Re: EMC compliance question

On Thu, 10 Oct 2024 13:41:07 -0700, Don Y
<blockedofcourse@foo.invalid> wrote:

On 10/9/2024 4:03 PM, bitrex wrote:

What's the deal with the "CPU board" exemption?
 
"CPU board. A circuit board that contains a microprocessor, or frequency
determining circuitry for the microprocessor, the primary function of which is
to execute user-provided programming, but not including:
A circuit board that contains only a microprocessor intended to operate under
the primary control or instruction of a microprocessor external to such a
circuit board; or
A circuit board that is a dedicated controller for a storage or input/output
device."
 
So if one sells a board that has say a PIC on it and some support logic, and
the 9kHz+ signals are all internal to the uP (self-clock), but it's otherwise a
functionally complete design other than it's not in a housing, is that an
exempt product?

>
Who is your customer?  If you are selling it as a *product*,
it is not a *compliant* product so your customer inherits
no certifications (because there are none).
>
If your customer integrates it into *his* product, then
the responsibility for "product certification" falls on him
(so, you have saved *yourself* a few pennies on the certification
process and left him with any "problems" that your board may
pose to *his* certification).

A few pennies for a certified test lab to do full certs?

>
If you are selling to hobbyists, you *may* be able to get by
as a noncompliant product (the first case, above) -- so long
as none of your (few?) customers finds themselves drawing
the ire of neighbors, etc. when your device interferes with
their pursuit of life, liberty and happiness.
>
But, you are still exposed as the seller of that noncompliant
product.  How likely will your customers "have your back"
if things get sticky?
>
In the latter case, your customer (integrator) will *likely*
be thankful for any steps you have taken to certify your
"component" as he goes about looking for certification on
*his* composite system.
>
Why do you think so many products are sold with El Cheapo,
off-brand wall warts instead of taking the power supply
design *into* the overall product?

A wart relieves one of all the AC-line safety certifications. There
are some big warts these days, including 48v ones.

One can resell a cheap wart with the usual molded-in (usually fake)
UN/CE/CSA markings, or let the customer buy their own wart.

>
Lastly, it's just "good engineering" -- and great experience -- to
go through the process so you know what to *avoid* in your
future designs.  (ditto for safety requirements)
>
Increasingly, I am seeing extra scrutiny on devices that CAN "talk"
to ensure they aren't talking to anyone that they can't *justify*.
"Why are you phoning home?"  "Why are you initiating HTTP requests?"
"Why are you trying to resolve some oddball domain name?"
>
[These, of course, are a lot harder to "guarantee" without (and
even *despite*!) releasing full sources.  Especially for OTS/FOSS
OSs that may have been preconfigured (for your convenience) to
support services having communications requirements that you
of which you may be ignorant!]

Software certs on top of hardware certs?

>
Assume your customer is going to need/want to certify his
use of your device and give him a leg up in that process,
pre-sale.

For a small company making a modest number of some test instrument,
full certs will multiply development cost. That may be why I don't see
a lot of small instrument companies in europe.

The guys I was working with in Oxford laughed at me when I asked if
our atom probe system would need to be CE tested. "CE means Cant
Enforce."