Noise Figure of Cascaded Preamplifiers

[sleroux]

I just read from a JuicedLink doc that:

The first amplifier in the signal chain sets the noise characteristics for the entire system. This is known as the "Noise Figure of Cascaded Amplifiers".

Does that mean that if I have a rather noisy preamp set on +50dB gain on a recorder, I would obtain better result by using a less noisy preamp first with gain of +25dB and cascading it with the noisy internal preamp of the recorder for the remaining +25dB?

This is rather counter intuitive to me, as at first sight, it feels to me like the second preamp will amplify the electronic noise of the first one and then add its own noise...

 

[Noiz2]

I think they are over stating it a bit, and it really depends on how the next stages work. So for instance if using an out board preamplifier lets you go line IN and bypass the internal preamplifier you will see a BIG improvement. If using the external lets you set the internal to a lower gain level you will generally see an improvement, though it might not be huge. This is because the noise is not generally linear with gain. Most preamps are fairly quiet at low gain and get worse at an increasing rate over 50% or so. However if the second preamp is a fixed gain and they are changing the level by padding the input you won't see any improvement.

So assuming scenario #2 by doing most of your gain in the first "quiet" stage you will not add nearly as much noise in the second "noisey" stage because you are down in the "cleaner" part of it's range.

 

[Richard Crowley]

Yes, I concur with Mr. Koue. There are some schemes for improving signal-to-noise ratio (SNR) by using external preamps, but there are many schemes out there that provide no real improvement depending on how the signal chain is designed, and the nature of the NEXT stage in the scheme. Of course, people selling gear want you to buy their gadget without regard to whether it will actually provide any improvement in YOUR signal path. caveat emptor.

 

[sleroux]

Thank you Scott and Richard for our answers.

Of course, people selling gear want you to buy their gadget without regard to whether it will actually provide any improvement in YOUR signal path. caveat emptor.

You are perfectly right. That's exactly why I was skeptical about this assertion.

So for instance if using an out board preamplifier lets you go line IN and bypass the internal preamplifier you will see a BIG improvement. If using the external lets you set the internal to a lower gain level you will generally see an improvement, though it might not be huge

[...]there are many schemes out there that provide no real improvement depending on how the signal chain is designed, and the nature of the NEXT stage in the scheme.

Yes, this is really the cascading-thing that has puzzled me the most. That's why in my hypothetical example I was comparing a one stage +50dB setup with a two stages +25dB/+25dB.


Since then, I found some evidences that tend to confirm the idea expressed by the JuicedLink guy. The Friis formula for noise factor clearly shows that the first stage is the most important regarding the overall noise level. Moreover it implies it is a good approach to have an high gain first stage, in order to reduce the contribution of the other stages to the noise. I also found a cascaded noise calculator. Playing a little with it confirms the same thing.

BUT, both the formula and the calculator are for the RF domain. And make use of the "Noise Figure" of each stage, which is something that can't (?) be easily derived from the usual specs of audio gears.


ANYWAY, if I understand it well, noise signals do not simply add. And with "clever" arrangement, cascading two stages is not as great heresy as I naively though it was.
Good to know as I am a little bit "short" while recording, and can't afford a low noise and high gain preamp to just bypass the recorder preamp. Now, I know that I must first hunt for a low noise preamp, set it at its max and cascade with the recorder preamp set to just enought gain to reach the desired level. I hope I don't make myself too unclear: it's late here and I read to much math for this evening...

- Sylvain

 

[Noiz2]

Basically you are right BUT all preamps tend to have this escalating noise at high gain. A really good preamp will make that much closer to linear than a bargain one will but it's rarely a good idea to have anything in the gain staging "maxed" out.

If you do some searches on "gain staging" you may find more that is helpful than "cascaded noise' since it's an old topic in analog sound.

As a really gross generalisation, most devices work at their optimum at about 75% of their gain. It's not always true but that is usually the sweet spot.

And you are right that it's complicated. Partly because it's really hard to find out sometimes how the gain works in various devices.

The good news is there is an extremely sensitive device that if used properly will tell you the optimum settings on any combination of devices.
And the better news is it's free!

Your ears and brain are very sensitive to noise and distortion etc. You can play with the settings till you get the best out of any combo if you just really listen.

That doesn't help you that much on devices you don't have access too but, in that case you can't use them either so...

Other than a decent mic I would put a good preamp as my #1 bit of sound gear. I had a nice one that used a TON of batteries and swapped it for a MixPre when it came out and the two were by far the best things I have bought. The MixPre has out lived a bunch of recorders ,and made them all perform above their pay rate.

I have had others and even built a few. It's not that easy to make a good sounding quiet mic preamp. I would be skeptical of anything that is fairly cheap and portable. You can get good pre's in mixers that plug in and cost less than a MixPre, but it's hard to find something cheap and baggable that is worth buying. Used is not a bad option on something like a SD device since you can get it repaired. New you are looking at close to $1K for anything really good.

There will be lots of opinions but to me IF you can't swing a real quality preamp then I would get a recorder with decent preamps OR a really cheap and OK mixer/ preamp. Behringer used to make a very small battery powered mixer for like $100 that was not great, but OK. Radio Labs makes a pretty decent mic pre module. You would have to make up the box for it and wire it to jacks and a power supply (and level adjustments would be done with a tuning screwdriver) but at around $100 you could have a decent preamp for under $150 and your labor (and no circuit building).

There are a lot of products in the middle that are not enough better than the Behringer to justify the cost. They are not "really good", but if you drop $300 - $500 your not going to feel good about having to go to a $1k pre to actually get what you thought you were going to get at a bargain.

A used MixPre can be had for $300 or so, sometimes, and that IS a good deal.

Juice Link might be OK? They started out as not so great and I haven't heard one in a long time so? BUT they are marketed to DSLR video shooters not sound people so I have my doubts.

And I understand the $ issue. I built my first mixer and mic pres because I couldn't afford to buy what I wanted. I bought the two above when I was making a living doing sound recording for films with budgets. But in hindsight I would have bought a good mic pre before I bought my first DAT.