How Good Does a DIY Microphone Sound?
Вставка
- Опубліковано 12 вер 2023
- Is it possible to build a good sounding DIY microphone?
-- Links --
My Website: sinelab.net
Schematic: sinelab.net/pdf/diy_mic.pdf
Buy me a coffee: www.buymeacoffee.com/thesinelab
Bitcoin Address: bc1qwlfr2ml9km4r5vhpfh7f3m60k80whhcq3ysths
Ethereum Address: 0x7e8B8Ff8D6bD8DfB48b4A139694BB343f040Bb3f
-- - Наука та технологія
It works very good for a 1st attempt. Software will remove lots of the background noise. Your design reminds me of microphones from the 50's. Good work.
Thanks for watching! Software definitely helps.
Look up ribbon microphones. They're really similar to condensers. But they use magnetism instead of electric plates
@@kevinclass2010 So not really similar then. ;)
They have a lot of useful features but they are abominably difficult to make. I know, I've made one from scratch... and it still needs a lot of work on the mounting and tension adjustment.
shielding body is a must, plastic housing won't do it, but even tin soda can will.
It is not apparent now probably because you loaded your capsule too much with that 2k2 R2. Usually it is in Mohm range, thus FETs.
Take a look at re-engineered Rode NT-1A preamp schematic floating around Internet, they even use 1GOhm resistor to achieve noise figures they are known for.
Thanks, I'll have a look at that schematic.
The JLI-2590A capsule that he used has a built in JFET, thus the 2k2 resistor which matches what is used by the manufacturer to measure the specs. I believe what you refer to is internal to the capsule and not changeable, nor is it a problem since the electret connects only to the gate.
@@jrstf true, somehow during the presentation of all the different schematics I totally forgot that we were dealing with internal FET.
Although, these type of capsules are much less dependent on additional shielding, there is still some noise floor performance to gain.
I've witnesses an example where certain greybeard made one of those 10mm electret capsules headphones-level of silent.
@@Mr.Leeroy Linkwitz mod? Just wait until you see what Doug Ford did and I've (just about) managed to duplicate. And it's almost all discrete components, save for the regulator!
DIY Microphone love that ❤
Great work! I did feel like the whole xlr/phantom power was glossed over a bit and would have loved to see a little more on that aspect, what it does, how it does, etc… ;) tons of work went into this and editing is not fun so I totally understand haha.
Next time XLR comes up I will spend more time on it :)
Shorting a line called +48V to one called +12V seems like a bad idea...
I made a mistake in editing and the schematic. Take a look at the full schematic and you will see that the 48V net is connected to those 1k resistors. I should have been more clear that the 48 volts is coming through XLR, not that net.
Thanks for pointing that out.
You forgot GS and GD resistors made of PCB blank spaces. GS "contact area" is bigger few times than GD "contact area", so base is biased with huge impedance to voltage between drain and source an closer to source. So internal n-type induced channel j-fet is partially opened on it working point.
Hey! I’ve been doing something very similar and facing similar problems, my goal is to be able to power the mic/preamp/amp from the 5volt vbus wire from a usb connection and use the data lines to send the isochronous transfers back to a PC. So I made a switching power supply to boost the voltage. And I was wondering how to cut noise from power supply when bypass caps aren’t enough, is it even possible to get the linearity required for hi fidelity microphone applications from a high frequency switching supply without too much voltage drop? might be a decent video idea, would love to see it. Loved the hardware usb video too, keep up the good work 👍
Maybe you just made a mistake in your schematic connecting the12v directly to 48v line. If not you need insert a resistor between those 2 voltages. Looking at the pcb you may have a resistor there. The Phantom power should be able to deliver about 15ma. Your 072 IC typically draws 2ma from the datasheet so what's drawing all the extra current?
I'll definitely add that USB microphone idea to my list. :)
I did a closer inspection on the board and, when working properly, the zener actually does drop 12 volts.
The interesting thing, however, is that the circuit is very unstable. If I probe something in an odd way or even plug in the circuit before activating the 48V phantom power, then there is a certain instability that I can't put my finger on yet. The voltage across the zener is very low, like I have shown in the video.
So, under nominal conditions, the voltage regulation is fine. Just, there is another issue that I haven't found the source of yet.
USB isolator will help with noise from pc power supply..
I originally made a 25mm design with a FET (LSK170) and a TL072 on there and while it worked find, the 40 dB gain was more hiss than speech.
This circuit distorts quite a bit (not that we'd notice) due to non-linearity in the JFET. I've adapted a design from Doug Ford (Rode) which uses negative feedback to deliver a very low noise, very linear response.
I've used a TL431 for regulation and ran into the same issue on my last prototype (because that's the one thing I took for granted). It makes a great 10KHz oscillator though. Oh dear. And no, smoothing capacitors won't fix it.
The answer is to drop your current down. The p48 standard can deliver (give or take) 10 mA at around 12V because of the current limiting resistors on each leg (6k8). I've learned a lot of this trying to miniaturise the wheel, while making a mic that can (and I think will) rival the Rode NT1a. Actual noise from the circuit is very low and the TL431 regulator should be better than a simple Zener because I've used a pass transistor on the latest designs.
The TL071/2/4 series have JFET inputs which are known to be very noisy indeed. Texas and others do a whole slew of far better amps with lower currents.
The whole thing is Open Source hardware and on the DIY Perks forum (but anyone can make one, improve it etc).
One thing to note is that you don't need the 1Gig load resistor if the capsule is an electret. That only appears to be the case with unbiased capsules where the capacitor between the gate and the supply is required. Doug Ford worked around this somewhat by driving the gate in series rather than parallel but I haven't tried that technique as yet.
I think the NT1a also uses a pseudo balanced output. It's still balanced, a single-signal balanced output might make more sense. This is another point from Doug: if you have a fairly quiet preamp, why double your signal (to make it differential) and add 100% more noise.
What matters on a balanced line is the AC impedances have to match.
As of this writing, I've just sent another batch to JLCPCB (they are the folks I use, I don't get anything from them) with the corrected power supply. My issue was the shunt regulator (as here) pulling the positive supply down so hard, the thing switched off and cycled. I tried cheating and it cost me weeks and several months of beer money.
The "final" design is larger (it now fits 25mm and 34mm capsules, but can be wired to almost any FET-free electret). But I have to stress that until I "sign off", I don't want anyone spending their own cash and finding I've goofed (more than one time I've mixed up an catalogue number and had inductors where resistors went, or 10 ohm resistors where I needed a 10K). So lots of magic smoke!
One design I did do (not currently published as it's rough) is a solid state Op Amp with performance like you wouldn't believe, even in the simplest form it's very low-noise and can be pushed into driving fairly significant loads like capacitors. The design has been adapted from the Jensen 990 which is decades old, but still holds up remarkably well. The "matched pair" isn't made any more so I've replaced it with an (optional) MAT02. The operating current required for the MAT02 is different to the discrete LTP so I've included a current setting resistor.
Anyone interested can find me making my cockups public (so you don't have to) on Matt's forum - but full disclosure - we live in the same country, and that's about all the connection we have.
You can get much better performance with wm61-a cheap and little electret mic. It needs only one common emitter amplification stage and all assembly can be few times bigger than bare electret capsule and it will provide you significantly better performance.
Check out some new and modern opamps. The onec from the 70s need to ve forgotten 😅 modern opamps make less noise, draws extremely little current and work well on a single voltage. Good luck with next iteration of your mic 😊
only BJT or Fets
and I thought I'm smart when I soldered like 0.5$ microphone to an old jack's wire xD
It would be nice to see a comparison between the microphone u made and the microphone u shown in the beginning without any electronics
👍
جميل جدا
Don't you know LF353 yet?
I cant make the microphone, but can I buy yours, I am a music producer and I really love weird microphones
Send me an email: contact@sinelab.net
both legs of differential amplifiers need to be as equal as possible .. feeding the input of the inverting amp from the output of the non-inverting amp is an issue .. the issue arises from propagation delay .. though very small and definitely freq dependent, this delay causes the signals to be out of phase so as not be exactly 180 degrees out of phase .. this phase difference creates noise which is then fed into the amplifier stages.. Hope this helps ... Cheers :)
That helps a lot! I'll keep this in mind when making another microphone in the future :)
From what you say, I'm guessing the effect would be non-flat frequency response at the high end, not sure if that can be called noise.
Seems like one option would be to eliminate the extra op amp, preserving R10, R11, and C10. All XLR requires is impedance matching, voltage drive not required on pin 2. Most XLR equipment (not mics) simply connect XLR pin 2 to ground through a resistor that matches the impedance of pin 3.
Some have done far worse. I remember looking at an LR Baggs Mixpro, reverse engineering the circuit, and finding they tried to use an incredibly complex differential driver circuit which was a monumental failure, leaving the device totally useless when used in non-ideal situations (very noisy, as in, 60 Hz and harmonics).
@@SineLab He's right, you want to treat the two out of phase signals as equally as possible in terms of impedance. The good news is, there's sort of an industry standard configuration that you'll find on pretty much any balanced audio input and it only takes one op-amp side to do it. It's typically done with a split supply. There's a way to do it with a single supply with a little trickery. Google "ST an4586" it should be the first result "differential to single-ended amplification" ap note. Page 2 figure 2 you'll see that industry standard configuration I was talking about. All resistors 10k will give you unity gain (A=1v/v aka 0dB) then use the other side of the op amp to gain the hell out of it. But yeah, you'll need a split supply for that . To do it with a single supply you'll have to replace that ground with a common mode reference voltage which would be 1/2 of your supply voltage. So if you have a +12V supply Vref= 6V. They give you 4 different ways to configure that on Page 6 Table 4. The first one using a zener is all you need. Or you can make an easy split supply by putting two batteries in series and grounding the middle node where they meet. That'll give you negative voltage off the cathode of the bottom battery and positive voltage off the anode of top battery. Just make sure to reference all 'grounds' to that 0V spot between the batteries. Alright, I feel like that's a good amount of info. All stuff I wish someone had told me. Didn't want to just tell you what NOT to do ya know? And now you kinda have to build the circuit since I spent so much of my precious time telling you. 😝
A lot of noise. Your tl072 may be lm358
Low bit rate!!! Sadly does not sound good on youtube because of the bit rate i hear is low.