开云体育

And how are you going to "shield" the touch screen? And those "thin
traces" to the SMAs? They are likely printed against a ground plane in the
layer beneath the surface layer and likely adjusted in thickness to be
nominally Zo = 50 ohms traces as stripline transmission lines. In the RF
world things are not always what they look like on first inspection.

Dave - W?LEV

On Sun, Jan 16, 2022 at 3:06 PM Jeff Green <Jeff.L.Green1970@...>
wrote:

Out and out heresy.
I'm making this a separate post because there have been several threads
covering the emi issues.

I had an interesting conversation with the ee I worked with. She explained
a bit about shielding and filtering for emi/emc.

As a rule of thumb, radiated emissions are a concern at 30MHz and above
and conducted emissions are a concern below 30MHz.

There is some overlap, but, as you go from say 100kHz up to 100MHz, you
will find the emi is caused by conduction at 100kHz and radiation at
100MHz. The transition zone depends on the physical size of the device
under test.

Devices that are physically larger will radiate at lower frequencies.

She used an example. "Say you have an object that is a 1" cube that
produces rf white noise at a constant amplitude. It won't become a problem
until high uhf because the small size simply can't radiate.

As a rule of thumb, significant radiation starts when an object is roughly
1/20 of a wavelength and almost always becomes an issue when an object is
1/10 of a wavelength in size."

The smaller an dut is, the higher the frequency has to be before radiation
becomes a concern.

She suggested I study point sources.

A NanaVNA has a USB cable and two rf cables, this is a near worst case
scenario for radiation because the cables will form the arms of a quasi
dipole.

It is extremely difficult to add enough shielding to control radiation
after a product is designed, and it's "damn near impossible to add enough
filtering adequate to control conducted emissions after a product is
designed."

The best, most effect, way to deal with emi is during the design phase.
Any effort after a product is designed will never be as effective as proper
steps during design and initial testing.

Ferrite inductors can help but, the lower the frequency of the emi you are
trying to attenuate, the more ferrite you will need.

At some point, you can't add enough ferrite to stop the emi.

She added, "Say you make your NanoVNA dead quiet, what about your PC?
Unless you have a laptop or desktop desktop that meets TEMPEST
specifications, your PC will almost certainly produce more emi then your
NanoVNA."

She looked up the NanoVNA and pointed out numerous issues. The main one is
the traces that connect the shells of the SMA connectors. For frequencies
below 30MHz they aren't an issue, as you move up from 30MHz, the narrow
traces will have enough inductance to become an issue for emi.

She is sending me an older edition of
Electromagnetic Compatibility Engineering 1st Edition, by Henry Ott. She
says it's the bible for emi/emc.

She also suggested that we not worry so much about emi because we probably
can't reduce the emi enough to be worth the effort and most likely, well
never be able to see any improvment.

She suggested we all study the manual at:


to see the steps required to really deal with emi. She also thought we'd
probably never see the difference in our test results if we were able to
make a case with perfect shielding because of the small size of the NanoVNA
and the conducted emi from the usb and rf cables.

"Far better to concentrate on a case that is robust enough to protect the
NanoVNA."

She is a smart gal, PHDs in EE and physics. Extra class ham, really smart.

"For the price, the NanoVNA is an amazing value, just don't expect the
same results you'd get with a Agilent 8563EC Spectrum Analyzer, used ones
start at 10K."

--
*Dave - W?LEV*
*Just Let Darwin Work*

On 1/16/22 9:12 AM, W0LEV wrote:

And how are you going to "shield" the touch screen?

in practice, touch screens usually have a transparent conductive layer on the top (indium tin oxide is common).? Or the actual LCD display has an ITO top electrode, and the touch screen is resistive or capacitive over the top of that, and runs at low frequencies which can be filtered (after all, you're probably not touching the screen a million times a second).

And those "thin
traces" to the SMAs? They are likely printed against a ground plane in the
layer beneath the surface layer and likely adjusted in thickness to be
nominally Zo = 50 ohms traces as stripline transmission lines. In the RF
world things are not always what they look like on first inspection.

Yes, but that transmission line still radiates a bit. Not all the field is contained in the dielectric, especially with microstripline. It's going to be a poor radiator/receiver, but that depends on what the need is.

Dave - W?LEV

On Sun, Jan 16, 2022 at 3:06 PM Jeff Green <Jeff.L.Green1970@...>
wrote:

Out and out heresy.
I'm making this a separate post because there have been several threads
covering the emi issues.

I had an interesting conversation with the ee I worked with. She explained
a bit about shielding and filtering for emi/emc.

As a rule of thumb, radiated emissions are a concern at 30MHz and above
and conducted emissions are a concern below 30MHz.

There is some overlap, but, as you go from say 100kHz up to 100MHz, you
will find the emi is caused by conduction at 100kHz and radiation at
100MHz. The transition zone depends on the physical size of the device
under test.

Devices that are physically larger will radiate at lower frequencies.

She used an example. "Say you have an object that is a 1" cube that
produces rf white noise at a constant amplitude. It won't become a problem
until high uhf because the small size simply can't radiate.

As a rule of thumb, significant radiation starts when an object is roughly
1/20 of a wavelength and almost always becomes an issue when an object is
1/10 of a wavelength in size."

The smaller an dut is, the higher the frequency has to be before radiation
becomes a concern.

She suggested I study point sources.

A NanaVNA has a USB cable and two rf cables, this is a near worst case
scenario for radiation because the cables will form the arms of a quasi
dipole.

It is extremely difficult to add enough shielding to control radiation
after a product is designed, and it's "damn near impossible to add enough
filtering adequate to control conducted emissions after a product is
designed."

The best, most effect, way to deal with emi is during the design phase.
Any effort after a product is designed will never be as effective as proper
steps during design and initial testing.

Ferrite inductors can help but, the lower the frequency of the emi you are
trying to attenuate, the more ferrite you will need.

At some point, you can't add enough ferrite to stop the emi.

She added, "Say you make your NanoVNA dead quiet, what about your PC?
Unless you have a laptop or desktop desktop that meets TEMPEST
specifications, your PC will almost certainly produce more emi then your
NanoVNA."

She looked up the NanoVNA and pointed out numerous issues. The main one is
the traces that connect the shells of the SMA connectors. For frequencies
below 30MHz they aren't an issue, as you move up from 30MHz, the narrow
traces will have enough inductance to become an issue for emi.

She is sending me an older edition of
Electromagnetic Compatibility Engineering 1st Edition, by Henry Ott. She
says it's the bible for emi/emc.

She also suggested that we not worry so much about emi because we probably
can't reduce the emi enough to be worth the effort and most likely, well
never be able to see any improvment.

She suggested we all study the manual at:


to see the steps required to really deal with emi. She also thought we'd
probably never see the difference in our test results if we were able to
make a case with perfect shielding because of the small size of the NanoVNA
and the conducted emi from the usb and rf cables.

"Far better to concentrate on a case that is robust enough to protect the
NanoVNA."

She is a smart gal, PHDs in EE and physics. Extra class ham, really smart.

"For the price, the NanoVNA is an amazing value, just don't expect the
same results you'd get with a Agilent 8563EC Spectrum Analyzer, used ones
start at 10K."

Jim, I've used that tin oxide for shielding of a few large products.
Visibly it's transparent which was a requirement. But I've found it's only
marginally effective, but it's better than nothing.

Dave - W?LEV

--
*Dave - W?LEV*
*Just Let Darwin Work*

Very informative, thank you for passing on the insight and summary from your friend. We are lucky to have such a functional device at the price point.