Mark A. Corio wrote:
...
> I have some sensitive analog circuitry that I need to do some relay switching
> in for varying some filter bandwidths and other things.  I am considering
> using reed relays.  Does anyone have any comments about this?  I would like
> to drive the relay coils direct from a PIC I/O pin (16C74A or 16C84.....not
> sure yet).  Will the protection diodes on the PIC I/O pins be adequate for
> coil suppression or do I need external diodes as with other relay types?  I
> have no experience with reed types but believe the inductive kick should be
> smaller from reeds.

Energy stored in relay coil is

1/2 * L * i ^ 2

where L is the coil inductance and i is the current at the moment you
switch off the voltage.  This energy must be absorbed by something, and
that something is usually either catch diodes and/or output
transistors.  Due to the large number of turns in many sensitive relays,
their inductance can be surprisingly large.

Reed relays are less of a problem for these reasons:

1. The fine wire used to wind reed relay coils has a significant
resistance, so a significant portion of the energy dissipates in the
coil itself.
2. Whereas, conventional open frame moving armature relays have a
magnetic circuit that closes upon actuate, reed relays usually have a
solenoid construction with a lower inductance.
3. Signal switching (as opposed to power switching) reed relays have a
low actuation force and consequently, a low operating current.

Consequently, output structure damage is less likely with signal
switching reed relays.  Having said the above, since PIC output
protection diodes are not rated for energy absorption, I would still
recommend that you incorporate at least 1 protection diode per coil.

>
> In the past, for this type of application I have used low signal level
> latching relays so as not to have a constant current flowing through the
> coils......I have had problems with noise on the coil drive signals being
> imparted on the signal in standard type relays.  This approach also let me
> use a common ground plane without the coil current causing problems in the
> return path.  Remember, I am very noise sensitive here.  In my current
> application the relay (reed or otherwise) will have the coil driven by a
> signal from another board (both ends of coil) without electrical connection
> the board with the signal.  Are reed relays more sensitive to coupling of
> noise to the signal from the coil since the contacts are surrounded by the
> actuating magnetic fields?  Any experiences with this are of most interest.

There may be significant capacitance between the coil and the contacts,
and your layout may aggravate the problem.  However, since the magnetic
is largely unaffected by non-magnetic materials in the path, it is
possible to electrostatically shield the contacts.  I haven't looked
lately, but it used to be possible to buy shielded reed relays.  If not,
it shouldn't be too hard to add some shielding to a reed relay.  Copper
foil wrapped around the glass envelope of the reed will work: ground it
to your analog ground.  For best results, avoid letting the shielding
act like a bunch of shorted turns: leave a narrow gap so that the shield
makes a 'C' rather than an 'O' around the reed.


--

Paul Mathews, consulting engineer
AEngineering Co.
optoeng@whidbey.com
non-contact sensing and optoelectronics specialists