[time-nuts] SMD TADD-1 distribution amplifier

Bruce Griffiths bruce.griffiths at xtra.co.nz
Fri Dec 18 17:50:05 EST 2015


On Friday, December 18, 2015 12:27:33 PM Charles Steinmetz wrote:
> Poul-Henning wrote:
> >A significant reason for the TADD-1 existing in the first place was to
> >break groundloops.  This is incompatible with tying all the BNC's together.
> 
> It is perfectly possible to ground coax connector bodies at RF
> without grounding them at power-line frequencies (and expected
> harmonics) or DC.  Indeed, that is what the TADD-1 attempts to
> do.  The BNC bodies are connected to the enclosure with 10nF capacitors.
> 
> However, the implementation in the TADD-1 is not as ideal as I'd
> like.  The bypass isn't directly to the enclosure, but rather to the
> ground plane of the PC card, which appears to be connected to the
> enclosure only by fasteners at the corners of the PCB -- some inches
> away from the BNCs.  That is why I will almost never approve a design
> that uses isolated, PC-mount coax connectors.  The loop from each BNC
> body, through the capacitor, to the enclosure should be no more than
> 1/2" (~ 1cm) long, preferably less.

Something like an annular ring capacitor that insulates the BNC body from the 
chassis whilst capacitively shorting it to the Chassis at RF ??.
Followed by a freeite sleeve/bead arround the connector body or the coax 
connecting the signal from the connector to the PCB?

> 
> I like to add a resistor in parallel with each capacitor, to shunt
> any unexpected DC or LF leakage current.  This resistor needs to be
> large enough (in value) to limit any circulating ground currents to
> low values that will not cause significant voltage drop in the
> shields (i.e., ground loops), but at the same time will effectively
> terminate any expected leakage currents.  Generally, this will be
> from 10 to 100 ohms, depending on the sensitivity of the most
> sensitive inputs in the system and whether guarding is employed.
> 
> Best regards,
> 
> Charles
> 
> 

Bruce





More information about the time-nuts mailing list