OPA1611: trying to simulate the effect of stray capacitances due to metallic shield results in irregular circuits

Morning Alberto, the 1st thing any simulator tries to do is find a DC operating point for every node, 

Your Tee feedback C has a floating node, add a large R there and it will be fine, 

Incidentally, you might not want to put that 50pF on the output pin - looks marginally stable, 

Hi Alosio,

TINA-TI doesn't like floating nodes:

alosio_amplifier_shield.TSC

Kai

Hi Alberto again,

I'm curious: Do you want to tell more about your shielding and stray capacitance issue?

Kai

Dear Kai,

Henry W. Ott wrote a very nice book on noise reduction. The title is

NOISE REDUCTION TECHNIQUES IN ELECTRONIC SYSTEMS

You can easily find many references by just *googling* 'Ott NOISE REDUCTION TECHNIQUES'

In the 'Grounding' chapter, he addresses the effect of  metallic enclosures on amplifiers, proposing a simple yet effective model of the stray capacitances between amplifier and metallic chassis. As reported in my post, such stray capacitors make a T network 'around' the amplifier input and output. I am now simulating the effect of the enclosure (according to the Ott's model) on my low noise amplifier. In the book you will also find a very VERY nice treatment of the best practices about coax cable grounding in low noise applcation. Ott's book is really great!

Thank you so much for all your time and help !

Best,

Alberto

Hi Aloiso,

yes, looks good this book

In my most critical circuits I use individual, local shields which are not connected to each other. I connect them to the local signal ground, best at the signal ground connections to the supply decoupling caps. The associated ground connections should form a ground plane or at least a local ground plane.

There's a nasty trap one can fall into: If you connect the overall shield or the local shield to signal ground via more than one connection, then a loop will be formed like a transformer winding which enters magnetic field changes (predominantly hum) into the signal chain. This might not play a significant role in circuits with high input signals, but with low or very low signals it will. So take care that you don't create loops with your shielding.

Another trap one can fall into is believing, that ground is ground. This is not the case. Not only for HF circuits but also for circuits dealing with very low signals. So it's a good idea to think in terms of local grounds and -consequently- the use of local shielding. This does not mean that there's only one shield for each circuit part or module. No, think of the onion layers. You can have multiple shields overlapping each other. But don't have them connected in a way that signal ground loops can be formed introducing hum into the signal path. 

A good idea of connecting circuits in bigger systems where ground cannot be gound just because of size of circuit, using differential signalling can dramatically help. Not just within one and the same modul, or course, but when connecting modules to each other.

Kai