Hi group,
Thank you everyone for your great responses on the subject.
Sorry I couldn't acknowledge your postings earlier than this and
sorry I haven't acknowledged your postings individually. Had
to go away somewhere - at short notice. Only just returned.
I believe the gist of your responses was that it is a bit of a tall
order to try and get a filter that will let 25kHz signals through
and reject RF at 150kHz.
To this end, I am thinking I may have to design a separate filter circuit
for the output while keeping my original filter strategies on the DC lines.
Any fundamental flaws in this reasoning?
Gert and others raise a few points I would like to comment/pursue/query
etc. Note that I have to assume screened cable is not an option initially
and further that the output it is a PWM square wave with a maximum
fundamental frequency of 25kHz and not 25kbit..
2/ It has not been proved that the sensor actually
degrades when exposed to the signals.
My fault. I didn't supply you with enough information. I have the
analog version of the sensor. Before addition of the pi-filters it
was very susceptible. With only LC filtering, it was much better
but still showed mild susceptibility during conducted immunity
tests although the radiated immunity was vastly improved. With
PI filters, it was immune over the full test spectrum 150k-1G.
3/ A CM coil may be applied with a asym signal if certain
conditions are met (esp. grounding )
I have had bad experiences with trying to use CM chokes on
asym. lines, but you make a point about grounding, so maybe
worth another look?
5/ standard EMI filters of the pi-type will probably
do the job. I recommend using off the shelve parts available easily.
I have heard conflicting reports on these devices. I have heard from
certain 'well-informed' sources that the SMT PI-filters which one can
get from the big manufacturers do not work quite as well as filters
made from discrete SM devices. Any experiences worth sharing -
anyone? Is it all about 'horses for courses' i.e. the right filters for the job.
7/ The value of reasonable impedance does not need to be
that high 300- 500 Ohms is sufficient. Don't over do.
That is a good point. I work to a minimum of 300 Ohms - and where I
have to squeeze every last bit of the spectrum into the useable range
on a ferrite, will accept slightly lower (usually means using larger
caps-to-case to maintain the impedance ratio).
Thank you again for the wonderful responses.
- Best regards
- Chris Chileshe
- Ultronics Ltd
-Original Message-
From: owner-emc-p...@majordomo.ieee.org
[mailto:owner-emc-p...@majordomo.ieee.org]On Behalf Of Chris Chileshe
Sent: Thursday, August 22, 2002 11:31 AM
To: 'emc-pstc'
Subject: Designing for low power conducted and radiated immunity
Dear all,
On the evidence of the quality of responses I have seen on more difficult
queries, I think this should be fairly straightforward for all you RF
engineers
out there.
My product is a small 3-wire pressure sensor which sends out a digital bit
stream encryption of the pressure measured. It's current draw is around
30mA max. The bit stream is around 25kHz, 5V TTL. The 3-wires are
simply supply (5-10VDC), Output (5V logic) and Ground.
I am trying to make the product pass conducted RF immunity tests to
EN 61000-4-6 (150kHz - 80Mhz) at 10V and radiated immunity tests to
EN 61000-4-3 (80Mhz - 1Ghz) at 30V/m. There is an intrinsic safety limit
on the overall capacitance I can use in the product.
My general thoughts are to use a PI-filter on each of the 3 lines with
the vertical section of the filter being ceramic capacitors to the metal
case enclosure. A limit of about 10nF exists (for other reasons) on
these capacitors.
The horizontal section of the PI will be made up of a series connection
of a ferrite (for high frequency suppression) and an inductor for the lower
frequencies where the ferrite is transparent - in that order.
My problems (sorry .. challenges) are as follows:
1 - I need to let a bit stream through at 25kHz, but reject RF at 150kHz
without a common mode choke (output is not differential). Is this a
tall order?
2 - The inductors that give me a reasonable value of impedance at these
sub 1Mhz frequencies tend to be largish and have Self Resonant Freqs.
in the test spectrum albeit some of them (SRFs) are in the radiated
immunity
band where I expect the ferrite to be in charge of attenuation. Is this
likely to cause me problems?
3 - Can I find a single ferrite that will cover the entire RF test spectrum
of
EN 61000-6-2 ( 150kHz - 1Ghz) or is it generally accepted that even the
so called 'wideband' ferrites (SMT 0603/0805 max) are good down to
about 5MHz but no lower?
Any relevant comments welcome.
Best regards
- Chris Chileshe
- Ultronics Ltd
This e-mail has been scanned for all viruses by Star Internet.
