desing for high strength RF immunity

[[email protected]]

I have a design for a simple addressable temperature probe (PIC and
dallas semi "1wire temperature probe") that will be placed near a high
strength RF transmiter. Transmitter is for long range coms spanning
several miles in the 500Mhz to 1Ghz frequency range (5 - 20 watts?).

The unit cannot be placed inside a shielded enclosure and the
electonics will recieve direct exposure to the RF.

What are some circuit design practices to harden the circuit form the
effects of the RF field.

Are there any components that are particularly sensitive to RF.

Mosfets (enhancement or delpetion) that may turn on unexpectadly?

BJTs?

PIC micros with internal RC oscillator?

opamps?

The board does have individual wires that extend off the board and I'm
sure would pickup significant RF.

ferrite beads?

diodes?


thanks

 

[Joerg]

I have a design for a simple addressable temperature probe (PIC and
dallas semi "1wire temperature probe") that will be placed near a high
strength RF transmiter. Transmitter is for long range coms spanning
several miles in the 500Mhz to 1Ghz frequency range (5 - 20 watts?).

The unit cannot be placed inside a shielded enclosure and the
electonics will recieve direct exposure to the RF.

What are some circuit design practices to harden the circuit form the
effects of the RF field.

Lots. For example a metal shield box inside (tuner shell). Good layout,
full ground plane, trace sandwiching, PCB inductors, caps, common mode
chokes, series resistors. Etc.

Are there any components that are particularly sensitive to RF.

Nearly all semiconductor parts are.

Mosfets (enhancement or delpetion) that may turn on unexpectadly?

That'll take a lot of RF. I'd be more worried about the substrate diode,
depending on what the FET drives.

BJTs?

Big time. They have an undesired built-in rectifier in the form of a
base-emitter junction.

PIC micros with internal RC oscillator?

Depends on how much you let get into it via ports and supplies.

opamps?

Yep :-(

The board does have individual wires that extend off the board and I'm
sure would pickup significant RF.

ferrite beads?

They help if placed strategically.

diodes?

They can protect against exceeding electric pain thresholds but can make
susceptibility worse because they act as rectifiers.

Seriously, if this is a critical project or one with a tough deadline
get professional help from an RF expert before the boss is breathing
down your neck or nervously pacing the hallways.

 

[Tam/WB2TT]

*****************************************
5 - 20 W is not remotely high power. Bypass all the low frequency leads from
the temperature probe to ground with .001 uF capacitors and short leads.
Assuming everything is CMOS, you could also put ferrite beads or 100 Ohm
resistors in series with the PIC leads that are exposed to the RF. This is
probably overkill, though. The main thing is to keep the leads short, and
use a PC board with a ground plane; this can even be done with a 2 layer
board. I am also assuming that the antenna is outdoors, and not connected
directly to the RF box.

Tam

 

[[email protected]]

Bypass all the low frequency leads from
the temperature probe to ground with .001 uF capacitors and short leads.

It can be helpful to select capacitors with a self resonant frequency
which is close to the transmitter frequency. They will then have the
lowest possible impedance.

Many manufacturers supply selection tools which allow the impedance to
be plotted as a function of frequency. I have used the Murata tool
"Murata Chip S-Parameter & Impedance Library" which despite the name is
such a graphical application.

John

 

[Telstar Electronics]

I have a design for a simple addressable temperature probe (PIC and
dallas semi "1wire temperature probe") that will be placed near a high
strength RF transmiter. Transmitter is for long range coms spanning
several miles in the 500Mhz to 1Ghz frequency range (5 - 20 watts?).

The unit cannot be placed inside a shielded enclosure and the
electonics will recieve direct exposure to the RF.

What are some circuit design practices to harden the circuit form the
effects of the RF field.

Are there any components that are particularly sensitive to RF.

Mosfets (enhancement or delpetion) that may turn on unexpectadly?

BJTs?

PIC micros with internal RC oscillator?

opamps?

The board does have individual wires that extend off the board and I'm
sure would pickup significant RF.

ferrite beads?

diodes?

thanks

5-20W direct RF exposure with no shield... you may have a big problem
on that one. Sounds to me like your signal level will be quite low...
making the situation that much worse. Why is shielding out of the
question???

www.telstar-electronics.com

 

[Tam/WB2TT]

I have a design for a simple addressable temperature probe (PIC and
dallas semi "1wire temperature probe") that will be placed near a high
strength RF transmiter.

OK, Lets back up. I assumed you were going to put your thing inside the
transmitter box. Perhaps not. Here is an example. I have an AC powered CO
detector that went off when a 1200W transmitter whose antenna is about 50
feet away is active. To fix it, I had to do two things.

To prevent the house AC wiring from acting as an antenna, I connected a .01
uFd 1000V capacitor across the AC line inside the CO detector. This almost
fixed the problem, but not quite.

An examination of the CO detecto's PC board showed no VCC bypassing other
than a 470 uFd capacitor at the rectifier. Adding a second bypass cap (also
..01) from VCC to ground fixed the problem.

Tam

 

[JoeBloe]

The unit cannot be placed inside a shielded enclosure and the
electonics will recieve direct exposure to the RF.

What are some circuit design practices to harden the circuit form the
effects of the RF field.

The shielded enclosure you mentioned. Don't place it in one. Place
one ON the board. If a quarter inch or half inch thick can is too
much, you have bigger problems.

The other way is to place only the transducer where the reading are
to be taken, and port the data/signal over to the electronics at a
distance.

You could make a little bluetooth (I'm sure they are out there)
transmitter for the transducer, and read the data anywhere within 100
feet or such.

 

[Joerg]

*****************************************
5 - 20 W is not remotely high power. ...

That will entirely depend on what the OP means by "near". I have had to
debug cases where the RF went smack dab into the bond wires and metal
layers of chips and nothing short of changing to metal shielding helped.

 

[LVMarc]

That will entirely depend on what the OP means by "near". I have had to
debug cases where the RF went smack dab into the bond wires and metal
layers of chips and nothing short of changing to metal shielding helped.

not at all! You absolutley need a differential signal path for your
littel temperature signal riding awash in a sea of common mode rf. just
encode temp differentially, run the wired pair through the hell of noise
and subtract, in a sheilded and qiet space!

ask me how @@

[email protected]

 

[Joerg]

not at all! You absolutley need a differential signal path for your
littel temperature signal riding awash in a sea of common mode rf. just
encode temp differentially, run the wired pair through the hell of noise
and subtract, in a sheilded and qiet space!

ask me how @@

Yes, differential will help with the RF portion that doesn't hit the
sensor directly. Best is both, shiedling plus differential like it's
done on aircraft (Twinax etc.).

 

[[email protected]]

I have a design for a simple addressable temperature probe (PIC and
dallas semi "1wire temperature probe") that will be placed near a high
strength RF transmiter.

The unit cannot be placed inside a shielded enclosure and the
electonics will recieve direct exposure to the RF.

I presume you don't mean 'near a transmitter' but rather 'near an
antenna'?
The transmitter, while undoubtedly a source that can cause
interference,
is not going to be radiating lots of RF. That's the antenna's job.

If you really want a thermometer insensitive to near-field antenna
placement, look at fiber optic solutions
<http://www.temperatures.com/fiberoptics.html>

Shielding can be a simple matter; have you considered the placement of
that
one-wire probe inside a copper tube? Regular plumbing fittings are
very
good shielding for the MHz frequency range. The only alternative to
shielding is
to keep the circuit very compact, then bypass (with capacitors) and
choke (with
ferrite beads) all the circuit paths sensitive to RF. If you can make
the circuit
slow enough with filtering, the (slow) temperature signal will dominate
the
total (signal + noise + RFartifacts).