Although I registered for this forum on 7/1, I was just "cleared" (must have been a very thorough background check) and I'd like to post more detail to issues raised in this thread. Please forgive me if I'm "style clumsy" at first. Here goes:
1. Shield Connections:
Faraday shields (inaccurately called "electrostatic" shields) are used to prevent capacitive coupling of AC electric fields. In an input-type transformer, they're intended to make ALL the coupling magnetic. Because the shield has capacitance to both primary and secondary windings, it matters whether the shield is tied to the incoming signal's ground or the SRG (signal reference ground) for the amplifier that follows. Most input transformers are driven by a balanced line, therefore the primary-to-shield capacitance is intentionally balanced as much as possible. Similarly, the output (secondary) is designed to drive a single-ended (ground referenced) amplifier. Therefore, no attempt is made to balance secondary-to-shield capacitances (it would generally compromise other design aspects. That's why it's important that both the designated "low" side of the secondary, as well as the shield, be tied to SRG for the amplifier. For typical non-inverting op-amp circuits, this point is the grounded end of the shunt resistor in the negative feedback. For typical vacuum-tube circuits, it's the grounded end of the cathode bypass capacitor (or resistor, if no bypass is used).
That being said, the balanced input signal to the transformer may have common-mode voltage (noise) with respect to SRG. The transformer is specifically designed to "reject" this noise (evidenced by its CMRR or common-mode rejection ratio figure). Therefore, connecting the Faraday shield to the signal source ground reference (pin 1 for an XLR connection) would result in the source's ground noise (relative to SRG) being capacitively coupled to the secondary winding. But connecting the shield to the amplifier's SRG makes the noise appear between primary and shield - where it can be rejected. This is the recommended connection.
For a typical mic input transformer, the Jensen JT-115K-E, the capacitances are: primary-to-shield = 475 pF (effectively 238 pF at each end); secondary-to-shield = 205 pF; and all windings and Faraday shield to case = 34 pF. While tying the can to SRG is ideal, the can must not touch the chassis as might happen via mounting to a metal chassis. SRG should be tied to other grounds at only one intended place. Therefore, connecting or mounting the case to chassis ground is generally recommended. It's very unlikely that the tiny voltage difference between SRG and chassis will cause noise. Remember that any noise between case and SRG would couple to the secondary through less than 34 pF - making the noise an order of magnitude lower than that caused by connecting the Faraday shield to chassis! Don't "float" the case connection - this disables it's ability to act as an external Faraday shield by allowing nearby electric fields (such as from the "hot" side of AC power) to couple to transformer windings inside. Although the can is a magnetic shield intended to route magnetic fields around the core/windings inside, why not make use of its electric field shielding, too?
2. Faraday "Shield-per-Winding" Feature:
To make "mic split" transformers more noise-immune, Jensen pioneered the "shield-per-winding" design. In some applications (live events where a mic must feed both a "house" and "remote truck" mixer), system ground-voltage-differences can be very high (often tens of volts). With a single shield between windings, there is no way to prevent this voltage from appearing between shield and one of the windings. Putting a shield on each winding now puts the voltage between shields, with each winding shielded by a "ground" voltage equal to its destination mixer's "ground" voltage. It neatly solves a pesky problem! It also drives home the point that ground is not ground is not ground.
3. The Pin 1 problem:
When system equipment is interconnected with cables, substantial currents can flow in the ground conductors of thos cables ... and, much more important, through the connector ground contacts at each end. Such currents have been observed at 100 mA or more - they are created by small voltage differences (quite normal) between the grounds of AC power outlets (I've written a paper about this, so I won't go into detail here). The current (noisy hum or buzz) flows in shield pins on balanced I/O of the equipment. Like all currents, it will flow back to the source of voltage that created it - in this case AC outlet ground. But, if these large, noisy currents are allowed to flow in wires or PCB traces that are part of internal equipment "ground," the resulting tiny voltage drops will inject noise into the signal path. Therefore, pin 1 (shield) for every XLR connector must have an exclusive, independent pathway to the chassis (where UL requires that the AC power ground be tied). In general, connecting pin 1 to any wire or PCB trace used by amplifier circuitry is a recipe for noise disaster! This problem exists in a lot of commercial equipment ... it's called the "Pin 1 problem." Pin 1 has absolutely nothing to do with a balanced signal! I'll spare the rant here ... but the misconceptions (even the very definition) of balanced interfaces has been a cause celeb with me for over 25 years!
4. About Ground Symbols:
I use ground symbols as specified by IEEE (Institute of Electrical & Electronic Engineers) Standard 315-1975 section 3.9 "Circuit Return." You can find this entire document at
https://www.ee.iitb.ac.in/~spilab/Tips/ansii_graphic_symbols_for_electrical_and_electronics_daigrams_1993.pdf
This standard has been adopted by: American Society of Mechanical Engineers, the American National Standards Institute, Canadian Standards Association, and US Department of Defense. The "chassis" symbol (as well as the "general" symbol using a stack of lines in a triangular shape) is also part of IEC standard symbols. The "triangle" symbol, as used in Jensen diagrams for an "isolated" signal ground, is one of the IEEE symbols "To be used when identically annotated common-return connections are at the same potential level."
Overall, as I always note in my seminars, the use of the words "ground" and "earth" to describe common circuit returns is unfortunate because many, if not most, electronic circuits are not connected to actual soil (battery-powered, aircraft, spacecraft, etc.). Soil connections are all about lightning and, secondarily, fault mitigation in high-voltage electrical distribution networks. It would be so much better if electronics engineers used the word "return" instead! It would encourage them to think in terms of current flow - instead of thinking of a soil connection as an infinite sink for all things noisy. My colleague, the late Neil Muncy, referred to this as the "sump theory of noise control."
5. I'm New Here:
I've been an analog circuit designer since about 1970 - serving as chief electronics engineer for Quad-Eight and Capitol Records, among others. In his will, Deane Jensen left me his company upon his death in 1989 and I served as owner/chief engineer until I sold it in 2014. Deane and I worked on many projects together before his tragic death ... and he inspired me to "dig deeper" into everything. I now do consulting out of my home in Ventura, CA (I can't allow myself to really retire - how boring!).
With some reservations, I joined this forum just today. The only other forum I've been able to deal with for many years is Syn-Aud-Con ... many others just seem like "opinion fests." But I'm optimistic about participating here ... I just hope my post isn't "too much information" and that it answers more questions than it raises. Flame suit on!!
Bill Whitlock, AES Life Fellow, IEEE Life Senior
Whitlock Consulting, Ventura, CA