Hello All,
Scratching my head here looking at single conductor unbalanced RCA interconnects.
If you dissect the typical RCA cord you will see a single conductor inside a grounded shield. The grounded shield is also a signal carrying conductor. That does not seem so different than a twisted pair. That grounded conductor has a nonzero impedance and likely be subject to capacitive of inductive electromagnetic interference. You know pick up Hum or the like from the environment.
Sometimes the the RCA plug ground connects directly to a metal chassis sometimes not.
Trying to keep all the grounds separate in my mind and on the PCB. You know keeping Power Supply Ground, Signal Ground, Digital Ground and equipment grounds separate to avoid common mode noise. Even trying to avoid ground loops.
Seem like there could be some improvements to the old school single conductor inside a grounded shield way of moving line level signals around?
Thanks DT
Scratching my head here looking at single conductor unbalanced RCA interconnects.
If you dissect the typical RCA cord you will see a single conductor inside a grounded shield. The grounded shield is also a signal carrying conductor. That does not seem so different than a twisted pair. That grounded conductor has a nonzero impedance and likely be subject to capacitive of inductive electromagnetic interference. You know pick up Hum or the like from the environment.
Sometimes the the RCA plug ground connects directly to a metal chassis sometimes not.
Trying to keep all the grounds separate in my mind and on the PCB. You know keeping Power Supply Ground, Signal Ground, Digital Ground and equipment grounds separate to avoid common mode noise. Even trying to avoid ground loops.
Seem like there could be some improvements to the old school single conductor inside a grounded shield way of moving line level signals around?
Thanks DT
It’s worse than that - the shield also carries current depending upon the power supply construction of the two devices. This explains it quite well: https://www.jensen-transformers.com/wp-content/uploads/2014/08/an004.pdf
Balanced connections are the best solution, but not easy to implement on most consumer devices
Brian
Balanced connections are the best solution, but not easy to implement on most consumer devices
Brian
You could look at JVC amplifier diagrams of the late 80's and 90's. They used RCA inputs, but their internal construction segregated signal returns from the shield, i.e. their hook-up wiring was indeed the same as balanced.
In general, studying Marantz, Rotel, Denon, Pioneer and similar PCB layouts will teach you a lot about how to segregate the grounds where they need to be separated, how to bring them all together and where exactly, how to reference them to the chassis, how to tie them (or not) to IEC ground, what type of wires and lengths to use, and how to keep that signal return current away from the shield.
RCA's approach is nothing more than drastic cost-cutting measures that seem like science fiction to me. To accept the audio signal return to travel via braided shield wire back to the source is simply unbelievable. This (RCA single-ended topology approach throughout the whole replay signal chain, including sources, wiring, and amplification) should have never been allowed into consumer audio gear.
When my PAS2 and ST70 were built 1961, Tip ring shaft 1/4" phone plugs were not even on the market yet. You wanted 3 pins+sockets, you bought Bendix or Cannon mil-spec plugs for about 20 hours pay each. (I was making $.25 an hours). RCA shielded cable works okay up to 6' distance in a low EMI environment. Police & fire did not even have personal radios, they were out in the street in the cars. No cell phones. BTW the PAS2 & St70 still work, fine, as long as the 1000 w CB radio emitting barking dogs continuously is not driving by. Those products had no safety ground to the chassis, 3 pin wall plug was not required by code yet. 1961 I was in 5th grade, but some kind audiophile bought and built the kits, then sloughed them off in greensheet in 1970 when I had a few dollars for something better than Mother's garbage stereo bought with Top Value stamps.
Last edited:
Yes - bad idea.
The single-ended interconnects which I make use 2x solid-core wires inside a braided shield (Belden 3079A).
Thus, signal & signal-ground are carried from one RCA socket to the next via the 2 wires ... and the shield is connected to the RCA barrel at one end only - so it doesn't form part of the signal chain.
And, of course, RCA sockets are insulated from the case!
Where do you get that distance from?
In my last house, I had 11m (36') RCA interconnects (shielded) - and they worked fine. They were low-C, though.
Then I can understand your comment - but it still surprises me (given a low-powered radio station). Here, the main radio station antennas are about 30 miles away ... but they are high power.
I am surprised that increasing the size of the input coupling caps in your amp solved the RFI problem. I would've thought parallelling the ... what, 4.7uF? 10uF? input coupling caps with some 0.1uF caps would've "caught" the RFI.
The required cable design is known for decades now, I'd think.
Henry W. Ott's "Electromagnetic Compatibility Engineering" covers it all.
- 100% shield coverage, that means heavy and multiple braided shields.
- lowest possible DC resistance of the shield.
- shortest possible length.
- preferably a high common-mode impedance. A coax cable already is a common-mode choke for higher frequencies but the effect can be augmented with clamp-on ferrites, even winding a few turns on a ferrite toroid core.
The required device design is also known.
- all inputs (and outputs) should have RF-blocking feed-through capacitors, doubling up with an input resistor as a RF line termination, as close as possible at the connector.
- all unbalanced connectors must be tightly grouped and must directly connect to the chassis or a copper bonding plate with a 360deg bond, RF-tight.
- that bonding plate is the local "star GND" reference. The supply (GND and rail voltages) point of entry must also be right at this point. No supply leakage currents and no GND balancing currents shall flow across the PCB.
- the impedance to mains grid should be as high as practical or allowed. Notably earth-grounded devices benefit from PE chokes (agency-rated ones) and good mains filters, and again benefit from additional ferrites as common mode chokes.
- there should be additional grounding posts (bonded to said copper plate) adjacent to the unbalanced I/O's, to have options to enforce the GND connections between devices beyond what is possible with the shields of the interconnects alone.
Last edited:
Sadly, Henry Ott passed away in January 2022.
and his web-page is no longer available.
but it might be on the Wayback Machine.
https://www.hottconsultants.com/
and his web-page is no longer available.
but it might be on the Wayback Machine.
https://www.hottconsultants.com/
Last edited: