naturerecordists
[Top] [All Lists]

Re: New Telinga PIP Stereo preamp. & EIN

Subject: Re: New Telinga PIP Stereo preamp. & EIN
From: "sainganrob" sainganrob
Date: Sun Jan 20, 2013 3:55 pm ((PST))

--- In  "John Crockett"  wrote:
>
> I think I have answered my questions to my own satisfaction.
>
> John
>

Hmmn, I should have typed the following faster then :)

E.I.N. (Equivalent Input Noise)

The EIN of a system is a convenient way of lumping together any noise volta=
ges generated within that system, and presenting it as though it were an Eq=
uivalent Noise voltage at the Input to a noisless system.

e.g. Assume you had a black box amplifier with a gain of 40dB

If you short circuit the input to ensure that no signal, either wanted or n=
oise is able to enter the box, then you could reasonably expect that the ou=
tput should remain at zero. i.e. minus infinity dB, afterall, (40dB gain =
=3D x100) ... (zero x100 =3D zero !)

In reality however, due to the various noise mechanisms within the amplifie=
r electronics, an output noise voltage will indeed exist.

Now, if we measure that output voltage at say 1mV (-60dBV) and we recall th=
at our amplifier has a gain of 40dB, then we can say that black box has an =
EIN of (-60dBV minus 40dB), i.e. -100dBV.

In other words, if we had applied a noise voltage of -100dBV to the input o=
f a noisless 40dB amplifier then we would get -60dBV out. Black box has an =
EIN of -100dBV.

Suppose we input a noisless signal of say -30dBV to the above black box.
Well, it would come out +10dBV, i.e. gained by 40dB.
The EIN (-100dB) would come out -60dBV
We would have a signal to noise ratio of +10 minus -60 i.e. SNR 70dB

Important : Although we used the gain of the amplifier to calculate the EIN=
, the gain itself has no bearing on the EIN figure. i.e. for the above box,=
 the EIN is always -100dBV.

If its gain were reduced to 20dB then the output noise would be -80dBV i.e.=
 -100dBV plus 20dB

If you take a second box (err, white box ?) with a better EIN performance o=
f -120dBV with a gain of say 35dB, and we input our same noisless signal of=
 -30dBV to it, then the output would be +5dBV, i.e. gained by 35dB and our =
output noise would be -120dBV gained by 35dB i.e. -85dBV
We would have a signal to noise ratio of +5dBV minus -85dBV i.e. SNR 90dB
So we can see that a 20dB improvement of EIN equates directly to a 20dB SNR=
 improvement in output, irrespective of gain settings.

If you were to combine these two systems such that the better performing wh=
ite box preceded the black box and adjusted their individual gains to achie=
ve an overall gain of 40dB to replicate the original black box setup, then =
...

white box (gain +35dB, EIN -120dBV) ... black box (gain +5dB, EIN -100dBV)

Input a -30dBV signal, comes out of white at +5dBV, noise at -85dBV.
Now, at this point we've got an interesting situation. We have a signal lev=
el of +5dBV, but we have two noise voltages!! one is the -85dBV white box o=
utput and the second is the black box EIN, -100dBV

So ... all three of these signals now go into the +5dB gain black box ampli=
fier, and come out as +10dBV signal, -80dBV white box noise and -95dBV blac=
k box noise. In reality the three outputs are summed and don't exist as sep=
arate entities. The two noise sources actually sum as the square root of th=
e sum of their squares, and become
 -79.87dBV

So, finally, we end up with a 40dB amplifier with an input of -30dBV and an=
 output of +10dBV signal and -79.87dBV noise i.e. a SNR of 89.87dB very nea=
rly the full 20dB improvement of the white box vs black box EIN individual =
figures.

Caveat : Although I said that the gain of an amplifier has no bearing on it=
s EIN, practically, due to the topology of common input stages, generally s=
peaking the EIN performance of an amplifier will improve with increasing ga=
in.

Rob








<Prev in Thread] Current Thread [Next in Thread>
Admin

The University of NSW School of Computer and Engineering takes no responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the naturerecordists mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU