Ground loop isolator

[R.G.]

JRC has just brought out an IC for the car-audio marked designed to break ground noise loops.

The NJM2794 is a ten pin package that contains two differential amplifiers with biasing and internal references. Surface mount only though.

To do a one-channel differential input, you only need five caps to block/filter voltages. No resistors.

See http://semicon.njr.co.jp/njr/hp/productDetail.do?_productId=1846

[Arn C.]

Thanks R.G. !

    I was looking at the pdf and this may be cool for some of the stuff we're doing here.  And only 5 caps needed!

Peace!
Arn C.

[zpyder]

Hmmm....

I see that the link you provided for the new NJM2794 lists "Common mode rejection ratio (CMRR=60dB typ.)"

By looking at the block diagram and pinouts, I'm thinking that this is this basically a balanced line (eg XLR) reciever that performs the CMR...(?)  And for two of them... If not, what is this??  If it is what I think it is, could this device be used in reverse?

thanks,
zpyder

[R.G.]

There are two channels in the chip. It's intended for car stereo, where ground noise is also a severe problem.

It's a classical dual buffer/balanced diffamp circuit on a chip. Two sets of them, actually. You can do the same thing with opamps, but it takes three opamps per channel to do it, as well as some precision resistors. Putting the whole mess on a chip is nice.

[brett]

Hi
I'm a newb about this kinds stuff, but I've noticed a couple of things in various circuits.

I've got an old power amp, designed by some top-notch EEs, where the signal input ground is connected via a 1 ohm resistor to the true ground.  Could this be a good thing by greatly reducing the earth loop currents?
I've got a direct injection box which allows an earth "lift", which disconnects the input ground with respect to DC, but leaves a 100uF cap in place to AC-couple the input ground.  That seems like a good idea (at least as an option).

But I don't see these options in general use.  Why not?

[d95err]

Hi
I'm a newb about this kinds stuff, but I've noticed a couple of things in various circuits.

I've got an old power amp, designed by some top-notch EEs, where the signal input ground is connected via a 1 ohm resistor to the true ground.  Could this be a good thing by greatly reducing the earth loop currents?
I've got a direct injection box which allows an earth "lift", which disconnects the input ground with respect to DC, but leaves a 100uF cap in place to AC-couple the input ground.  That seems like a good idea (at least as an option).

But I don't see these options in general use.  Why not?

Iside the DI-box, the signal is isolated from ground by an audio transformer or by an opamp network. In normal use you'd have the ground "lifted" to break ground loops. In some cases it is good to leave the ground connection on (e.g. if the ground is lifted at the other end of the cable too). The balanced circuit (opamps or transformer) is required to make the isolation work. Just adding a series resistor to ground in one end is not a complete solution for ground loops, although I've heard that it may improve things a bit.

In the poweramp, it could be that the whole signal ground of the amp is lifted from the safety/chassis/true ground. I've got something similar in my DIY tube amp.

[R.G.]

I've got an old power amp, designed by some top-notch EEs, where the signal input ground is connected via a 1 ohm resistor to the true ground.  Could this be a good thing by greatly reducing the earth loop currents?
I've got a direct injection box which allows an earth "lift", which disconnects the input ground with respect to DC, but leaves a 100uF cap in place to AC-couple the input ground.  That seems like a good idea (at least as an option).

But I don't see these options in general use.  Why not?

They're not in general use because in some countries such practice is illegal - it violates electrical safety codes. In the USA, it is illegal by some city ordinances although not by national law. And whether it is actually illegal or not, it violates safety codes in all countries that have them.

All electrical safety codes have the requirement that any equipment that has a third wire safety ground have it connected to any accessible metal parts. "Accessible metal" means if you can touch it, even with considerable force, and in recent European regulations, even by poking your hand through a speaker cone. The ground bonding has to be less than 0.1 ohm, and has to carry a 25A current to pass the test. One ohm resistors, capacitors, etc. do not pass.

The "top-notch EEs" are probably not aware of the safety rules, expecially if they really are top notch EEs, because except in very small companies, only the second tier EEs have to deal with electrical safety. It's often assigned as punishment for a poor reputation in big companies.

One modestly safer way to do things is to connect a 35A diode bridge with + and - terminals shorted between signal ground and safety ground, paralleled by a one to ten ohm resistor. This still will not meet safety regs, but it has a better chance of protecting you if something fails in the AC power wiring. Currents less than 35A will be limited to +/-1.4V, arguably safer than power line voltages, and as long as the current surge does not burn out the diodes you're modestly protected. A competent safety engineer would be horrified. One could argue that coupling this with a fuse of less than the diode rating in both incoming AC lines adds another layer of safety, protecting the diodes, but it still will not meet safety testing practice.

The single IC and capacitors is probably cheaper than the diode bridge and power resistor, and still meets electrical safety codes.

I think...     

[zpyder]

It's a classical dual buffer/balanced diffamp circuit on a chip. Two sets of them, actually.

Sweet.... I don't know exactly what that means  , but I'm getting there.  The more datasheets I read, the more I start to understand them.  I love learning...

R.G.: You titled this topic "Ground loop isolator" - would a more appropriate title be "Ground loop hum canceller/rejector"

cheers,
zpyder

[ADR]

Hi,

I've looked at the pdf for the IC on the website listed above. I'm trying to figure out how to try this out for myself.

Is there a topic on geofex about this opamp ground loop breaker circuit? I'm looking for something like it, but all I see involve xformers. Is there another resource? The search here is yeilding lots of links, but have not sifted thru anything yet that matches or explains what the idea shown in the pdf.

Cheers

A

[brett]

Hi
RE: That thing about using a 1 ohm resistor

The resistors in question were either 1 or even 0.1 ohm, and from the little that I understand were used to "lift" the signal earth above the power supply earth. Presumably, ground currents in the chassis relate to very small (milliohm ?) resistances. So 1 ohm would be a very high-resistance pathway, and many dB of noise reduction could be achieved (?).  Or is this fanciful?

I've seen this method used in the Playmaster hi-fi (1970s ) amps. They were designed by the engineers at Electronics Australia (who now publish Silicon Chip).
cheers

[Processaurus]

A great add on for stereo effects, for people that are crazy enough to use two amps, which inevitably get plugged into different outlets.