What is dBr ??

[ruffrecords]

I have just been reading the specification of the Amek channel in a box (CIB) designed by Rupert Neve:

http://www.studiomanuals.com/docs/amek/amek_cib_manual.pdf

Several equivalent input noise and input balance are measured in units of dBr. That's aunit I have never heard of. What does the R stand for? - relative, resistive, Rupert?

Cheers

Ian

 

[[silent:arts]]

I (and my Neutrik A2) do know dBr as dB relative.

 

[ruffrecords]

I (and my Neutrik A2) do know dBr as dB relative.

Which begs the question, relative to what?

Edit: I just checked a tech glossary that says is is "relative to a reference level" which does not help much because the AMEK manual does not specify a reference level. If we assume it is +4dBu then it makes all the noise measurements 4dB better tha if they were specified as dBu. Sounds like a convenient marketing tool.

Cheers

Ian

 

[[silent:arts]]

I can set the reference to whatever I want - just push one button where ever you are, and it is 0 dBr.
Useful if you are more interested in dB difference than absolute values.

 

[JohnRoberts]

Yup dB relative, so it suggests relative to something other than conventional 0VU standards.

JR

 

[zebra50]

I thought dB was a relative scaling unit. That's the point!

 

[[silent:arts]]

dBr is the true and only dB 

 

[JohnRoberts]

OK I looked this up in my copy of IEEE S-100 (standards dictionary.

There are several flavors of dBr

1- dBr: power difference in dB between any point and a reference point.
2-dBRAP: dB above reference acoustical defines as  10^16 Watt
3-dBRN : dB above reference noise
4-dBrn (144 line) : Noise power measured by a set with 144 line weighting
5-dbrn(f1-f2): Flat noise power in dBrn
6-dBRNC: dB above reference noise , C message weighted
7-dBrnc0: noise power in dBrnc referred to or measured at 0TLP

I can't say that I recognize (or understand) most of the engineering jargon, and my copy of S-100 is from the '80s so that list might be a little dated.

Yes all dB are relative (power if you want to be strict) and the suffix following the dB is supposed to denote some reference power.

That said, decibels are too damn convenient to not use for voltage ratios. The qualification to excuse such behavior is to pretend that both voltages are terminated with the same resistance so compare similarly for power. 

Even this contortion doesn't work for describing voltage ratios of transformers using dB since input power and output power for transformers are always roughly the same.

Sorry for TMI...

JR

PS: Back decades ago when I first learned dB, dBu represented the rough voltage equivalent of dBm, while dBV meant that it was a voltage ratio, and proper use involved specifying the reference voltage too, i.e. 0dBV(1V) concisely described  0dB re 1V. Decible use has been pretty well bastardized for voltage ratio use over the decades but it's the only way to visualize large voltage ratios (like s/n or dynamic range) in our heads.

 

[ruffrecords]

I thought dB was a relative scaling unit. That's the point!

It is and when specifying something that is a simple ratio like gain then it is fine. However, when it is used to specify an absolute measurement you must specify what your measurement is relative to otherwise it is meaningless. That's why we have dBu - which is relative to 0.775V rms. In the old days, when everything has a 600 ohm source and load impedance, measurements were made in dBm which is relative to 1mW into a specified load.

The odd thing is, the AMEK spec uses dBu in most places but dBr (without a reference) in others. I don't mind what units they use as long as they use them properly.

Cheers

ian

 

[JohnRoberts]

+1

I recall having similar discussions back in the 80s before we had agreement on XLR wiring polarity. As I said back then, it doesn't matter which polarity you choose, as much as being clear to publish (or screen on the back panel) which you used.

Same for relative dB references. Just tell us what you are referring to.

JR

 

[moamps]

I thought dB was a relative scaling unit. That's the point!

It is and when specifying something that is a simple ratio like gain then it is fine. However, when it is used to specify an absolute measurement you must specify what your measurement is relative to otherwise it is meaningless. That's why we have dBu - which is relative to 0.775V rms. In the old days, when everything has a 600 ohm source and load impedance, measurements were made in dBm which is relative to 1mW into a specified load.

The odd thing is, the AMEK spec uses dBu in most places but dBr (without a reference) in others. I don't mind what units they use as long as they use them properly.

Cheers

ian

AMEK CiB is, as they say, "inspired by the success of the 9098i cosole". For dual mic channel Amek System 9098 - Dual Mic Amp, there is EIN:
Noise - 150R source. Figures measured with RMS rectifier, 22Hz-22kHz filter
Equivalent input noise (66dB gain) RMS -128dBu .
So, the numbers you are chasing are there.

Btw, look into specs from some another microphone preamplifiers:

Millenia HV-37
Noise
60 dB Gain, 10 Hz - 30 kHz, Inputs common, -130 dB EIN

Millenia M-2B
Noise
50 dB Gain, 10 Hz - 30 kHz, 150 ohm Source, -116 dB EIN

Avalon VT737
Noise microphone EIN -116dB, 22Hz to 22kHz unweighted

Avalon M5
Noise EIN Unweighted: -126dB 150 ohm

Focusrite ISA ONE
EIN (Equivalent Input Noise)
-126dB measured at 60dB of gain with 150 Ohm terminating impedance and 22Hz/22kHz band-pass filter

Focusrite OCTOPRE
Noise: EIN > 125dB
measured at ~60dB of gain with 150 Ohm termination (20Hz/22kHz bandpass filter)


Neve 1073
EIN: Better than -125dBu @ 60dB gain

Neve 1081
Output noise better than -42dBm from Zin 1200 Ohms and -80dB input,
giving an equivalent noise of -125dBm referred to 600 Ohms input impedance fed from 100 Ohms

Neve 1081R
Noise: (22Hz to 22kHz bandwidth) EIN better than -123dB @60dB gain

API 3124
Noise EIN: -129 Mic, -125 Un-Bal. Actual

SSL Alpha Channel
Noise Figure
Equivalent Ein:
Mic, Zin = 10R -130.5 dBU
Mic, Zin =150R -127.5 dBU


As you can see there are lot of preamps with specs  where EIN isn't referenced to anything. But, the most interesting is Neve where dB, dBm and dBu are used

At the end, Rane suggest (note 145)  that the EIN specification should include measurement conditions (bandwidth, max gain and  Rs).

So, I will not blame Amek more than others.

 

[JohnRoberts]

I recall an article that Paul Buff wrote in RE/P back in the '80s talking about the difficulty in making absolute noise specifications.  IIRC there was even some debate about dBm input power being shared between source and input terminations  with some funny math to game a dB or two.?  ;D

I long ago embraced specifying input noise, using NF (Noise Figure). Simply stated noise figure is the noise of an active gain stage as compared to a perfect, noiseless gain stage.  A good mic preamp is low single digit dB NF (SOTA is probably close to 1.0 dB NF, to do much better you need to bring out the liquid nitrogen and what's the point since no room is that quiet?). 

It might be interesting to run that group of preamps across a single test bench with identical terminations and test conditions to see what is really what.

In general all else equal a wide-band noise measurement will read significantly higher than a properly band limited noise spec, like 22-22kHz (AP?). Since most solid state preamps can have response up to and even above 100kHz  a wide band number will include 5x the noise bandwidth with a majority of that noise above 20kHz that isn't audible. The frequency response above 20kHz can impact wide band noise numbers. 

Measurements at 60 dB of gain will generally be better than even the exact same preamp at 50 dB because output referred noise remains constant but it's contribution matters less when divided by the higher forward gain for ein calculations (EIN measures output noise then divides that by the forward gain).

Input termination matters... I noticed one preamp provided EIN for 10 ohm and 150 ohm terminations. 150 ohm represents the typical low Z mic.  The input of an active preamp stage will have both a noise voltage component, and a noise current component. The measurement using a 10 ohm termination reflects both the lower Johnson noise of the termination (source) resistance, and the reduced noise contribution  from noise current times the input resistance.  I would take that 10 ohm spec with a gain of salt, while lower is always better. I've seen some preamps specified with shorted inputs which is complete nonsense.  No zero ohm mics out there.

JR