Is it safe to isolate circuit 0V from chassis/PE.

Hey all, I've signed up to ask this as I've read conflicting advice on what is probably quite an important subject.

(I know that the chassis must be connected to the earth of the mains input wire)

I've read that the chassis (and by extension, protective earth) need not be connected to the circuit ground and that it is often only done so to keep the chassis and circuity at the same potential and it is perfectly fine to leave the circuity floating within the chassis. Therefore some suggest the use of 100ohm and 100nF caps in parallel between 0V and earth in order to reduce ground loop effects.

Is this really safe though? What happens if mains voltage gets onto the audio circuity somehow?

Not sure I understand your question.

The audio lines can not be low impedance to safety ground, so if mains voltage gets into audio lines it will generally result in smoke. One possible exception is transformer isolated audio, but I don't think that is your question.

That said shield grounds need to be bonded to chassis and safety ground to drain away noise.  Professional interfaces use separate + and - audio lines in addition to the shield which can be ignored by the audio that does not share a common connection.

JR

For example... If we assume the chassis and PE are joined and that audio 0v is left floating (power supply has no connection to earth)- Suppose the mains live wire somehow comes loose, if it touches the chassis we are OK, if it touches audio 0V we are probably less OK, assuming for example there are unbalanced connections where the cable plug might be common to audio 0V?

Alternatively, if there is a primary to secondary mains transformer fault (somehow) or a faulty SMPS module that allows mains through to the outputs.

Perhaps these are such hypothetical situations that they would never occur, or couldn't occur in the manner I describe.

Everyone talks about chassis to PE as crucial but I don't see 0V to PE being mentioned nearly as often,.

sharkee said:

For example... If we assume the chassis and PE are joined and that audio 0v is left floating

Click to expand...

I ASSume PE is the same thing as safety ground, the third wire in most outlets (here in the US)?

(power supply has no connection to earth)-

Click to expand...

can but doesn't have to. Most consumer products use 2 wire line cords and double insulated power transformers to reduce the risk of mains voltage leakage to acceptable hazard levels.

Suppose the mains live wire somehow comes loose, if it touches the chassis we are OK,

Click to expand...

OK for a safety grounded chassis (fuse/breaker will trip), not so safe for a 2-wire consumer product (floating chassis will become energized), but primary wiring is double insulated to reduce the probability of that happening.

if it touches audio 0V we are probably less OK, assuming for example there are unbalanced connections where the cable plug might be common to audio 0V?

Click to expand...

Maybe... 

Alternatively, if there is a primary to secondary mains transformer fault (somehow) or a faulty SMPS module that allows mains through to the outputs.

Click to expand...

UL and various safety agencies have strict requirements for power transformer insulation. In production products often get hi-pot tested,  testing for leakage at elevated voltage

Perhaps these are such hypothetical situations that they would never occur, or couldn't occur in the manner I describe.

Click to expand...

They occur but safety grounded chassis helps (by tripping fuse/breaker). Additionally modern GFCI/RCD mains power protection circuits can detect when all the mains current is not returning to the outlet it left from. In US as little as 5 mA of mains leakage can trip the GFCI and remove power (I believe RCD thresholds are higher because they protect the entire branch of outlets).

Everyone talks about chassis to PE as crucial but I don't see 0V to PE being mentioned nearly as often,.

Click to expand...

I still don't understand your question.

JR

Yes, the entire circuit can be "floating" relative to earth. In fact, most consumer electronics are floating. If it's power cable doesn't have an earth prong how could the circuit be connected to earth?

From a safety perspective the critical thing is to have all metal enclosure parts connected to an earth wire so that if some voltage touches anything metal inside it will just short and blow a fuse and not pass the voltage to the chassis and have someone get shocked.

If the circuit PSU is floating and some voltage touches the chassis it will become that voltage but without a ground connection between the PSU and earth, there will be no path for current to flow. Your hair might stand on end though.

Regarding faulty SMPS, no modern SMPS worth it's weight in Hydrogen is going to pass mains to the outputs. They're just not designed that way [1]. In fact, a remotely good one will use a small transform instead of an inductor so that the outputs are both floating (which is very handy because they can be "stacked" to make + and - voltages).

Note that a lot of DIY circuits and pro-audio circuits assume the circuit ground is connected to earth. So you would need to be very conscientious about that to make it work. For one thing all of your jacks would have to be isolated. That is pretty much unheard of in the pro-audio world. That's the sort of thing cheap consumer electronics do and it shows when you connect more than one "floating" piece of gear together because it will tend to hum from each circuit being at different potentials.

[1] Disclaimer: I am not an EE and I am not responsible for any safety related issues that you might encounter from following my  advice on a "DIY" website.

sharkee said:

...... Therefore some suggest the use of 100ohm and 100nF caps in parallel between 0V and earth in order to reduce ground loop effects.

Click to expand...

Bryston uses 100ohms/100nF/2xdiodes  as shown in the picture.  The diodes should withstand the current higher than rated by the fuses in the primary  of the transformer.  It's good solution, IMO.

http://www.rane.com/note151.html is a good article about grounding.

The same as the loop breaker circuit described here http://sound.whsites.net/earthing.htm
Which is what got me asking the question really...If the following quote is true and the diodes are necessary, how is it then also safe to have no connection there at all, or just the 100r/100nF with no diodes.
" the diode bridge provides the path for fault currents. The use of a large chassis mounting (35A) type is suggested, since this will be able to handle the possibly very high fault currents that may occur without becoming open circuit. "

sharkee said:

The same as the loop breaker circuit described here http://sound.whsites.net/earthing.htm
Which is what got me asking the question really...If the following quote is true and the diodes are necessary, how is it then also safe to have no connection there at all, or just the 100r/100nF with no diodes.
" the diode bridge provides the path for fault currents. The use of a large chassis mounting (35A) type is suggested, since this will be able to handle the possibly very high fault currents that may occur without becoming open circuit. "

Click to expand...

It seems like you may be overestimating the risk from audio lines.

0V is generally not treated as a safety ground. Years ago when designing a small install amp that had an external screw terminal labelled "ground", UL insisted on testing it for safety ground bonding. Their bonding test involved passing something like 50 amps with less than a 10V rise, long enough to trip the fuse/breaker.  My original PCB trace vaporized, and I could have just relabelled the screw terminal 0V or something other than ground, but I decided to beef up the PCB trace (when in doubt do what is right).

The trick of using anti-parallel diodes, across a modest R and C between circuit ground and safety ground works. I used it back in the 80s to manage ground conduction paths between rack mount gear screwed into the same rack. Later while at Peavey I had our in house agency guy talk to UL about getting the diode trick approved by UL. They were receptive to review it but I couldn't justify spending the couple ten thousand dollars to pay for all the UL testing. You can just tie shield ground to chassis and treat the audio lines differentially for next to no cost.

JR

PS: The suggestion to use a healthy diode bridge for diode shunt trick is valid if pursuing that approach. Using small diodes will vaporize before tripping the fuse/breaker. Using a larger bridge the diodes will still melt and fail as short circuit but they won't vaporize and open the safety ground path.

JohnRoberts said:

It seems like you may be overestimating the risk from audio lines.

Click to expand...

Possibly, however I was quite happy using just the 100r and 100nF until I saw people advocating 35A diode bridges, so now I'm thinking what I've doing isn't safe.

Lets change the question a bit, is it safe to use that loop breaker circuit without the diodes?

sharkee said:

Possibly, however I was quite happy using just the 100r and 100nF until I saw people advocating 35A diode bridges, so now I'm thinking what I've doing isn't safe.

Lets change the question a bit, is it safe to use that loop breaker circuit without the diodes?

Click to expand...

If concerned about safety, using GFCI/RCD mains leakage cut out devices will provide adequate human safety protection and not involve changing audio circuits.

AFAIK, the diode safety ground lift trick is NOT UL approved (for safety ground bonding) and probably not widely used in mass production. I seem to recall one company selling a commercial gadget that uses the diode trick inside and last I checked they were not UL listed.

Many manufacturers use the modest value resistor/cap ground lift. I am not aware of injuries that resulted because of that. More dangerous IMO are stinger caps in old guitar amps, ground lift mains plug adapters, etc.

JR

Just try a 10R/10W resistor in parallel to the 100nF.

Sometimes I go as low as 1R/5w with no HUM.

If properly done, even shorted 0v to chassis shouldn't cause problems. I did it many times.

Why 10W resistor, what happens if I use 1/4W?

sharkee said:

Why 10W resistor, what happens if I use 1/4W?

Click to expand...

it will be smaller and cheaper... 8)

JR

So if that's safe to do, I really don't see why a 35A bridge is then recommended to handle the fault currents.

sharkee said:

So if that's safe to do, I really don't see why a 35A bridge is then recommended to handle the fault currents.

Click to expand...

If that path is considered a "Safety Ground" bond. i.e. the path must provide a low impedance current path back to the mains power panel safety ground sufficient to trip a branch fuse/breaker, a small diode will not work. A diode will generally fail as a short circuit, but the several tens of amps from a direct mains fault will vaporize a small diode, making it an open circuit.

In my bench testing several decades ago I determined that modest diode bridges had enough thermal mass to not vaporize before the fuse/breaker tripped. I didn't use 35A bridges while something heavier than 1A diodes was required. 

I repeat I stopped using this approach decades ago, because proper differential treatment of audio signals allows shields to be directly connected to safety ground. I only say that "IF" you use the diode trick, use a suitably sized diode bridge.

Note: not every ground path inside an audio product need pass UL ground bonding tests (something like 50A with less than 10V rise.).

JR 

Sorry If I'm being particularly thick...I really don't want to get this misunderstood.

JohnRoberts said:

A diode will generally fail as a short circuit, but the several tens of amps from a direct mains fault will vaporize a small diode, making it an open circuit.

Click to expand...

So would a 1/4w resistor but that's apparently fine to use? (in the same path)

I repeat I stopped using this approach decades ago, because proper differential treatment of audio signals allows shields to be directly connected to safety ground. I only say that "IF" you use the diode trick, use a suitably sized diode bridge.

JR

Click to expand...

It seems the advice is if you use a diode, use a big one but if you don't use a diode, a 1/4w resistor is fine. I just don't see the logic. If we need to use large current diodes so that they don't blow and go open circuit surely the same applies with the resistor. Why is it fine for a resistor to blow open but not a diode, we'd surely get much the same result?