Help Needed: Detecting High-Pitched Whining from LED Bulbs – Best Mic & Setup?

[itchaboyagain]

Hi everyone,

I'm working on a project to detect and document high-pitched whining sounds from LED light bulbs. These noises are subtle, and I want to ensure that I capture them accurately if/when they’re there.

Audio science is not my area of expertise so I was hoping some of you kind folk could help set me on the right path!

Here are some key details:

Goal: Detect and analyze faint high-frequency noise (potentially in the range of 15kHz to 20kHz) emitted by some LED light bulbs.
Requirements: Low self-noise microphone with flat frequency response.
Budget: I'm open to spending more if it means higher accuracy, but I'd prefer to keep costs reasonable.

I've been considering options like the Earthworks M23 or MiniDSP UMIK-1. I’d also love some advice on spectral analysis software that would allow be to export graphs of any readings I take. I’ve dabbled a bit into this world and BOY am I overwhelmed. I basically just need to show where the peak frequencies are (if there are any). I’ve stumbled across Open Sound Meter which so far has been a breathe of fresh air due to its simplicity, but if there are better options I’m open to it.

Has anyone here tackled a similar project or have experience with microphones and setups suited for detecting high-pitched sounds? Any recommendations or tips would be greatly appreciated!

Some of these lights are smart lights and emit some EMF, so a setup that’s not prone to EMF interference would also be ideal.

Thanks in advance!

 

[DVDdoug]

I can't make a recommendation...

Measurement mics aren't necessarily flat but they come with a calibration curve (file) so the measurement ends-up flat.

It's probably not that important for it to be flat, since you are only interested in the high frequency range. You'll need filtering or spectral analysis or else you'll mostly be measuring low-frequency noise.

USB measurement mics are calibrated to a known sensitivity so they can easily measure the SPL level. But they also tend to have low(er) sensitivity so they don't overload with loud sounds.

With analog mics, the preamp gain has to be calibrated (with a USB mic or SPL meter) to get a known SPL level.

Most SPL meters are A-weighted so they roll-off the highest & lowest frequencies. Software with a calibrated mic is likely to be better.

 

[Sokel]

At first I thought it was about the electrical interference shown at line-level measurements like this:




(That's with the Multitone Analyzer which you can use)

Which is near impossible to get rid off other than using a different phase of your electrical installation (as with HVACs,motors,etc no matter how good your grounding is or using unbalanced or balanced gear.

On the other hand I have never catch any kind of interference like this with acoustic measurements.
Next time I'll get close to see,I have a bunch of dim-able leds that I can measure at different intensities so to also see the resistor contribution to this.

 

[kemmler3D]

I've been considering options like the Earthworks M23 or MiniDSP UMIK-1.

I am assuming that the Earthworks has lower noise, so I'd go with that. Noise is typically going to be your biggest obstacle when recording quiet stuff. You will also want to find an interface with low noise, depends on your budget but MOTU M2 / M4 aren't bad, RME might be better but is $$$.

I’d also love some advice on spectral analysis software that would allow be to export graphs of any readings I take. I’ve dabbled a bit into this world and BOY am I overwhelmed. I basically just need to show where the peak frequencies are (if there are any).

REW will do this for you and it's free, just open up the RTA tab and hit the red button. It should show you the peak frequencies as long as they're above noise.

Can I ask what the end goal of this project is?

 

[Blumlein 88]

Lewitt LCT 540 subzero is about the lowest self noise mic without getting exotic. I believe it also has higher than average sensitivity. REW or Multitone software are free and good. You will need an audio interface and to calibrate for SPL. Other low noise mics are Rode NT1, and Shure KSM44a. Earthworks are probably not lowest noise or highest sensitivity due to smaller diaphragm size.

 

[itchaboyagain]

I am assuming that the Earthworks has lower noise, so I'd go with that. Noise is typically going to be your biggest obstacle when recording quiet stuff. You will also want to find an interface with low noise, depends on your budget but MOTU M2 / M4 aren't bad, RME might be better but is $$$.

REW will do this for you and it's free, just open up the RTA tab and hit the red button. It should show you the peak frequencies as long as they're above noise.

Can I ask what the end goal of this project is?

We do LED testing, see here, and have had more than a few people mention they wish we'd tested for audible whine, something I hadn't noticed or considered. So I thought to myself how hard can it be?

Seems like the approach is a low-noise setup, just gotta figure out the whole software thing, calibration files, etc. It's all foreign and new to me. It looks like measuring the visual spectrum is easier than the audible one haha

 

[itchaboyagain]

Lewitt LCT 540 subzero is about the lowest self noise mic without getting exotic. I believe it also has higher than average sensitivity. REW or Multitone software are free and good. You will need an audio interface and to calibrate for SPL. Other low noise mics are Rode NT1, and Shure KSM44a. Earthworks are probably not lowest noise or highest sensitivity due to smaller diaphragm size.

Oh, that does look nice! Low noise seems like it would be an advantage for this type of work. I hadn't considered something like this.

 

[kemmler3D]

We do LED testing, see here, and have had more than a few people mention they wish we'd tested for audible whine, something I hadn't noticed or considered. So I thought to myself how hard can it be?

Seems like the approach is a low-noise setup, just gotta figure out the whole software thing, calibration files, etc. It's all foreign and new to me. It looks like measuring the visual spectrum is easier than the audible one haha

Oh, very cool! I think this is very doable, although in this context I'm starting to wonder if there isn't a standard testing protocol for this kind of thing. Maybe ASTM has something on the subject? https://www.astm.org/e1686-16.html

Depending on how far you want to go with this, it might be worth investing in a vocal / recording / isolation booth to knock down the background noise even more. But if people can hear these things from across the room I don't think you'll have any trouble recording and characterizing the noise at a practical level, with a decent mic and interface.

 

[EERecordist]

Hi everyone,

I'm working on a project to detect and document high-pitched whining sounds from LED light bulbs. These noises are subtle, and I want to ensure that I capture them accurately if/when they’re there.

Audio science is not my area of expertise so I was hoping some of you kind folk could help set me on the right path!

Here are some key details:

Goal: Detect and analyze faint high-frequency noise (potentially in the range of 15kHz to 20kHz) emitted by some LED light bulbs.
Requirements: Low self-noise microphone with flat frequency response.
Budget: I'm open to spending more if it means higher accuracy, but I'd prefer to keep costs reasonable.

I've been considering options like the Earthworks M23 or MiniDSP UMIK-1. I’d also love some advice on spectral analysis software that would allow be to export graphs of any readings I take. I’ve dabbled a bit into this world and BOY am I overwhelmed. I basically just need to show where the peak frequencies are (if there are any). I’ve stumbled across Open Sound Meter which so far has been a breathe of fresh air due to its simplicity, but if there are better options I’m open to it.

Has anyone here tackled a similar project or have experience with microphones and setups suited for detecting high-pitched sounds? Any recommendations or tips would be greatly appreciated!

Some of these lights are smart lights and emit some EMF, so a setup that’s not prone to EMF interference would also be ideal.

Thanks in advance!

Send a note to Earthworks support on your electrical noise concerns. You can do

Hi everyone,

I'm working on a project to detect and document high-pitched whining sounds from LED light bulbs. These noises are subtle, and I want to ensure that I capture them accurately if/when they’re there.

Audio science is not my area of expertise so I was hoping some of you kind folk could help set me on the right path!

Here are some key details:

Goal: Detect and analyze faint high-frequency noise (potentially in the range of 15kHz to 20kHz) emitted by some LED light bulbs.
Requirements: Low self-noise microphone with flat frequency response.
Budget: I'm open to spending more if it means higher accuracy, but I'd prefer to keep costs reasonable.

I've been considering options like the Earthworks M23 or MiniDSP UMIK-1. I’d also love some advice on spectral analysis software that would allow be to export graphs of any readings I take. I’ve dabbled a bit into this world and BOY am I overwhelmed. I basically just need to show where the peak frequencies are (if there are any). I’ve stumbled across Open Sound Meter which so far has been a breathe of fresh air due to its simplicity, but if there are better options I’m open to it.

Has anyone here tackled a similar project or have experience with microphones and setups suited for detecting high-pitched sounds? Any recommendations or tips would be greatly appreciated!

Some of these lights are smart lights and emit some EMF, so a setup that’s not prone to EMF interference would also be ideal.

Thanks in advance!

Fun project! Looking forward to results, you are welcome to post them here.

Get the microphone as close as possible to the device under test and in as best sound insulated environment you can find. Some universities may have mini-anechoic chambers, and big cities/very big companies/some audio transducer companies have them in product test and certification labs.

The mounting of the bulb is going to have an effect. Any kind of dimmer supplying it is going to have an effect.

When you find noise, consider doing and documenting a teardown.

 

[itchaboyagain]

At first I thought it was about the electrical interference shown at line-level measurements like this:


View attachment 390118

(That's with the Multitone Analyzer which you can use)

Which is near impossible to get rid off other than using a different phase of your electrical installation (as with HVACs,motors,etc no matter how good your grounding is or using unbalanced or balanced gear.

On the other hand I have never catch any kind of interference like this with acoustic measurements.
Next time I'll get close to see,I have a bunch of dim-able leds that I can measure at different intensities so to also see the resistor contribution to this.

Yeah, I'm not sure how common it is for LEDs to emit any kind of noise but apparently, it does happen. I never noticed it and I've tested well over 300 different kinds fairly extensively. Not sure if this project is work the time, effort, and money, but figured I'd see what it would take.

Thank you for the link!

 

[itchaboyagain]

Oh, very cool! I think this is very doable, although in this context I'm starting to wonder if there isn't a standard testing protocol for this kind of thing. Maybe ASTM has something on the subject? https://www.astm.org/e1686-16.html

Depending on how far you want to go with this, it might be worth investing in a vocal / recording / isolation booth to knock down the background noise even more. But if people can hear these things from across the room I don't think you'll have any trouble recording and characterizing the noise at a practical level, with a decent mic and interface.

Oh duh! Good idea. I'll check for industry testing standards and see if I can find anything there. Thank you for the idea.

I was thinking of putting the mic inside of my integration sphere to cut down on exterior "noise", although I'm not sure just how much sound that blocks.

 

[itchaboyagain]

Send a note to Earthworks support on your electrical noise concerns. You can do


Fun project! Looking forward to results, you are welcome to post them here.

Get the microphone as close as possible to the device under test and in as best sound insulated environment you can find. Some universities may have mini-anechoic chambers, and big cities/very big companies/some audio transducer companies have them in product test and certification labs.

The mounting of the bulb is going to have an effect. Any kind of dimmer supplying it is going to have an effect.

When you find noise, consider doing and documenting a teardown.

Will do! I'd love to document it here if/when this project takes shape. I hadn't considered sending Earthworks an email haha I guess they'd be the ones to know! Thank you.

 

[kemmler3D]

although I'm not sure just how much sound that blocks.

Well, once you have the mic you can stick it in there and see how many dB lower the background noise is. I think you can get 9-12dB lower without using anything too fancy.

I promise that testing this stuff (at least at a casual level) isn't actually very hard. Plenty of people around here do similar stuff just to get their subwoofers sounding good.

 

[Timcognito]

Have you considered building a small Anechoic Chamber. Seems like that would be helpful.
Looks like kemmler beat me to it.

 

[Timcognito]

McMaster-Carr

McMaster-Carr is the complete source for your plant with over 595,000 products. 98% of products ordered ship from stock and deliver same or next day.

www.mcmaster.com

 

[Blumlein 88]

Oh, that does look nice! Low noise seems like it would be an advantage for this type of work. I hadn't considered something like this.

The Lewitt is not only low noise, but about among the most sensitive. Those are the two things you most need.

The free software is pretty easy to learn. You might want to read up on FFTs if that is not something you are familiar with. High count FFTs let you look well below the normal noise floor. It is something REW and Multitone let you do. The people who wrote both are members here.

As for calibration an advantage of the Umik 1 is it is self calibrated in REW. As they are less than $100, I might suggest getting one of those and see how far you can get in your testing with it. Then if you later go with a better microphone this one can be used to calibrate the other microphones for SPL. You also might want to get it from Cross Spectrum labs as they calibrate them more carefully. While not the most sensitive or lowest noise the Umik 1 is good enough if the noise is loud enough to be heard you'll likely pick it up. It does have some noise spikes in its noise profile that might be an issue, but still pretty cheap to take a first shot at it without spending too much.

 

[Blumlein 88]

Yeah, I'm not sure how common it is for LEDs to emit any kind of noise but apparently, it does happen. I never noticed it and I've tested well over 300 different kinds fairly extensively. Not sure if this project is work the time, effort, and money, but figured I'd see what it would take.

Thank you for the link!

I think it is most common in dimmable bulbs. I've heard a few that emit a barely audible high pitched whine when they are dimmed. Of course I'm older and younger people likely hear it more than I do. You can look at dimmable LED reviews on Amazon and cheaper ones often have people complain of the whine. I had one that would only do this while my HVAC was on (lower voltage I guess). I measured a low level signal around 11 khz. The fixture it was in made it a bit directional so you didn't hear it unless right below it.

 

[Beave]

A few years ago my girlfriend bought a pack of four non-dimmable LED lights. She quickly got annoyed by the high-pitched sound they emitted, so she gave them to me (?!?!). She was using them in a fixture right by her couch, where the sound would annoy her when she was reading.

I put them in my bathroom, heard a faint high-pitched sound (11kHz might be right), but I didn't care since it's the bathroom.

 

[restorer-john]

It's not the LEDs themselves, it's the inverters in the screw/bayonet caps.

And yes, they can be annoying, just as various phone chargers, laptop supplies and SMPS wall warts can all make noises varying from subtle to downright annoying.

I'd also consider the power supply itself, as distorted mains can make LEDs more or less noisy depending on the situation. I would also be looking a voltage variation (supply) and how that can affect the volume of the choke/txf noise in the inverters.

 

[Blumlein 88]

Tried a few lamps with LED bulbs easy to reach with the Umik 1.

Only found one with noise. You'll never hear this. White is with the bulb on and purple with it off. Mike was 2 cm from the bulb. 64k FFT. 44khz sampling. Utilitech bulb from Lowes. 60 watt equivalent 2700k color temp.