DC Power Supply Noise elimination for Electret Microphone

LowQCab

Joined Nov 6, 2012
4,023
Resistors create noise when Current flows through them.
It's not easy to measure, and it covers a very broad Frequency-range.

The larger the Resistance-Element is, the less Noise it makes.

The particular Construction-Techniques used to make the Resistor also create
noticeably different amounts of Noise,
That's why a "Metal-Film" type Resistor makes less Noise than a "Carbon-Composition" Resistor.
Precision "Metal-Film" Resistors may cost ~5 to ~10 times what a Carbon-Resistor costs.
It really depends on just how much background Noise, ( mostly "HISSSSS" ), is acceptable to You.

The "Signal-to-Noise-Ratio" ( SNR ) is what really counts.

If You are always recording very loud Sounds, then the Noise will be so much
less than the sounds that You want, that the background Noise may become almost irrelevant.
But if You are recording moderate-level sounds, especially with a very dead and silent background,
the Hiss may become quite noticeable, and irritating, especially when listening with Headphones.

Knowing the expected circumstances of usage of the Microphone will help to determine
whether or not "Noise" will be a factor that must be considered in the design of the Amplifier.
.
.
.
 

Thread Starter

rpschultz

Joined Nov 23, 2022
416
Resistors create noise when Current flows through them.
The larger the Resistance-Element is, the less Noise it makes.
Assuming all resistors are 1% metal film, which ones would benefit most from being 1/2W instead of 1/4W to reduce noise in an audio circuit?
 

PaulEE

Joined Dec 23, 2011
474
" The larger the Resistance-Element is, the less Noise it makes. " - I'm almost certain he meant to say the more noise it makes.

I have some...comments...I am a little late to the party; forgive me if I repeat something. I read most of it.

1) The DC power with the noise is some DC level, with some noise superimposed onto it. The folks at Shure, who MAKE MICROPHONES, published a circuit that anyone can build with common components that would've worked JUST FINE...as you found out by simply building it. High frequency switching harmonics, except for the very high ones, will be absolutely smashed into oblivion with the 3-stage RC filter that they had in their schematic. Each successive stage smashes a little more until all you have left is clean DC, with perhaps a hint of switching noise with the worst components and worst construction. Stick a ferrite bead on the input and you're good to go (for the "very high ones").

2) An electret microphone is interesting. Also known as a condenser microphone (condenser old name for capacitor), these are basically tiny capacitors whose capacitance changes when they experience varying pressure differences (sound). If the charge on the plates is fixed and you vibrate the front of the capacitor (change the capacitance by varying the space between parallel plates), C=Q/V and V must change. This small V is your audio, which is typically pre-amplified by a high-impedance FET, which requires a particular load resistor external to the microphone, and a DC supply. It is important that the DC supply is quiet, because as you can imagine, the signal (before the FET) is quite small, and FETs have their own capacitive coupling, etc, etc. By the time you take your small signal supplied by the FET and amplify it by x10 or x100, all of a sudden that low amount of noise on the power supply is hurting your ears!

3) There are a lot of ways to get a quiet voltage from a "loud" (noisy) voltage. One way is to make a dirt-cheap RC filter, and then try it. And when that doesn't work, add stages until it does. This is the original schematic you posted, and because the microphone has such small power requirements, it works great. Yay.

3A) If however, you want to be an audiophile-level nerd about it, you can pre-regulate the incoming power with a linear regulator whose power supply rejection ratio (PSRR) is very high to the frequencies you are likely to encounter coming from your unregulated source. In other words, stick a ferrite bead between the unregulated power and your regulator, and any off-the-shelf regulator will beat the tar out of the 3xRC filter arrangement. Typical PSRR values in the audio range are 60-90dB of attenuation. The 7805 +5v regulator, for example, goes from 80dB to 60dB from between 10 Hz and 20 kHz...and they're so popular you could probably walk down a sidewalk somewhere and trip over one.

3B) If however, you want to be anal-retentive, you can use a REF01, REF02, REF03, AD780 (with x2 gain stage ahead of it), etc. as the power supply for the microphone. These are voltage references, and their sole purpose in life is to give you V +/- some small percent error with minimum noise and near-zero temperature drift. I think the parts above are on the order of 10,20,30 microvolts of noise, and their output deviation versus input deviation is measured in parts per million (like percentage, except instead of X/100 it's X/1,000,000...small stuff!).

3C) If however...just kidding.

Also, noise in resistors. Resistors, or anything really above absolute zero, have noise. The question is how much, and should you care? ( 1672145899310.png is how much (voltage noise) a resistor of resistance R, at temperature T (Kelvin) (k is Boltzmann's constant) produces per sqrt(Hz). In other words, if you stick with resistors in the 10's of K and below, you're fine. For reference, a 1K resistor at room temperature and through a band from just above DC to 100 kHz is around 1.3 uV of noise. 10K is 4 uV. Metal film resistors are typically well-behaved compared to their historical ancestors. Running current through raises the temperature and they become noisier due to the temperature rise, but still quite small numbers compared to audio output levels. Resistor Noise Calculator (muzique.com) was what I used, and the 4kTR was from an analog devices article. Sometimes the bandwidth "B" is within the top sqrt as root(4KTBR), but anyhow...

Oh, as for the 1/2 or 1/4 watt question...if the resistors aren't dissipating any power, it doesn't, practically speaking, matter. If they are getting warm, 1/2 watt are typically bigger, which means heat is spread over a larger area, which means the 1/2 watt would be a little more favorable since it may run at a lower temperature. For your purposes, I do not think it matters.

Hope this helps,
Paul
KI5VNH
 

Thread Starter

rpschultz

Joined Nov 23, 2022
416
@PaulEE
Hey thanks for the thoughts. A lot of good stuff I'll have to dig through. I'm an engineer, but not EE, so I'm learning a lot of this stuff.
A few comments:
1) that musique reference, I've built some of this stuff!
2) This circuit has morphed into a mic power/preamp/HPF, over here on this thread. Latest schematic at the end.
3) The application is an internal guitar mic (2 conductor condenser). For the sake of this thread, most guitarist use isolated 9v power supplies, like this Cioks DC7 that I am using. So it's good and quiet and each output is isolated, but a mic circuit still benefits from the 47u and 100u caps at the beginning to remove AC ripple from the mic.
4) Ferrite bead... never considered that.
 

PaulEE

Joined Dec 23, 2011
474
I forgot to mention this. For measurement of anything, particularly low-noise or high-impedance...the Keithley Low-Level Measurements Handbook is a great reference: LowLevelHandbook_7Ed.pdf (tek.com)
1672148112654.png

As for the project, I figured it morphed into something more complex than "can I use this to quiet this power supply down?".

I'll have to take a look at what you just sent and see if I have any nuggets of info to share. I got into engineering initially for audio stuff, and have dabbled with effects and amps, so...right down my alley, so-to-speak.

Best regards,
Paul
KI5VNH
 

Thread Starter

rpschultz

Joined Nov 23, 2022
416
Yeah. I'm a mechanical engineer. But I'm a musician who has dabbled in modding pedals and amps for years. What fun! Now I'm actually trying to LEARN what the heck is going on electronically and design something that really isn't available for under $500.
 

PaulEE

Joined Dec 23, 2011
474
You will find that the actual parts required are a lot cheaper than the price tag. Also, an aside; your little box you 3D-printed - I would paint the inside with carbon paint and attach it to circuit ground. This will prevent noise from getting into your little box as well. I think the original schematic source does mention this, and I forgot to in my "mondo-post" above.

Paul
KI5VNH
 

Thread Starter

rpschultz

Joined Nov 23, 2022
416
3D printed boxes are super great for prototyping. But eventually when I get the layout like I want, I'll go metal - it shields and grounds whereas plastic doesn't. Right now I'm working on sticking my circuit in a 1590A (4 x 1.5"). So I'll print it to get the holes right, etc. but eventually make it metal.

I've noticed a lot of the smaller pedals (4x1.5, 2x2"), etc, the pedalmakers are soldering the jacks onto the board. But to do that, they have to slot the sides of the metal box, which requires a mill instead of just a drill press. It's thin wall, so I probably could mill those slots with my drill press. Like below 1590A example:

IMG_3717.jpg

IMG_3718.jpg
 

MisterBill2

Joined Jan 23, 2018
18,175
As far as noise goes, current also has some effect, and at least in microphones using phantom power, that is not very much current. And if the noise is in MICROVOLTS, consider that the amplifier output is in the tens of MILLIVOLTS, so that the resistor noise is at least 40 dB down, often more like 60 dB down. And that is before any effort to reduce the noise is considered.
As for the Shure circuits, it looks to me like they put functional quality first, and cost a distant third. So Shure is a good quality product to learn from.
 

PaulEE

Joined Dec 23, 2011
474
I am thoroughly puzzled by that last one, MisterBillyDos. Yes, there is voltage noise and current noise, but there is barely any current to speak of in any of this, so I didn't bother going into that.

He asked about filtering noise on power supply rail and he asked about resistor noise...so I told him about both and stuck some numbers on so that he could compare the levels of absurdity between various solutions...while proving the point that the circuit he started with was fine to begin with. I'm aware that we were debating curtains on the Titanic here, but I wanted the OP to 1) be aware that we were and 2) understand why.

Also, Audioguru again and LowQCarb (and others) are definitely speaking the language you need, rpschultz!

I'll review that other thread and see how far you've gotten on the overall beast.

Oh, to add to something Audioguru again had mentioned. Since you're a MechE, you may not know. When you use a sensor or transducer, or an active amplifying circuit element, you typically have to "bias" them; in other words, in their steady-state (no perturbations), they're at some specified point, and when they are perturbed, this phenomenon causes them to fluctuate around this point. This was why he suggested 4.5v on the mic, and 4.5v on the resistor. This gives you MaxHeadRoom (pun intended) in the positive and negative directions, reference being 1/2 Vsupply, or 4.5v.

The reason the Electret has a built-in field-effect transistor is that it has quite a high input impedance; it is a field (voltage) controlled device. The field in the gate controls the current that flows through the other two legs (drain and source). These devices have characteristic curves (you are probably familiar with what these look like) that vary according to (Vgs-Vthres)^2, in other words, equal incremental changes on the gate allow ever-increasing amounts of current flow that follow a square-law instead of being linear. Since the signal from the microphone is very small, a linear approximation is OK, and the FET works great.

This will then be AC-coupled to your next stage, which LowQCarb has you all set up with above. Properly configured op-amp pre-amps are a great choice. As MisterBill2 pointed out, you're still in the 10's of mV range (estimating, could be more or less depending), so a gain of x10 or x100 will surely uncover other noise sources. But implementing filtering techniques / band-limiting the pre-amp as Audioguru again and others touched on will get you all sorted out.

And good call on the metal box; don't forget to ground it :)

Good stuff! I have to stop writing novels as responses...I enjoy helping where I can!

Paul
KI5VNH
 

MisterBill2

Joined Jan 23, 2018
18,175
Resistor noise was mentioned and I pointed out the relative amplitudes so that the TS could see in plain words that it is not an issue here. The supply filtering had been quite well covered and so I made no mention of it.
 

Audioguru again

Joined Oct 21, 2019
6,673
the electret mic also, produces noise and the amount is listed in its datasheet.
But since the mic is inside a guitar then I think overload distortion will be produced, not noise.
Using the Linkwitz mic mod converts the Jfet from a common source with gain to a source follower is used for electret mics inside pianos and drums to eliminate overload distortion.
 

Attachments

Thread Starter

rpschultz

Joined Nov 23, 2022
416
the electret mic also, produces noise and the amount is listed in its datasheet.
But since the mic is inside a guitar then I think overload distortion will be produced, not noise.
Using the Linkwitz mic mod converts the Jfet from a common source with gain to a source follower is used for electret mics inside pianos and drums to eliminate overload distortion.
I have read about this before and considered it. Unfortunately, having 3 wires is kind of a deal breaker for this situation without getting crazy with multiple ports (drilling holes) in the guitar, etc. TRS cables and jacks are commonly available, this supports a main pickup (tip) and a secondary pickup (ring) which is a mic in this case.

That said, most stereo guitar jacks are actually TRRS jacks that support a switch to turn on/off an internal battery - maybe this could be used as TS for the main and RRS for the 3 conductor mic. The challenging part is to find a 1/4" TRRS cable. 1/8" ones are often used for phone/video, etc. with L, R, Video signals. Guitarists typically need 10-15' cables. I haven't been able to find a 1/4" TRRS cable.

@Audioguru again ; Question. To utilize the Linkwitz mod, does the condenser mic itself need modified? Or can the original 3 conductors be used?
 

Audioguru again

Joined Oct 21, 2019
6,673
The two-wires electret mic has modified wiring for the Linkwitz mod.
One printed wire is cut on the rear of the mic to separate the Jfet source pin from the ground wiring and metal case.
Two wires plus cable shield wire connect to the modified mic.
 

Attachments

MisterBill2

Joined Jan 23, 2018
18,175
To power any circuit inside a guitar can be simple with a standard TRS shielded cable using phantom power and a balanced output. Nothing new except the application to a guitar instead of a regular microphone. And best of all, the arrangement is noise resistant. With no battery in the guitar, but a handy amount of power, a multi-stage amplifier can be used.
 

MrSalts

Joined Apr 2, 2020
2,767
3D printed boxes are super great for prototyping. But eventually when I get the layout like I want, I'll go metal - it shields and grounds whereas plastic doesn't. Right now I'm working on sticking my circuit in a 1590A (4 x 1.5"). So I'll print it to get the holes right, etc. but eventually make it metal.

I've noticed a lot of the smaller pedals (4x1.5, 2x2"), etc, the pedalmakers are soldering the jacks onto the board. But to do that, they have to slot the sides of the metal box, which requires a mill instead of just a drill press. It's thin wall, so I probably could mill those slots with my drill press. Like below 1590A example:

View attachment 283977

View attachment 283978
You can get these boxes at TaydaElectronics.com - and they'll custom drill them for $4 extra per box (up to 40 holes per box).
they can also do slots and non-round holes. They have a drawing tool to define the hole position in each box model they sell.
Also, they can do UV printing and powered coating after the drilling. Super cheap solution to many of your issues.

Note, powder coat thickness fills in the hole so you have to make holes slightly larger - instructions in the design website.

https://www.taydaelectronics.com/hardware/enclosures/enclosure-custom-drill-service.html
The design tool site (https://drill.taydakits.com) needs to have the same username as the account for taydaelectronics.com site.
 

Thread Starter

rpschultz

Joined Nov 23, 2022
416
A few months ago, I was studying this circuit created by @Audioguru again , and ultimately some student from the University of Washington created a PCB of it. The CAM files were posted, so I sent them off to Osh Park and created 3 prototypes for $12. Recently I found time to solder one of them up, but I get a buzzing sound. It doesn't matter whether I have the mic plugged in or not. Of note:
  1. It shouldn't matter but the output polarity is labeled backwards - and thus I hooked it up backwards.
  2. The 9v supply is center negative, but the jacks I have were prewired so the colors are backwards. But I measured the voltage at pin 7 of the TL071 and got +9.4vdc.
  3. I socketed and substituted a 10k for R5 to reduce the max gain to 11. Gain = 100k pot / R5 + 1.

Any ideas?

U-Wash-preamp.png
1675212482589.png

IMG_3818.jpg

IMG_3821.jpg
 

LowQCab

Joined Nov 6, 2012
4,023
A Single-Sided PCB is not going to work very well for a Mic Preamp.
At the very least it needs a 2-sided Board with a mostly solid Ground-Plane on the bottom,
and it needs to be in a Metal-Box.
How particular Parts are arranged on the Board can make all the difference.
It's possible to not only get hideous Noise,
but You could actually turn the Circuit into an Oscillator.
.
.
.
 
Top