Low Noise opamps for 2014?

Status
Not open for further replies.
Jump to Latest
#45
Are you claiming your AD797 uses input bias current cancellation? The AD Datasheet implies not and IIRC so did my Jurassic tests.
Click to expand...
scott wurcer said:
Sorry yes, simply a misunderstanding on your part. We engineers are clever little b*****s.
Click to expand...
Gasp! Shock! Horror!

AD datasheet speak with forked tongue! 😱

Surely its usual to denote an input bias current that could be either +ve or -ve as +/-50nA as on the LT1115 datasheet?

AD797 input bias spec (in da evil false datasheet) at 500nA to 3uA is also as large as several OPAs without input bias cancelling.

I'm still reeling in shock. Good think I never put anything with AD797 into production based on my limited tests. 🙂
 
#51
I plugged the OPA627 into the 'visualizer' and it is very good above about 250 Ohms - so its the champion above this Rsource so far (ref 1 kHz). But, the 1/f noise seems to start quite high, so you would need to watch out for that in RIAA applications.
 
#52
so its the champion above this Rsource so far
Click to expand...

not at all - look at my graph, the AD743/745 is plotted along with the OPA627

for either the i_n is no problem below 100s of kOhm

and I already mentioned paralleling the 4 op amps in the quad OPA1644 - which would then better even the AD743 - and take up less space if you can deal with tssop
 
#54
Show what your LN uber OPA can do

There's loadsa stuff that is more important for LN design than choosing the latest uber OPA from their Datasheets.

At the risk of hijacking this thread, who wants to show a real life application circuit that gets within 1dB of AD797's 1nV/rt(Hz) using it or some other uber OPA?

You can specify the application etc but the circuit must include ALL the stuff for the practical application.

Detailed noise analysis please.

Real measured data preferred but fully worked out flights of fancy allowed if detailed analysis provided.

Guru Wurcer, you can't enter cos you're the judge.

We do expect you to show us how it should be done after you've awarded the prize .. which is a glass of Guru Wurcer's finest Burgundy if the winner is ever lucky enough to visit him.
 
#55
BTW, the most important things I've got out of this thread is the realization that
  • even AD datasheets can lie. 😱
  • AD797 has evil input bias compensation
  • and I must now find a 1nV/rtHz OPA to replace it in my Bumper Book of Supa Dupa OPAs 😡
 
#56
Just to demonstrate it is possible to parallel an OPA1644 TSSOP within a DIP8 footprint (10x10mm).
Each opamp has its own 0603 SMD output resistor (say 3.9R).

Patrick
 

Attachments

  • XEN OPA1644 to DIP8.png
    XEN OPA1644 to DIP8.png
    21 KB · Views: 503
#57
EUVL said:
There are better FET input opamp these days than OPA132, OPA134 from TI.
OPA1641 and OPA140, for example.
Close to OPA627 specs but do not cost a fortune.
Patrick
Click to expand...

OPA4140 in multiples, with cross-correlation, are being used by physicists to study low frequency semiconductor noise. I simulated their results in TINA down to 0.010Hz and the devices really work well for instrumentation/measurement.
 
#58
jcx said:
not at all - look at my graph, the AD743/745 is plotted along with the OPA627

for either the i_n is no problem below 100s of kOhm

and I already mentioned paralleling the 4 op amps in the quad OPA1644 - which would then better even the AD743 - and take up less space if you can deal with tssop
Click to expand...

I am only taking the data from the data sheet jcx - worst case at 1k Ohms. I think you are plotting 1/f. I am plotting nV/rt Hz.
 
#60
EUVL said:
Just to demonstrate it is possible to parallel an OPA1644 TSSOP within a DIP8 footprint (10x10mm).
Each opamp has its own 0603 SMD output resistor (say 3.9R).

Patrick
Click to expand...

Before you get too enthusiastic, check figure 10 on datasheet. Looks like bad non-linear input C. In fact bad at rsource = 600 Ohms I find hard to believe. Extrapolate to 47K for 4 in parallel !!.
 
Status
Not open for further replies.
Top Bottom