 -16 352
WIRE 64 368 -16 368
WIRE -224 384 -224 224
WIRE -144 384 -144 288
WIRE -16 384 -16 368
WIRE 64 384 64 368
WIRE 176 384 176 352
WIRE 304 384 304 352
WIRE -320 480 -320 416
WIRE -224 480 -224 464
WIRE -224 480 -320 480
WIRE -144 480 -144 464
WIRE -144 480 -224 480
WIRE -16 480 -16 464
WIRE -16 480 -144 480
WIRE 64 480 64 448
WIRE 64 480 -16 480
WIRE 176 480 176 464
WIRE 176 480 64 480
WIRE 304 480 304 464
WIRE 304 480 176 480
WIRE 304 496 304 480
FLAG -336 224 input
IOPIN -336 224 In
FLAG 336 352 output
IOPIN 336 352 Out
FLAG -224 64 b+
IOPIN -224 64 In
FLAG 304 496 0
FLAG -96 288 follow
FLAG 64 368 base
FLAG 96 288 amp
SYMBOL npn -208 176 R0
SYMATTR InstName Q1
SYMBOL npn 112 240 R0
SYMATTR InstName Q3
SYMBOL npn -64 352 R270
SYMATTR InstName Q2
SYMBOL res -240 80 R0
SYMATTR InstName R1
SYMATTR Value 10Meg
SYMBOL res -240 368 R0
SYMATTR InstName R2
SYMATTR Value 1Meg
SYMBOL cap -240 208 R90
WINDOW 0 0 32 VBottom 2
WINDOW 3 32 32 VTop 2
SYMATTR InstName C1
SYMATTR Value 100n
SYMBOL cap 272 336 R90
WINDOW 0 0 32 VBottom 2
WINDOW 3 32 32 VTop 2
SYMATTR InstName C3
SYMATTR Value 100n
SYMBOL res 160 368 R0
SYMATTR InstName R7
SYMATTR Value 4.7k
SYMBOL res 288 368 R0
SYMATTR InstName R8
SYMATTR Value 100k
SYMBOL res 48 144 R0
SYMATTR InstName R6
SYMATTR Value 1Meg
SYMBOL res -160 368 R0
SYMATTR InstName R3
SYMATTR Value 10k
SYMBOL res -32 368 R0
SYMATTR InstName R5
SYMATTR Value 20k
SYMBOL default\\voltage -320 320 R0
WINDOW 3 -63 183 Left 2
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName VG
SYMATTR Value PULSE(0 100m 0 1m 1m 1n 2m)
SYMBOL res -32 144 R0
SYMATTR InstName R4
SYMATTR Value 180k
SYMBOL cap 48 384 R0
SYMATTR InstName C2
SYMATTR Value 1µ
TEXT -200 48 Left 2 !V1 b+ 0 DC 9
TEXT 314 320 Left 2 !.tran 10m

This circuit needs additional bypass capacitors to eliminate radio interference and I haven't bothered to measure the bandwidth (AC analysis) or trim anything.



You can see I've adjusted it assuming a typical input level near 100 millivolts with input impedance near 1 meg. The non-linearity is already very clear with about 20x (26 dB) gain. It has a much higher component count than I was thinking. Three transistors, three capacitors (not counting bypass), eight resistors.

It probably shows horrible performance and definitely needs a lot of tweaking, but it demonstrates the non-linear behavior I was talking about while providing a little gain and (?) very low noise. The symmetry can be trimmed by adjusting R2 or R4, and the circuit is already acting as a high-pass which can be trimmed by adjusting C2. Feed-forward only and zero signal path impedance.

[aciddose]

Here's a minor follow-up with the tweaking I didn't bother to do in the first post.



This version has added bypass capacitors (I'm awful at this, I may have done it wrong), input impedance should be exactly 1 meg, bandwidth (-3 db) should be ~20 to ~40k and gain is near 40x (32 dB). It's also trimmed for far higher linearity but may draw twice as much current. Since the gain is twice as high it should give near twice the noise performance... but resistors + capacitors + RFI are the main noise sources and generally speaking noise performance is a trade-off with power draw / battery life.

Code: Select all

Version 4
SHEET 1 880 680
WIRE -224 80 -224 64
WIRE -144 80 -224 80
WIRE -16 80 -144 80
WIRE 64 80 -16 80
WIRE 176 80 64 80
WIRE -224 96 -224 80
WIRE -16 160 -16 80
WIRE 64 160 64 80
WIRE -144 176 -144 80
WIRE -448 224 -464 224
WIRE -432 224 -448 224
WIRE -336 224 -368 224
WIRE -224 224 -224 176
WIRE -224 224 -256 224
WIRE -208 224 -224 224
WIRE 176 240 176 80
WIRE -144 288 -144 272
WIRE -96 288 -144 288
WIRE -64 288 -96 288
WIRE 64 288 64 240
WIRE 64 288 32 288
WIRE 96 288 64 288
WIRE 112 288 96 288
WIRE -448 336 -448 224
WIRE -224 352 -224 224
WIRE -224 352 -304 352
WIRE -16 352 -16 240
WIRE 176 352 176 336
WIRE 208 352 176 352
WIRE 304 352 272 352
WIRE 336 352 304 352
WIRE -16 368 -16 352
WIRE -16 368 -80 368
WIRE 64 368 -16 368
WIRE 112 368 112 288
WIRE -304 384 -304 352
WIRE -224 384 -224 352
WIRE -144 384 -144 288
WIRE -80 384 -80 368
WIRE -16 384 -16 368
WIRE 64 384 64 368
WIRE 176 384 176 352
WIRE 304 384 304 352
WIRE -448 480 -448 416
WIRE -304 480 -304 448
WIRE -304 480 -448 480
WIRE -224 480 -224 464
WIRE -224 480 -304 480
WIRE -144 480 -144 464
WIRE -144 480 -224 480
WIRE -80 480 -80 448
WIRE -80 480 -144 480
WIRE -16 480 -16 464
WIRE -16 480 -80 480
WIRE 64 480 64 448
WIRE 64 480 -16 480
WIRE 112 480 112 432
WIRE 112 480 64 480
WIRE 176 480 176 464
WIRE 176 480 112 480
WIRE 304 480 304 464
WIRE 304 480 176 480
WIRE 304 496 304 480
FLAG -464 224 input
IOPIN -464 224 In
FLAG 336 352 output
IOPIN 336 352 Out
FLAG -224 64 b+
IOPIN -224 64 In
FLAG 304 496 0
FLAG -96 288 follow
FLAG 64 368 base
FLAG 96 288 amp
SYMBOL npn -208 176 R0
SYMATTR InstName Q1
SYMBOL npn 112 240 R0
SYMATTR InstName Q3
SYMBOL npn -64 352 R270
SYMATTR InstName Q2
SYMBOL res -240 80 R0
SYMATTR InstName R1
SYMATTR Value 5.5Meg
SYMBOL res -240 368 R0
SYMATTR InstName R2
SYMATTR Value 1.2Meg
SYMBOL cap -368 208 R90
WINDOW 0 0 32 VBottom 2
WINDOW 3 32 32 VTop 2
SYMATTR InstName C1
SYMATTR Value 100n
SYMBOL cap 272 336 R90
WINDOW 0 0 32 VBottom 2
WINDOW 3 32 32 VTop 2
SYMATTR InstName C3
SYMATTR Value 100n
SYMBOL res 160 368 R0
SYMATTR InstName R7
SYMATTR Value 4.7k
SYMBOL res 288 368 R0
SYMATTR InstName R8
SYMATTR Value 100k
SYMBOL res 48 144 R0
SYMATTR InstName R6
SYMATTR Value 220k
SYMBOL res -160 368 R0
SYMATTR InstName R3
SYMATTR Value 4.7k
SYMBOL res -32 368 R0
SYMATTR InstName R5
SYMATTR Value 20k
SYMBOL default\\voltage -448 320 R0
WINDOW 3 -63 183 Left 2
WINDOW 123 24 32 Left 2
WINDOW 39 0 0 Left 0
SYMATTR Value PULSE(0 50m 0 1m 1m 1n 2m)
SYMATTR Value2 AC 1 0
SYMATTR InstName VG
SYMBOL res -32 144 R0
SYMATTR InstName R4
SYMATTR Value 100k
SYMBOL cap 48 384 R0
SYMATTR InstName C2
SYMATTR Value 10µ
SYMBOL cap 96 368 R0
SYMATTR InstName Cb3
SYMATTR Value 33p
SYMBOL res -240 208 R90
WINDOW 0 0 56 VBottom 2
WINDOW 3 32 56 VTop 2
SYMATTR InstName Ri1
SYMATTR Value 47k
SYMBOL cap -320 384 R0
SYMATTR InstName Cb1
SYMATTR Value 33p
SYMBOL cap -96 384 R0
SYMATTR InstName Cb2
SYMATTR Value 33p
TEXT -200 48 Left 2 !V1 b+ 0 DC 9
TEXT 312 320 Left 2 !.tran 20m

Again I'm terrible at this, don't risk wasting your time on this assuming it works right