power supply 3

[kjnangre]

Hi,
I have a simple power supply circuit (please see:

http://www.nangreaves.ca/ps.jpg).

This circuit supplies 5V for the rest of my board. The PS input is AC line voltage. If someone were to plug my board into the wall socket backwards (switching hot and neutral) then my whole ground plane would suddenly be at line voltage. Is that a problem and what can I do about it?

Thanks

 

[IRstuff]

The simple answer is to NOT DO THAT! Use a grounding polarized plug and make sure that the circuit chassis is grounded.

And for what possible reason are you tying DC parts of your system to the AC at all? Even if the user doesn't screw it up, there's no guarantee that the AC socket is wired correctly. Ignoring the human danger, there's the possibility of frying any equipment that's attached to your circuit, such as an oscilloscope (speaking from personal experience).

There's nothing quite so adrenaline pumping as having a piece of an IC go shooting by your ear when you connect the scope ground to the chassis, only the chassis is hot.

TTFN

 

[kjnangre]

Thanks IRstuff.

Part of my circuit (not shown) needs to know the AC voltage that is being applied to the board (this ia a power measurement device). That is done very simply by inserting a voltage divider across the AC live voltage and then into an A/D converter. But this forces me to tie my DC and AC grounds to keep a constant reference.

I am using a polarized plug, but, in the unlikely case that it does get plugged backwards, will haveing my ground plane at line voltage be a problem. My board is not user accessible, unless someone takes my box apart.

 

[MacGyverS2000]

Please do not allow this design out of the house for public consumption...it's downright dangerous. BTW, what company will these be branded as? Just want to know so I can steer clear.

 

[kjnangre]

Alright MacGyver, why don't you suggest an improvement then?

 

[IRstuff]

Why is this necessary? You already have an isolated and reduced version of the AC main through the transformer.

As to your main question, yes it's a BAD idea, because you'll likely cook the crap out of your DC circuits. Transient delays within the circuit will likely result in large AC voltages appearing where they don't belong and damage parts. It's also unlikely that you can adequately isolate the chassis from AC coupling and the user will get a nasty surprise.

TTFN

 

[kjnangre]

Because the secondary voltage of the transformer is a function of the current draw, which is not constant, so measuring the voltage here could not give me the high accuracy I need.

Since my device has a 3-pronged plug, an improperly wired outlet is really the only way it could get plugged in backwards. Does anyone how common backwards outlets are?

Thanks again

 

[felixc]

You are going to need a way to reduce the voltage down to what a converter can measure. If you want any precision, you may also need a means to calibrate your system. The use of high-voltage, high value resistors to step the voltage down to what your converter can read, may let you escape the need to connect your chassis to the mains.
If your appliance has its case fully isolated, a reverse connection will cause no harm to the user. (but you have to be cautious during development)
If you want to sell this product, you will need a UL approval. They will tell you all that you need to implement to be legally safe. In addition to protecting the user after some known rules, such a label will protect you legally in showing that you did your homework.

 

[logbook]

First of all, do not even consider making the circuit as shown. It is unnecessarily lethal. If you consider standard EN60950 (or IEC60950) for example there is a clause about not killing service technicians. Your circuit will do exactly that.

Secondly, you have drawn an Earth symbol on your circuit, which nobody has commented on. That symbol means the protective conductor of the mains power supply, it does not mean a chassis connection. Whilst SPICE simulators use this symbol for signal 0V connections, it must never be used incorrectly in mains based circuitry as you have done. The chassis symbol is a collection of straight lines from the base plate at a 45 degree angle.

Now it would be less than ideal for me to just tell you what you can’t do. It is very straight forward to earth the DC supply 0V, in fact that is a very safe thing to do. OK. Then pot the mains signals down relative to earth as well. Use high value resistors so your earth leakage current is say less than 1mA. Use high integrity resistors since you don’t want to introduce an earth fault current. These resistors would be rated in excess of 3000V for example. Pot down both the live and the neutral conductors. Safety standards like you to treat live and neutral conductors with equal respect, just in case.

Now the signal has been potted down to a safe level you can use an off the shelf differential amplifier to measure the difference between the two potted down mains signals. This will give you a true scaled version of the mains voltage. If you need more than a couple of percent accuracy then put in a pot to adjust the gain. This makes a very safe scheme at low cost.

 

[buzzp]

Measuring mains voltage is often done through direct connections via resistors to scale the voltage down. This is not uncommon. Some choose to run this through a diff amp to eliminate common mode noise. When this is done, the ground on the secondary of your transformer is not necessary and allows for some isolation between your DC and AC circuits. High impedance is recommended but also need to watch the input impedance to your A/D converter (some want 10k ohms or less). Generally, the higher resistance longer conversion time (depends on converter type).
Eliminate any pots you have in the circuit as these are huge sources of error in a voltage divider circuit that have to operate over a decent temperature range.

 

[nbucska]

Your supply -- unless you connect GND-s -- separates
the circuit from the line. You may use insulated circuit
with transformer or optocoupler coupling.

What do you have to measure on the line ( V,I,etc ) and with what kind of accuracy? How much delay ? How many samples per sec?



<nbucska@pcperipherals DOT com> subj: eng-tips

 

[buzzp]

To answer your question: First eliminate the ground connection on the DC side. Also, not sure what your trying to do with the resistors on the primary side of the transformer but a series resistor after your choke along with two resistors for a voltage divider is all you need (I assume this is where your A/D input will come from). Run the signal from the voltage divider (usually scaled to produce no more than 5V at full scale-depends on Vcc and Vref)to a diff amp (other input from neutral on transformer). Then your circuit will be fine and you will not care if the polarity is reversed. I guess the real question is if your circuit/device will require grounding of the enclosure (some do NOT-phenolic or in general, plastic). If it does, then the only thing you can do is document the polarity as best as you can and follow all the standards when it comes to warning labels, instructions, and labels required at the terminals. No design is fool proof. You have to assume some intelligence on the end user. You can try and try and try but someone will find a way to hook it up wrong. As engineers, one of our jobs is to decide how much fool proofing do we need to implement based on the end user.
Anyway, I do not see anything real wrong with your circuit except for the ground on the DC side of the circuit (ignoring the resistor configuration on the primary side of the transformer).

 

[kjnangre]

Hey all.

The resistors on the primary (R1,2,3) are just jumpers that allow the circuit to be configured for 120V or 240V while the transformer outputs 10VAC regardless.

The case is plastic, and has no electrical connection. The ground symbol in the schematic just means my board's ground plane.

I am using a shunt resistor and a voltage divider for the I and V measurements. This is then fed into the ADE7753.
The ADE7753 has two fully differential voltage input channels. The maximum differential input voltage is ±0.5V. In addition, the maximum signal level is ±0.5 V with respect to GND.

The 0.5V differential requirement is easily met by selecting the right resistor values. But in order to meet the second requirement (sinal level within 0.5V of ground) I have to tie AC neutral to ground.

 

[kjnangre]

For the sake of clarity, I drew up a quick schematic:

http://www.nangreaves.ca/ps1.jpg

 

[walker1]

Just for the fun of it.

Have you thought about using measurement transformers ?
(That is the way it usually is done at the power plant, but yes, somewhat higher voltages and currents

You may have to do the current transformer yourself (remember to have the secondary side shunted with a few ohms), but you should be able to find, say, 1-5 VA mains transformers for the voltage.

If the transformer is for 115V, you can put a capacitor or a resistor in series with it to the AC supply.

Signal transformers can be used the same way, but be sure they follow the same standards isolationwise.

 

[buzzp]

If this is the case then I guess you have to tie the AC neutral to DC ground. This is still okay. You will have more issues with noise and the like but nothing that can not be overcome.
I would take off the earth grounds shown on your schematic since they will not actually be there (at least in your design). Since this is not actually earth ground at your device, then there is no issue with polarity. Maybe I am missing something but I see no problems as long as that earth ground on your jpg file is not really an earth ground but a signal ground. As long as the A/D uses this same 'signal ground' then your okay (you may want to protect the input with a zener or something especially if the application can draw large currents on starting (motor), unless of course your continuous operation level is relatively low on the full scale you set up with your resistors). Good luck.

 

[logbook]

Ok, treat the whole circuit as being live. Mark it internally as such. Use a different symbol for the signal 0V of the ground plane. You will find all sort of triangular or flat lines that you can use as a 0V interconnect, anything but the safety earth symbol! Anyone working on this circuit has to know not to hook up their (earthed) scope to the 0V line.

TVs have run with a "hot" chassis for years, so it is not of itself a problem.

Since you have said that there are no connections to this circuit, I assume that there is a visual display interface, otherwise where is the output? Make sure you don’t accidentally connect up an interface such as an RS432 etc unless it goes through an isolation barrier.

 

[kjnangre]

Yes, the only output is an LCD interface, there is no electrical connection with the outside other than the AC mains input.

I considered measurement transformers, but size and cost are both critical design constraints for this project, and resistors beat transformers on both counts, hands down. Admitably, transformers have the benefit of isolation.

Thanks for the note about the ground symbol.

Now then, my last question. If the device were plugged in backwards, my local ground plane would be connected to the hot conductor. That local ground plane serves several digital logic chips which communicatewith each other through a serial bus. My assumption is that since those chips have no knowledge of earth ground, they will happily do their thing and function normally, not knowing that their local ground is sinusoidal with respect to earth. Simiarly, since my 5V linear regulator has no knowledge of earth ground, it will happily create a 5V power plane which is at all times 5V higher than the ground plane. Is this assumption reasonable?

This has been an excellent discussion so far, thank you all.

 

[melone]

Could you get away with using a PTC (Postive Temperature Coefficent Thermister)? They have low resistance until a certain threshold, then their impedance rises very quickly, as the part heats up. This way the circuit shuts itself off.

http://www.circuitprotection.com/lvr/

Just a thought.

 

[walker1]

In case the capacity between your internal 'low-voltage level' and Mother Earth isn't too big and that the user has absolutely no way to tuch the internal voltage level, you should be home free.
At 50-60 Hz you will need some uF (many ?) before it could have an influence, and by then you would be way above the safety limit.
Not knowing your mechanics, I would say 1 nF would be a better guess for the coupling at hand.

I once had a TV set that worked that way.
It was before SCART was invented, the RF input was sent through certified capacitors, and the headphone output came through a transformer