I need a Solid State Relay that is:
1) NC (normally closed)
2) has less than 0.1 Ohms (more like 0.02 Ohm is desired) resistance
over the "switch"
3) can support 2-3A over the "switch" (often called Switching Current?)
4) Is in a convient package like DIP or surface-mount (no hockey
pucks) for PC board mounting (0.1" spacing)
5) Whos reaction times are fast (t_on and t_off <= 2ms)
6) Are available in small quantities (we need only 10 or so) and are
less than 7$US each
Note: The switching mechanism (sometimes called Input Current?) should
only draw something like 10-50mA

---

Clarification of Question by bunghole-ga on 14 Jan 2005 14:20 PST

Indeed it is needed in the US, basically via mail. And the supply is
DC. Thanks for pointing out those oversights.

The fear with reed relays is that the device it to be integrated in a
real time data acquisition device in harsh physical environments (100
G accel). A reed relay will not guarantee a continuous circuit in this
environment, or we are not convinced of this anyway. I have a solution
using traditional MOSFETS, however simplicity is preferred and the
MOSFET solution has many more components (diodes, zeener-diodes). I
will post a diagram of the circuit and would gladly pay the reward
money to someone that offers an equally simple solution as the one I
had in mind with the relay.

This is the best I can Do with ASCII, I can probably use this in
conjunction with some explanation to accurately describe the necessary
conditions.

                          ______
    V_BATT---------------| Relay| ------------------ V_Regulator-->Instrument
                         |      |               |
    V_DC_SUPPLY----------|______|------------   |
                    |                       |   |
                    |                      GND  |
                    |___________________________|


We are essentially making a UPS for the instrument. When the DC-Supply
power is disconnected the relay (normally closed) will then switch to
Battery power. I understand the need for some decent caps to buffer
the t_off time. We don't want to integrate complex IC's or too many
components as this is a real time one shot kind of project and we
cannot afford failure. Hope this helps. Also note that we are using
passive regulators for noise reasons, so solutions that create power
or RF noise are not going to be acceptable.

---

Clarification of Question by bunghole-ga on 17 Jan 2005 12:05 PST

Your help is appreciated indeed. Our supply is 4 AA (~6.3V fresh)
which will drop as they are used. The regulators we are using are
already dropping that 0.3V so we have about a volt of room, but as you
surely know, as batteries drain they also decrease in voltage. Since
we are trying to get as much time out of the batteries as possible we
cannot afford any more of a drop. I have considered a simple diode and
increasing the supply voltage, but like I said any voltage drop
decreases the time we have. The other problem is we are trying to
loose some weight on the instrument, which means loose some more
battery packs, so we have no room to waste energy. We are avoiding
switching regulators beacuse we are powering analog devices off the
supply, and feeding there outputs to a 16bit A/D (and we need all
16bits), I believe the power would generate noise in this case. I will
post the schematic of the power board on the web when I get home, say
7PM mountain time. I will also post a purposed MOSFET design, perhaps
I will have to go with the additional components... I will probably
need help finding the MOSFET that meets my necessary criteria as well.

---

Clarification of Question by bunghole-ga on 18 Jan 2005 12:57 PST

Alright I guess I lied about the scan, I cant find a scanner here at
work so lets say Ill post a comment when the scan is up with the
appropiate address. Sorry about the delay.

You didn?t mention country but if US check out recommendations from
?gopman? and alan7002 on this link for electronic suppliers :-

http://answers.google.com/answers/threadview?id=441457

However, need to know voltage requirement and is it AC or DC? You will
appreciate that for low resistance you are talking MOSFET rather than
SCR. Thyristors and triacs do of course exhibit very low dynamic
resistance but I suspect you couldn?t live with the voltage drop. This
is an audio application ? Switching speed with virtually all solid
state relays is fast, sometimes in the microseconds for MOSFET types.

A really big problem is the NC state. I very much doubt that you will
find a high(ish) current device NC. Can?t invert the drive? Failing
that, reed relays? Too slow for your application?

Best

Reeds at 100G. Agreed, nah. And SCR?s are u/s for DC (or rather
impractical) though a blocking SCR would work in your app were it not
for the voltage drop and it would still need driven. But why the
requirement for such low on-resistance? Are you very short of voltage
headroom? What is the voltage though. As you will be aware there are
MOSFETs in SO223 (and similar) packages which would meet your
resistance spec at low voltage but component count is as you say high.
I don?t know how hard you are into electronics so forgive me if I
oversimplify, but you are aware of the parasitic diode in MOSFETs? You
have to use two in series (N and P channel or two the same but one
reversed) to prevent back-flow. The voltage levels are messy. I
wondered about high side drivers and had a bit of a look -- there may
be but I didn?t want to spend time if you have already explored. Won?t
be really cheap though, especially if they incorporate on-chip voltage
generators.

Many systems simply use a rectifier but presumably you can?t tolerate
the drop. Not even schottky, less than 0.5 volts? -- or perhaps your
supply might be less than the back-up so a diode wouldn?t work. And do
you have to switch the high side? -- with your noise concerns you
probably have a high integrity ground.

Unfortunately GA doesn?t support pictures. Is there a web site which
you can post the diagram on?

I?ll keep a watch on this if you think I can assist.

Best

Sounds like a fun project. I confess to being slightly obsessive in
battery efficient design so this appeals. Switchers could of course
extract the last remnants, but as you say 16 bits. They *can* be
isolated but even with best practice you?d probably have to do
extensive analysis before you were absolutely confident. So, agreed,
it is expedience to go with linear.

There are several queries but I shall wait for the circuit to come up.
Be tomorrow night for me ?cos my ?mountain time? is in approximate
quadrature to yours.

Best

No probs. My (excellent) scanner came from a rubbish skip.

Will try to remember to keep a watch. If I don?t reply to your posting
could you post a new one so it?ll be near the top of the list.
Virtually any circuit of this nature which you throw at me could be
slapped together from box-of-bits to check it out. Actually it?s more
like a house-of-bits and an outhouse and garage :-)

Best