conversion from A-weighted to flat RMS

[tschrama]

Hi all,

Most mics have self noise specifications around 15dB to 30dB equivalent SPL but this is weighted with a curve. There are
DIN IEC 651
CCIR 468-3
A-weighted

This makes it difficult to compare with noise level that I usally calculate from the trace on my scope (Vnoise rms = 6 x Vnpp on scope usually works pretty good to check opamp noise specs).

Does anyone know how to comvert A-weighted noise to 'flat'-weighted noise? Is it possible, just in the right ballpark?

Is A-weighted noise something like 20dB lower that 'flat' weighted noise?

Kind regards,
Thijs

[peranders]

If you are interested in 20 - 20 000 Hz the difference is only a couple of dB's.

It's hard to make a judgement with an oscilloscope 100 MHz bandwidth and peak values only.

[tschrama]

Thanks Peranders.


I actually find it very easy to use a osciloscope. But you made a good reamark. In my case the bandwidth of the scope doesn't matter as long as the scope noise is (much) less the the device under test and the bandwidth of the device is know. I allways use Line-trigger mode to differentiate between humm and noise.

I still have a question about the various weighting curves.

The difference between
DIN IEC 651
and
CCIR 468-3
is about 10dB.

How do they compare with A-weighted?

Reagrds,
Thijs

[peranders]

The incoming bandwidth matters. 20-20kHz or 20-100 kHz or even more will matter compared to A-weighted values.

[fmak]

[QUOTE]Originally posted by peranders
[B]If you are interested in 20 - 20 000 Hz the difference is only a couple of dB's.
---------------------------------------------------
How have you concluded this. Look at the A weighting curve.

[tiroth]

Anecdotally it often seems that unweighted band-limited noise (20-20kHz) is generally about 3-6dB higher than A-weighted, but this is at best a poor rule of thumb. Peranders and I discussed this a little recently and I gave some examples where you could see big differences with A-weighted versus unweighted (but band-limited).

I believe Peranders is cautioning you that at 100MHz you are integrating noise over a much broader bandwidth, so the numbers will not be comparable in any way to A-weighted numbers. Even if the noise is the same, at 100MHz the total value will be much higher than at 20kHz.

Something like a Tek 7A22 comes in handy for this kind of work.

[tschrama]

Thanks for all the replies.

Let's not talk about the scope anymore. I realy do have that under controll.

This all concerns a Sennheiser MKH-70 microphone that seems to be off-specifications and needs to be recallibrated....

Let's talk about A-weighting and the other norms.

3dB to 6dB is a something to start with. I have the A-weighting curve in front of me and I can see the attenuation at the low and very high frequencies. I also understand why this is done and why it is therefore only suiteble for low hearing level application like background noise.

As I understand now, the A-weigthing curve is limmiting the bandwidth to 1/2 or even 1/4 of the unweighted bandwidth?
And the 1/f noise that might be will no longer be of any significance when A-weighting is applied.

But I can't seem te find information about those other two norms:
DIN IEC 651
CCIR 468-3


Anyone got some eperince or info about those....

Google turn up with all those german pages that I have trouble reading with,


Thanks,
Thijs

[Pjotr]

Originally posted by tiroth

I believe Peranders is cautioning you that at 100MHz you are integrating noise over a much broader bandwidth, so the numbers will not be comparable in any way to A-weighted numbers. Even if the noise is the same, at 100MHz the total value will be much higher than at 20kHz.

Duhhh, an oscilloscope is not integrating like a moving coil meter. It just shows you the momentary value.

With noise it is not easy to compare pp values as seen on an oscilloscope to RMS values measured with a meter. Lots depend on the “crest” factor.

BTW if you Google on noise+measurement+weighting there are tons of information.

Cheers

[Pjotr]

Hi Thijs,

Didn’t realise it, but you can also do it the other way around and use an A-weighting filter for your measurements:

http://sound.westhost.com/project17.htm

Cheers

[tschrama]

Thanks guys,

I'll have a look.

Gr,
T