40m QIL Cryo_Lab CTN SUS_Lab TCS_Lab OMC_Lab CRIME_Lab FEA ENG_Labs OptContFac Mariner WBEEShop
  40m Log  Not logged in ELOG logo
Entry  Fri Jul 17 18:04:54 2009, Clara, Update, PEM, Guralp Box Fail VERT2_missing_resistors.pngresistors_diagram.pngresistors_board.png
    Reply  Sat Jul 18 19:49:48 2009, rana, Update, PEM, Guralp Box Fail Picture_1.pngPicture_2.pngAD620.pdf
       Reply  Mon Jul 20 17:06:29 2009, Jenne, Update, PEM, Guralp Box Fail AD620noise_BenAbbott.pdf
Message ID: 1765     Entry time: Mon Jul 20 17:06:29 2009     In reply to: 1761
Author: Jenne 
Type: Update 
Category: PEM 
Subject: Guralp Box Fail 

Quote:
That's terrible: R5 & R8 should definitely be 100 Ohm and not 100kOhm. 100k would make it a noise disaster. They should also be metal film (from the expensive box, not from the standard box). This is the same for all channels so might as well stuff them.

The circuit diagram between TP3 and TP4 appears to be designed to make the whitening not work. That's why R6 & R7 should be 100k. And R2 should be metal film too.

Basically, every time we want good low frequency performance we have to use the metal film or metal foil or wirewound resistors. Everything else produces a lot of crackling noise under the influence of DC current.

I'm also attaching the voltage and current noise spectra for the AD620 from the datasheet. These should allow us to compare our measurements to a reasonable baseline.


While we're comparing things to other things, Ben Abbott just emailed me his measurement of the AD620 from back in the day. Clara's going to use this along with the specs to make sure that (a) we're not taking crazy measurements and (b) our AD620s aren't broken and in need of replacement. In this plot, we're looking at the GOLD trace, which has the AD620 set up with a gain of 10, which is how our AD620's are set up in the Guralp breakout box.

Just picking a single point to compare, it looks like at 1Hz, Ben saw ~130dBVrms/rtHz. Converting this to regular units [ 10^(#dB/20)*1Vrms = Vrms ], this is about 3*10^-7 Vrms. That means that Clara's measurements of our AD620 noise is within a factor of 2 of Ben's. Maybe the way we're connecting them up just isn't allowing us to achieve the ~50nV/rtHz that is claimed.
Attachment 1: AD620noise_BenAbbott.pdf  173 kB  | Hide | Hide all
AD620noise_BenAbbott.pdf
ELOG V3.1.3-