1. Home
  2. Electronics Design
  3. Low noise amplification

Low noise amplification

Hi all,

I'm new to the group... I'm designing a low-noise amplification system at work. The signal's output is ultimately digitized with a 16-bit A/D (Analog Devices AD977A).

We use 3 stages of 2nd-order Chebyschev filters to isolate the frequency of interest (e.g., a 15 kHz band surrounding a 100 kHz carrrier) prior to digitization. We currently use the Analog Devices OP467 (6 nV/rtHz noise density) but I am thinking of moving to the Linear Technology LT6232 (1.1 nV/rtHz).

We must amplify the input signal by 42 dB (a factor of 128) as part of the signal chain. This is done with an Analog Devices AD829 right now (1.7 nV/rtHz) but I may switch over to the LT6232 here as well.

My question is: should we filter then amplify (G=128) or amplify (G=128) then filter? The previous design filtered then amplified. Ultimately, the question boils down to where the most noise is coming from. In a low-pass situation, I know it's best to filter then amplify, but I wondered if bandpass might be different. We have 200 noisy counts on our A/D's and we need to get this down to around 15-20.

Thanks in advance for the help!

-Todd

Reply to
tschoepflin
Loading thread data ...

If it's genuinely a *preamp* there's no question. Amplify then filter.

Graham

Reply to
Pooh Bear

a écrit dans le message de news: snipped-for-privacy@y43g2000cwc.googlegroups.com...

Since you're planing using LN opamp, you don't have much noise riding on your signal (it wouldn't make sense otherwise).

Placing your BPF on front, will make you reduce the signal noise but amplify the BPF + amplifier noise, which you don't want.

The best is to amplify first, then filter. That way, you reduce all the system noise with your BPF. Also, only your first gain stage needs to be low noise.

Where did you get the impression that LPF are best placed before the amplifier? Other reasons might mandate this, but from a noise POV that's just wrong.

--
Thanks,
Fred.
Reply to
Fred Bartoli

If the noise won't ever clip the amplifier, amplify first. If that's not the case (and it probably isn't) you could amplify *some* first (to improve s/n, but not enough to saturate), maybe 4:1 or something, and then filter, then amplify the rest of the way.

Note that, if your signal source impedance isn't optimal for the first opamp, you might be better off filtering first, using a bandpass filter with different impedances at opposite ends. This will give a free, noiseless voltage gain ahead of the first gain stage and also zap noise that could clip the amp.

John

Reply to
John Larkin

Many active filters are pretty noisy, even when using low-noise opamps. Do a simulation on various active filters to get an appreciation of how noisy they can be. Thus, best to amplify first, then filter unless your noise is going to clip your amplifiers. If you need to use a BPF near the front-end (minimal gain before the filter), then consider a gyrator. They tend to be quieter and have a lower component sensitivity than state variable filters.

--
Mark
Reply to
qrk

That would of course be a passive LC filter.

John

Reply to
John Larkin

I'm glad someone made that point !

Graham

Reply to
Pooh Bear

digital passive or analogue passive?

martin

Reply to
martin griffith

Quick calcs suggest you have 26uV of noise at the moment. Better amps and putting the X128 amp first, only improves it to 18uV. You're looking for

10:1 noise reduction. Assuming all else is equal, you need to be looking elsewhere. john
Reply to
John Jardine.

"John Jardine."

** Huh?

Bandwidth = 15 kHz.

Noise = 6nVrtHz (OP467)

Gain = 128.

Works out to 94 uV.

** That would be a nice improvement.

....... Phil

Reply to
Phil Allison

Yes indeed. A good change. I'd started out with the lower noise amp. :( john

Reply to
John Jardine.

ho hum, who cares?

John

Reply to
John Larkin

In article , snipped-for-privacy@gmail.com wrote: [....]

Neither and both.

The gain stages should provide some of your filtering and then be followed by the rest of the filter.

As others have suggested, you may need to put enough filtering early in the circuit to prevent clipping,

A multistage filter always has some stages with higher "Q"s and some with lower "Q"s. You want some low "Q" stages to be the last ones in the filter. The higher "Q" stages tend to create noise at their cut off frequency. The lower "Q" stages will tend to roll off that noise.

--
--
kensmith@rahul.net   forging knowledge
Reply to
Ken Smith

Be aware that the low noise amplification is only achieved with a sufficiently low source impedance. See page 11 of the LT6230 datasheet : 100 Ohms or less. Another matter is the layout, the digital part should be isolated from the analog part. Especially tricky are the SPI chains. The SPI to the ADC should be quiet during the ADC operation. And the relevant low amplitude signals should be handled differential.

Rene

--
Ing.Buero R.Tschaggelar - http://www.ibrtses.com
& commercial newsgroups - http://www.talkto.net
Reply to
Rene Tschaggelar

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.