Author Topic: Low-Noise Front End Protection  (Read 2321 times)

0 Members and 1 Guest are viewing this topic.

Offline Evan.CornellTopic starter

  • Regular Contributor
  • *
  • Posts: 189
  • Country: us
Low-Noise Front End Protection
« on: February 12, 2020, 08:56:03 pm »
Any recommendations for low-noise/leakage protection diodes for non-inverting input of LNA?
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 15156
  • Country: de
Re: Low-Noise Front End Protection
« Reply #1 on: February 12, 2020, 09:03:02 pm »
A common modern low leakage diode is BAV199.

If tested low leakage is required, 1N4117 or 4118  JFETs are an option too.

Another point is keeping the temperature low and the voltage small, by bootstrapping the diodes.

Usually the diodes are not causing much noise, but they can cause leakage - so it depends on the impedance level.
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17428
  • Country: us
  • DavidH
Re: Low-Noise Front End Protection
« Reply #2 on: February 13, 2020, 02:47:14 am »
For RF, it is common to use microwave schottky or PIN diodes for input protection because they have lower capacitance than common diodes which would otherwise be suitable.  You said LNA so I assume it is an RF application.
 

Offline Evan.CornellTopic starter

  • Regular Contributor
  • *
  • Posts: 189
  • Country: us
Re: Low-Noise Front End Protection
« Reply #3 on: February 13, 2020, 02:46:54 pm »
Not RF, signal range is only up to 200kHz. I am mostly concerned about noise contributions of any amplifier input protection circuitry to the actual input signal.
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17428
  • Country: us
  • DavidH
Re: Low-Noise Front End Protection
« Reply #4 on: February 14, 2020, 03:46:28 am »
Then Kleinstein covered two of the good options, low leakage BAV199 diodes which are not tested and low leakage JFETs like the 2N4117 which are.  I like using 2N3904 emitter-base or base-collector junctions myself but they also are not tested for low leakage so you have to do that yourself.  Any of these options can be good to 100s of MHz.
 

Online fourfathom

  • Super Contributor
  • ***
  • Posts: 2005
  • Country: us
Re: Low-Noise Front End Protection
« Reply #5 on: February 14, 2020, 06:52:47 am »
Perhaps the Texas Instruments TPD1E04U04 or TPDxE05U06 -series ESD protection devices?  These clamp at around -1V / +6V, have a 10nA max leakage spec and 0.5pF typ capacitance.  They include a clamp zener, so for the "TPD1E04U04" single-channel device it's just a two-pin part (if you use your supply rail for the clamp diode connection you need to make sure you don't glitch the supply).  I don't know about noise though.
Would anyone care to comment on this part?  I am considering it for input protection in a new design.
We'll search out every place a sick, twisted, solitary misfit might run to! -- I'll start with Radio Shack.
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7044
  • Country: nl
Re: Low-Noise Front End Protection
« Reply #6 on: February 14, 2020, 06:42:29 pm »
Just put a LMC662 buffer in parallel with your own amplifier.
 

Offline Evan.CornellTopic starter

  • Regular Contributor
  • *
  • Posts: 189
  • Country: us
Re: Low-Noise Front End Protection
« Reply #7 on: February 14, 2020, 07:13:17 pm »
Just put a LMC662 buffer in parallel with your own amplifier.

My current LNA selection is AD8655 (Vn=2.7 nV/√rHz @ f = 10 kHz).. I don't think || LMC662 gains me anything other than almost 10x higher input referred voltage noise...
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7044
  • Country: nl
Re: Low-Noise Front End Protection
« Reply #8 on: February 14, 2020, 07:26:21 pm »
It can already take 5 mA current on the inputs and 3kV HBM, why isn't that enough protection?
 

Offline Evan.CornellTopic starter

  • Regular Contributor
  • *
  • Posts: 189
  • Country: us
Re: Low-Noise Front End Protection
« Reply #9 on: February 14, 2020, 07:49:45 pm »
It can already take 5 mA current on the inputs and 3kV HBM, why isn't that enough protection?

The sensor I'm amplifying is piezoelectric, and can easily generate several volts or more if dropped, knocked, or something like that. Supply rails are +-2.5V, so I need to make sure input pin doesn't exceed voltage rails during an event like that.
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7044
  • Country: nl
Re: Low-Noise Front End Protection
« Reply #10 on: February 14, 2020, 07:57:17 pm »
The sensor I'm amplifying is piezoelectric, and can easily generate several volts or more if dropped, knocked, or something like that. Supply rails are +-2.5V, so I need to make sure input pin doesn't exceed voltage rails during an event like that.

It's not about voltage, for a short amount of time it can already take 3kV ... unless the piezo generates enough power to heat up the AD8655 ESD protection to destruction the voltage generated is irrelevant.
 

Offline Evan.CornellTopic starter

  • Regular Contributor
  • *
  • Posts: 189
  • Country: us
Re: Low-Noise Front End Protection
« Reply #11 on: February 14, 2020, 08:03:20 pm »
It's not about voltage, for a short amount of time it can already take 3kV ... unless the piezo generates enough power to heat up the AD8655 ESD protection to destruction the voltage generated is irrelevant.

I understand that, and the energy profile of piezo pulse is undefined, so I want to operate on the safer side. I am thinking series R = 10ohm or thereabouts with BAV199 to each rail, and another series R ~10ohm before AD8655 input.
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 15156
  • Country: de
Re: Low-Noise Front End Protection
« Reply #12 on: February 14, 2020, 08:53:11 pm »
The BAV199 diodes have a rather high forward voltage ( e.g. ~ 1 V), especially with a high current peak. So it would take considerably more than 10 Ohms for protecting the OP.

It would likely need more something like 2 of the diodes in series directly parallel to the piezo. This would limit the voltage to some +-1 V without much loading and some +-2.5 V with higher current. For lower leakage current one could use bootstrapping from the amplifier for the 1st. pair of diodes.

Chances are the diodes can take enough of a current pulse so one may not need an extra current limiting resistor.
 

Offline Evan.CornellTopic starter

  • Regular Contributor
  • *
  • Posts: 189
  • Country: us
Re: Low-Noise Front End Protection
« Reply #13 on: February 14, 2020, 09:00:15 pm »
It would likely need more something like 2 of the diodes in series directly parallel to the piezo.

Do you mean like this?
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 15156
  • Country: de
Re: Low-Noise Front End Protection
« Reply #14 on: February 14, 2020, 09:15:22 pm »
That would be the minimal version.  Another pair of diodes in series to the diodes would be the next step up.
 
The following users thanked this post: Evan.Cornell

Offline splin

  • Frequent Contributor
  • **
  • Posts: 999
  • Country: gb
Re: Low-Noise Front End Protection
« Reply #15 on: February 14, 2020, 10:47:08 pm »
I understand that, and the energy profile of piezo pulse is undefined, so I want to operate on the safer side. I am thinking series R = 10ohm or thereabouts with BAV199 to each rail, and another series R ~10ohm before AD8655 input.

https://www.analog.com/media/en/reference-design-documentation/reference-designs/CN0350.pdf

It shows a 1K series input resistor for protection. It's only a reference design and not guaranteed to be a robust, production ready design of course. 1K has approx 4nV/sqrt(Hz) which is similar to the opamp's noise, but noise and bandwidth limitations allowing, somewhat higher resistance could be used.

[EDIT] Of course the circuit note author may save assumed a securely mounted sensor not susceptible to hard knocks.
« Last Edit: February 14, 2020, 11:12:52 pm by splin »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf