Opa Alice build noise floor

[wamosjk]

I just did my first build with an OPA Alice circuit with the hex board from JLI Electronics. It has the TSC1 capsule and the body is just one of those cheap u87 knockoffs from aliexpress. So my question is what is a typical noise floor for this build. I really dont have a frame of reference because all of my other mics are dynamics and I heard the OPA Alice circuit is supposed to be extremely quiet. Right now the self noise is faint but it is there and since this is my first build I just wanted to make sure that I don't have an issue somewhere that needs to be addressed. Here is a raw recording demonstrating the noise floor. I will link to the .wav files and also attach them to this post.


View attachment opa-alice-noise-floor.wav

As you can tell from the recording the noise floor is really faint but is there if you turn up the recording. Everything seems to be clean on the board as I cleaned it several times with 99% alcohol and a microfiber cloth. I also saw in the troubleshooting the following:

Scenario two: Noise. Our probable faulty functions for this one are wiring or a short of some kind to the shell. IfXLR_2 or XLR_3 is shorted to ground, you can have noise.This is caused by not enough supply current and running the operational amplifier below rated voltage. You canhear what this sounds like if you have the mic connected, and then turn phantom power off. As voltage drops, atsome point noise kicks in, then falls off as voltage completely goes away.

So I made a recording of me turning off the 48v power.

View attachment opa-alice-circuit-turning-48v-off.wav

This seem like it might be the issue but I'm not sure. So I just wanted to post on here to see if anyone has had any experiences and could point me in the right direction. Am I just being paranoid and the self noise is totally normal for this build or do I have an issue that needs to be addressed either in the circuit or the cheap shell its currently in? If you need more info let me know. Here are some images as well.

Edit: here is a recording of the sm58 with a mic booster at the same gain level.

View attachment sm58-noise-floor.wav

 

[mrgrooves666]

Well, you need to share more "in wich context" details, like how much gain do you have at the preamp. The circuit is pretty low noise yes, but depending on how much gain are you using at the preamp, that noise will come up nevertheless. But in what context...meaning is the level at the preamp actual real scenario, or practical level, or are you testing @ 58+ db gain? if its the latter, then yeah, it might be normal....links to files don't work btw...

 

[wamosjk]

Well, you need to share more "in wich context" details, like how much gain do you have at the preamp. The circuit is pretty low noise yes, but depending on how much gain are you using at the preamp, that noise will come up nevertheless. But in what context...meaning is the level at the preamp actual real scenario, or practical level, or are you testing @ 58+ db gain? if its the latter, then yeah, it might be normal....links to files don't work btw...

Im running through the tascam 1x2 hr interface and the knob is at 75%. The max gain on the interface is 58db. But if i plug in a shure sm58 i dont really hear the noise floor but its also a dynamic so it doesnt have a self noise. It should be noted that im on linux and on windows i can use less gain but my main computer that has dual boot to windows has like 9 fans so the fans make more noise than the noise floor of the mic. As far as the links they seem to be working fine for me. They are on google drive and i have it set so anyone can view them. I also attached the same files to the post as well so you can just view them that way.

 

[ccaudle]

Probably reasonable noise level. I assume you were not speaking very loudly, the timbre of your voice sounded like normal level for a close conversation. I have trouble guessing what that would translate to in SPL, but the noise floor was just audible at that gain setting.
Did you record the audio so you can get an idea of the SNR at that level?

if i plug in a shure sm58 i dont really hear the noise floor

How does the signal level from the SM58 compare to the condenser mic at the same gain level? Difference between signal level and noise floor is what you typically care about.

im on linux and on windows i can use less gain

That doesn't really make sense unless there is software gain on the windows system or software attenuation on the linux system. If there is software gain on the Windows system the end result probably isn't very different, because that gain would bring up the level of the signal and the noise floor. If there is attenuation happening on your Linux system then that would be a disadvantage because increasing the preamp gain to make up for it would increase the preamp noise floor as well.
Understanding how to optimize your software settings is a side topic from the original question, so to answer that, the noise floor does not sound unexpectedly high to me. To get objective numbers would require a much more rigorous measurement setup, either an acoustic noise source of a known SPL, or a preamp with accurate gain settings and a way to measure the electrical noise level.

 

[wamosjk]

That doesn't really make sense unless there is software gain on the windows system or software attenuation on the linux system. If there is software gain on the Windows system the end result probably isn't very different, because that gain would bring up the level of the signal and the noise floor. If there is attenuation happening on your Linux system then that would be a disadvantage because increasing the preamp gain to make up for it would increase the preamp noise floor as well.

Ya for some reason linux just doesn't like this Tascam 1x2hr audio interface. I had a different interface and didn't have that problem with but that interface really sucked. I like this interface I just have to turn it up a little more on linux. It really never presented an issue (until now). The Tascam 1x2hr does have proprietary windows drivers but no linux drivers so there may be something they're doing there to boost the signal in windows.

How does the signal level from the SM58 compare to the condenser mic at the same gain level? Difference between signal level and noise floor is what you typically care about.

So I made a recording with the noise floor of the sm58. I attached it to the original post. I also removed the links from the post and inserted the audio into the post (My first posts here and didn't see the insert button on the file attachment). I won't attach it here too just to save the site from duplicate uploads At first I just used the sm58 but since its a dynamic it was much quieter so I decided to just hook up my mic booster and when I do that the signals are relatively the same and the mic booster should introduce clean gain and not introduce any added noise. So there does seem to be some noise you can see the noise seems quite a bit less. Not sure if this seems normal or not because like I said all of my frame of referece is with dynamic mics so I never really had to deal with condenser self noise.

I did stick both samples in my DAW and then boosted the sm58 sample 6db and only listened to the quiet parts and they may be similiar in noise then so it might be that the self noise of the OPA Alice is right around where it needs to be.

I really just wanted to make sure I built the mic correctly and didn't have any issues that I need to address and everywhere says the OPA Alice circuit is really quiet (around 6db). So if I am getting any extra noise I would like to fix it. I was a bit concerned about the noise turning off the 48v power because in the troubleshooting it states.

Scenario two: Noise. Our probable faulty functions for this one are wiring or a short of some kind to the shell. IfXLR_2 or XLR_3 is shorted to ground, you can have noise.This is caused by not enough supply current and running the operational amplifier below rated voltage. You canhear what this sounds like if you have the mic connected, and then turn phantom power off. As voltage drops, atsome point noise kicks in, then falls off as voltage completely goes away.

So I'm not sure if that noise I posted is the noise they are talking about here. If so I may need to get a different body for it. But all that said the noise floor is pretty quiet and is much quieter than computer fan noise so its very workable.

 

[rogs]

I use a slightly different form of the OPA circuit. See here: https://www.jp137.com/lts/OPIC.43.pdf

The audio output on my circuit is single sided to help keep the noise to a minimum.
(The addition of the differential audio output in the commercial Alice OPA circuit will add an extra couple of dB to the noise floor).

The Equivalent Noise Level from that single sided design is around 3 dB(A) worse than my reference Rode NT1 microphone, which has a published noise figure of 4.5dB(A). The NT1 is considered to be a pretty low noise microphone.

Even including the extra couple of dB of noise added by the differential output of the 'Alice' circuit, it will still be classed as 'extremely low noise' according to Neumann's comments on THIS PAGE

HERE IS a short sample .wav file of the relative noise levels from a single sided and differential OPA circuit.

Both are refererenced to a -9dB tone. .... The mic capsule is not connected, and is replaced by a capacitor to connect the reference signal generator.
So this is only noise from the circuitry itself. .... As you can hear, there's not much in it!

 

[Khron]

... Plus wiring the capsule signal via the PCB itself (with zero attempt to mitigate anything) is... questionable to say the least, and may or may not add extraneous noise to begin with.

 

[ccaudle]

for some reason linux just doesn't like this Tascam 1x2hr audio interface

Check the alsamixer settings, perhaps the default is slightly attenuated for some reason. Unexpected differences in alsamixer settings between different interfaces is a relatively common occurrence (perhaps related to changes from default being saved to interface specific settings, so a new interface returns to defaults).
Depending on what audio server you are using (Linux has multiple available, Windows and MacOS just have one) there may be other places you need to double check, like pavucontrol.
Like I said, slightly off topic, but relevant to what you are trying to do because you need to control as many variables as possible to make valid comparisons.

The Tascam 1x2hr does have proprietary windows drivers

That is mostly because after two decades Microsoft still does not include a standard USB2 audio class driver with Windows, the way MacOS and Linux (and derivatives Android and ChromeOS) do.

or the cheap shell its currently in?

Problems with chassis usually show as power line related noise, and I do not hear a strong 50Hz or 60Hz component in the noise floor. There is nothing wrong with the mic body.

IfXLR_2 or XLR_3 is shorted to ground, you can have noise.This is caused by not enough supply current and running the operational amplifier below rated voltage.

That is ultra simple to just check. Do you not own a voltmeter of any kind? Measure the voltage on pin 2 and pin 3, and at the zener diode (using pin 1 as the reference terminal). It would be really obvious if one of the output pins had a short to 0V.
The OPA1642 op-amp will run down to under 5V supply voltage, so you might not notice even if that did occur, so easiest to just check.

I don't think there is a way to download files from embedded links, is there? I was trying to think how to compare the files from the OP to the file from rog to compare. My initial impression is that maybe the JLI build is not quite as optimized as rogs's build, but it is not too much different than other condenser mics I have (which probably have a little more circuity due to pattern switching, switchable high-pass, etc.).

 

[oscarhuang]

I also heard that the noise of this circuit board is very low and the sound is very clean. As a result, after using it, I just felt that the noise floor was a bit loud. The noise floor is much louder than the tubes I made myself. It cost me about a hundred Canadian dollars plus shipping and I was very disappointed.

 

[RuudNL]

I just listened to your recording:

https://cdn.imagearchive.com/groupdiy/data/audio/88/88853-ae606bafd522298b78be4a4c839f66f2.wav

I can't say that the noise is excessive.
Also compared to the SM58 the level seems to be higher with the same setting of the record volume.
This is normal, because in general a condenser microphone has a higher output at the same SPL than a dynamic microphone.

 

[wamosjk]

Check the alsamixer settings, perhaps the default is slightly attenuated for some reason. Unexpected differences in alsamixer settings between different interfaces is a relatively common occurrence (perhaps related to changes from default being saved to interface specific settings, so a new interface returns to defaults).
Depending on what audio server you are using (Linux has multiple available, Windows and MacOS just have one) there may be other places you need to double check, like pavucontrol.

Ya so in windows the slider only goes to 100% and in pavucontrol the slider goes to 200%. I guess I just assumed that 100% would be the same in both but it may be that the windows 100% is pavucontrol 200%.

Do you not own a voltmeter of any kind?

Yes I just pulled it out and the battery is dead in it. It uses a 9v sqare battery so I'll have to go get one and check back in on that.

I don't think there is a way to download files from embedded links, is there?

Are you talking about downloading the audio from the embeded audio player on the post? You should just be able to right click on the embedded audio and there should be an option to "save audio as". At least thats the case on desktop in chromium based browsers and firefox.

If you check the tone of the noise on both the sm58 recording and the Alice recording they seem to be the same which could indicate that it may just be preamp noise. It should be noted that I am using a mic booster with the sm58 so I could get the levels similar which also requires phantom power so maybe it is just he phantom power noise from the preamp but Idk. I also realized that when I plug in my audio interface to my laptop there is a faint high pitch noise coming from the interface itself that isn't present when I plug it into my main computer. The interface is bus powered and it might be that it's not producing the cleanest power and maybe this is contributing to the noise I am hearing in the recordings. I would check it against results from my main computer but the noise floor would increase due to fan noise of my main computer so you couldn't hear the noise floor of the actual mic anymore you would hear fan noise instead.

All in all the noise floor is more that workable I just wanted to make sure I am not missing something that is causing noise from a build standpoint since its my first one. Thanks and I will check back in when I get a battery for my multimeter.

 

[wamosjk]

Just checking back in here. I ordered a different body because I thought maybe I was getting some interferance somewhere due t othe cheap body I put the mic in. and I re wired the xlr jack with new capacitors and wires. I am still getting the same results. I did a couple more recordings here just to be sure of my earlier findings. The gain dial on my Tascam 1x2hr is set at around 70% for both recordings.

Alice Mic Recording

View attachment alice3_session.wav
So I load this recording into ardour and I do a loudness analysis on it I am getting around -50dBFS peaks. I did a loundess analysis on a few sections of the recording in several different spots just to be sure I wasn't hitting a spike somewhere and they are all around -50dBFS give or take a little. Then I did a recording of the sm58 with a mic booster. The mic booster gives the output levels to be about the same on both mics and it introduces 48v phantom power which I wanted to see if maybe that had anything to do with the noise from the preamp. Here is that recording.

SM58 Recording
View attachment 58_session.wav
When I do a loudness analysis of this recording on several different spots I am getting anywhere from -69dBFS to -65dBFS. So I can ony assume that the self noise level of the alice mic is in the -15 to -20 range which is way above the -6db that the circuit is supposed to be. Am I correct in this or am I understanding this wrong? I know they say db A weighted for self noise so I may be unserstanding this wrong. All I know is it's very difficult to keep the Alice mic under 60db on my meter when recording which is what I like to keep recordings at. There may be something else at play here but I am pretty much out of ideas as to what they could be other than the circuit itself. I still like the way the mic sounds so I have to just run it through a denoise plugin but I wish I didn't have to.

... Plus wiring the capsule signal via the PCB itself (with zero attempt to mitigate anything) is... questionable to say the least, and may or may not add extraneous noise to begin with.

I'm not really sure what this means. Mitigate what exactly? I followed the instructions on the wiring diagram that JLI provides.

I can't say that the noise is excessive.
Also compared to the SM58 the level seems to be higher with the same setting of the record volume.
This is normal, because in general a condenser microphone has a higher output at the same SPL than a dynamic microphone.

Ya I'm aware that the noise floor should be more in a condenser. The problem is that I am getting considerably more. Like I said before it seems to be -15db to -20db more noise. This circuit is supposed to be a quiet circuit.

That is ultra simple to just check. Do you not own a voltmeter of any kind? Measure the voltage on pin 2 and pin 3, and at the zener diode (using pin 1 as the reference terminal). It would be really obvious if one of the output pins had a short to 0V.
The OPA1642 op-amp will run down to under 5V supply voltage, so you might not notice even if that did occur, so easiest to just check.

I checked the voltages and everything seems to be fine. I also put it in a new body just to be sure and rewired the xlr connection and am still getting the same results as you can see by my latest recordings. I'm beginning to think this is just the noise level I am going to get from this circuit which is unfortunate because it is very difficult to keep under the -60db noise floor I like to keep recordings at.

Also wanted to say thanks to all of the replies here everyone has been very helpful.

 

[rogs]

Most of the noise from your 'Alice' mic recording is low frequency hum and 'room rumble'.
Simply applying an 'A' weighted EQ curve (which is how most mic noise figures are quoted) will dramatically improve your measured noise figure.

Here's your Alice file with A weighting applied......
View attachment OPA A weighted.wav
The perceived noise floor is dropped by around 18dB. Doesn't seem an excessive noise floor to me?

 

[wamosjk]

Most of the noise from your 'Alice' mic recording is low frequency hum and 'room rumble'.
Simply applying an 'A' weighted EQ curve (which is how most mic noise figures are quoted) will dramatically improve your measured noise figure.

Here's your Alice file with A weighting applied......
The perceived noise floor is dropped by around 18dB. Doesn't seem an excessive noise floor to me?

Thanks for this. So your saying more than likely my actual problem is the untreated space and the ambient room noise that the condenser is picking up. Like I said in earlier posts on this thread I have always used dynamics and this is the first condenser I've used in this space (I recently moved and haven't gotten around to treating the room yet). So the condenser is more sensative and is probably picking up a lot more of the room noise and that is why there is so much more noise than the sm58. I am not really sure if this noise floor seems excessive. The only thing I know is I notice it in my recordings. I've been planning on making some sound panels so I will probably get to doing that and see if it helps. It will definitely help with reflections because they are pretty bad right now.

 

[rogs]

Thanks for this. So your saying more than likely my actual problem is the untreated space and the ambient room noise that the condenser is picking up.

Yes

 

[Voyager10]

Listening to the original clips in Post #1 it seems to me that the speech level is quite a but louder on the Alice than the SM58. The noise is also louder but it's hard to say whether the signal to noise ratio is actually worse.

To get a better handle on this, you could set up both mics next to each other in front of a speaker (say 30cm from the centre of the woofer) and play a test tone through it. The mics would then run into two input channels and you adjust the preamp gains until they both read the same value (e.g. -10dBFS) on the meters.

The mics+preamp channels will then be matched for sensitivity so you can then record the noise and compare levels.

As @rogs pointed out, there's a lot more bass response on the Alice than the SM58, so if there's any LF noise in your room you'll be hearing it much better now .

 

[ccaudle]

So I can ony assume that the self noise level of the alice mic is in the -15 to -20 range which is way above the -6db that the circuit is supposed to be.

dB is a dimensionless quantity, so for a description like -15 dB or -6 dB to mean anything you have to have a reference level.
In the case of negative dB values it is values below a reference level, so -6dB would be 9dB higher level than -15dB. I assume that what you were trying to say is that you heard the OPA circuit should have around a 6 dB equivalent noise level, and you are guessing it is actually in the 15dB to 20dB range.

In the case of microphones the reference level is the acoustic SPL which would produce that output voltage. So when a microphone is referred to having a 6dB equivalent noise, it means that the electrical noise produced by the microphone circuitry (or coil resistance in the case of a dynamic mic) is the same as would be produced by acoustic noise at 6dB SPL.

To calculate that you need to know the sensitivity of the microphone, which is how much voltage it produces at the output for a given SPL acoustic signal.
That is the parameter you are missing in order to calculate the effective noise level of the microphone. Well, one of the parameters, in order to have a meaningful number using the audio software you would also need to know the conversion from voltage to full scale digital so that you would be able to work backwards from the dB FS display of the software to the electrical level, and from there to the equivalent acoustic level.

I'm not really sure what this means [wiring the capsule signal via the PCB itself (with zero attempt to mitigate anything)]. Mitigate what exactly?

Mitigate the parasitic capacitance. Any time there are two conductors in proximity to each other there is a capacitance between those two conductors. A condenser microphone is another way of saying a capacitance microphone, and any additional capacitance on the connection from capsule to buffer amplifier is an impedance in parallel to the capsule which reduces the signal level from the capsule. The finite resistance of the circuit board material is also in parallel with that connection and will reduce the signal level.
Microphones will often have the gate of the JFET, or the input pin of the op-amp, lifted into the air and soldered to a teflon stand-off to reduce the leakage current across the circuit board and to reduce the capacitance to nearby copper on the circuit board.
That is really inconvenient for mass production, so there are other things you can do, like surround the connection with a low impedance copy of the signal (from the op-amp output) to eliminate any voltage difference between the capsule connection and the closest conductor. Zero voltage difference means zero current flow, effectively nullifying the parasitic impedances. That has to be done at the circuit design stage, so nothing you needed to do differently, I think Khron was just pointing out that the JLI implementation did not do everything it is possible to do relatively easily to maximize the signal-to-noise ratio.

 

[AUDIO FREQ]

I have the same build, it's the power supply board. The one with the hex chip. I swapped it out with a different power supply board and the noise was virtually gone. No idea why, but that board adds some noise...

 

[Khron]

I have the same build, it's the power supply board. The one with the hex chip. I swapped it out with a different power supply board and the noise was virtually gone. No idea why, but that board adds some noise...

The layout is........ "sub-optimal", let's call it (although the first term that came to mind was more in the direction of "apalling")

 

[jp8]

The layout is........ "sub-optimal", let's call it (although the first term that came to mind was more in the direction of "apalling")

Yes, it violates almost any EMC design guideline that one can think of. Horrible design. It should be designed as compact as possible, with RF/EMC design rules in mind. The circuit I use measures ~10x17mm or so with "big" 0805 SMTs. Zero RF residuals or noise in the audio signal.

Jan