Maker Pro
Maker Pro

Advise needed on EMC

S

siliconmike

Jan 1, 1970
0
I'm building a small device that
- runs on a 9V battery
- includes 9V to 4.5V voltage splitter IC running at few hundred
kiloherts
- includes 8 MHz microcontroller with in-built oscillator, operating on
4.5V
- The microcontroller uses PWM to control intensity of 4 LEDs connected
to its port pins.

Thats about it, the device has 4 LEDs that dance at varied intensity.

Well, this device would be submitted for EMC testing as per CE
requirements.

I would appreciate advise and comments on whether this low voltage
device would pose a EMC failure risk, or whether there is absolutely no
risk.

I would also be greatful for advise on how to minimize risk of EMC
failure.

There are 2 pcbs between which the battery is sandwiched. The top PCB
has 4 LEDs. The bottom PCB has the electronics. The pcbs are connected
with perpendicular wires.

| | | | leds
----------------- top pcb
[ ] battery
----------------- bottom pcb

Thanks,
Mike
 
P

Pooh Bear

Jan 1, 1970
0
siliconmike said:
Well, this device would be submitted for EMC testing as per CE
requirements.

What country are you in btw ?
I would appreciate advise and comments on whether this low voltage
device would pose a EMC failure risk, or whether there is absolutely no
risk.

Oh there's a risk for sure. Is the enclosure plastic or metal ?
I would also be greatful for advise on how to minimize risk of EMC
failure.

How long have you got ! ?

If you can find it, Philips had a useful app note on minimising EMC issues for
things like remotes esp wrt pcb layout.
Here you go. 1st link. The other are likely useful too.
http://www.google.com/search?hs=NTw...s+emc+application+note+pcb+layout&btnG=Search

Graham
 
S

siliconmike

Jan 1, 1970
0
Pooh said:
What country are you in btw ?

India. The device will be exported to the UK.
Oh there's a risk for sure. Is the enclosure plastic or metal ?
Plastic.


How long have you got ! ?

3 months.
 
L

Leon

Jan 1, 1970
0
siliconmike said:
I'm building a small device that
- runs on a 9V battery
- includes 9V to 4.5V voltage splitter IC running at few hundred
kiloherts
- includes 8 MHz microcontroller with in-built oscillator, operating on
4.5V
- The microcontroller uses PWM to control intensity of 4 LEDs connected
to its port pins.

Thats about it, the device has 4 LEDs that dance at varied intensity.

Well, this device would be submitted for EMC testing as per CE
requirements.

I would appreciate advise and comments on whether this low voltage
device would pose a EMC failure risk, or whether there is absolutely no
risk.

I would also be greatful for advise on how to minimize risk of EMC
failure.

There are 2 pcbs between which the battery is sandwiched. The top PCB
has 4 LEDs. The bottom PCB has the electronics. The pcbs are connected
with perpendicular wires.

| | | | leds
----------------- top pcb
[ ] battery
----------------- bottom pcb

Thanks,
Mike

It doesn't cost much to hire the facilities at a test house for a
couple of hours for preliminary testing, they do that where I used to
work. A unit I designed failed on leakage from the RS-232 cables, it
just needed a simple filter consisting of a few ferrite beads and
capacitors to be incorporated to pass the tests.

Leon
 
T

Terry Given

Jan 1, 1970
0
siliconmike said:
I'm building a small device that
- runs on a 9V battery
- includes 9V to 4.5V voltage splitter IC running at few hundred
kiloherts
- includes 8 MHz microcontroller with in-built oscillator, operating on
4.5V
- The microcontroller uses PWM to control intensity of 4 LEDs connected
to its port pins.

Thats about it, the device has 4 LEDs that dance at varied intensity.

Well, this device would be submitted for EMC testing as per CE
requirements.

I would appreciate advise and comments on whether this low voltage
device would pose a EMC failure risk, or whether there is absolutely no
risk.

I would also be greatful for advise on how to minimize risk of EMC
failure.

There are 2 pcbs between which the battery is sandwiched. The top PCB
has 4 LEDs. The bottom PCB has the electronics. The pcbs are connected
with perpendicular wires.

| | | | leds
----------------- top pcb
[ ] battery
----------------- bottom pcb

Thanks,
Mike

You need to do a few things:

1) ensure the PCB with the electronics on it has a good 0V plane
underneath the micro, crystal etc. Ideally, lay your PCB out as single
sided (often easy with smt) and have the other side as a solid 0V plane.

using a single-sided PCB makes life a lot more difficult, and means you
need to pay a lot of attention to the frequency content of every signal.
a solid 0V plane allows you to pretty much ignore that stuff.

2) keep fast edges well away from the interconnecting wires and LEDs. If
you trace the physical path that LED current flows through, you will
find a big loop. Current always flows in loops, and that loop is a
radiating antenna, and you specifically dont want that. You have 2 ways
to recduce the problem: reduce the size of the loop, and/or reduce the
high frequency content of the current flowing in that loop.

In your case the separation between the 2 PCBs and the pin spacing
determines the minimum loop size you can have. Your main pcb with a
solid 0V plane has essentially zero loop area; if all the LEDs have a
common pin, and you make that common (eg Anode +5V) one side of a
2-layer PCB, then the LED pcb also contributes essentially zero loop
area, and all thats left is the interconnection between the two pcbs

If you put the LED series resistor on the control pcb, then split it
into 2 series resistors, and stick an ~100nF cap to 0V in the middle of
them. That will filter out almost all of the HF content, so the LED
current will look roughly DC. By doing that you dont really care about
the LED current loop anymore, so dont need a 2-sided led pcb.

A similar argument holds for putting an L-C filter between the battery
and split-supply chip, to ensure the current that flows in the battery
wires is also DC. make sure your L is not a bobbin core (they have a
huge external air gap, so convert conducted noise into radiated noise)
but is a toroid, little E-core etc. also ensure your LC filter is
damped, either with an R-C damper, or a small R = sqrt(L/C) in series
with the capacitor.

Then your product will be extremely likely to pass first time.

it is quite feasible to look at the areas of the loops, and the actual
currents, and calculate whether or not the device will pass or fail. But
you really need to know what you are doing.
 
S

siliconmike

Jan 1, 1970
0
Leon said:
siliconmike said:
I'm building a small device that
- runs on a 9V battery
- includes 9V to 4.5V voltage splitter IC running at few hundred
kiloherts
- includes 8 MHz microcontroller with in-built oscillator, operating on
4.5V
- The microcontroller uses PWM to control intensity of 4 LEDs connected
to its port pins.

Thats about it, the device has 4 LEDs that dance at varied intensity.

Well, this device would be submitted for EMC testing as per CE
requirements.

I would appreciate advise and comments on whether this low voltage
device would pose a EMC failure risk, or whether there is absolutely no
risk.

I would also be greatful for advise on how to minimize risk of EMC
failure.

There are 2 pcbs between which the battery is sandwiched. The top PCB
has 4 LEDs. The bottom PCB has the electronics. The pcbs are connected
with perpendicular wires.

| | | | leds
----------------- top pcb
[ ] battery
----------------- bottom pcb

Thanks,
Mike

It doesn't cost much to hire the facilities at a test house for a
couple of hours for preliminary testing, they do that where I used to
work. A unit I designed failed on leakage from the RS-232 cables, it
just needed a simple filter consisting of a few ferrite beads and
capacitors to be incorporated to pass the tests.

Leon

I was wondering if I could send them the device and obtain preliminary
results. That way probably I could save on thousands of dollars charged
by government testing labs.

Could you please suggest the name of the lab?
 
S

siliconmike

Jan 1, 1970
0
Terry said:
siliconmike said:
I'm building a small device that
- runs on a 9V battery
- includes 9V to 4.5V voltage splitter IC running at few hundred
kiloherts
- includes 8 MHz microcontroller with in-built oscillator, operating on
4.5V
- The microcontroller uses PWM to control intensity of 4 LEDs connected
to its port pins.

Thats about it, the device has 4 LEDs that dance at varied intensity.

Well, this device would be submitted for EMC testing as per CE
requirements.

I would appreciate advise and comments on whether this low voltage
device would pose a EMC failure risk, or whether there is absolutely no
risk.

I would also be greatful for advise on how to minimize risk of EMC
failure.

There are 2 pcbs between which the battery is sandwiched. The top PCB
has 4 LEDs. The bottom PCB has the electronics. The pcbs are connected
with perpendicular wires.

| | | | leds
----------------- top pcb
[ ] battery
----------------- bottom pcb

Thanks,
Mike

You need to do a few things:

1) ensure the PCB with the electronics on it has a good 0V plane
underneath the micro, crystal etc. Ideally, lay your PCB out as single
sided (often easy with smt) and have the other side as a solid 0V plane.

using a single-sided PCB makes life a lot more difficult, and means you
need to pay a lot of attention to the frequency content of every signal.
a solid 0V plane allows you to pretty much ignore that stuff.

2) keep fast edges well away from the interconnecting wires and LEDs. If
you trace the physical path that LED current flows through, you will
find a big loop. Current always flows in loops, and that loop is a
radiating antenna, and you specifically dont want that. You have 2 ways
to recduce the problem: reduce the size of the loop, and/or reduce the
high frequency content of the current flowing in that loop.

In your case the separation between the 2 PCBs and the pin spacing
determines the minimum loop size you can have. Your main pcb with a
solid 0V plane has essentially zero loop area; if all the LEDs have a
common pin, and you make that common (eg Anode +5V) one side of a
2-layer PCB, then the LED pcb also contributes essentially zero loop
area, and all thats left is the interconnection between the two pcbs

If you put the LED series resistor on the control pcb, then split it
into 2 series resistors, and stick an ~100nF cap to 0V in the middle of
them. That will filter out almost all of the HF content, so the LED
current will look roughly DC. By doing that you dont really care about
the LED current loop anymore, so dont need a 2-sided led pcb.

A similar argument holds for putting an L-C filter between the battery
and split-supply chip, to ensure the current that flows in the battery
wires is also DC. make sure your L is not a bobbin core (they have a
huge external air gap, so convert conducted noise into radiated noise)
but is a toroid, little E-core etc. also ensure your LC filter is
damped, either with an R-C damper, or a small R = sqrt(L/C) in series
with the capacitor.

Then your product will be extremely likely to pass first time.

it is quite feasible to look at the areas of the loops, and the actual
currents, and calculate whether or not the device will pass or fail. But
you really need to know what you are doing.

Thank you for the detailed response. I intend to do exactly as you
suggested: reduce loop areas, incorporate filters with damper and use
DC planes.

I was also wondering if anyone knows of a test house who would do the
testing for me for a reasonable fee.

Thank you.
-Mike
 
P

Pooh Bear

Jan 1, 1970
0
siliconmike said:
I was wondering if I could send them the device and obtain preliminary
results. That way probably I could save on thousands of dollars charged
by government testing labs.

Could you please suggest the name of the lab?

In India ? You're joking. Well actually our subcontractor used ETL right next to
SEEPZ in Mumbai.

Simplest answer is to approach your test lab and ask about simple
'pre-compliance testing'. I suspect you're going to be most interested in
radiated emissions. I'm unsure how you'd test such a device for static discharge
immunity but you should ask them.

It pays to buy all the relevant standards btw. I just hope it's not as expensive
over there as it is here.

Graham
 
R

Rich Grise

Jan 1, 1970
0
India. The device will be exported to the UK.


Plastic.

Plate the inside of the case with metal, and put a spring contact on each
side of the PCB directly to ground, to ground both case halves without
any bolting or anything.

Good Luck!
Rich
 
R

Rich Grise

Jan 1, 1970
0
Terry said:
siliconmike said:
I'm building a small device that
- runs on a 9V battery
- includes 9V to 4.5V voltage splitter IC running at few hundred
kiloherts
- includes 8 MHz microcontroller with in-built oscillator, operating on
4.5V
- The microcontroller uses PWM to control intensity of 4 LEDs connected
to its port pins.

Thats about it, the device has 4 LEDs that dance at varied intensity.

Well, this device would be submitted for EMC testing as per CE
requirements.

I would appreciate advise and comments on whether this low voltage
device would pose a EMC failure risk, or whether there is absolutely no
risk.

I would also be greatful for advise on how to minimize risk of EMC
failure.

There are 2 pcbs between which the battery is sandwiched. The top PCB
has 4 LEDs. The bottom PCB has the electronics. The pcbs are connected
with perpendicular wires.

| | | | leds
----------------- top pcb
[ ] battery
----------------- bottom pcb

Thanks,
Mike

You need to do a few things:

1) ensure the PCB with the electronics on it has a good 0V plane
underneath the micro, crystal etc. Ideally, lay your PCB out as single
sided (often easy with smt) and have the other side as a solid 0V plane.

using a single-sided PCB makes life a lot more difficult, and means you
need to pay a lot of attention to the frequency content of every signal.
a solid 0V plane allows you to pretty much ignore that stuff.

2) keep fast edges well away from the interconnecting wires and LEDs. If
you trace the physical path that LED current flows through, you will
find a big loop. Current always flows in loops, and that loop is a
radiating antenna, and you specifically dont want that. You have 2 ways
to recduce the problem: reduce the size of the loop, and/or reduce the
high frequency content of the current flowing in that loop.

In your case the separation between the 2 PCBs and the pin spacing
determines the minimum loop size you can have. Your main pcb with a
solid 0V plane has essentially zero loop area; if all the LEDs have a
common pin, and you make that common (eg Anode +5V) one side of a
2-layer PCB, then the LED pcb also contributes essentially zero loop
area, and all thats left is the interconnection between the two pcbs

If you put the LED series resistor on the control pcb, then split it
into 2 series resistors, and stick an ~100nF cap to 0V in the middle of
them. That will filter out almost all of the HF content, so the LED
current will look roughly DC. By doing that you dont really care about
the LED current loop anymore, so dont need a 2-sided led pcb.

A similar argument holds for putting an L-C filter between the battery
and split-supply chip, to ensure the current that flows in the battery
wires is also DC. make sure your L is not a bobbin core (they have a
huge external air gap, so convert conducted noise into radiated noise)
but is a toroid, little E-core etc. also ensure your LC filter is
damped, either with an R-C damper, or a small R = sqrt(L/C) in series
with the capacitor.

Then your product will be extremely likely to pass first time.

it is quite feasible to look at the areas of the loops, and the actual
currents, and calculate whether or not the device will pass or fail. But
you really need to know what you are doing.

Thank you for the detailed response. I intend to do exactly as you
suggested: reduce loop areas, incorporate filters with damper and use
DC planes.

I was also wondering if anyone knows of a test house who would do the
testing for me for a reasonable fee.

Thank you.
-Mike

Have you checked the India yellow pages? ;-)

Good Luck!
Rich
 
L

Leon

Jan 1, 1970
0
siliconmike said:
Leon said:
siliconmike said:
I'm building a small device that
- runs on a 9V battery
- includes 9V to 4.5V voltage splitter IC running at few hundred
kiloherts
- includes 8 MHz microcontroller with in-built oscillator, operating on
4.5V
- The microcontroller uses PWM to control intensity of 4 LEDs connected
to its port pins.

Thats about it, the device has 4 LEDs that dance at varied intensity.

Well, this device would be submitted for EMC testing as per CE
requirements.

I would appreciate advise and comments on whether this low voltage
device would pose a EMC failure risk, or whether there is absolutely no
risk.

I would also be greatful for advise on how to minimize risk of EMC
failure.

There are 2 pcbs between which the battery is sandwiched. The top PCB
has 4 LEDs. The bottom PCB has the electronics. The pcbs are connected
with perpendicular wires.

| | | | leds
----------------- top pcb
[ ] battery
----------------- bottom pcb

Thanks,
Mike

It doesn't cost much to hire the facilities at a test house for a
couple of hours for preliminary testing, they do that where I used to
work. A unit I designed failed on leakage from the RS-232 cables, it
just needed a simple filter consisting of a few ferrite beads and
capacitors to be incorporated to pass the tests.

Leon

I was wondering if I could send them the device and obtain preliminary
results. That way probably I could save on thousands of dollars charged
by government testing labs.

Could you please suggest the name of the lab?

One does the tests oneself, with their facilities. I think. There are
lots of labs here in the UK that provide those facilities, I don't know
about the USA.

Leon
 
M

martin griffith

Jan 1, 1970
0
siliconmike wrote:

In India ? You're joking. Well actually our subcontractor used ETL right next to
SEEPZ in Mumbai.

Simplest answer is to approach your test lab and ask about simple
'pre-compliance testing'. I suspect you're going to be most interested in
radiated emissions. I'm unsure how you'd test such a device for static discharge
immunity but you should ask them.

It pays to buy all the relevant standards btw. I just hope it's not as expensive
over there as it is here.

Graham

Someone should scan the "relevent docs", put them on a non EU/RIAA
server....

Might be cheaper to fly to India/ Thailand get the testing done there,
than use a UK test house. One outsourcing I could live with.


martin
 
Top