Current Leakage Problwm with Switch Power Supply

[JimboJones21]

Hello All,

I have an interesting problem with a power supply I put together. I bought 12 switching power supplys from a company called Sinpro. The output is 48V and 1.7A. The input is 120V. I connect 2 in parallel through a diode so that I can diliver more current to my load, and I have 6 loads (2*6=12).

The interesting problem is that if you measure the voltage between the ground pin (all the switching boards are grounded together) and the ground of another device pluged into the same outlet I get 116V at 4mA!! I found this out initially buy touching the case of this supply and the case of another device!! Zap!

Any ideas on how to stop the current flow or make it flow easier, if that makes sense? Or is this inherent with a switching power supply?
Thanks.

 

[itsmoked]

First you need to figure out if you are seeing a capacitive effect or a conducted problem.

Hook your meter up between the receptacle ground and your boards. Got 116V? Okay while leaving the meter hooked up and untouched wire something like a light bulb between these same two points. A 7W night light would be best but any bulb will work. As soon as you hook up the bulb what happens to the reading?

Well?

Don't sit there... Make the measurement!

 

[JimboJones21]

Thanks itsmoked, I will try that tomorrow What would be the remedy if it were capacitive or conductive?
I'm guessing the light will flash if it's capacitive since this would be the built-up charge discharging, correct?

 

[itsmoked]

Actually if it's capacitive the light won't go on and the meter will drop to nil.

The solution will depend on how the test comes out.

Another question: Is there a line filter feeding the switchers?

Is there a ground to the switchers?

 

[fsmyth]

Sounds like you need to distinguish between SAFETY
ground and SYSTEM ground as a first step. On a
truly isolated SMPS, there should be NO, repeat NO,
interaction of the two. Whether or not to hold
them at the same potential is usually a noise issue.
In general, most SMPS (switching supplies) incorporate
a noise filter in the input circuit which should be
referenced to SAFETY ground (since this is the reference
used on the supply side. If this is done, there will be
NO potential difference between case grounds until there
is a fault. Note that some SMPS manufacturers use a
common ground (the supply side and the converter output
grounds are connected internally), and you have no
choice in the matter.
So, make sure the input filters have a ground path to
the incoming supply ground, and that all equipment cases
are connected to the same (SAFETY) ground. Whether or
not the output (SYSTEM) ground(s) are connected to the
chassis (SAFETY) ground is a noise issue only.
<als>

 

[JimboJones21]

Thanks itsomked and fsmyth. I still haven't had to time to do the test!! I'm new to the working world and didn't know there was a safety ground and system ground. I will have to look into this further. I know that there is 1 ground pin on the board and that's what I connect to.
I contacted a sales engineer at Sinpro and they told me that they have an EMI filter board designed to solve this problem.
Would this EMI filter be complex? I'm familiar with filter design from one of my electronics classes, but they were pretty simple single pole filters.
Any tips on how to go about designing an EMI filter?
Thanks again.

 

[coconutalley]

The question I would ask is: is the power supplies isolated?
If you measure from the AC/safety ground to an non-isolated supply, you'll see 120V.
If you are measuring from + and - of the supplies you should see 48V.
If the supplies are isolated, then unless you connect the return on the output to AC/safety ground, there should be no ac voltage. One possiblility that you could be seeing 120V is through your scope ground. Actually your scope ground is eventually connected to AC/safety ground and you could be making the connection there unintentionally. You used differential probes for this type of measurements. You could use a cheater plug but then you are defeating the safety intention of safety ground.

Good luck.

 

[OperaHouse]

"I get 116V at 4mA!! "

You may get both readings, but I doubt that you get both at the same time. The EMI filter may be part of the problem. These usually have a cap to ground that pass higher frequencies to ground. An agency approved supply will have leakage well below the 4 ma you have read. I think you have to look and see if you really have a a ground on one of your two seperate devices. While you mighr get some induced voltage from switching, it won't be a potential that high. They make hugh current sourcing milliohm meters like HYPATIA 309 that can test groundsat 30 or more amps. Amazing how many ground connections aren't that good!

 

[itsmoked]

Yeah, I had an S-100 computer that had a standard emi filter in front of the 8V @ 40A and +/- 19V @ 10A linear unregulated supplies. One day the ground wire plug/socket failed. If you touched the 20" rack mount computer it would blast you out of your chair.

Investigation showed the standard 3 cap, 2 inductor filter
with capacitors tied to the safety ground. This means if the safety ground is not hooked up then the case of the filter and hence the case of the computer were set to 120Vac X 50% or 57-60Vac.

Note in the picture below C2 and C2 make a voltage divider to the product case if that G on the left is not hooked up.

I am wondering if this isn't what JimboJones21 has been experiencing.

 

[hottip]

itsmoked, I think you are right! There is an issue in the electrical design literature if you are using earth leakage devices (or if you want so called RCCB) that the computers are having 3.5 mA earth leakage to earth all the time(because of filters). This is the reason that it is not recommended to put in the same socket circuit protected by a RCCB (10-30mA earth leakage protection) more than 3-4 computers. And switching power supplies are equiped with such a filter wich has a constant earth leakage of 3.5 mA .

 

[JimboJones21]

Thank you all this has been very informative. I ordered some of these 'EMI' filters from Sinpro becaause they claim it will reduce my leakage current from 4mA to 0.5mA. I don't understand how a filter can block leakage current? The signal is AC so you could design a filter to cut-off before that frequency, but I believe its 60Hz and that would be blocking my incomming signal from hydro? Or would this filter be connected between ground and neutral? Or am I on the wrong track?

After reading hottip's post I'm wondering if the filter they are sending is to compensate for the filter that the switching power supply came with? So is the only way to stop the leakage current to dissconnect the EMI filter that the switch supply comes with?

I still don't undertand Safety ground and System ground. Is the safety groung the earth ground, and the system ground the ground that the 12 switching supplies are connecter too? In that case I have the safety ground connected to the system ground. All the grounds are connected to the case and the case is connect to the earth ground.
Would one solution be put the 12 switching supplies on there own metal base, isolate this from the case, and connect this metal base to the earth ground? This way nothing flows through the case.

 

[itsmoked]

JimboJones21; Model number and mabe a link to the supplies?? we can get to the bottom of this..

 

[fsmyth]

Here... this might help explain the situation:

http://people.txucom.net/fsmyth/pages/sw-regs1.html#nec01

Just an example picked at random; it may or may not be
similar to your supply. Note jumper J1 - it allows the
output common (system ground) to be tied to the line input
(safety) ground. Pretty much standard in OEM SMPS offerings. There are reasons not to common SYSTEM and
SAFTEY grounds, but they do not seem to apply to you.

Adding the additional EMI filter on the line input will NOT
alleviate your problem; in fact, it will probably aggravate
it.

Step 1: make sure that the power supply board is grounded
properly. Some boards do not have a pin on the input
connector, but rely on one (or more!) of the mounting holes
for the board to be in contact with the chassis, which is
assumed to be at line ground (i.e., the green wire of the
power cord is bolted to the chassis). With this done, you
will now have a return circuit for the filtering included
within the SMPS. This is your SAFETY ground.
Step 2: if your output common is to be connected to the
chassis (as seems to be the case), it can be done at any
point on the chassis. This is your SYSTEM ground, which
is now at the same potential as the SAFETY ground.
Step 3: your EMI filter will have a lug or bolt welded to
its case; it and the case should be solidly attached to
the same point that the input line (SAFETY) ground is
attached, which is almost always the chassis/case enclosing
the equipment. In general, it is not necessary, but
will not hurt. (In fact, it will furthur isolate the
SMPS-generated noise/harmonics from the utility power).

If all this is done, there will NEVER be an instance in
which you will feel (by touch) a voltage on the grounds,
even in a fault condition --- always assuming that the
plant power wiring is correct, i.e., that the ground lead
in the receptacle has continuity all the way back to the
utility/supplier ground/earth (which should always be the
case), and that the branch circuit is properly protected,
preferably by GFCI.

Now, back to your original post:
" I found this out initially buy touching the case of this supply and the case of another device!!"

There may not be a problem with the wiring in your
power supplies at all...............
Sounds like one of the equipment cases is NOT properly
grounded to the receptacle. Check the cords; make sure
that they are all 3-prong (with ground), and there is
continuity from the ground pin (the round one) to the
equipment chassis. If this not the case, make it so.

<als>

 

[JimboJones21]

Thanks for the help. So far I haven't had time to work on the power supply problem. But I did try messing around with the grounds and I noticed that if I twisted the plug connections on the extension cords and outlets the voltage across the 2 cases decreased, but from 115V to 75V, so there's still a problem with the leakage current.
I noticed that the schematics you sent has an isolation TX at the input. Interestingly the supply cards I have do not have any isolation TX. I have a feeling that I have chosen a too leaky SMPS design for my idea. I'm curious to see what Sinpro sends me. I will report on it when it's received to see what you guys think.

 

[fsmyth]

Transformer irrelevant (here). There will always be
isolation between the line input and the D.C. output
in the types of supplies examined here.

Another example (similar, with no input transformer:

http://people.txucom.net/fsmyth/pages/sw-regs3.html#sanps3

If you mean bt "twist" that you swapped the cord placement,
then it is almost a surety that one of the chassis is NOT
properly grounded. Again, check continuity between the
ground prong on EACH cord and its associated housing.

<als>

 

[coconutalley]

To answer your question on whether to ground the DC common or return to the safety ground: answer is NO if you are trying to isolate 120vac from dc. This is usually needed to keep the noise level from passing through one or the other. The emi filter is useless here if you connect safety ground to the DC common. This defeats the isolation because there is a path via safety ground for noise, bypassing the filter. Most robust designs require isolation.
If you don't need isolation between AC and DC, then you can just rectify the 120vac, filter it, and use cheaper buck converters.
A lot of low power consumer products have the dc isolated for safety reasons,e.g. if any of the ac filter caps shorts to safety ground, the floating dc side wouldn't see those voltages. I'd check with applicable UL for safety requirements if this is going to be a consumer product application.

 

[JimboJones21]

Thanks coconutalley, what I've done is used the chasis as common ground for both the safety ground and DC ground. This is as you described my problem.
I bought pre-made switching power supplies (SMPS) and hooked them in parallel and the current leak to ground is very high.
How can I isolate these SMPS from the AC ground and their DC outputs? An isolation transformer on the input? How do you isolate the AC and DC and give a path back to ground if there is a fault in the SMPS?