Dave said:
Jon Slaughter said:
To me it sounds mainly like a way for them to get money(assuming
those costs are correct). If some company fabs and ic and creates a
certain for testing it and that circuit passes the FCC's tests then
that circuit to almost exact specs(anything within reason) should be
valid.
My recollection is that most of the money goes to the (privately
operated) test lab, not to the FCC?
If it is so critical to keep noise down then they should designate a
band for commercial use but free to use without restriction. This
way you use at your own risk.
"Should" is a rather subjective judgement. Yes, it would be nice...
but they haven't done so to the extent that you desire (although the
various ISM [Industrial, Scientific, and Measurement] bands come
fairly close in many ways.)
As I said... the testing isn't only done to ensure that the device
does not interfere *within* its intended frequency band of operation.
It's also done to ensure that the device does not emit interference
*outside* of its intended freequency band.
If the circuit is is virtually the same(no major major modifications) as the
one that was used by the IC manufacturer for their FCC testing and if it
passed then it should pass for all time. This is like saying that every
single circuit board must be tested because slight changes could drastically
change it's behavior. That simply isn't the case... or if it is they still
allow it. It is the design that is important(which I'm including the
physical layout) rather the exact device. They obviously understand this to
a degree.
They do allow "minor" modifications as I just found out without requiring
any notifications.
... and you might be surprised how messily an RF device can behave, if
it's hooked up to a load (e.g. antenna) it wasn't designed to handle,
or as a result of parasitic reactances or unexpected cross-coupling on
the circuit board.
Ok, but then thats not following the original design spec. Unexpected
cross-coupling? Parasitic reactances? If the circuit is routed by exactly
what was used by the IC's fcc testing(which had to have such a circuit to
test with) then it should behave almost exactly the same. Obviously there
will be minute differences and some things could potentially make a big
difference such as dielectric of the pcb. But basically if you copy their
design on the RF side then it should pretty much be functionally equivalent.
Any actual differences shouldn't cause any real problems.
Something as simple as the length of the transmission line (coax or
PC-board traces) between an amplifier stage, and a filter which
follows it, can cause a transmitter to go from clean-and-stable to
oscillating-like-a-banshee.
Your not changing the length. Or, if it is a separate IC and antenna as I
mentioned somewhere, the FCC could specify a maximum allowed distanced.
(which is based on the original design)
I'm rather a tyro at this stuff myself (Amateur radio license, and
several years of experience in maintaining a multi-band repeater
system) but I've learned that making RF work right can be a lot more
subtle than one might expect.
Yes, but I'm not talking about RF from scratch. I'm basically talking about
copying an already proven design.
My advice is: if you want to use RF, make sure that your design is
being done by an RF-competent engineer and tested in accordance with
good engineering practice. You can let that be done by the module
company (and accept the pass-through licensing) or you can do it
yourself and hire it done (and pay the R&D and certification costs).
Well, since the design was done by some TI guys then it is good enough? So I
have to waste money to simply duplicate what they have already done? Hence
the "scam". If I were to get their exact gerber documents and used the same
components listed in their BOM(in the datasheet) then I haven't changed
anything except possibly the pcb thickness/dielectric and copper thickness.
Obviously I could use the wrong valued cap, for example... but that could
potentially happen with a certified design when reproduced.
How about "A PC board trace which is an inch too long, causes some
interference which jams a medical sensor in the next room."
An inch pretty big. If you've changed something like that from spec then
it's not really an unessential modification. If the band is only for
commerial unlicensed use then I would imagine medical sensors should not be
using it. As I said... non-critical apps such as your TV communicating with
your comp.
Again, I don't know a lot about RF and I realize the potential for major
problems. My point is that if you are basically copying the original design,
which is FCC approved and the band is for non-critical use then it shouldn't
be an issue. Given that the range would be only, max, a few hundred feet
LOS in 2.4ghz at 0.1dBm at most, if someone screws up or intentionally makes
major changes, then it will only effect things in that range. If it becomes
a serious problem, say they overpowered the device, then the FCC will come
in as normal and be able to see what they did and that they made changes
from original spec and fine them.
I just can't imagine, for example, if I took TI's gerber(they give the
copper layout in their datasheets) and BOM and follow the rules given in the
datasheet(after all, they want you to pass the tests) that I'd be so far off
to cause any problems at all. It might not be as efficient but I doubt it
would fail any tests.
Of course, again, there are people that can't even tie their shoelaces but
somehow got a degree in engineering... so...
But again, I'm basically ignorant on the subject. I just feel that it can't
be too critical to design such a thing... one because relatively speaking,
it is a very simple circuit just from the number of components and layout.
Second, the manufacturers pretty much tells you exactly what to do.
Am I wrong to assume that if you follow there method to the T and don't make
any blunders that you could bring down the space shuttle?
... or "causes a wandering 'birdie' signal which interferes with
police and fire-department radio systems." I recall reading about a
case a couple of years ago, in which a temporary traffic-signal system
(which used RF between the lights at the end of a road segment with
poor visibility) seriously interfered with a small town's police radio
system, and cut off communication with the police cars for several
days. Turned out that the road-repair company had bought a
non-certificated (in the U.S.) model - good RF hardware, but it was
designed for the EU and was transmitting on the wrong band.
Hehe, well, thats a pretty significant error. Huge relative to what I'm
saying.
... and that's why there's a testing requirement.
Then there should be a "Free" band. That people can use at their own risk
and not critical. Say I make a design and follow spec's exactly but old joe
engineer wants to cut costs so he uses some cheaper caps but at the wrong
value. Joe sold some products that were used in the same location as mine
and it caused mine to screw up... since it's non-critical it's no big deal.
Obviously something is wrong. At some point the FCC comes out and then gets
one of Joe's devices and find out what he did and fines him.
The FCC's *primary* job, as defined by Congress, is to manage the
airwaves and spectrum-space so as to prevent harmful interference.
That's why they were originally set up, back in The Day.
A lot of things have changed though. Back in the day you didn't have
integrated RF modules that did 90% of the work. In those cases it was much
easier to screw stuff up.
Their primary job is *not* to make life convenient for you and me.
True... I guess that is the job of any government. But they shouldn't stifle
entrepreneurship or small business growth nor techological advancement...
that is, if it's not going to cause problems.
... and that basic capability is *already* available to a significant
extent, in at least some Part 15 applications.
Rather, I think it is a *process* intended to reduce the risk to
consumers, of having unacceptable interference which requires
expensive efforts to mitigate (e.g. product recalls).
The product recalls is the companies responsibility. If I only have 100
devices sold a year then there is not much recall.
If you want a regulatory change in this regard, you're free to
do what any of us U.S. citizens is: file a petition to request a
change in the rules. The FCC *has* accepted some petitions with
regard to liberalizing Part 15 radio setup and operation over the past
few years (particularly with respect to WiFi).
Hehe... well, as I said... I really don't know enough about it. I'm just
venting my frustration. I really wanted to implement my own RF device for my
new nifty little widget to give it some very competative features. Luckily
I have found some RF modules that can do the trick but they are a bit
limited.
The main point is that I see a logical gap in the regulation. Basically it
is overregulated to some degree. Regulation is good in moderation but there
is such a thing as too much. But as I have said many times... I know very
little about it. But the ideas I have stated are either wrong or right(for
hte most part). If I were to produce an exact copy of the fcc tested design
then it should behave exact. (I know, nothings exact) What the FCC could do
then is allow for a quick test of the device for a much lower cost and or
create a band for such non-tested or "quick-tested" devices.
The IC's and design used would have to be fully tested and would be the
responsibility of the manufacturer. But any "exact" implementation of that
would then fall under that testing and only a simple "compliance" test could
be required(or none at all if a proper spec for the band exists). The IC's
could have a "compliance" test mode which would create a test situation so
that the testers would have to do very little work. Would probably take < 5
mins to do such a test.
The results could be compared not only against the FCC's requirements but
against original chips spec. If they are not with reason do to component
variations then it could be rejected.
Again, my point is simply that the situation could probably be improved
unless I'm missing something that invalidates my logic. (in theory it makes
sense)