Understanding noise specs

[Audiop]

So, if a preamp has equivalent input noise spec like: -130dBV @60dB gain.

What does that mean? I can crank the gain to 60dB and measure 0.00003V RMS on the output? Or that the "-130dBV" should be raised with 60dB to get the noise on the output? Meaning -70dBV.

Seems odd these numbers never come with a bandwith spec.

I did crank a preamp to 60dB (input shorted) and recorded it into a AD and when playing back and looking at the meters the noise hovers around -60dBFS.

What is it that I'm missing?


/Peter

[JohnRoberts]

One could write a book, or at least a chapter on noise specs.

ein "equivalent" input noise is measured at the output, then mathematically adjusted by subtracting the nominal gain to come up with an input noise number.

Indeed bandwidth is important, as well as input termination, including source impedance of input connection.

Another consideration for S/N is that signal is almost always a sinewave, while noise is random or quasi random so it will have differing peak, average, and spectral content.

Another (lesser) consideration is the frequency response of the gain stage.

Simple answer is -130dBV @60dB gain, maps out to -70dBV at output.

Note: ein of high gain preamps also differs with gain setting. They are often dialed in for best S/N at high gain. ein is not strictly "input" noise, but all noise referred to the input. In well designed circuits, that will be dominated by input stage noise, but at lower gains noise may be dominated by intermediate stages.

JR

[Audiop]

Thanks for confirming. Seems I had some grip on the issue.

I have a high quality preamp here at my bench and was surprised about the noise. Sure the mic noise will normally dominate over preamp noise but somehow I thought that the preamp noise was lower.

Seems like preamp noise is a problem (even with the best units) when recording quiter sources and/or at a distance.

I'm about to get into location recording and added a Earthworks 1024 to my Lynx Aurora 8. And the noise was higher than I did expect. Also there is some hum at high gain and 30dB higher at the channel that is closest to the powersupply.

I wonder if there's other preamps that have better performance.


/Peter

[tINY]

Remember, too, that there is background noise "on location". It's rare to have lower than 30dB background noise and audiences can increase this to 50dB easily.



-tINY

[nosebleedaudio]

A good real world test is a 150 ohm resistor across the input, pin 2, 3 and crank it to max.
But a noise spec without all the required termination(shorted input,150ohm) , noise filter ect is meaningless..

[JohnRoberts]

Originally Posted by Audiop

Thanks for confirming. Seems I had some grip on the issue.

I have a high quality preamp here at my bench and was surprised about the noise. Sure the mic noise will normally dominate over preamp noise but somehow I thought that the preamp noise was lower.

Seems like preamp noise is a problem (even with the best units) when recording quiter sources and/or at a distance.

I'm about to get into location recording and added a Earthworks 1024 to my Lynx Aurora 8. And the noise was higher than I did expect. Also there is some hum at high gain and 30dB higher at the channel that is closest to the powersupply.

I wonder if there's other preamps that have better performance.


/Peter

Another tidbit about ein.. is that even a perfect preamp is amplifying the thermal noise (noise caused by motion of electrons due to heat) in the source impedance of the microphone. I prefer to look at preamps using NF or a comparison to actual vs. theoretical noiseless... FWIW the general performance of good modern preamps is in the 1-2dB NF range or within less than 2dB of perfect. Then you add in room noise, etc and the mic preamps are rarely the limiting factor for noise floor.

But marketers still persist is arguing how much quieter their new improved preamps are. The SOTA for nosie floor hasn't really changed in decades.

JR

ps: the thermal noise is relative to absoulte zero so to reduce that noise you need to soak your mics in liquid nitrogen... bet your tounge would stick to that.

[Audiop]

Would be reasonable to assume that 60dB gain may be necessary once in a while doing location recordings? (I think so)

And hat preamp noise will add a bit to the actuall audible noise? (if the hall/room noise is relatively low)

Thing is when I crank the gain of the pre until I hear some noise, I hear smooth wideband noise but also more of a 50Hz hum (with odd multiples up to 2kHz or so) which is more than a little annoying. There is also a 30dB (aprox.) difference in level of the peaks of this hum/buzz (spectral analysis with ARTA, nice software) between the channel that is closest to the powersupply and the one that is most distant to it.

So I'm thinking on returning the preamp. I want the cleanest most noisefree preamp that can be found.


/Peter

[tINY]

Be ready to dig deep into your pockets with that criteria....

I would expect the Earthworks stuff to be better than that. Have you called costomer service?

There are ways to aleviate power supply noise.



-tINY

[Audiop]

Originally Posted by tINY

Be ready to dig deep into your pockets with that criteria....

I would expect the Earthworks stuff to be better than that. Have you called costomer service?

There are ways to aleviate power supply noise.



-tINY

How deep? :-) Seriously, can you suggest other pre's that are in the top?

I will contact EW tomorrow. I talked to my dealer today.

Aleviate PS noise you say. Building an outboard supply? That was what I was thinking but it doen't seem ok for this type of product... or this price I should say. I have connections so I think I could manage to build a complete amp from scratch with good performance but I choosed to take the cost by going with the 1024 to save time and get things going. Now I'm dissapointed since I did expect perfection as far as the technology allows.

For a while I thought there could be some interfacing problems with the Aurora but after talking to the Lynx people I doubt that.

I think the specs on the ZDT line is done on a lab unit with one or two channels. Cramming four channels in a 1ru box with a EI trafo seems like a bad idea. I don't know.

The thing is the 50Hz spikes (and the odd multiples) stick up really high above the basic noisefloor. It hurt my eyes and ears. :-)

Wouldn't a toroid be a better option?

Aaahrg! I'll go design a pre with battery supply.


/Peter

[tINY]

Sometimes xformers are just bad (or go bad after getting jarred in shipping). If the unit isn't performing to thier spec, return it.

Earthworks is in the running for the big bucks award, but things like hand assembly, indivudual, exhaustive testing, isolated power supplies and Jensen input transformers can run the price up even higher.

If you are in England, you may check with D.A.V. Elelctronics. Then there is AMS Neve...



-tINY

[Audiop]

I wrote a deailed mail to Earthworks today and we'll see what they say.

I came to think of something. Regarding the hum/buz, could it be that my unit, which is a 220-240V, version has more inductive and capacitive coupling due to the higher voltage?

Hmm.. should be more capacitive coupling in such case since higher voltage means lower current.


/Peter

[tINY]

Actually, the primary is usually split into two isolated windings. Then the 240 is wired series and the 120 is wired parallel.

Should have the same charateristics from a voltage and current point of view.

But, that silly 50 cycle stuff may have an effect.



-tINY

[Audiop]

EW will take the unit in for repair or exchange. They agre that the unit is performing bad according to their standards and spec. I did send them detailed measurements graphs so they could see what was going on. They do suspect a faulty PS.


/Peter

[Geoff_T]

Hi

I've always considered noise specs a bit like the old saying... "lies, damn lies, and statistics"... just substitute noise specs for statistics.

This attitude is born of working 14+ years at Neve and then, after moving on, seeing how other companies and industries specify noise.

There are two camps...

1. You state the noise relative to the normal listening level of 0dBu or thereabouts.... and even then, there are scores of filter weighting options.

2. You state the noise referenced to the absolute clipping point of your equipment.

Neve chose version #1, the power amplifier and digital industries chose #2.

#2 can obviously produce grander numbers than #1 but it all means diddly squat.

Neve specs would have an amplifier running at 80dB gain with a 200 ohm input termination. The noise, measured 20Hz to 20KHz was typically around -48dBu so you add that figure to the gain to get an EIN of -128

Had they measured it down from the greater than +26dBu clipping point... say +28dBu, they would have got a number like -156 but this is meaningless... it's what you hear when you stand next to the loudspeaker... not the numbers.

I once dealt with an amp manufacturer that sold a 500 watt power amp with a -90dB noise figure. You could hear the noise from the other side of the room, never mind having your ear up against the speaker!

That's because it was -90 down from around +40 (very rough brain calculation) to give a real world noise of -50dB that anyone could hear.

So, unless there's a level playing field and everyone is using the same reference (both level and reference like dBm, dBu, dBV), terminations and filters, the numbers you read are pretty academic.

It's what you hear coming out of the loudspeaker that matters!