Gearslutz

Gearslutz (https://www.gearslutz.com/board/)
-   All Things Technical (https://www.gearslutz.com/board/all-things-technical/)
-   -   Sennheiser mkh 8040 noise over 20kHz (https://www.gearslutz.com/board/all-things-technical/1081741-sennheiser-mkh-8040-noise-over-20khz.html)

skyb 22nd April 2016 10:38 AM

Sennheiser mkh 8040 noise over 20kHz
 
1 Attachment(s)
Hi!

I am interested in hearing about others experience with the mkh 8040 (or other in the same series) and self noise in the mic. It is rated at 13dBa and used in a traditional recording sense it is a beautiful microphone with low self noise. My issue is with the self noise over 20 kHz . Sennheiser rates it to go to 50kHz but over 20kHz in my experience it is extremely noisy. This is of course only possible to hear when the recording is slowed down, bringing the high frequency info down to the audible range, something i like to do in sound design.

This post on stackexchange discusses the matter and one guy( Pavel Doreuli) even posted that on some gear it wasn't noisy. I have not managed to contact him to get an answer to what gear it was. pitch shifting - Wide Frequency Response Microphones - Sound Design Stack Exchange

I mean even if your not pitching down recordings for that effect, but you bought these mics because of the advantages of the harmonic content recorded over 20khz creating subharmonics in the audible range..isnt all this noise up there interfering with that? just a thought.

anyway I have borrowed a mic from sennheiser to check mine aren't faulty and it was the same thing with my setup: RME UC (except the mic i borrowed was about 5-6db lower overall). I am taking my microphone to work today to test it on other preamps. ill post the results here.

TMetzinger 22nd April 2016 11:10 AM

Unless the rest of your signal chain is guaranteed to be clean at 20-50 KHz, you can't be sure where the noise is. Preamp, the analog part of your ADC can all be culprits.

skyb 22nd April 2016 01:44 PM

1 Attachment(s)
Well yes this is true but I have now tested several other preamps and I`m pretty sure its the microphone. This is because the noise is only apparent when that microphone is connected to any preamp i have tested (ssl, apollo, RME). As you can see on the screenshot from RXdenoiser which goes over 20khz there is a steep slope from under 20k and up towards 50k. Thats the mike in recording room with gain of about 46dB on an UA Apollo Quad running at 96kHz samplingrate.

Obviously the noise is more noticeable when recording quiet subjects and not so much on loud noises, but its still there...

:)

dseetoo 22nd April 2016 04:19 PM

The noise you refer to is real. It is a dirty secret microphone manufactures would not tell you when they extended the frequency range of their microphones. The listed noise spec in their literature won’t reflect the noise once the noise measurement is weighted. All that comes from very aggressive analog high frequency boost in the mic circuitry in to order to achieve extended high frequency response. This not only happens with Sennheiser microphones but also happen with Schoeps CMC6 xt body, and Sanken CO-100K.

When 96KHz recording technology became widely available every recording engineer wanted to have microphones that can take advantage of the newly available bandwidth so the microphone manufactures had to re-invent the wheels, so to speak. Since nobody can hear anything above 20 KHz, the huge amount of noise above 20 KHz won’t matter, thus nobody noticed it, either. Not until someone looked at the spectrogram of their recordings, that is.



Best regards,

Da-Hong Seetoo

Jim Williams 22nd April 2016 05:06 PM

It is also done on the Pearl rectangular capsules from Sweden. Those have a natural roll-off and need corrective EQ in the mic's amp. Better caps and transistors can lower it.

MKH are radio frequency biased designs. It's natural that high frequency noise will increase with rising frequency in those designs. Shifting down the frequency makes it more apparent as you pushed that noise down an octave into the audible range.

Sweep these mics on a decent analyzer like Audio Precision and you will see the rising noise vs frequency. It can be problematic on wide bandwidth 192k recordings as those upper frequency noises can modulate down to the audible range.

Best to use these for their attributes, a resistance to moisture caused noise in humid places.

Audiop 22nd April 2016 05:45 PM

Quote:

Originally Posted by skyb (Post 11853091)
I mean even if your not pitching down recordings for that effect, but you bought these mics because of the advantages of the harmonic content recorded over 20khz creating subharmonics in the audible range..isnt all this noise up there interfering with that? just a thought.

There's no advantage capturing ultra sound for "subharmonics".

If it would exist (to any practical/significant degree) this would be captured in the air, in the audible range, during recording.

What you can get dealing with strong ultra sound is intermodulation distortion in following stages. That's not hifi though and either you lowpass your signal and/or you make sure the whole chain has high performance linearity wise (read low distortion a decent way up above 20kHz). Right Jim? ;-)

Quote:

anyway I have borrowed a mic from sennheiser to check mine aren't faulty and it was the same thing with my setup: RME UC (except the mic i borrowed was about 5-6db lower overall). I am taking my microphone to work today to test it on other preamps. ill post the results here.
Are you sure you made no mistake during those tests? A noise difference of 5-6dB is huge. If Senn QC is that weak I would want to know it.

Deleted User 22nd April 2016 07:00 PM

Quote:

Originally Posted by dseetoo (Post 11853553)
When 96KHz recording technology became widely available every recording engineer wanted to have microphones that can take advantage of the newly available bandwidth so the microphone manufactures had to re-invent the wheels, so to speak.

Not every recording engineer. :)

skyb 22nd April 2016 09:35 PM

Quote:

Originally Posted by dseetoo (Post 11853553)
The noise you refer to is real. It is a dirty secret microphone manufactures would not tell you when they extended the frequency range of their microphones. The listed noise spec in their literature won’t reflect the noise once the noise measurement is weighted. All that comes from very aggressive analog high frequency boost in the mic circuitry in to order to achieve extended high frequency response. This not only happens with Sennheiser microphones but also happen with Schoeps CMC6 xt body, and Sanken CO-100K.

When 96KHz recording technology became widely available every recording engineer wanted to have microphones that can take advantage of the newly available bandwidth so the microphone manufactures had to re-invent the wheels, so to speak. Since nobody can hear anything above 20 KHz, the huge amount of noise above 20 KHz won’t matter, thus nobody noticed it, either. Not until someone looked at the spectrogram of their recordings, that is.



Best regards,

Da-Hong Seetoo

I was fearing this was the case.. oh well at least they are also fantastic mics. :)

Quote:

There's no advantage capturing ultra sound for "subharmonics".

If it would exist (to any practical/significant degree) this would be captured in the air, in the audible range, during recording.

What you can get dealing with strong ultra sound is intermodulation distortion in following stages. That's not hifi though and either you lowpass your signal and/or you make sure the whole chain has high performance linearity wise (read low distortion a decent way up above 20kHz). Right Jim? ;-)
Ah yes. I wasn't thinking straight on that one. Thanks for clarifying on that.

Quote:

Are you sure you made no mistake during those tests? A noise difference of 5-6dB is huge. If Senn QC is that weak I would want to know it.
I'm pretty sure i didn't mess the test up as I had the loaned 8040 connected to one of the channels on my rycote mzl ortf setup which i know is in top condition. The RME was set to Stereo link on 1/2 so gain was exactly the same. And I placed myself in the center between the mikes and sang them a song. The channel with the loaned mkh8040 was significantly lower. Bringing it up about 5-6dB in post made the stereo image balanced. The loaned mike is of a much lower serialnr than mine are. It even has a larger font printed on the side, and the insides where the connection between mike amp and xlr connector is looked slightly different. Well not different, but worse manufactured, like the copper rings weren't centered on the mike body.


Thanks for everyones input on this. It has been bugging me for a while... :)

David Spearritt 22nd April 2016 11:05 PM

I am grateful for the OP and posters in this thread. I have always steered clear of these extended FR microphones but didn't suspect the 8000 range. Will strike them off any consideration in the future.

There are lots of these ultrasonic traps in high end audio too these days. Its somewhat ironic how the hi-res audio guys are trying to prove that you can hear above 20kHz FR and IR, and hence need high sample rates, but that the ultrasonic noise shaping in DSD cannot be heard. What this ultrasonic hash is doing to in-band audio is probably more significant than most are willing to admit, especially if it originates well upstream in the chain.

Still, I am happily using Hypex Class D amps for my main monitors and there is an awful lot of ultrasonic hash coming out of those. But they are at the end of the chain, and I know the drivers cannot follow the signal in any way.

jetam 23rd April 2016 12:59 AM

Quote:

Originally Posted by David Spearritt (Post 11854363)
I am grateful for the OP and posters in this thread. I have always steered clear of these extended FR microphones but didn't suspect the 8000 range. Will strike them off any consideration in the future.

There are lots of these ultrasonic traps in high end audio too these days. Its somewhat ironic how the hi-res audio guys are trying to prove that you can hear above 20kHz FR and IR, and hence need high sample rates, but that the ultrasonic noise shaping in DSD cannot be heard. What this ultrasonic hash is doing to in-band audio is probably more significant than most are willing to admit, especially if it originates well upstream in the chain.

Still, I am happily using Hypex Class D amps for my main monitors and there is an awful lot of ultrasonic hash coming out of those. But they are at the end of the chain, and I know the drivers cannot follow the signal in any way.

Just because there is ~10 dB more noise above 20 Khz than below, it doesn't mean that the mic is bad.

matucha 23rd April 2016 02:10 AM

This is a spectrogram of clock ticking. MKH80, MKH405 and gefell M300. Recorded with Nagra LB. Both MKHs have higher noisefloor at both extremes of the spectrum. Though MKH80 is perhaps a bell shape EQ while MKH405 is a shelf?

https://www.gearslutz.com/board/atta...0_405_m300.jpg

To my ears gefell is the noisiest of the three. MKH80 the cleanest (by a lot). While it doen't suppose to record ultrasonic freqs there is quite a lot of relatively healthy signal up to at least 35khz. Perhaps not flat though.

MKH80 and 60 have similar noise characteristics at the top, though there is no signal above 20khz. At least I've never seen anything on spectrogram, unlike on 80 a 405.

John Willett 23rd April 2016 09:44 AM

Quote:

Originally Posted by matucha (Post 11854583)
This is a spectrogram of clock ticking. MKH80, MKH405 and gefell M300. Recorded with Nagra LB. Both MKHs have higher noisefloor at both extremes of the spectrum. Though MKH80 is perhaps a bell shape EQ while MKH405 is a shelf?

https://www.gearslutz.com/board/atta...0_405_m300.jpg

To my ears gefell is the noisiest of the three. MKH80 the cleanest (by a lot). While it doen't suppose to record ultrasonic freqs there is quite a lot of relatively healthy signal up to at least 35khz. Perhaps not flat though.

MKH80 and 60 have similar noise characteristics at the top, though there is no signal above 20khz. At least I've never seen anything on spectrogram, unlike on 80 a 405.

Sennheiser MKH 20/30/40 series, right up to the MKH 80, all have a steep roll-off above 20kHz.

The MKH 406 is a very old mic. - the forerunner of the MKH 40 and discontinued about 1985 - also rolls off steeply above 20kHz.

The above are both RF condensers.

The M 300 is an AF condenser with a capsule that rolls off naturally above 20kHz.

Ulrich 23rd April 2016 10:07 AM

Quote:

Originally Posted by David Spearritt (Post 11854363)
I am grateful for the OP and posters in this thread. I have always steered clear of these extended FR microphones but didn't suspect the 8000 range. Will strike them off any consideration in the future.

Huh? The MKH 8040 is a very impressive mic. I will stay with omnis for my main tasks, but these cardiods are really remarkable. I think there is no evidence that the HF noise causes any problems in real world.

Rolo 46 23rd April 2016 10:24 AM

Its all noise over 20Khz and most cant hear it
Hands up who can ?

John Willett 23rd April 2016 11:25 AM

Quote:

Originally Posted by Rolo 46 (Post 11855014)
Its all noise over 20Khz and most cant hear it

It only becomes a possible problem if you are using the high freqencies for special effects - eg: recording at 96 or 192kHz and then playing back at 48kHz.

Me, my hearing probably rolls off at about 12kHz now, maybe lower wworried

tailspn 23rd April 2016 03:55 PM

Quote:

Originally Posted by David Spearritt (Post 11854363)
....but that the ultrasonic noise shaping in DSD cannot be heard. What this ultrasonic hash is doing to in-band audio is probably more significant than most are willing to admit, especially if it originates well upstream in the chain.

The unfiltered noise of 64fs DSD is in the area of -110dB at 20KHz, and rises to about -55dB at 100KHz. That noise envelope shifts up an octave at 128fs (~-110db at 40KHz), and two octaves at 256fs (~-110dB at 80KHz). The most important consideration though is the noise is uncorrelated, so does not produce modulation sidebands within the audio range.

aracu 23rd April 2016 07:00 PM

In order to get the high frequency noise from the mics on the actual recording the
recorder and preamps must have a similar frequency range.

David Spearritt 24th April 2016 12:59 AM

Quote:

Originally Posted by Rolo 46 (Post 11855014)
Its all noise over 20Khz and most cant hear it
Hands up who can ?

We can't hear it, but the preamps, and playback amplifier can and possibly distort in-band in response to it, which we can hear.

David Spearritt 24th April 2016 01:01 AM

Quote:

Originally Posted by tailspn (Post 11855363)
The unfiltered noise of 64fs DSD is in the area of -110dB at 20KHz, and rises to about -55dB at 100KHz. That noise envelope shifts up an octave at 128fs (~-110db at 40KHz), and two octaves at 256fs (~-110dB at 80KHz). The most important consideration though is the noise is uncorrelated, so does not produce modulation sidebands within the audio range.

and it's independent of signal level, so in quiet pianissimo, or reverb tails or lute recordings, it is maybe only 30dB below signal level.

tailspn 24th April 2016 02:13 AM

Quote:

Originally Posted by David Spearritt (Post 11856144)
and it's independent of signal level, so in quiet pianissimo, or reverb tails or lute recordings, it is maybe only 30dB below signal level.

Yes, isn't that terrific? Three octaves above where it becomes measurable, and -30dB below. If it was tape, the noise/hiss would be within the audio band, and 20 dB above. Sounds pretty good to me, and that's at 64fs DSD.

skyb 25th April 2016 07:43 AM

Quote:

Originally Posted by John Willett (Post 11855078)
It only becomes a possible problem if you are using the high freqencies for special effects - eg: recording at 96 or 192kHz and then playing back at 48kHz.

Me, my hearing probably rolls off at about 12kHz now, maybe lower wworried

This is what I was saying, only a problem when pitching down high samplerate recordings. I dont feel it is audible when playing back at same samplerate. I just feel that when you're selling a mic at this price, and stating that it captures up to 50kHz, your buyers are going to expect the same quality within the frequency range stated. I probably should of done more research on the matter and also thought twice about what self noise stated as dB(a) actually means. I see that the self noise induced by the mike starts to rise exactly where the dBa curve starts to curve downwards...

studer58 25th April 2016 09:34 AM

It's not just MKH mics that display HF noise...there was a class of AD chips that had a similar characteristic and were used in the Echo Audiofire interfaces, and perhaps others of the era as well ?

You can read a review of how it manifested itself here...in particular the section titled 'To Infinity': Echo Audiofire 12

Jim Williams 29th April 2016 05:26 PM

Quote:

Originally Posted by aracu (Post 11855616)
In order to get the high frequency noise from the mics on the actual recording the
recorder and preamps must have a similar frequency range.

My mic preamps are all current feedback designs and do 30 mhz, far beyond any ADC. Track at 192k hz and you can encode a good amount of untaimed high frequency noise. No, you won't hear it directly (my cat can) but those random noise frequencies can enharmonically beat against audible frequencies causing intermodulation distortion. That is another reason why I avoid class D monitor amps here.

Examine any of your condenser mics on a scope and you will find the other dirty little secret: many popular brands have considerable polarization osciallator leakage onto the audio. It looks like a fat band riding over the audible waveforms. When timebased it shows the oscillator frequency.

That also causes intermodulation distortion and adds more noise to the upper regions. It has in severe cases overloaded my mic preamps at higher gains as that leakage is not attenuated with my wideband mic preamp designs.

Careful layout and screening can reduce that leakage to almost nothing. Many of my mics have been modified to do just that. Sometimes a small piece of copper foil will do the trick if placed over the oscillator transistor and underneath it.

sd270 29th April 2016 06:03 PM

I use 8040 and 8020 into Nagra recorders at 24/96 for location recordings and don't hear any audible artifacts, not a problem for me.

jimjazzdad 30th April 2016 01:12 PM

Quote:

Originally Posted by Jim Williams (Post 11868433)
...Examine any of your condenser mics on a scope and you will find the other dirty little secret: many popular brands have considerable polarization osciallator leakage onto the audio. It looks like a fat band riding over the audible waveforms. When timebased it shows the oscillator frequency...

Jim, does that apply to RF mikes, like Sennheiser produces, as well?

dseetoo 30th April 2016 03:55 PM

No. Because the frequency Sennheiser uses is too high (8-10MHz) for it to survive the trip out of the mic body.

Plush 1st May 2016 08:41 PM

Using Sanken CO-100K and Senn. 8020--no problems here.

Sanken sound especially good. Very open sound--no brightness though.

skyb 1st May 2016 08:43 PM

Quote:

Originally Posted by sd270 (Post 11868553)
I use 8040 and 8020 into Nagra recorders at 24/96 for location recordings and don't hear any audible artifacts, not a problem for me.

If you had a recording with the 8040 I could listen to and test, i would be grateful.

sd270 2nd May 2016 03:21 AM

The recording with MKH8040 will be on a cd to be released, can't put samples on forum.

pieter k 2nd May 2016 06:01 PM

Very interesting discussion, especially as I'm possibly just a few days away from ordering a pair of MKH8040s, as much for their studio flexibility as for their field advantages.

So out of curiosity, does this type of extended HF response automatically indicate the raised noise floor, or is there some manufacturer or model of SDC that has managed to both extend HF and keep the noise floor in line with audible spectrum specs?